Tmg-A 92

UTOMATIC
YNAMIC
NCREMENTAL
ONLINEAR
NALYSIS
Theory and Modeling Guide

Volume I: ADINA
ADINA 9.2 October 2016
ADINA R & D, Inc.

ADINA
Volume I: ADINA Solids & Structures
October 2016
ADINA R & D, Inc.

71 Elton Avenue
Watertown, MA 02472 USA
tel. (617) 926-5199

telefax (617) 926-0238
www.adina.com
Notices
ADINA R & D, Inc. owns both this software program system and its documentation. Both the
program system and the documentation are copyrighted with all rights reserved by ADINA R & D,
Inc.
The information contained in this document is subject to change without notice.
ADINA R & D, Inc. makes no warranty whatsoever, expressed or implied that the Program and its
documentation including any modifications and updates are free from errors and defects. In no
event shall ADINA R & D, Inc. become liable to the User or any party for any loss, including but
not limited to, loss of time, money or goodwill, which may arise from the use of the Program and its
documentation including any modifications and updates.
Trademarks
ADINA is a registered trademark of K.J. Bathe / ADINA R & D, Inc.
All other product names are trademarks or registered trademarks of their respective owners.
Copyright Notice
ADINA R & D, Inc. 1987 - 2016

October 2016 Printing
Printed in the USA
Table of contents
Table of contents
1. Introduction ......................................................................................................... 1
1.1 Objective of this manual.................................................................................. 1
1.2 Supported computers and operating systems .................................................. 2
1.3 Units ................................................................................................................ 2
1.4 ADINA System documentation....................................................................... 4
2. Elements ............................................................................................................... 7
2.1 Truss and cable elements................................................................................. 7
2.1.1 General considerations ............................................................................. 7
2.1.2 Material models and formulations............................................................ 9
2.1.3 Numerical integration............................................................................... 9
2.1.4 Mass matrices ......................................................................................... 10
2.1.5 Element output ....................................................................................... 11
2.1.6 Recommendations on use of elements.................................................... 12
2.1.7 Rebar elements ....................................................................................... 12
2.2 Two-dimensional solid elements................................................................... 15
2.2.1 General considerations ........................................................................... 15
2.2.2 Material models and formulations.......................................................... 22
2.2.3 Numerical integration............................................................................. 24
2.2.4 Mass matrices ......................................................................................... 26
2.2.5 Element output ....................................................................................... 27
2.3 Three-dimensional solid elements................................................................. 37
2.3.1 General considerations ............................................................................ 37
2.3.2 Material models and nonlinear formulations.......................................... 45
2.3.3 Numerical integration............................................................................. 47
2.3.4 Mass matrices ......................................................................................... 50
2.3.5 Element output ....................................................................................... 50
2.4 Beam elements .............................................................................................. 60
2.4.1 Beam geometry....................................................................................... 62
2.4.2 Beam cross-section geometric properties ............................................... 67
2.4.3 Beam formulation................................................................................... 69
2.4.3.1 Beam displacements ........................................................................ 69
2.4.3.2 Displacement derivatives and strains .............................................. 71
2.4.3.3 Stresses ............................................................................................ 72
2.4.3.4 Principle of virtual work.................................................................. 72
2.4.3.5 Torsional response........................................................................... 72
ADINA R & D, Inc. iii

Table of contents
2.4.3.6 Strain energy for a linear elastic material........................................ 74

2.4.3.7 Wagner effect .................................................................................. 76
2.4.3.8 Kinetic energy and the mass matrix ................................................ 81
2.4.3.9 The warping function ...................................................................... 82
2.4.4 Beam interpolations................................................................................ 90
2.4.4.1 Standard beam ................................................................................. 90
2.4.4.2 Warping beam ................................................................................. 91
2.4.4.3 Comparison of standard beam and warping beam........................... 92
2.4.5 Beam element implementation ............................................................... 93
2.4.5.1 Linear formulation........................................................................... 94
2.4.5.2 Materially-nonlinear-only formulation............................................ 94
2.4.5.3 Large displacement formulation...................................................... 94
2.4.5.4 Mass matrices .................................................................................. 98
2.4.5.5 Elastic beam element....................................................................... 99
2.4.5.6 Elastic-plastic beam element ......................................................... 100
2.4.5.7 Element output .............................................................................. 102
2.4.6 Cross-sections....................................................................................... 110
2.4.6.1 General cross-section .................................................................... 110
2.4.6.2 Rectangular cross-section.............................................................. 111
2.4.6.3 Pipe cross-section .......................................................................... 116
2.4.6.4 Box cross-section .......................................................................... 119
2.4.6.5 I cross-section................................................................................ 124
2.4.6.6 U cross-section .............................................................................. 129
2.4.6.7 L cross-section............................................................................... 134
2.4.7 Moment-curvature beam element......................................................... 141
2.4.8 Additional features ............................................................................... 152
2.4.9 Beam element modeling hints .............................................................. 157
2.4.9.1 Hints concerning the warping degrees of freedom........................ 159
2.5 Iso-beam elements; axisymmetric shell elements........................................ 162
2.5.1 General considerations ......................................................................... 162
2.5.2 Numerical integration........................................................................... 166
2.5.3 Linear iso-beam elements..................................................................... 168
2.5.4 Nonlinear iso-beam elements ............................................................... 171
2.5.5 Axisymmetric shell element ................................................................. 172
2.5.6 Element mass matrices ......................................................................... 173
2.5.7 Element output ..................................................................................... 173
2.6 Plate elements............................................................................................... 176
2.7 Shell elements.............................................................................................. 177
2.7.1 Basic assumptions in element formulation........................................... 177
2.7.2 Material models and formulations........................................................ 185
2.7.3 Shell nodal point degrees of freedom................................................... 188
iv ADINA Structures Theory and Modeling Guide

Table of contents
2.7.4 Transition elements .............................................................................. 195

2.7.6 Composite shell elements ..................................................................... 200
2.7.7 Mass matrices ....................................................................................... 205
2.7.8 Anisotropic failure criteria ................................................................... 206
2.7.9 Element output ..................................................................................... 214
2.7.10 Selection of elements for analysis of thin and thick shells................. 220
2.7.11 3D-shell element................................................................................. 225
2.8 Pipe elements............................................................................................... 237
2.8.1 Material models and formulations........................................................ 246
2.8.2 Pipe internal pressures.......................................................................... 247
2.8.4 Element mass matrices ......................................................................... 253
2.8.5 Element output ..................................................................................... 254
2.8.6 Recommendations on use of elements.................................................. 256
2.9 General and spring/damper/mass elements ................................................. 257
2.9.1 General elements .................................................................................. 257
2.9.2 Linear and nonlinear spring/damper/mass elements ............................ 260
2.9.3 6DOF Spring Element........................................................................... 266
2.9.4 User-supplied element .......................................................................... 270
2.9.4.1 User-supplied element example ..................................................... 276
2.10 Displacement-based fluid elements ........................................................... 277
2.11 Potential-based fluid elements................................................................... 280
2.11.1 Theory: Subsonic velocity formulation.............................................. 282
2.11.2 Theory: Infinitesimal velocity formulation ........................................ 289
2.11.3 Theory: Infinite fluid regions ............................................................. 291
2.11.4 Theory: Ground motion loadings ....................................................... 295
2.11.5 Theory: Static analysis ....................................................................... 296
2.11.6 Theory: Frequency analysis ............................................................... 299
2.11.7 Theory: Mode superposition .............................................................. 300
2.11.8 Theory: Response spectrum, harmonic and random vibration analysis302
2.11.9 Modeling: Formulation choice and potential master degree of freedom
........................................................................................................................ 304
2.11.10 Modeling: Potential degree of freedom fixities................................ 305
2.11.11 Modeling: Elements ......................................................................... 306
2.11.12 Modeling: Potential-interfaces ......................................................... 306
2.11.13 Modeling: Interface elements ........................................................... 308
2.11.14 Modeling: Loads .............................................................................. 315
2.11.15 Modeling: Phi model completion ..................................................... 317
2.11.16 Modeling: Considerations for static analysis ................................... 333
2.11.17 Modeling: Considerations for dynamic analysis .............................. 334
ADINA R & D, Inc. v

Table of contents
2.11.18 Modeling: Considerations for frequency analysis and modal

participation factor analysis............................................................................ 334
2.11.19 Modeling: Element output................................................................ 335
2.12 Alignment elements................................................................................... 336
2.12.1 Overview ............................................................................................. 336
2.12.2 Triads................................................................................................... 344
2.12.3 Notations ............................................................................................. 346
2.12.4 Translation alignment theory............................................................... 348
2.12.5 Distance alignment theory................................................................... 351
2.12.6 Rotation alignment theory ................................................................... 355
2.12.7 Triadsets and node-triadset pairs ......................................................... 361
2.12.8 Align-translation specification ............................................................ 363
2.12.9 Align-distance specification ................................................................ 364
2.12.10 Align-rotation specification............................................................... 364
2.12.11 Element group specification .............................................................. 365
2.12.12 Element specification ........................................................................ 365
2.12.13 Input examples .................................................................................. 366
2.12.14 Element output .................................................................................. 378
2.12.15 Recommendations ............................................................................. 380
2.13 Cohesive elements...................................................................................... 381
2.13.1 General considerations ........................................................................ 381
2.13.2 Element formulation............................................................................ 381
2.13.3 Constitutive law of cohesive element.................................................. 382
2.13.4 Numerical integration.......................................................................... 385
2.13.5 Element output .................................................................................... 385
3. Material models and formulations................................................................. 387
3.1 Stress and strain measures in ADINA ......................................................... 387
3.1.1 Summary ............................................................................................... 387
3.1.2 Large strain thermo-plasticity analysis with the ULH formulation...... 393
3.1.3 Large strain thermo-plasticity analysis with the ULJ formulation ....... 398
3.1.4 Thermal strains ..................................................................................... 400
3.2 Linear elastic material models..................................................................... 402
3.2.1 Elastic-isotropic material model........................................................... 403
3.2.2 Elastic-orthotropic material model ....................................................... 404
3.3 Nonlinear elastic material model................................................................. 413
3.3.1 Nonlinear elastic material for truss element ......................................... 417
3.4 Isothermal plasticity material models.......................................................... 420
3.4.1 Plastic-bilinear and plastic-multilinear material models ...................... 421
3.4.2 Mroz-bilinear material model............................................................... 432
3.4.3 Plastic-orthotropic material model ....................................................... 440
vi ADINA Structures Theory and Modeling Guide

Table of contents
3.4.4 Gurson material model ......................................................................... 446

3.4.5 Plastic-cyclic material model ............................................................... 448
3.4.5.1 Fundamental concepts ................................................................... 449
3.4.5.2 Specification of input .................................................................... 467
3.4.5.3 Output variables ............................................................................ 471
3.5 Thermo-elastic material models .................................................................. 472
3.6 Thermo-elasto-plasticity and creep material models................................... 474
3.6.1 Evaluation of thermal strains................................................................ 478
3.6.2 Evaluation of plastic strains ................................................................. 479
3.6.3 Evaluation of creep strains ................................................................... 483
3.6.4 Computational procedures.................................................................... 490
3.6.5 Shifting rule for the irradiation creep model ........................................ 492
3.7 Concrete material models ............................................................................ 493
3.7.1 Concrete material model....................................................................... 494
3.7.2 Data fitted concrete material model...................................................... 513
3.7.3 Creep/Shrinkage effects ....................................................................... 534
3.8 Rubber and foam material models............................................................... 539
3.8.1 Isotropic hyperelastic effects................................................................ 540
3.8.1.1 Mooney-Rivlin model ................................................................... 542
3.8.1.2 Ogden material model ................................................................... 545
3.8.1.3 Arruda-Boyce material model ....................................................... 547
3.8.1.4 Hyper-foam material model .......................................................... 549
3.8.1.5 Sussman-Bathe material model ..................................................... 551
3.8.1.6 Eight-chain material model ........................................................... 559
3.8.1.7 Curve fitting .................................................................................. 569
3.8.2 Viscoelastic effects............................................................................... 578
3.8.2.1 Holzapfel viscoelasticity ............................................................... 579
3.8.2.2 Bergstrm-Boyce viscoelasticity................................................... 588
3.8.3 Mullins effect ....................................................................................... 592
3.8.4 Orthotropic effect ................................................................................. 597
3.8.5 Thermal strain effect ............................................................................ 600
3.8.6 TRS material......................................................................................... 602
3.8.7 Temperature-dependent material.......................................................... 602
3.8.8 Rubber stability indicators.................................................................... 604
3.8.8.1 Introduction ................................................................................... 604
3.8.8.2 Stability ......................................................................................... 606
3.8.8.3 Stability in uniaxial tension........................................................... 608
3.8.8.4 Stability in pure shear.................................................................... 608
3.8.8.5 Stability in equibiaxial tension ...................................................... 610
3.8.8.6 Comments...................................................................................... 610
3.9 Geotechnical material models ..................................................................... 610
ADINA R & D, Inc. vii

Table of contents
3.9.1 Curve description material model ........................................................ 610

3.9.2 Drucker-Prager material model ............................................................ 616
3.9.3 Cam-clay material model ..................................................................... 621
3.9.4 Mohr-Coulomb material model............................................................ 625
3.9.4.1 Mohr-Coulomb material model with temperature effects ............. 631
3.10 Fabric material model with wrinkling ....................................................... 633
3.11 Viscoelastic material model ...................................................................... 634
3.12 Porous media formulation ......................................................................... 638
3.13 Gasket material model............................................................................... 643
3.14 Shape Memory Alloy (SMA) material model ........................................... 652
3.15 The Anand material model for soldering................................................... 661
3.16 Piezoelectric material model ..................................................................... 664
3.16.1 Constitutive equation.......................................................................... 665
3.16.2 Orthotropic properties ........................................................................ 667
3.16.3 Strain-form coupling matrix ............................................................... 668
3.16.4 Polarization direction of the piezoelectric material............................ 669
3.16.5 Orthotropic axis system for the piezoelectric material model ............ 670
3.16.6 Electrical loading and boundary condition......................................... 673
3.17 Parallel-Network Framework .................................................................... 674
3.17.1 Network-specific kinematics .............................................................. 674
3.17.2 Power expenditure, virtual velocities, and force balance ................... 677
3.17.3 Energy dissipation .............................................................................. 680
3.17.4 Stress update procedure...................................................................... 680
3.17.5 Three-network model ......................................................................... 681
3.17.5.1 Rheological representation .............................................................. 682
3.17.5.2 Thermoelastic response ................................................................... 682
3.17.5.3 Viscous response ............................................................................. 684
3.17.5.4 Constants and their interpretations .................................................. 685
3.17.5.5 Additional notes .............................................................................. 688
3.18 User-coded material model........................................................................ 688
3.18.1 General considerations ....................................................................... 689
3.18.2 Input and output parameters ............................................................... 692
3.18.3 Stress integration procedure ............................................................... 697
3.18.3.1 Subincrementation....................................................................... 699
3.18.4 Coding guidelines............................................................................... 701
3.18.5 1-D truss elements .............................................................................. 704
3.18.6 2-D solid elements.............................................................................. 704
3.18.7 3-D solid elements.............................................................................. 705
3.18.8 User-coded material models supplied by ADINA R&D, Inc. ............ 705
3.18.8.1 Viscoplastic material model ........................................................ 706
3.18.8.2 Plasticity model ........................................................................... 713
viii ADINA Structures Theory and Modeling Guide

Table of contents
3.18.8.3 Thermo-elasticity and creep material model ............................... 714

3.18.8.4 Concrete material model, including creep effects ....................... 715
3.18.8.5 Viscoelastic material model ........................................................ 719
3.18.8.6 Ramberg-Osgood material model with mixed hardening............ 720
3.18.8.7 Nonlinear elastic material model.................................................. 722
4. Contact conditions........................................................................................... 725
4.1 Overview ..................................................................................................... 726
4.2 Contact algorithms for implicit analysis....................................................... 736
4.2.1 Constraint-function method.................................................................. 737
4.2.2 Segment (Lagrange multiplier) method................................................ 739
4.2.3 Rigid target method .............................................................................. 739
4.2.4 Selection of contact algorithm.............................................................. 739
4.3 Contact algorithms for explicit analysis ....................................................... 740
4.3.1 Kinematic constraint method................................................................ 740
4.3.2 Penalty method ..................................................................................... 742
4.3.3 Rigid target method .............................................................................. 743
4.3.4 Selection of contact algorithm for explicit analysis ............................. 743
4.4 Contact group properties ............................................................................. 743
4.5 Friction ........................................................................................................ 759
4.5.1 Basic friction models............................................................................ 759
4.5.2 User-supplied friction models (not available in explicit analysis) ....... 761
4.5.3 Frictional heat generation..................................................................... 764
4.6 Contact definition features .......................................................................... 765
4.6.1 Analytical rigid targets .......................................................................... 765
4.6.2 Contact body ........................................................................................ 765
4.6.3 Automatic generation of contact surfaces between Parasolid bodies... 766
4.7 Contact analysis features ............................................................................. 766
4.7.1 Implicit dynamic contact/impact .......................................................... 766
4.7.2 Explicit dynamic contact /impact ......................................................... 768
4.7.3 Contact detection.................................................................................. 768
4.7.4 Suppression of contact oscillations ...................................................... 769
4.7.5 Restart with contact .............................................................................. 770
4.7.6 Contact damping................................................................................... 770
4.8 Modeling considerations ............................................................................. 771
4.8.1 Contactor and target selection .............................................................. 771
4.8.2 General modeling hints ........................................................................ 773
4.8.3 Modeling hints for implicit analysis..................................................... 776
4.8.4 Modeling hints for explicit analysis ..................................................... 778
4.8.5 Convergence considerations (implicit analysis only)........................... 780
4.8.6 Handling improperly supported bodies ................................................ 782
ADINA R & D, Inc. ix

Table of contents
4.9 Rigid target contact algorithm..................................................................... 784

4.9.1 Introduction .............................................................................................. 784
4.9.2 Basic concepts ...................................................................................... 787
4.9.2.1 Contactor surfaces ......................................................................... 787
4.9.2.2 Target surfaces .............................................................................. 790
4.9.2.3 Determination of contact between contactor and target ................ 790
4.9.2.4 Frictional contact........................................................................... 798
4.9.2.5 Drawbeads ..................................................................................... 800
4.9.3 Modeling considerations ...................................................................... 805
4.9.4 Rigid target contact reports for static and implicit dynamics............... 815
4.9.5 Rigid target contact report for explicit dynamics ................................. 818
4.9.6 Modeling hints and recommendations.................................................. 819
4.9.7 Conversion of models set up using the rigid target algorithm of ADINA
8.3................................................................................................................... 822
5. Boundary conditions/applied loading/constraint ......................................... 825
5.1 Introduction ................................................................................................. 825
5.2 Concentrated loads ...................................................................................... 830
5.3 Pressure and distributed loading.................................................................. 833
5.3.1 Two- and three-dimensional pressure loading ..................................... 836
5.3.2 Hermitian (2-node) beam distributed loading ...................................... 839
5.3.3 Iso-beam and pipe distributed loading ................................................. 840
5.3.4 Plate/shell pressure loading .................................................................. 841
5.3.5 Shell loading......................................................................................... 841
5.3.6 Contact surface modification of prescribed loads feature .................... 842
5.4 Centrifugal, rotational and mass proportional loading ................................ 849
5.4.1 Overview ............................................................................................... 849
5.4.2 Centrifugal loading................................................................................ 851
5.4.3 Rotational loading ................................................................................. 854
5.4.4 Mass proportional loading..................................................................... 857
5.5 Prescribed displacements, velocities & accelerations ................................. 860
5.6 Prescribed temperatures and temperature gradients .................................... 867
5.7 Pipe internal pressure data........................................................................... 870
5.8 Electromagnetic loading.............................................................................. 871
5.9 Poreflow loads............................................................................................. 872
5.10 Phiflux loads.............................................................................................. 872
5.11 Contact slip loads ...................................................................................... 873
5.12 Surface tension boundary .......................................................................... 873
5.13 Initial load calculations ............................................................................. 874
5.14 User-supplied loads ................................................................................... 875
x ADINA Structures Theory and Modeling Guide

Table of contents
5.14.2 Usage of the user-supplied loads........................................................ 875

5.14.3 Example: Hydrodynamic forces ......................................................... 878
5.15 Constraint equations .................................................................................. 881
5.15.2 Rigid links .......................................................................................... 883
5.16 Mesh glueing .............................................................................................. 887
6. Eigenvalue problems ....................................................................................... 893
6.1 Linearized buckling analysis ....................................................................... 893
6.1.1 General considerations ......................................................................... 893
6.1.2 Introduction of geometric imperfections .............................................. 898
6.2 Frequency analysis ...................................................................................... 898
6.2.1 Determinant search method................................................................... 902
6.2.2 Subspace iteration method.................................................................... 903
6.2.3 Lanczos iteration method ..................................................................... 904
6.2.4 Modal stresses ...................................................................................... 906
7. Static and implicit dynamic analysis ............................................................. 909
7.1 Linear static analysis ................................................................................... 909
7.1.1 Sparse and skyline solvers.................................................................... 910
7.1.2 Iterative and multigrid solvers.............................................................. 912
7.1.3 3D-iterative solver................................................................................ 914
7.2 Nonlinear static analysis.............................................................................. 916
7.2.1 Solution of incremental nonlinear static equations .............................. 918
7.2.2 Line search ........................................................................................... 920
7.2.3 Low speed dynamics feature ................................................................ 922
7.2.4 Automatic-Time-Stepping (ATS) method............................................ 923
7.2.5 Total Load Application (TLA) method and Stabilized TLA (TLA-S)
method ............................................................................................................ 927
7.2.6 Load-Displacement-Control (LDC) method ........................................ 929
7.2.7 Convergence criteria for equilibrium iterations ................................... 933
7.2.8 Selection of incremental solution method ............................................ 938
7.2.9 Example................................................................................................ 942
7.3 Linear dynamic analysis.............................................................................. 948
7.3.1 Step-by-step implicit time integration .................................................. 948
7.3.2 Time history by mode superposition .................................................... 951
7.3.3 Lumped and consistent damping; Rayleigh damping .......................... 953
7.3.4 Modal damping ratios based on strain energy proportional damping .. 955
7.4 Nonlinear dynamic analysis ........................................................................ 959
7.4.1 Step-by-step implicit time integration .................................................. 959
7.4.2 Time history by mode superposition .................................................... 962
ADINA R & D, Inc. xi

Table of contents
7.4.3 Global mass matrix............................................................................... 963

7.5 Choosing between implicit and explicit formulations................................. 964
7.6 Tracking solution progress .......................................................................... 966
7.6.1 .rto file ................................................................................................... 967
7.7 Solution monitoring..................................................................................... 968
8. Explicit dynamic analysis ............................................................................... 971
8.1 Formulation ................................................................................................. 973
8.1.1 Central difference method (CDM) ...................................................... 974
8.1.2 Noh-Bathe method ............................................................................... 976
8.1.3 Mass matrix .......................................................................................... 978
8.1.4 Damping central difference method .................................................. 978
8.1.5 Damping Noh-Bathe method............................................................. 979
8.2 Stability ....................................................................................................... 980
8.3 Time step management................................................................................ 983
8.4 Tracking solution progress .......................................................................... 984
9. Frequency domain analysis ............................................................................ 987
9.1 Response spectrum analysis ........................................................................ 987
9.2 Fourier analysis ........................................................................................... 998
9.3 Harmonic vibration analysis...................................................................... 1001
9.4 Random vibration analysis ........................................................................ 1008
9.5 SDOF system response.............................................................................. 1014
10. Fracture mechanics ..................................................................................... 1027
10.1 Overview ................................................................................................. 1027
10.2 The energy release rate............................................................................ 1030
10.3 Stress intensity factors............................................................................. 1031
10.4 Calculation of the J-integral using the line contour integration method . 1035
10.5 Calculation of the J-integral using the virtual crack extension method .. 1037
10.6 Calculation of stress intensity factors using the virtual crack extension
method .............................................................................................................. 1043
10.6.1 Overview of theory........................................................................... 1043
10.6.2 Calculation of stress intensity factors............................................... 1050
10.6.3 Discussion ........................................................................................ 1051
10.7 Fracture control parameters..................................................................... 1052
10.8 2-D crack geometry and meshing............................................................ 1053
10.8.1 Overview .......................................................................................... 1053
10.8.2 Meshing recommendations............................................................... 1056
10.8.3 Definition of the crack propagation line, no nodes on the line of self-
similar crack advance ................................................................................... 1059
xii ADINA Structures Theory and Modeling Guide

Table of contents
10.8.4 Definition of the crack propagation line, nodes on the line of self-similar
crack advance ............................................................................................... 1059
10.9 Definition and use of line contours for 2-D cracks ................................. 1060
10.9.1 Definition of line contours ............................................................... 1060
10.9.2 Output............................................................................................... 1063
10.10 Definition and use of virtual shifts for 2-D cracks ................................ 1063
10.10.1 Domain of the virtual shift ............................................................. 1065
10.10.2 Virtual shift vector.......................................................................... 1068
10.10.3 Output............................................................................................. 1069
10.11 Crack propagation for 2-D cracks ......................................................... 1069
10.12 3-D crack geometry and meshing.......................................................... 1079
10.12.1 Overview ........................................................................................ 1079
10.12.2 Meshing recommendations............................................................. 1080
10.13 3-D crack geometry and meshing using CRACK-M............................. 1083
10.14 NVS method for virtual crack extension for 3-D cracks ....................... 1094
10.14.1 Overview ........................................................................................ 1094
10.14.2 Definition of the crack propagation surface, CRACK-M features are not
used............................................................................................................... 1103
10.14.3 Definition of the virtual shifts, CRACK-M features are not used.. 1105
10.14.4 Definitions of the virtual shifts when CRACK-M features are used1109
10.14.5 Fracture mechanics output.............................................................. 1111
10.15 SVS method for virtual crack extension for 3-D cracks........................ 1112
10.15.1 Overview ........................................................................................ 1112
10.15.2 Definitions of the virtual shifts....................................................... 1113
10.15.3 Calculation of J-integral and stress intensity factors ...................... 1122
10.15.4 Calculation of G using numerical integration ................................ 1123
10.15.5 Definitions of the cracks................................................................. 1129
10.15.6 Virtual shift numbers assigned to the virtual shifts ........................ 1132
10.15.7 Modeling hints................................................................................ 1134
10.15.8 Fracture mechanics output.............................................................. 1136
11. Additional capabilities ................................................................................ 1141
11.1 Substructuring ......................................................................................... 1141
11.1.1 Substructuring in linear analysis ...................................................... 1141
11.1.2 Substructuring in nonlinear analysis ................................................ 1143
11.1.3 Substructuring restrictions................................................................ 1144
11.1.4 Additional substructuring modeling techniques............................... 1145
11.2 Cyclic symmetry analysis........................................................................ 1153
11.3 Reactions calculation............................................................................... 1159
11.4 Element birth and death feature............................................................... 1159
11.5 Element "death upon rupture" ................................................................. 1166
ADINA R & D, Inc. xiii

Table of contents
11.6 Initial conditions...................................................................................... 1167

11.6.1 Initial displacements, velocities and accelerations ........................... 1167
11.6.2 Initial temperatures and temperature gradients ................................ 1168
11.6.3 Initial pipe internal pressures ........................................................... 1169
11.6.4 Initial strains ..................................................................................... 1169
11.6.5 Initial stresses ................................................................................... 1177
11.7 Energy calculation................................................................................... 1181
11.7.1 Energy printout................................................................................. 1181
11.7.2 Strain energy densities...................................................................... 1188
11.8 Element group inertial properties ............................................................ 1190
11.9 Restart...................................................................................................... 1192
11.10 Parallel processing................................................................................. 1196
11.11 Usage of memory and disk storage ....................................................... 1197
11.12 Miscellaneous........................................................................................ 1198
11.13 Remeshing options ................................................................................ 1200
11.14 Analysis zooming .................................................................................. 1202
11.15 Stiffness stabilization ............................................................................ 1203
11.16 Geometric imperfections ....................................................................... 1206
11.17 Bolt feature ............................................................................................ 1207
11.17.1 Beam-bolt element .......................................................................... 1209
11.17.2 3D-bolt element group..................................................................... 1213
11.17.3 Usage of bolt loadings..................................................................... 1217
11.17.4 Usage of beam-bolts........................................................................ 1219
11.17.5 Usage of 3D-bolts............................................................................ 1220
11.17.6 Modeling issues............................................................................... 1225
11.17.7 Post-processing variables ................................................................ 1226
11.18 Component mode synthesis (CMS) method........................................... 1227
12. Heat transfer capabilities in the ADINA program .................................... 1233
12.1 Analysis options ...................................................................................... 1233
12.2 Modeling ................................................................................................. 1235
12.3 Soil consolidation analysis ...................................................................... 1237
12.4 Piezoelectric analysis .............................................................................. 1241
12.5 Thermal coupling between solids and fluids ........................................... 1242
13. Postprocessing considerations .................................................................... 1245
13.1 Calculation of results within the AUI...................................................... 1245
13.1.1 What results are available................................................................. 1251
13.1.2 Where results can be obtained.......................................................... 1288
13.1.3 How results are evaluated (including smoothing) ............................ 1299
13.1.4 When (for which load step, etc.) are results available...................... 1303
xiv ADINA Structures Theory and Modeling Guide

Table of contents
13.2 Evaluation of meshes used ...................................................................... 1309

Index .................................................................................................................... 1311
ADINA R & D, Inc. xv

1.1 Objective of this manual
1. Introduction
1.1 Objective of this manual
The objective of this manual is to give a concise summary and
guide for study of the theoretical basis of the finite element
computer program ADINA Solids & Structures (ADINA in short).
The program ADINA is used for displacement and stress analysis.
Since a large number of analysis options is available in this
computer program, a user might well be initially overwhelmed with
the different analysis choices and the theoretical bases of the
computer program. A significant number of publications referred
to in the text (books, papers and reports) describe in detail the finite
element analysis procedures used in the program. However, this
literature is very comprehensive and frequently provides more
detail than the user needs to consult for the effective use of
ADINA. Furthermore, it is important that a user can identify easily
which publication should be studied if more information is desired
on a specific analysis option.
The intent with this Theory and Modeling Guide is
To provide a document that summarizes the methods and

assumptions used in the computer program ADINA
To provide specific references that describe the finite

element procedures in more detail.
Hence, this manual has been compiled to provide a bridge

between the actual practical use of the ADINA system and the
theory documented in various publications. Much reference is
made to the book Finite Element Procedures (ref. KJB) and to
other publications but we endeavored to be specific when
referencing these publications so as to help you to find the relevant
information.
ref. K. J. Bathe, Finite Element Procedures, 2nd ed.,

Cambridge, MA: Klaus-Jrgen Bathe, 2014.
Following this introductory chapter, Chapter 2 describes the

elements available in ADINA. The formulations used for these
elements have been proven to be reliable and efficient in linear,
large displacement, and large strain analyses. Chapter 3 describes
ADINA R & D, Inc. 1

Chapter 1: Introduction
the material models available in ADINA. Chapter 4 describes the

different contact formulations and provides modeling tips for
contact problems. Loads, boundary conditions and constraints are
addressed in Chapter 5. Eigenvalue type analyses such as
linearized buckling and frequency analyses are described in
Chapter 6. Chapter 7 provides the formulations used for static and
implicit dynamic analysis, while Chapter 8 deals with explicit
dynamic analysis. Several frequency domain analysis tools are
detailed in Chapter 9. Fracture mechanics features are described in
Chapter 10. Additional capabilities such as substructures, cyclic
symmetry, initial conditions, parallel processing and restarts are
discussed in Chapter 11. Chapter 12 describes the heat transfer
capabilities in the ADINA program. Finally, some post-processing
considerations are provided in Chapter 13.
We intend to update this report as we continue our work on the
ADINA system. If you have any suggestions regarding the
material discussed in this manual, we would be glad to hear from
you.
1.2 Supported computers and operating systems

Table 1.2-1 shows all supported computers/operating systems.
1.3 Units
When using the ADINA system, it is important to enter all
physical quantities (lengths, forces, masses, times, etc.) using a
consistent set of units. For example, when working with the SI
system of units, enter lengths in meters, forces in Newtons, masses
in kg, times in seconds. When working with other systems of units,
all mass and mass-related units must be consistent with the length,
force and time units. For example, in the USCS system (USCS is
an abbreviation for the U.S. Customary System), when the length
unit is inches, the force unit is pound and the time unit is second,
the mass unit is lb-sec2/in, not lb.
Rotational degrees of freedom are always expressed in radians.
Most angular input parameters are expressed in degrees.
Table 1.3-1 gives some of the more commonly used units
needed for ADINA input.
2 ADINA Structures Theory and Modeling Guide

1. 4 ADINA System documentation
Table 1.2-1: Supported computers/operating systems
Platform Operating system Version ADINA- Parallelized Fortran

M assembly1 compiler
Linux 2.6.18 and higher, 64-bit Parasolid Yes Intel ifort
x86_64 / glibc 2.3.4 and and Open 11.1
Opteron higher Cascade
IBM AIX 5.1 64-bit Parasolid No IBM xlf90
AIX 8.1
PC Windows XP, 64-bit Parasolid Yes Intel Visual
x86_64 / Vista, 7, 8 and Open Fortran 11
Opteron Cascade
1) All program versions have parallelized solvers. Only ADINA and ADINA-
Thermal have parallelized assembly.
2) The Intel 64 and AMD Opteron implementations of the x86_64 architecture are
supported on Linux and PC (Windows).
ADINA R & D, Inc. 3

Table 1.3-1: Units
SI SI (mm) USCS (inch) USCS (kip)
length meter (m) millimeter inch (in) inch (in)

(mm)
force Newton (N) Newton (N) pound (lb) kip (1000 lb)
time second (s) second (s) second (sec) second (sec)
mass kilogram (kg) N-s2/mm lb-sec2/in kip-sec2/in
= N-s2/m (note 1)
pressure, Pascal (Pa) = N/mm2 psi = lb/in2 ksi = kip/in2

stress, N/m2
Youngs
modulus,
etc.
density kg/m3 N-s2/mm4 lb-sec2/in4 kip-sec2/in4
(note 2)
1) A body that weighs 1 lb has a mass of 1/386.1 = 0.002590 lb-sec2/in.

2) A body with weight density 1 lb/in3 has a density of 0.002590 lb-sec2/in4.
1.4 ADINA System documentation

At the time of printing of this manual, the following documents
are available with the ADINA System:
Installation Notes
Describes the installation of the ADINA System on your
computer. Depending on the platform, the appropriate
installation notes in pdf format can be printed or downloaded
from http://www.adina.com
ADINA Handbook
Written as a task-oriented desktop reference, the ADINA
handbook helps users to quickly and effectively leverage
ADINA's advanced geometric modeling, meshing, and
visualization features.

1. 4 ADINA System documentation
ADINA User Interface Command Reference Manual
Volume I: ADINA Solids & Structures Model Definition
Volume II: ADINA Thermal Model Definition
Volume III: ADINA CFD & FSI Model Definition
Volume IV: ADINA EM Model Definition
Volume V: Display Processing
These documents describe the AUI command language. You

use the AUI command language to write batch files for the AUI.
ADINA Primer
Tutorial for the ADINA User Interface, presenting a sequence
of worked examples which progressively instruct you how to
effectively use the ADINA System.
Volume I: ADINA Solids & Structures
Volume II: ADINA Thermal
Volume III: ADINA CFD & FSI
Volume IV: ADINA EM
Provides a concise summary and guide for the theoretical basis

of the analysis programs. The manuals also provide references
to other publications which contain further information, but the
detail contained in the manuals is usually sufficient for effective
understanding and use of the programs.
ADINA Verification Manual

Presents solutions to problems which verify and demonstrate the
usage of the ADINA System. Input files for these problems are
distributed along with the ADINA System programs.
ADINA R & D, Inc. 5

ADINA-Nastran Interface Manual

Describes the ADINA-AUI Nastran Interface. This guide is
available as a pdf file. The Nastran Case Control Commands,
Parameters, and Bulk Data Entries that are supported by the
AUI are documented.
TRANSOR for I-DEAS Users Guide

Describes the interface between the ADINA System and NX
I-deas. This guide is available in html format and is directly
accessible from the TRANSOR interface within I-deas. The use
of TRANSOR for I-deas to perform pre-/post-processing and
ADINA analysis within the I-deas environment, is described.
TRANSOR for Femap Users Guide

Describes the interface between the ADINA System and Femap.
This guide is available as a pdf file. The use of TRANSOR for
Femap to perform pre-/post-processing and ADINA analysis
within the Femap environment, is described.
ADINA System 9.2 Release Notes

Provides a description of the new and modified features of the
ADINA System 9.2.

2.1 Truss and cable elements
2. Elements
This chapter outlines the theory behind the different element
classes, and also provides details on how to use the elements in
modeling. This includes the materials that can be used with each
element type, their applicability to large displacement and large
strain problems, their numerical integration, etc.
2.1.1 General considerations
The truss elements can be employed as 2-node, 3-node and

4-node elements, or as a 1-node ring element. Fig. 2.1-1 shows the
elements available in ADINA.
wi
l i
l l
vi
l l ui
Z l
l
l
l
Y
X (u, v, w) are global displacement
degrees of freedom (DOF)
(a) 2-, 3- and 4-node elements
Z
i vi
l
One DOF per node
Y
(b) 1-node ring element
ref. KJB
Sections Figure 2.1-1: Truss elements available in ADINA
5.3.1,
6.3.3
ADINA R & D, Inc. 7

Chapter 2: Elements
Note that the only force transmitted by the truss element is the
longitudinal force as illustrated in Fig. 2.1-2. This force is constant
in the 2-node truss and the ring element, but can vary in the 3- and
4-node truss (cable) elements.
Z P
Y
X
area A
s
Stress constant over
P cross-sectional area
(a) 2- to 4-node elements
Young's modulus E
Z area A
P
R
R
Z 1 radian Y
l
Y
s (hoop stress)
X
Hoop strain = v/R Elastic stiffness = EA/R (one radian)
Circumferential force P = sA
(b) Ring element
Figure 2.1-2: Stresses and forces in truss elements
The ring element is formulated for one radian of the structure

(as is the axisymmetric 2-D solid element, see Section 2.2.1), and
this formulation is illustrated in Fig. 2.1-2.
The local node numbering and natural coordinate system for the
2-node, 3-node and 4-node truss elements are shown in Fig. 2.1-3.

r r=0 r
r=-1 r=+1 r=-1 r=+1
l l l l l
1 2 1 3 2
2-node element 3-node element
r=-1/3 r r=1/3
r=-1 r=1
l l l l
1 3 4 2
4-node element
Figure 2.1-3: Local node numbering; natural coordinate system
2.1.2 Material models and formulations
The truss element can be used with the following material

models: elastic-isotropic, nonlinear-elastic, plastic-bilinear,
plastic-multilinear, plastic-cyclic, thermo-isotropic, thermo-
plastic, creep, plastic-creep, multilinear-plastic-creep, creep-
variable, plastic-creep-variable, multilinear-plastic-creep-
variable, viscoelastic, shape-memory alloy, user-coded.
The truss element can be used with the small or large

displacement formulations. In the small displacement formulation,
the displacements and strains are assumed infinitesimally small. In
the large displacement formulation, the displacements and rotations
can be very large. In all cases, the cross-sectional area of the
element is assumed to remain unchanged, and the strain is equal to
the longitudinal displacement divided by the original length.
All of the material models in the above list can be used with
either formulation. The use of a linear material with the small
displacement formulation corresponds to a linear formulation, and
the use of a nonlinear material with the small displacement
formulation corresponds to a materially-nonlinear-only
formulation.
2.1.3 Numerical integration
The integration point labeling of the truss (cable) elements is

given in Fig. 2.1-4.
ADINA R & D, Inc. 9

Chapter 2: Elements
r r
l l l l
1 1 2
One-point integration Two-point integration
r r
l l l l
1 2 3 1 2 3 4
Three-point integration Four-point integration
Figure 2.1-4: Integration point locations
Note that when the material is temperature-independent, the 2-

node truss and ring elements require only 1-point Gauss numerical
integration for an exact evaluation of the stiffness matrix, because
the force is constant in the element. However, 2-point Gauss
numerical integration may be appropriate when a temperature-
dependent material is used because, due to a varying temperature,
the material properties can vary along the length of an element.
2.1.4 Mass matrices
The consistent mass matrix is calculated using Eq. (4.25) in ref.

KJB, p. 165, which accurately computes the consistent mass
distribution.
The lumped mass matrix is formed by dividing the elements mass

i
among its nodes. The mass assigned to each node is M ,
L
where M = total mass, L = total element length, i = fraction of
the total element length associated with element node i (i.e., for the
L L
2-node truss element, 1 = and 2 = , and for the 4-node
2 2
L L
truss 1 = 2 = and 3 = 4 = ). The element has no
6 3
rotational mass.

2.1.5 Element output
Each element outputs, at its integration points, the following

information to the porthole file, based on the material model. This
information is accessible in the AUI using the given variable
names.
Elastic-isotropic, nonlinear-elastic: FORCE-R, STRESS-RR,

STRAIN-RR
Elastic-isotropic with thermal effects: FORCE-R, STRESS-RR,

STRAIN-RR, THERMAL_STRAIN,
ELEMENT_TEMPERATURE
Thermo-isotropic: FORCE-R, STRESS-RR, STRAIN-RR,

ELEMENT_TEMPERATURE
Plastic-bilinear, plastic-multilinear: PLASTIC_FLAG,

FORCE-R, STRESS-RR, STRAIN-RR,
PLASTIC_STRAIN-RR
SMA: EFFECTIVE_STRESS, STRESS-RR,

ACCUM_EFF_TRANSF_STRAIN, AUSTENITE_FRACTION,
DETWINNED_MARTENSITE_FRACTION, STRAIN-RR,
THERMAL_STRAIN-RR, TRANSFORMATION_STRAIN-RR,
TWINNED_MARTENSITE_FRACTION, SMA_FLAG
Thermo-plastic, creep, plastic-creep, multilinear-plastic-creep:

PLASTIC_FLAG, NUMBER_OF_SUBINCREMENTS,
FORCE-R, STRESS-RR, STRAIN-RR,
ACCUMULATED_EFFECTIVE_PLASTIC_STRAIN,
PLASTIC_STRAIN-RR, CREEP_STRAIN-RR,
THERMAL_STRAIN-RR, ELEMENT_TEMPERATURE,
FE_EFFECTIVE_STRESS, YIELD_STRESS,
EFFECTIVE_CREEP_STRAIN
Viscoelastic: PLASTIC_FLAG, STRESS-RR, STRAIN-RR,

THERMAL_STRAIN-RR, ELEMENT_TEMPERATURE
User-coded: STRESS-RR, STRAIN-

RR,USER_VARIABLE_I, INT_USER_VARIABLE_I
ADINA R & D, Inc. 11

Chapter 2: Elements
See Section 13.1.1 for the definitions of those variables that are not
self-explanatory.
2.1.6 Recommendations on use of elements
Of the 2-, 3- and 4-node truss elements, the 2-node element is

usually most effective and can be used in modeling truss structures,
cable structures, linear springs (the spring stiffness k is equal to
EA/L where L is the length of the element) and nonlinear elastic
gap elements (see Section 3.3).
The 3- and 4-node elements are employed to model cables and

steel reinforcement in reinforced concrete structures that are
modeled with higher-order continuum or shell elements. In this
case, the 3- and 4-node truss elements are compatible with the
continuum or shell elements.
The internal nodes for the 3-node and 4-node truss elements are
usually best placed at the mid- and third-points, respectively, unless
a specific predictive capability of the elements is required (see ref.
KJB, Examples 5.2 and 5.17, pp. 348 and 370).
2.1.7 Rebar elements
The AUI can generate truss elements that are connected to the
2D or 3D solid elements in which the truss elements lie. There are
several separate cases, corresponding to the type of the truss
element group and the type of the solid element group in which the
truss elements lie.
Axisymmetric truss elements, axisymmetric solid elements: The

AUI can connect axisymmetric truss elements that exist in the
model prior to data file generation to the axisymmetric 2D elements
in which the truss elements lie. The AUI does this connection
during data file generation as follows. For each axisymmetric truss
element that lies in an axisymmetric 2D element, constraint
equations are defined between the axisymmetric truss element node
and the three closest corner nodes of the 2D element.

Axisymmetric truss element Constraint equation

l l l l l l
l l
l l l l l l
Axisymmetric 2D solid element

(a) Before data file generation (b) After data file generation
Figure 2.1-5: Rebar in axisymmetric truss, 2D solid elements
3D truss elements, planar 2D solid elements: The AUI can

generate 3D truss elements and then connect the truss elements to
the planar 2D elements in which the truss elements lie. The AUI
does this during data file generation as follows. For each rebar
line, the AUI finds the intersections of the rebar line and the sides
of the 2D elements. The AUI then generates nodes at these
intersections and generates truss elements that connect the
successive nodes. The AUI also defines constraint equations
between the generated nodes and the corner nodes of the 2D
elements.
Rebar line Truss element
l l l l l l
l
l l
l l l l l l
Planar 2D solid element Constraint equation
Figure 2.1-6: Rebar in 3D truss, planar 2D solid elements
3D truss elements, 3D solid elements: The AUI can generate 3D

truss elements and then connect the truss elements to the 3D
elements in which the truss elements lie. The AUI does this during
data file generation as follows. For each rebar line, the AUI finds
the intersections of the rebar line and the faces of the 3D elements.
The AUI then generates nodes at these intersections and generates
truss elements that connect the successive nodes. The AUI also
defines constraint equations between the generated nodes and three
closest corner nodes of the 3D element faces.

Chapter 2: Elements
Rebar line Constraint equation

l l
l l l l
l
l l l l l l
l
l l l l l l
l
l l l l
l
3D solid element l Truss element
Figure 2.1-7: Rebar in 3D truss, 3D solid elements
The rebar option is intended for use with lower-order solid

elements (that is, the solid elements should not have mid-side
nodes).
For 3D truss element generation, if the rebar lines are curved,

you should specify enough subdivisions on the rebar lines so that
the distance between subdivisions is less than the solid element
size. The more subdivisions that are on the rebar lines, the more
accurately the AUI will compute the intersections of the rebar lines
and the solid element sides or faces.
To request the connection:
AUI: Click the Define Element Group icon to open the Define
Element Group window. In this window, check that the Element
Type has been set to Truss. Click on the Advanced tab. Select
the Use as Rebars option from the Element Option drop-down,
and then fill in the Rebar-Line Label box below the Element
Option drop-down box.
Command Line: Set OPTION=REBAR in the EGROUP TRUSS

command. For 3D truss elements, use the REBAR-LINE command
and RB-LINE parameter of the EGROUP TRUSS command to
define the rebar lines.

2.2: Two-dimensional solid elements
2.2 Two-dimensional solid elements

The following kinematic assumptions are available for two-

dimensional elements in ADINA: plane stress, plane strain,
axisymmetric , generalized plane strain and 3-D plane stress
(membrane). Figures 2.2-1 and 2.2-2 show some typical 2-D
elements and the assumptions used in the formulations.
l
l
l
l l
l
L
l
l
Recommendation:
l
h h/L > 1/10

(a) 8- & 9-node elements
l l
l
l l
l
l l l
l
l l
l l l
l l l l l
l l
(b) 3- & 4-node elements (c) 6- & 7-node triangles
Figure 2.2-1: 2-D solid elements
The plane stress, plane strain, generalized plane strain, and

axisymmetric elements can be defined in the X-Y, Y-Z or X-Z
planes. The axisymmetric element can have x or y or z as the
rotating axis and does not need to be located in the +X, +Y or +Z
half planes.
The 3-D plane stress element can lie in general three-

dimensional space. The element can be initially flat or can be
curved. This element is sometimes called a membrane element.
The axisymmetric element represents one radian of the

ref. KJB structure. The stiffness, mass and loads are defined accordingly.
Sections 5.3.1
and 5.3.2 Hence, when this element is combined with other elements, or
when concentrated loads are defined, these must also refer to one
radian, see ref. KJB, Examples 5.9 and 5.10, p. 356.

Chapter 2: Elements
z l l s xx = 0 l l e xx = 0
z
y s xy = 0 g xy = 0
x y
l l s xz = 0 x l l g xz = 0
a) Plane stress element b) Plane strain element

in y-z plane in y-z plane
exx = u/y e xx = function of (y, z),
l l l l
z (linear analysis) see text
z
g xy = 0 g xy = 0
x y y
l l
g xz = 0 x l l
g xz = 0
c) Axisymmetric element d) Generalized plane strain

in y-z plane in y-z plane
Z l l s xx = 0 (Stresses computed
s xy = 0 in element local
X Y l l s xz = 0 coordinate system)
e) 3-D plane stress (membrane) element
Figure 2.2-2: Basic assumptions in 2-D analysis
The plane strain element represents a unit thickness of the

structure. The stiffness, mass and loads are defined accordingly.
The plane stress element is either a constant thickness element

or a varying thickness element. In a varying thickness element, the
thickness at each element local node can be different.
The generalized plane strain element is intended to be used to

model beam-like structures which have a constant cross-section and
are submitted to a loading independent of the axial coordinate of
the beam. In such cases, the deformations of the structure are
independent of the axial coordinate of the structure. This situation
is shown in Fig. 2.2-7(d).
The solution is valid only for the section of the structure that are
far enough from the ends; otherwise local 3-D effects usually
render the generalized plane strain assumptions invalid.
The elements usually used in ADINA are isoparametric

displacement-based finite elements, and their formulation is

described in detail in ref. KJB, Section 5.3.
The basic finite element assumptions are, see Fig. 2.2-3, for the
coordinates
q q
y = hi yi ; z = hi zi
i =1 i =1
for the displacements in plane stress, plane strain, and

axisymmetric elements
q q
v = hi vi ; w = hi wi
i =1 i =1
for the displacements in generalized plane strain elements
u = xU p x( y y p ) zp + x( z z p ) yp
q q
1 2 p 1 2 p
v = hi vi + x z w = hi wi x y
i =1 2 i =1 2
where
hi(r,s) = interpolation function corresponding to node i
(r,s) = isoparametric coordinates
q = number of element nodes, 3 q 9, excluding
auxiliary node in generalized plane strain
yi, zi = nodal point coordinates
vi, wi = nodal point displacements
U p , yp , zp = degrees of freedom of the auxiliary node
xp, yp, zp = coordinates of the auxiliary node
For generalized plane strain the elements are assumed to have a

constant curvature in the X-direction, and to lie between two planes
that can move relative to each other. One plane must be the X = 0
plane where the elements must be located. The location of the other
plane is specified via the auxiliary point which must lie on that
plane.
An axial force and bending moments can be applied to the
auxiliary point, and these forces/moments can be used to model the

Chapter 2: Elements
corresponding forces/moments acting onto the beam structure.
Z
s
2l l
5 1
l
Local displacement DOF

6l l
9 8 r
l
w4
3l l
7 4 v4
l
z4
X Y
y4
a) Plane strain, plane stress, and axisymmetric elements
Section of the structure

(must lie in Y-Z plane) Fzp
Np Fyp
Up
Z, w w4 w3
4 v4 3 v3
Y, v
w1 v w2 v2
X, u 1
1 2
The coordinates of the auxiliary node Np are (xp, yp, zp).
The element has unit length in the X direction.
b) Generalized plane strain element
Figure 2.2-3: Conventions used for the nodal coordinates and

displacements of the 2-D solid element
The displacements in the element can be interpreted as the sum

of the 2-D in-plane displacements (due to in-plane forces and

boundary conditions) plus additional displacements which come
from the out-of-plane rigid body motions of the cross-section (due
to the axial force and bending moments applied to the structure).
From the displacement interpolations, the out-of-plane normal
strain is seen to be
xx = U p ( y y p ) zp + ( z z p ) yp
In addition to the displacement-based elements, special mixed-

interpolated elements are also available, in which the displacements
and pressure are interpolated separately. These elements are
effective and should be preferred in the analysis of incompressible
media and inelastic materials (specifically for materials in which
Poisson's ratio is close to 0.5, for rubber-like materials and for
elasto-plastic materials). The mixed formulation is only available
for the plane strain, axisymmetric and generalized plane strain
elements. It is not available (and not needed) for plane stress and
3-D plane stress (membrane) elements.
The mixed formulation is the default setting for the materials in

which either the entire response is incompressible (for example, the
rubber material models), or in which the inelastic response is
incompressible (for example, the plastic and creep material
models).
Table 2.2-1 shows the default number of pressure degrees of

freedom for each 2-D element type. For more details on the mixed
interpolation of pressure and displacement degrees of freedom for
3-D solids, see Section 4.4.3, p. 276, and Tables 4.6 and 4.7,
pp.292-295 in ref. KJB.

Chapter 2: Elements
Table 2.2-1: Mixed formulation settings for 2-D solid elements
Number of default
2-D solid element
pressure DOFs
3-node triangle 1*
4-node quadrilateral (4/1) 1
6-node triangle (6/3) 3
7-node triangle (7/3) 3
*
To use this element, the user must use the displacement-
based element with 1-point integration.
The default number of pressure degrees of freedom shown in

Table 2.2-1 is optimal for most applications. However, it can be
modified if needed. In some problems where rubber is subjected to
severe compression, numerical instability can be avoided by using
4 pressure degrees of freedom for the 8 and 9-node quadrilaterals.
The following choices for the number of pressure degrees of

freedom are valid; 1 for a constant pressure, 3 for additional linear
pressure terms, 4 for an additional bilinear pressure term (involving
rs), and 6 for additional quadratic pressure terms (involving r2, s2).
In general, using too many pressure degrees of freedom leads to

volumetric locking, while using too few pressure degrees of
freedom leads to spurious zero-energy modes. See Sections 4.4.2
and 4.5 in ref. KJB for more details.
ref. T. Sussman and K.J. Bathe, "A Finite Element
Formulation for Nonlinear Incompressible Elastic and
Inelastic Analysis," J. Computers and Structures, Vol.
26, No. 1/2, pp. 357-409, 1987.
In addition to the displacement-based and mixed-interpolated

elements, ADINA also includes the possibility of including
incompatible modes in the formulation of the 4-node element.

Within this element, additional displacement degrees of freedom

are introduced. These additional displacement degrees of freedom
are not associated with nodes; therefore the condition of
displacement compatibility between adjacent elements is not
satisfied in general. However the additional displacement degrees
of freedom increase the flexibility of the element, especially in
bending situations.
Figure 2.2-4 shows a situation in which the incompatible modes
elements give an improved solution.
For theoretical considerations, see reference KJB, Section 4.4.1.
Note that these elements are formulated to pass the patch test. Also
note that element distortions deteriorate the element performance.
The incompatible modes feature cannot be used in conjunction
with the mixed-interpolation formulation.
The incompatible modes feature is only available for the 4-node
element; in particular note that the incompatible modes feature is
not available for the 3-node triangular element.
First bending mode Second bending mode
4-node elements,
134.5 Hz 358.0 Hz
no incompatible modes
4-node elements, 108.6 Hz 294.0 Hz
incompatible modes
107.1 Hz 279.9 Hz
8-node elements
Plane strain conditions, Third bending mode First axial mode

beam length = 5 m
beam height = 0.5 m
beam width = 1.0 m 674.8 Hz 540.6 Hz
566.3 Hz 540.5 Hz
E=2.07 1011 N/m2
u = 0.29 513.9 Hz 537.9 Hz
r = 7800 kg/m3
Figure 2.2-4: Frequency analysis of a free-free beam

with three types of finite elements
The elements can be used with 3 to 9 nodes. The interpolation

functions are defined in ref. KJB, Fig. 5.4, p. 344.
ref. KJB Triangular elements are formed in ADINA by assigning the

Section 5.3.2

Chapter 2: Elements
same global node number to all the nodes located on one side. It is
recommended that this "collapsed" side be the one with local end
nodes 1 and 4.
The triangular elements can be of different types:
A collapsed 6-node triangular element with the strain

1
singularity (see Chapter 10)
r
1
A collapsed 8-node element with the strain singularity
r
(see Chapter 10)
A collapsed 6-node spatially isotropic triangular element
A collapsed 7-node spatially isotropic triangular element
These element types are described on pp. 369-370 of ref. KJB. The
strain singularities are obtained by collapsing one side of an 8-node
rectangular element and not correcting the interpolation functions.
The 6-node spatially isotropic triangle is obtained by correcting
the interpolation functions of the collapsed 8-node element. It then
uses the same interpolation functions for each of the 3 corner nodes
and for each of the midside nodes. Note that when the corrections
to the interpolation functions are employed, only the element side
with end nodes 1 and 4 can be collapsed.
The 3-node triangular element is obtained by collapsing one
side of the 4-node element. This element exhibits the constant
strain conditions (except that the hoop strain in axisymmetric
analysis varies over the element) but is usually not effective.
Linear and nonlinear fracture mechanics analysis of stationary

or propagating cracks can be performed with two-dimensional
elements (see Chapter 10).
The 2-D elements can be used with the following material

models: elastic-isotropic, elastic-orthotropic (with or without
wrinkling), nonlinear-elastic, plastic-bilinear, plastic-

multilinear, Drucker-Prager, Mroz-bilinear, plastic-

orthotropic, thermo-isotropic, thermo-orthotropic, thermo-
plastic, piezoelectric, creep, plastic-creep, multilinear-plastic-
creep, concrete, curve-description, Ogden, Mooney-Rivlin,
Arruda-Boyce, hyper-foam, Sussman-Bathe, creep-variable,
plastic-creep-variable, multilinear-plastic-creep-variable,
Gurson-plastic, Cam-clay, Mohr-Coulomb, viscoelastic, creep
irradiation, gasket, shape-memory alloy, plastic-cyclic, user-
coded.
In plane strain, generalized plane strain and axisymmetric

analysis, the mixed interpolation formulation should be used (and
is the default) whenever using the plastic-bilinear, plastic-
multilinear, Mroz- bilinear, plastic-orthotropic, thermo-plastic,
creep, plastic-creep, multilinear-plastic-creep, Ogden, Mooney-
Rivlin, Arruda-Boyce, Sussman-Bathe, creep-variable, plastic-
creep-variable, multilinear-plastic-creep-variable, plastic-cyclic or
viscoelastic material models.
In addition the mixed interpolation formulation should be
used (but is not the default) when using the elastic-isotropic
material and the Poissons ratio is close to 0.5.
The two-dimensional elements can be used with a small

displacement/small strain, large displacement/small strain or a
large displacement/large strain formulation.
The small displacement/small strain and large
displacement/small strain formulations can be used with any
material model, except for the Ogden, Mooney-Rivlin, Arruda-
Boyce, hyper-foam and Sussman-Bathe material models. The use
of a linear material with the small displacement/
small strain formulation corresponds to a linear formulation, and
the use of a nonlinear material with the small displacement/small
strain formulation corresponds to a materially-nonlinear-only
formulation. The program uses the TL (total Lagrangian)
formulation when you choose a large displacement/small strain
formulation.
The large displacement/large strain formulations can be used
with the plastic-bilinear, plastic-multilinear, Mroz-bilinear, plastic-
orthotropic, thermo-plastic, creep, plastic-creep, multilinear-plastic-
creep, creep-variable, plastic-creep-variable, multilinear-plastic-
creep-variable, viscoelastic, plastic-cyclic and user-coded material

Chapter 2: Elements
models. The program uses either a ULH or ULJ formulation when

you choose a large displacement/large strain formulation with these
material models.
A large displacement/large strain formulation is used with the
Ogden, Mooney-Rivlin, Arruda-Boyce, hyper-foam and Sussman-
Bathe material models. The program uses a TL (total Lagrangian)
formulation in this case.
ref. KJB
The basic continuum mechanics formulations are described in
Sections 6.2 and ref. KJB, pp. 497-537, and the finite element discretization is given
6.3.4 in ref. KJB pp. 538-542, 549-555.
ref. KJB Note that all these formulations can be used together in a single
Section 6.8.1 finite element mesh. If the elements are initially compatible, then
they will remain compatible throughout the complete analysis.
For the calculation of all element matrices and vectors,

ref. KJB
Sections 5.5.3, numerical Gauss integration is used. You can use from 22 to 66
5.5.4 and 5.5.5 Gauss integration. The numbering convention and the location of
the integration points are given in Fig. 2.2-5 for 22 and 33 Gauss
integration. The convention used for higher orders is analogous.
The default Gauss integration orders for rectangular elements are
22 for 4-node elements and 33 otherwise.

s
INR=2 INS=2 s INS=3
INR=3
INR=1 INR=2
INR=1 INS=2
r r
INS=1
INS=1
a) Rectangular elements
2
1
1 4
3
r
1-point 4-point
3 3 4
2 6 7 8
5 2 10
6 12
1 7
4 1 11 13
5 9
7-point 13-point
b) Triangular elements
Figure 2.2-5: Integration point positions for 2-D solid elements
For the 8-node rectangular element, the use of 22 Gauss

integration corresponds to a slight under-integration and one
spurious zero energy mode is present. In practice, this particular
kinematic mode usually does not present any problem in linear
analysis (except in the analysis of one-element cases, see ref. KJB,
Fig. 5.40, p. 472) and thus 22 Gauss integration can be employed
with caution for the 8-node elements.

Chapter 2: Elements
The 3-node, 6-node and 7-node triangular elements are spatially

isotropic with respect to integration point locations and
interpolation functions (see Figure 2.2-5).
The integration order used for triangular elements depends upon

the integration order used for rectangular elements as follows.
When rectangular elements use 22 Gauss integration, triangular
elements use 4-point Gauss integration; when rectangular elements
use 33 Gauss integration, triangular elements use 7-point Gauss
integration; when rectangular elements use 44 Gauss integration
or higher, triangular elements use 13-point Gauss integration.
However, for 3-node triangular elements, when the
axisymmetric subtype is not used, then 1-point Gauss integration is
always used.
Note that in geometrically nonlinear analysis, the spatial

positions of the Gauss integration points change continuously as
the element undergoes deformations, but throughout the response
the same material particles are at the integration points.
The order of numerical integration in the large displacement

ref. KJB
Section 6.8.4 elastic analysis is usually best chosen to be equal to the order used
in linear elastic analysis. Hence, the program default values should
usually be employed. However, in inelastic analysis, a higher order
integration should be used when the elements are used to model
thin structures (see Fig. 6.25, p. 638, ref. KJB), or large
deformations (large strains) are anticipated.
2.2.4 Mass matrices
The consistent mass matrix is always calculated using either

33 Gauss integration for rectangular elements or 7-point Gauss
integration for triangular elements.
The lumped mass matrix of an element is formed by dividing

the elements mass M equally among its n nodes. Hence, the mass
assigned to each node is M / n. No special distributory concepts
are employed to distinguish between corner and midside nodes, or
to account for element distortion.
Note that n is the number of distinct non-repeated nodes in the

element. Hence, when a quad element side is collapsed to a single

node, the total mass of the element is equally distributed among the
final nodes of the triangular configuration.
You can request that ADINA either print or save stresses or forces.
Stresses: Each element outputs, at its integration points, the

following information to the porthole file, based on the material
model. This information is accessible in the AUI using the given
variable names.
Notice that for the orthotropic and piezoelectric material
models, you can request that the stresses and strains be saved in the
material coordinate system.
For the 3-D plane stress (membrane) elements, results with
indices YY, ZZ, etc are saved in the element local coordinate
system (see Figure 2.2-6).
Z The local y direction

is determined by local
Y nodes 1 and 2.
z
X x The local z direction
4 lies in the plane defined
3 by local nodes 1, 2, 3,
1 and is perpendicular to the
2 local y direction.
y
Figure 2.2-6: Local coordinate system for 3-D plane stress

(membrane) element
Elastic-isotropic, elastic-orthotropic (no wrinkling, results saved in

global system): STRESS(XYZ), STRAIN(XYZ)
Elastic-isotropic with thermal effects: STRESS(XYZ),

STRAIN(XYZ), THERMAL_STRAIN,
ELEMENT_TEMPERATURE

Chapter 2: Elements
Elastic-orthotropic (no wrinkling, results saved in material system):

STRESS(ABC), STRAIN(ABC)
Thermo-isotropic, thermo-orthotropic (results saved in global

system): STRESS(XYZ), STRAIN(XYZ),
THERMAL_STRAIN(XYZ), ELEMENT_TEMPERATURE
Piezoelectric (results saved in global system): STRESS(XYZ),

STRAIN(XYZ), ELECTRIC_FIELD-Y,
ELECTRIC_FIELD-Z,
ELECTRIC_DISPLACEMENT_FIELD-Y,
ELECTRIC_DISPLACEMENT_FIELD-Z.
Piezoelectric (results saved in material system):

STRESS(ABC), STRAIN(ABC), ELECTRIC_FIELD-A,
ELECTRIC_FIELD-B,
ELECTRIC_DISPLACEMENT_FIELD-A,
ELECTRIC_DISPLACEMENT_FIELD-B.
Curve description: CRACK_FLAG, STRESS(XYZ),

STRAIN(XYZ), FE_SIGMA-P1, FE_SIGMA-P2,
FE_SIGMA-P1_ANGLE, GRAVITY_IN-SITU_PRESSURE,
VOLUMETRIC_STRAIN
Concrete: CRACK_FLAG, STRESS(XYZ), STRAIN(XYZ),

FE_SIGMA-P1, FE_SIGMA-P2, FE_SIGMA-P1_ANGLE,
FAILURE_STRESS, ELEMENT_TEMPERATURE,
THERMAL_STRAIN(XYZ)
Small strains: Plastic-bilinear, plastic-multilinear, Mroz-bilinear:

PLASTIC_FLAG, STRESS(XYZ), STRAIN(XYZ),
PLASTIC_STRAIN(XYZ), FE_EFFECTIVE_STRESS,
YIELD_STRESS,
SMA: EFFECTIVE_STRESS, STRESS-XX, STRESS-YY,

STRESS-YZ, STRESS-ZZ,
DETWINNED_MARTENSITE_FRACTION, STRAIN-XX,
STRAIN-YY, STRAIN-YZ, STRAIN-ZZ,
THERMAL_STRAIN-XX, THERMAL_STRAIN-YY,
THERMAL_STRAIN-ZZ, TRANSFORMATION_STRAIN-XX,
TRANSFORMATION_STRAIN-YY,

TRANSFORMATION_STRAIN-YZ,
TRANSFORMATION_STRAIN-ZZ,
Large strains: Plastic-bilinear, plastic-multilinear, Mroz-bilinear:

PLASTIC_FLAG, STRESS(XYZ),
DEFORMATION_GRADIENT(XYZ),
Small strains: Anand Model

STRESS(XYZ), STRAIN(XYZ),
VISCOPLASTIC_STRAIN (XYZ),
FE_EFFECTIVE_STRESS,
ACCUMULATED_EFFECTIVE_VISCOPLASTIC_STRAIN,
EFFECTIVE_VISCOPLASTIC_STRAIN_RATE,
Large strains: Anand Model

STRESS(XYZ), DEFORMATION_GRADIENT(XYZ),
Plastic-orthotropic (results saved in global system):

PLASTIC_STRAIN(XYZ), HILL_EFFECTIVE_STRESS,
YIELD_STRESS,
Plastic-orthotropic (results saved in material system):

PLASTIC_FLAG, STRESS(ABC), STRAIN(ABC),
PLASTIC_STRAIN(ABC), HILL_EFFECTIVE_STRESS,
YIELD_STRESS,
THERMAL_STRAIN(ABC), ELEMENT_TEMPERATURE
Small strains: Drucker-Prager: PLASTIC_FLAG,
PLASTIC_FLAG-2, STRESS(XYZ), STRAIN(XYZ),
PLASTIC_STRAIN(XYZ), CAP_LOCATION,
YIELD_FUNCTION

Chapter 2: Elements
Large strains: Drucker-Prager: PLASTIC_FLAG,

PLASTIC_FLAG-2, STRESS(XYZ),
DEFORMATION_GRADIENT(XYZ), CAP_LOCATION,
YIELD_FUNCTION
Small strains: Thermo-plastic, creep, plastic-creep, multilinear-

plastic-creep, creep-variable, plastic-creep-variable, multilinear-
plastic-creep-variable: PLASTIC_FLAG,
NUMBER_OF_SUBINCREMENTS, STRESS(XYZ),
STRAIN(XYZ), PLASTIC_STRAIN(XYZ),
CREEP_STRAIN(XYZ), THERMAL_STRAIN(XYZ),
ELEMENT_TEMPERATURE,
Large strains: Thermo-plastic, creep, plastic-creep, multilinear-

THERMAL_STRAIN(XYZ), ELEMENT_TEMPERATURE,
Mooney-Rivlin , Ogden, Arruda-Boyce, hyper-foam (strains

saved): STRESS(XYZ), STRAIN(XYZ)
Mooney-Rivlin, Ogden, Arruda-Boyce, hyper-foam (deformation

gradients saved): STRESS(XYZ),
DEFORMATION_GRADIENT(XYZ).
Small strains: User-supplied: STRESS(XYZ), STRAIN(XYZ),

USER_VARIABLE_I
Large strains: User-supplied: STRESS(XYZ),

DEFORMATION_GRADIENT(XYZ), USER_VARIABLE_I
Elastic-orthotropic (wrinkling, results saved in global system):

WRINKLE_FLAG, STRESS(XYZ), STRAIN(XYZ)

Elastic-orthotropic (wrinkling, results saved in material system):

WRINKLE_FLAG, STRESS(ABC), STRAIN(ABC)
Thermo-orthotropic (results saved in material system):

STRESS(ABC), STRAIN(ABC),
Small strains: Gurson plastic: PLASTIC_FLAG,

YIELD_STRESS,
VOID_VOLUME_FRACTION
Large strains: Gurson plastic: PLASTIC_FLAG,

Small strains, Cam-clay: PLASTIC_FLAG, STRESS(XYZ),

YIELD_SURFACE_DIAMETER_P, YIELD_FUNCTION,
MEAN_STRESS, DISTORTIONAL_STRESS,
VOLUMETRIC_STRAIN, VOID_RATIO,
EFFECTIVE_STRESS_RATIO, SPECIFIC_VOLUME
Large strains, Cam-clay: PLASTIC_FLAG, STRESS(XYZ),

Mohr-Coulomb: PLASTIC_FLAG, STRESS(XYZ),

YIELD_FUNCTION
Small strains, viscoelastic: STRESS(XYZ), STRAIN(XYZ),


Chapter 2: Elements
Large strains, viscoelastic: STRESS(XYZ),

Small strains: creep-irradiation: STRESS(XYZ),

EFFECTIVE_STRESS, YIELD_STRESS,
EFFECTIVE_CREEP_STRAIN,
IRRADIATION_STRAIN(XYZ).
Large strains: creep-irradiation: STRESS(XYZ),

In the above lists,
STRESS(XYZ) = STRESS-YY, STRESS-ZZ, STRESS-

YZ, STRESS-XX
STRESS(ABC) = STRESS-AA, STRESS-BB, STRESS-

AB, STRESS-CC
with similar definitions for the other abbreviations used above. But
the variable DEFORMATION_GRADIENT(XYZ) is interpreted as
follows:
DEFORMATION_GRADIENT(XYZ) =
DEFORMATION_GRADIENT-YY,
DEFORMATION_GRADIENT-ZZ,
DEFORMATION_GRADIENT-XX,
DEFORMATION_GRADIENT-YZ,
DEFORMATION_GRADIENT-ZY
Also note that you can request stretches instead of deformation

gradients, and in this case STRETCH(XYZ) replaces
DEFORMATION_GRADIENT(XYZ) in the above lists.

See Section 13.1.1 for the definitions of those variables that are
not self-explanatory.
ref. KJB
Section 4.2.1, In ADINA, the stresses are calculated using the strains at the
6.6.3 point of interest. Hence, they are not spatially extrapolated or
smoothed. However, the AUI can be employed to calculate
smoothed stresses (see Section 13.1.3).
You can also request that strain energy densities be output along
with the stresses.
Nodal forces: The nodal forces which correspond to the element

stresses can also be requested in ADINA. This force vector is
calculated in a linear analysis using
F = BT dV
V
where B is the element strain displacement matrix, and is the

stress vector. The integration is performed over the volume of the
element; see Example 5.11 in ref. KJB, pp. 358-359.
The nodal forces are accessible in the AUI using the variable
names NODAL_FORCE-X, NODAL_FORCE-Y,
NODAL_FORCE-Z.
The same relation is used for the element force calculation in a

nonlinear analysis, except that updated quantities are used in the
integration.
The 9-node element is usually most effective (except in explicit

dynamic analysis).
The 8-node and 9-node elements can be employed for analysis

of thick structures and solids, such as (see Fig. 2.2-7):
Thick cylinders (axisymmetric idealization)

Dams (plane strain idealization)
Membrane sheets (plane stress idealization)

Chapter 2: Elements
Turbine blades (generalized plane strain idealization)
and for thin structures, such as (see Fig. 2.2-8):
Plates and shells (axisymmetric idealization)

Long plates (plane strain idealization)
Beams (plane stress idealization)
The 8- and 9-node elements are usually most effective if the

element is rectangular (non-distorted).
The 4-node rectangular and 3-node triangular elements should

only be used in analyses when bending effects are not significant.
If the 4-node rectangular elements are used when bending effects
are significant, consider the use of the incompatible modes element
(see above).
When the structure to be modeled is axisymmetric and has a

dimension which is very small compared with the others, e.g., thin
plates and shells, the use of axisymmetric shell elements is more
effective (see Section 2.5).
Geometrically nonlinear incompatible modes elements with

large aspect ratio should not be used, because spurious modes may
be present in the finite element solution.

Axis of revolution
l l l
p
l l
CL l l l
a) Thick cylinder (axisymmetric idealization)
l
l
l
l l
l
1 unit
b) Long dam (plane strain idealization)

l
l
l
l
l
c) Sheet in membrane action (plane stress idealization)
Figure 2.2-7: Use of 2-D solid element for

thick structures and solids

Chapter 2: Elements
p
Mz p
My
Z
Np
Y Fp
X
d) Turbine blade (generalized plane strain idealization)

Figure 2.2-7: (continued)
1 radian 1 radian
l
l l
l l
l l
l l
l
l
l l
l l l
L
t
Circular plate Spherical shell
a) Axisymmetric idealizations

P
l l l
l l l l l
l l l 1 unit l l t
l l l
L

b) Plane strain idealization (long plate) c) Plane stress idealization (cantilever)
Figure 2.2-8: Use of 2-D solid element for thin structures

2.3: Three-dimensional solid elements
2.3 Three-dimensional solid elements

The three-dimensional (3-D) solid element is a variable 4- to

20-node or a 21- or 27-node isoparametric element applicable to
general 3-D analysis. Some typical 3-D solid elements are shown
in Fig. 2.3-1.
l l
l l l l l l l l
l
l l l l l l l
l
l l l
l l l l l l l
l l l l l l l
l l l l
l
l l l l l l l
l l l l l l l l l
l l l l l l l
l
l l l l l l l l l
(a) 20-, 21- & 27-node elements
l l l l
l l
l ll l
l l
l
l l l
(b) 8-node elements

l l l
l l
l l l l
l l l l
l l l l
l l l l l l
l l
(c) 4-, 10- and 11-node tetrahedral elements
Figure 2.3-1: Some 3-D solid elements
The 3-D solid element should be employed in analyses in which

the three-dimensional state of stress (or strain) is required or in
which special stress/strain conditions, such as those given in
Section 2.2, do not exist (see Figs. 2.3-2 and 2.3-3).

Chapter 2: Elements
l
l l
l l l
l
l l
l
l l l
l
l
l l
l l
l
(d) 6- and 15- node prisms
l l
l l
l
l
l
l
l l l
l
l l l
l l
l
l l l
l l
l
l
l
l l
l l
l l l l
l
l
l l
l
l l
(e) Elements for transition zones
l l l
l l
l l
l l
l l l l
l l l l l
l l
l l
l
(f) 5-, 13-, and 14-pyramid elements
(f) 5, 13-node and 14-node pyramid elements
Figure 2.3-1 (continued)
ref. KJB The elements usually used are isoparametric displacement-based

Section 5.3

finite elements, and the formulation of the elements used in

ADINA is described in ref. KJB, Ch. 5.
The basic finite element assumptions are (see Fig. 2.3-4):
for the coordinates

q q q
x = hi xi y = hi yi z = hi zi
i =1 i =1 i =1
E,A
P
(a) Physical problem considered
P Plane sections remain

l l plane. d is constant
L for the entire section.
d
(b) 1-D model
P
Plane sections do
B not remain plane.
A dB dA dB
L
dA
(c) 3-D model
Figure 2.3-2: Illustration of use of 3-D elements
and for the displacements

Chapter 2: Elements
q q q
u = hi ui v = hi vi w = hi wi
i =1 i =1 i =1
where
hi (r, s, t) = interpolation function corresponding to node i
r, s, t = isoparametric coordinates
q = number of element nodes, 4 q 27
xi, yi, zi = nodal point coordinates
ui, vi, wi = nodal point displacements
No axial displacement
should be possible Line loads One quarter of
(uniform) the structure
is modeled
l
l l
l
l
l l
l
l l
l
l
Y Z
X
Z 2-D model
Y
a) Plane strain 2-D analysis applicable
One quarter of the

structure is modeled
l
l l
l l l l
l l l
l l l
l
l l
l
l l
l l
Fixed end
Z 3-D model
Y
b) Three-dimensional model is used
Figure 2.3-3: Examples of 3-D solid modeling versus

2-D solid modeling

In addition to the displacement-based elements, special mixed-

interpolated elements are also available, in which the displacements
and pressure are interpolated. These elements are effective and
should be preferred in the analysis of incompressible media and
inelastic materials (specifically for materials in which Poisson's
ratio is close to 0.5, for rubber-like materials and for elasto-plastic
materials).
The mixed formulation is the default setting for the materials in

which either the entire response is incompressible (for example, the
rubber material models), or in which the inelastic response is
incompressible (for example, the plastic and creep material
models).
Table 2.3-1 shows the default number of pressure degrees of

freedom for each 3-D element type. For more details on the mixed
interpolation of pressure and displacement degrees of freedom for
3-D solids, see Section 4.4.3, p. 276, and Tables 4.6 and 4.7,
pp.292-295 in ref. KJB.
The default number of pressure degrees of freedom shown in

Table 2.3-1 is optimal for most applications. However, it can be
modified if needed. In some problems where rubber is subjected to
severe compression, numerical instability can be avoided by using
8 pressure degrees of freedom for the 20, 21 and 27-node bricks.
For brick elements, the following choices for the number of

pressure degrees of freedom are valid; 1 for a constant pressure, 4
for additional linear pressure terms, 7 for additional bilinear
pressure terms (involving rs, rt, st), 8 for an additional tri-linear
pressure term (involving rst), and 20 for additional quadratic
pressure terms (involving r2, s2, t2). For tetrahedral elements, the
following choices for the number of pressure degrees of freedom
are valid; 1 for a constant pressure, 4 for additional linear pressure
terms, 10 for additional quadratic terms, 11 for an additional
bubble term at the center of the element and 15 for additional
cubic terms.
In general, using too many pressure degrees of freedom leads to

volumetric locking, while using too few pressure degrees of
freedom leads to spurious zero-energy modes. See Sections 4.4.2

Chapter 2: Elements
and 4.5 in ref. KJB for more details.
Table 2.3-1: Mixed u/p formulations available for 3-D solid elements
Number of default
3-D solid element
pressure DOFs
4-node tetrahedron 1*
10-node tetrahedron (10/4) 4
11-node tetrahedron (11/4) 4
6-node prism 1
15-node prism 4
21-node prism 4
8-node brick (8/1) 1
21-node brick 4
5-node pyramid 1
13-node pyramid 4
14-node pyramid 4
*
To use this element, the user must use the displacement-
based element with 1-point integration.

Structure and loading are axisymmetric

Axisymmetric elements
l l l
l l
l l l
l l 1 radian
Z l l l
l l
l l l of structure
is modeled
X
Y
Z Axisymmetric model
Y
c) Axisymmetric analysis applicable
Structure axisymmetric,
loading non-axisymmetric 3-D elements
l
l l l l l
l l
l l
l l l
l
l l
l
l l
l l
One quarter
Z
of structure
Y is modeled
d) Three-dimensional model is used
In addition to the displacement-based and mixed-interpolated

elements, ADINA also includes the possibility of including
incompatible modes in the formulation of the 8-node element. The
addition of the incompatible modes increases the flexibility of the
element, especially in bending situations. This element is
analogous to the 4-node 2-D solid element with incompatible
modes discussed in Section 2.2.1, see the comments in that section
for theoretical considerations.
The incompatible modes feature cannot be used in conjunction
with the mixed-interpolation formulation.

Chapter 2: Elements
t
Z 3
l
W4 l
10
11
l
l 26 2
4 V4 19 l
l l 23
12 l l
l l 9
24 21 l 1 18
l l l
U4 l 20 22
7 l
l
l l 14
15 25 l 17 l
8l l 6
27 s
l
l 13
Z8 r 16
l
5
X8 Y
Y8
X
Figure 2.3-4: Conventions used for the nodal coordinates
and displacements of the 3-D solid element
The incompatible modes feature is only available for the 8-node

element, and elements degenerated from the 8-node element,
except for the 4-node tetrahedral element.
The elements can be used with 4 to 20 or with 21 or 27 nodes

(tetrahedra, pyramids or prisms are derived from the degeneration
of the 4 to 20-node rectangular elements, see Example 5.16 on pp.
366-367, ref. KJB). The interpolation functions for q 20 are
shown in Fig. 5.5, ref. KJB, p. 345.
Degenerated elements such as prisms, pyramids or tetrahedra

are formed in ADINA by assigning the same global node to the
local element nodes located along the same side or on the same
face.
The triangular prism as a degenerated 20-node element (see Fig.
2.3-1(d)) can be used in ADINA in three different ways:
As a 15-node triangular prism with the strain singularity

1
. This is an extension of the techniques described on pp.
r

368-376 of ref. KJB (see also Chapter 10);

1
As a 20-node triangular prism with the strain singularity
r
(see also Chapter 10);
As a spatially isotropic triangular prism where corrections

are applied to the interpolation functions of the collapsed 20-
node element.
The 10-node tetrahedron (see Fig. 2.3-1(c)) is obtained by

collapsing nodes and sides of rectangular elements. A spatially
isotropic 10-node tetrahedron is available in ADINA. Similarly, an
11-node spatially isotropic tetrahedron is available in ADINA.
The 13-node pyramid and the 14-node pyramid (see Fig. 2.3-
1(f)) are degenerate solid elements from 20-node and 21-node
bricks, respectively. They can be used as spatially isotropic
elements.
The 4-node tetrahedron (see Fig. 2.3-1(c)) is obtained by
collapsing nodes and sides of the 8-node rectangular element. This
element exhibits constant strain conditions.
The elements in Fig. 2.3-1(e) are not spatially isotropic and
should usually only be employed in transition regions.
Linear and nonlinear fracture mechanics analysis of stationary

cracks can be performed with three-dimensional elements (see
Chapter 10).
2.3.2 Material models and nonlinear formulations
The 3-D elements can be used with the following material

models: elastic-isotropic, elastic-orthotropic, nonlinear-elastic,
plastic-bilinear, plastic-multilinear, Drucker-Prager, Mroz-
bilinear, plastic-orthotropic, thermo-isotropic, thermo-
orthotropic, thermo-plastic, piezoelectric, creep, plastic-creep,
multilinear-plastic-creep, concrete, curve-description, Ogden,
Mooney-Rivlin, Arruda-Boyce, hyper-foam, Sussman-Bathe,
creep-variable, plastic-creep-variable, multilinear-plastic-
creep-variable, Gurson-plastic, Cam-clay, Mohr-Coulomb,
viscoelastic, irradiation creep, gasket, shape-memory alloy,
plastic-cyclic, user-coded.

Chapter 2: Elements
The mixed interpolation formulation should be used (and is the

default) whenever using the plastic-bilinear, plastic-multilinear,
Mroz- bilinear, plastic-orthotropic, thermo-plastic, creep, plastic-
creep, multilinear-plastic-creep, Ogden, Mooney-Rivlin, Arruda-
Boyce, Sussman-Bathe, creep-variable, plastic-creep-variable,
multilinear-plastic-creep-variable, plastic-cyclic or viscoelastic
material models.
In addition the mixed interpolation formulation should be
used (but is not the default) when using the elastic-isotropic
material and the Poissons ratio is close to 0.5.
The 3-D elements can be used with a small displacement/small

strain, large displacement/small strain or a large displacement/
large strain formulation.
The small displacement/small strain and large
displacement/small strain formulations can be used with any
material model, except for the Ogden, Mooney-Rivlin, Arruda-
Boyce, hyper-foam and Sussman-Bathe material models. The use
of a linear material with the small displacement/
small strain formulation corresponds to a linear formulation, and
the use of a nonlinear material with the small displacement/small
strain formulation corresponds to a materially-nonlinear-only
formulation. The program uses the TL (total Lagrangian)
formulation when you choose a large displacement/small strain
formulation.
The large displacement/large strain formulations can be used
with the plastic-bilinear, plastic-multilinear, Mroz-bilinear, plastic-
orthotropic, thermo-plastic, creep, plastic-creep, multilinear-plastic-
creep, creep-variable, plastic-creep-variable, multilinear-plastic-
creep-variable, viscoelastic, plastic-cyclic and user-coded material
models. The program uses either a ULH or ULJ formulation when
you choose a large displacement/large strain formulation with these
material models.
A large displacement/large strain formulation is used with the
Ogden, Mooney-Rivlin, Arruda-Boyce, hyper-foam and Sussman-
Bathe material models. The program uses a TL (total Lagrangian)
formulation in this case.
ref. KJB The basic continuum mechanics formulations are described in

Sections 6.2 ref. KJB, pp. 497-568. The finite element discretization is
and 6.3.5
summarized in Table 6.6, p. 555, ref. KJB.

ref. KJB Note that all these formulations can be used in one finite
Section 6.8.1 element mesh. If the elements are initially compatible, then they
will remain compatible throughout the analysis.
Hexahedral (brick) elements, including collapsed hexahedral

elements
ref. KJB For the calculation of element matrices, Gauss numerical

Sections 5.5.3, integration is used. The same integration order (RSINT) is always
5.5.4 and 5.5.5 assigned to the r- and s-directions, and can be from 2 to 6. The
integration order in the t-direction (TINT) can, however, be
assigned independently, and can also be from 2 to 6. The default
Gauss integration orders are 222 for the 8-node (cube or prism)
elements and 333 otherwise, except for tetrahedra.
Except for the 4-node, 10-node and 11-node tetrahedra, the

convention for the integration point numbering used in the stress
output is as follows: The first integration point is the point with the
most negative location in r, s and t. The next integration points are
located by increasing t (and label INT) successively up to its
maximum positive value, then increasing s (and label INS) one
position towards positive and varying t again from its maximum
negative to its maximum positive values, and so on (Fig. 2.3-5).

Chapter 2: Elements
INR=1,INS=1,INT=3
s

INR=1,INS=1,INT=1
r
INR=2,INS=1,INT=1
Integration order in r-s plane=2

Integration order in t-direction=3
a) All elements except tetrahedra
Figure 2.3-5: Example of integration point labeling for

3-D solid elements
4, 10 and 11-node tetrahedral elements
Tetrahedral elements are spatially isotropic with respect to

integration point locations and interpolation functions (see Figure
2.3-5(b)). For the 4-node tetrahedral element, 1-point Gauss
integration is used. For the 10-node and 11-node tetrahedral
elements, TETINT Gauss integration points are used, where
TETINT can be 1, 4, 5, 17.
TETINT=DEFAULT is also allowed, and is the default. When
TETINT=DEFAULT, the number of tetrahedral integration points
is set based on the maximum number of nodes for any element in
the element group:
Max number of nodes Max number of nodes in TETINT

in any brick element any tetrahedral element
8 4 1
20 10 5
27 11 17

t t
5
1 2 1 4
s s
1-point r r 5-point
3
16 17
4 15 5
4 13
5 5
4 s r
8 6 12 14 9 10
3 2 2 3
7 2 11
t t
17-point (unfolded) (Point 1 at centroid not shown)

b) Tetrahedral elements
General Comments
Note that in geometrically nonlinear analysis, the spatial

positions of the Gauss integration points change continuously as
the element undergoes deformations, but throughout the response
the same material particles are at the integration points.
The order of numerical integration in large displacement elastic

ref. KJB
Section 6.8.4 analysis is usually best chosen to be equal to the order appropriate
in linear elastic analysis. Hence the default values are usually
appropriate. However, in certain inelastic analyses, a higher order
should be used (see Fig. 6.25, p. 638, ref. KJB).

Chapter 2: Elements
2.3.4 Mass matrices
The consistent mass matrix is always calculated using 333

Gauss integration except for the tetrahedral 4-node, 10-node and
11-node elements which use a 17-point Gauss integration.
The lumped mass matrix of an element is formed by dividing

the elements mass M equally among each of its n nodal points.
Hence the mass assigned to each node is M / n. No special
distributory concepts are employed to distinguish between corner
and midside nodes, or to account for element distortion.
Note that n is the number of distinct non-repeated nodes in the

element. Hence, when an element side or face is collapsed to a
single node, the total mass of the element is divided among the
unique nodes in the element.
You can request that ADINA either print or save stresses or

forces. However ADINA cannot output both stresses and forces in
the same run.
Stresses
Each element outputs the following information to the porthole file,

based on the material model. This information is accessible in the
AUI using the given variable names.
Notice that for the orthotropic and piezoelectric material models,

you can request that the stresses and strains be saved in the material
coordinate system.
Elastic-isotropic, elastic-orthotropic (results saved in global

system): STRESS(XYZ), STRAIN(XYZ)

STRAIN(XYZ), THERMAL_STRAIN,
ELEMENT_TEMPERATURE

Orthotropic (results saved in material system): STRESS(ABC),

STRAIN(ABC)
Thermo-isotropic, thermo-orthotropic (results saved in global

system): STRESS(XYZ), STRAIN(XYZ),
Piezoelectric (results saved in global system): STRESS(XYZ),

STRAIN(XYZ), ELECTRIC_FIELD-X, ELECTRIC_FIELD-Y,
ELECTRIC_FIELD-Z, ELECTRIC_DISPLACEMENT_FIELD-X,
ELECTRIC_DISPLACEMENT_FIELD-Y,
ELECTRIC_DISPLACEMENT_FIELD-Z.
Piezoelectric (results saved in material system): STRESS(ABC),

STRAIN(ABC), ELECTRIC_FIELD-A, ELECTRIC_FIELD-B,
ELECTRIC_FIELD-C, ELECTRIC_DISPLACEMENT_FIELD-A,
ELECTRIC_DISPLACEMENT_FIELD-B,
ELECTRIC_DISPLACEMENT_FIELD-C.
Curve description: CRACK_FLAG, STRESS(XYZ),

STRAIN(XYZ), FE_SIGMA-P1, FE_SIGMA-P2,
FE_SIGMA-P3, GRAVITY_IN-SITU_PRESSURE,
VOLUMETRIC_STRAIN, FE_SIGMA-P1_DIRECTION-X,
FE_SIGMA-P1_DIRECTION-Y,
FE_SIGMA-P1_DIRECTION-Z,
FE_SIGMA-P2_DIRECTION-X,
FE_SIGMA-P3_DIRECTION-Z
Concrete: CRACK_FLAG, STRESS(XYZ), STRAIN(XYZ),

FE_SIGMA-P1, FE_SIGMA-P2, FE_SIGMA-P3,
FAILURE_STRESS, ELEMENT_TEMPERATURE,FE_SIGMA-
P1_DIRECTION-X,

Chapter 2: Elements
FE_SIGMA-P3_DIRECTION-Z, THERMAL_STRAIN(XYZ)
Small strains: Plastic-bilinear, plastic-multilinear, Mroz-bilinear:

YIELD_STRESS,
Large strains: Plastic-bilinear, plastic-multilinear, Mroz-bilinear:

PLASTIC_FLAG, STRESS(XYZ),
SMA: EFFECTIVE_STRESS, STRESS-XX, STRESS-XY,

STRESS-XZ, STRESS-YY, STRESS-YZ, STRESS-ZZ,
DETWINNED_MARTENSITE_FRACTION, STRAIN-XX,
STRAIN-XY, STRAIN-XZ, STRAIN-YY, STRAIN-YZ,
STRAIN-ZZ, THERMAL_STRAIN-XX, THERMAL_STRAIN-
YY, THERMAL_STRAIN-ZZ, TRANSFORMATION_STRAIN-
XX, TRANSFORMATION_STRAIN-XY,
TRANSFORMATION_STRAIN-XZ,
TRANSFORMATION_STRAIN-YY,
TRANSFORMATION_STRAIN-YZ,
TRANSFORMATION_STRAIN-ZZ,
Small strains: Anand Model

VISCOPLASTIC_STRAIN (XYZ),
Large strains: Anand Model


Plastic-orthotropic (results saved in global system):

YIELD_STRESS,
Plastic-orthotropic (results saved in material system):

PLASTIC_FLAG, STRESS(ABC), STRAIN(ABC),
PLASTIC_STRAIN(ABC), HILL_EFFECTIVE_STRESS,
YIELD_STRESS,
Small strains: Drucker-Prager: PLASTIC_FLAG,

PLASTIC_FLAG-2, STRESS(XYZ), STRAIN(XYZ),
PLASTIC_STRAIN(XYZ), CAP_LOCATION,
YIELD_FUNCTION
Large strains: Drucker-Prager: PLASTIC_FLAG,

PLASTIC_FLAG-2, STRESS(XYZ),
DEFORMATION_GRADIENT(XYZ), CAP_LOCATION,
YIELD_FUNCTION
Small strains: Thermo-plastic, creep, plastic-creep, multilinear-

Large strains: Thermo-plastic, creep, plastic-creep, multilinear-


Chapter 2: Elements
Mooney-Rivlin , Ogden, Arruda-Boyce, hyper-foam (strains

saved): STRESS(XYZ), STRAIN(XYZ)
Mooney-Rivlin , Ogden, Arruda-Boyce, hyper-foam (deformation

gradients saved): STRESS(XYZ),
DEFORMATION_GRADIENT(XYZ)
Small strains: User-supplied: STRESS(XYZ), STRAIN(XYZ),

USER_VARIABLE_I
Large strains: User-supplied: STRESS(XYZ),

DEFORMATION_GRADIENT(XYZ), USER_VARIABLE_I
Thermo-orthotropic (results saved in material system):

THERMAL_STRAIN(ABC), FE_EFFECTIVE_STRESS,
ELEMENT_TEMPERATURE
Small strains: Gurson plastic: PLASTIC_FLAG,

YIELD_STRESS,
Large strains: Gurson plastic: PLASTIC_FLAG,

Small strains: Cam-clay: PLASTIC_FLAG, STRESS(XYZ),


Large strains: Cam-clay: PLASTIC_FLAG, STRESS(XYZ),

Mohr-Coulomb: PLASTIC_FLAG, STRESS(XYZ),

YIELD_FUNCTION
Small strains: viscoelastic: STRESS(XYZ), STRAIN(XYZ),

Large strains: viscoelastic: STRESS(XYZ),

Small strains: creep-irradiation: STRESS(XYZ),

Large strains: creep-irradiation: STRESS(XYZ),

In the above lists,
STRESS(XYZ) = STRESS-XX, STRESS-YY, STRESS-

ZZ, STRESS-XY, STRESS-XZ, STRESS-YZ
STRESS(ABC) = STRESS-AA, STRESS-BB, STRESS-

CC, STRESS-AB, STRESS-AC, STRESS-BC

Chapter 2: Elements
with similar definitions for the other abbreviations used above. But
the variable DEFORMATION_GRADIENT(XYZ) is interpreted as
follows:
DEFORMATION_GRADIENT(XYZ) =
DEFORMATION_GRADIENT-XX,
DEFORMATION_GRADIENT-XY,
DEFORMATION_GRADIENT-XZ,
DEFORMATION_GRADIENT-YX,
DEFORMATION_GRADIENT-YY,
DEFORMATION_GRADIENT-YZ,
DEFORMATION_GRADIENT-ZX,
DEFORMATION_GRADIENT-ZY,
DEFORMATION_GRADIENT-ZZ,
Also note that you can request stretches instead of deformation

gradients, and in this case STRETCH(XYZ) replaces
DEFORMATION_GRADIENT(XYZ) in the above lists.
See Section 13.1.1 for the definitions of those variables that are not
self-explanatory.
You can also request that strain energy densities be output along
with the stresses.
Variable categories
The variables are grouped into the following categories.
Stress (no saving, basic saving, all saving), default all

saving
Strain (no saving, all saving), default no saving
Inelastic (no saving, basic saving, all saving), default basic

saving
Thermal (no saving, basic saving, all saving), default basic
saving
Energy (no saving, all saving), default no saving

Electromagnetic (no saving, all saving), default no saving

User-coded variables (no saving, all saving), default no
saving
Misc (no saving, all saving), default no saving
Notice that some categories allow "basic" saving, namely only

those results that are commonly used in post-processing are saved.
The intent of the variable category feature is to decrease the

number of results written to the porthole file. In particular notice
that not all results are saved to the porthole file by default.
The RESULTS-ELEMENT command is used to control the

saving of each variable category. To set the saving for a given 3D
solid element group in the model, use the command RESULTS-
ELEMENT ... GROUP=(group number). To set the saving for all
remaining 3D solid element groups in the model, use the command
RESULTS-ELEMENT ... GROUP=0.
The category saving feature can be turned off using the

command PORTHOLE ... RESULTS-ELEMENT=NO. When
PORTHOLE ... RESULTS-ELEMENT=NO, all results are saved at
each element integration point. The default is PORTHOLE ...
RESULTS-ELEMENT=YES, so that the category saving feature is
used by default.
In order to save all results, either set each category in

RESULTS-ELEMENT to "all", or use the command PORTHOLE
... RESULTS-ELEMENT=NO.
ref. KJB In ADINA, the stresses are calculated using the strains at the
Section 4.2.1 point of interest. The stresses are not smoothed. The AUI can be
employed to calculate smoothed stresses from the results output by
ADINA.
Locations where results are saved
The results are saved either at the element integration points or at

the element corner nodes. When the results are saved at the

Chapter 2: Elements
integration points, the stresses are calculated using the strains at the
point of interest. Hence, they are not spatially extrapolated or
smoothed. The AUI can be employed to calculate smoothed
stresses from the results output by ADINA.
When the results are saved at the element corner nodes, the
number of points per element at which results are saved is
significantly reduced. For example, if results are saved at the corner
nodes of a 27-node brick element, there are 8 points per element, as
opposed to 27 integration points; and if results are saved at the
corner nodes of an 11-node tetrahedral element, there are 4 points
per element, as opposed to 17 integration points.
The RESULTS-ELEMENT command is used to control where

the results are saved. The default is to save the results at the
element integration points.
Nodal forces
The nodal forces that correspond to the element stresses can also be
requested in ADINA. This force vector is calculated in a linear
analysis using
F = BT dV
V
where B is the element strain-displacement matrix and is the

stress vector. The integration is performed over the volume of the
element; see Example 5.11 in ref. KJB, pp. 358-359.
The nodal forces are accessible in the AUI using the variable names
NODAL_FORCE-X, NODAL_FORCE-Y, NODAL_FORCE-Z.
The same relation is used for the element force calculation in a

nonlinear analysis, except that updated quantities are used in the
integration.
The 27-node element is the most accurate among all available

elements. However, the use of this element can be costly.

The 20-node element is usually the most effective (except in

wave propagation analysis using the central difference method and
a lumped mass idealization, see Chapter 7).
The 20-node element can be employed for analysis of thick

structures and solids; some examples are given in Fig. 2.3-6.
The 20-node element is usually most effective if the element is

rectangular (undistorted).
The 11-node tetrahedral element should be used when

tetrahedral meshing is used and a mixed interpolated formulation is
required (as in incompressible analysis).
Dam
Vessel
Tunnel
Figure 2.3-6: Some structures for which 3-D solid

elements can be used
The 8-node brick element and 4-node tetrahedral element

should only be used in analyses when bending effects are not
significant. If the 8-node brick element must be used when
bending effects are significant, consider the use of the incompatible
modes element.

Chapter 2: Elements
When the structure to be modeled has a dimension which is

extremely small compared with the others, e.g., thin plates and
shells, the use of the 3-D solid element usually results in too stiff a
model and a poor conditioning of the stiffness matrix. In this case,
the use of the shell or the plate/shell elements (see Sections 2.7 and
2.6) is more effective.
2.4 Beam elements

The beam element is a 2-node Hermitian beam with a constant
cross-section. The element is initially straight.
The beam element can be employed in the following analysis

conditions:
Linear analysis, in which case the displacements, rotations

and strains are infinitesimally small, and the material is linear
elastic.
Materially nonlinear only analysis, in which case the

displacements, rotations and strains are infinitesimally small,
but the material is nonlinear.
Large displacement/large rotation analysis, in which case the

displacements and rotations can be large, but the strains are
small. The material can either be linear or nonlinear.
The beam element can be used in statics, implicit dynamics,

explicit dynamics and frequency analysis.
The material behavior of the beam can be described using either

a cross-section shape and a material model, or a moment-curvature
input.
The beam element can optionally include warping degrees of

freedom. This option is suitable when modeling thin-walled cross-
sections, especially when the angle of twist per unit length is not

2.4: Beam elements
constant along the beam.Throughout this section, the beam element

without warping degrees of freedom is referred to as the standard
beam element, and the beam element with warping degrees of
freedom is referred to as the warping beam element.
The available beam capabilities are summarized in Table 2.4.1.
The beam element can optionally include a bolt feature. The

bolt feature is fully described in Section 11.17.
Throughout this section, the element formulations of the current

ADINA system are described. It is, however, possible to choose the
element formulations used in ADINA version 8.7, using the
command KINEMATICS BEAM-ALGORITHM=V87. (However,
if the elasto-plastic material model is chosen, then the current beam
element kinematics and material algorithms are always used.)
The results obtained will vary depending upon whether the
current formulations or the 8.7 formulations are used.

Chapter 2: Elements
Table 2.4-1: Table of beam capabilities
Formulation Cross-section Cross-sections Cross- Cross-

and material PROPERTIES RECTANGULAR, section sections
model PIPE BOX U, I, L
1 2 3
Linear > > > >4
elastic
Large >1 >2 >3 >4
displacement
elastic
MNO plastic --- >2 >3 >4
Large --- >2 >3 >4
displacement
plastic
MNO >2 --- --- ---
moment-
curvature
Large >2 --- --- ---
displacement
moment-
curvature
1. Either the standard or warping beam can be used. However, if the properties
correspond to an open thin-walled section, the warping beam is recommended for
general use.
2. Only the standard beam can be used.
3. Either the standard or warping beam can be used.
4. Either the standard or warping beam can be used, however the warping beam
element is recommended for general use.
2.4.1 Beam geometry
Fig 2.4-1 shows the beam element along with its local
coordinate system (r,s,t). The r direction always lies along the
neutral line of the beam. The orientations of the s and t directions
are defined using either an auxiliary node K (as in Fig. 2.4-1(a)) or
an orientation vector (XO,YO,ZO) (as in Fig. 2.4-1(b)). If node K

2.4: Beam elements
is specified in the input, it is always used as the auxiliary node. If

node K is not specified, then the orientation vector is used.
When the orientation vector is used, the components
(XO,YO,ZO) are interpreted either in the skew system of local
node 1 (the default), or in the global system.
Notice that, for the Hermitian beam element, (r,s,t) are not
isoparametric coordinates, rather (r,s,t) have the same units as the
global coordinates.
Fig 2.4-1 also shows the degrees of freedom at the local nodes.
These degrees of freedom are defined in the local coordinate
system. The degree of freedom is used only for the warping
beam element.
The forces/moments in the beam element are shown in Fig. 2.4-

2. These forces/moments are also defined in the element local
coordinate system. The bimoment forces are used only for the
warping beam element.
The s and t directions give the orientation of the beam element

cross-section. Care must be used in defining the s and t directions
so that the beam element cross-section has the desired orientation.

Chapter 2: Elements
K qs
l
a
qr
v u
s
l
r 2 w
z
l qt
1 t
x y
The s-direction lies in the plane defined by

nodes 1, 2, K, and the t-direction is perpendicular
to the r-s plane.
a : warping DOF
(a) Geometry definition, using auxiliary node K
Orientation vector (XO,YO,ZO) qs

a
qr
v
u
s
l
z r 2 w
l qt
1 t
x y
The s-direction lies in the plane defined by nodes 1, 2

and the orientation vector. The t-direction is perpendicular
to the r-s plane.
(b) Geometry definition, using orientation vector
Figure 2.4-1: Degrees of freedom and local axes for beam element

2.4: Beam elements
s
S11
S 14
S5 S8 S 10
S7
l
S2
S1 2 S9
S4 l S12
S13
1 S3
S6 t
z
y
x
S1 = r-direction force at node 1 (axial force, positive in compression)

S2 = s-direction force at node 1 (shear force)
S3 = t-direction force at node 1 (shear force)
S4 = r-direction moment at node 1 (torsion)
S5 = s-direction moment at node 1 (bending moment)
S6 = t-direction moment at node 1 (bending moment)
S7 = r-direction force at node 2 (axial force, positive in tension)

S8 = s-direction force at node 2 (shear force)
S9 = t-direction force at node 2 (shear force)
S10 = r-direction moment at node 2 (torsion)
S11 = s-direction moment at node 2 (bending moment)
S12 = t-direction moment at node 2 (bending moment)
S13 = bimoment at node 1

S14 = bimoment at node 2
Figure 2.4-2: Element end forces/moments
The beam element cross-section with its coordinate system is

shown in Fig. 2.4-3. The cross-section coordinate system has its
origin on the line connecting the beam end-nodes. (Note that the
cross-section coordinate system uses the same letters (s,t) as the
beam element coordinate system, however this should not cause
confusion.)

Chapter 2: Elements
t
Node 1
s
C
s
Node 2
r
(a) Isometric view
^t
t
ct P
s^
.
r cs s
C
C: Centroid of the cross-section

P: Shear center of the cross-section
(b) s-t plane view, showing location of shear center
Figure 2.4-3: Beam cross-section coordinate system
Fig 2.4-3(b) shows the cross-section with its centroid and shear
center. The centroid of the cross-section is assumed to be at the
origin of the ( s, t ) system. Also notice that a coordinate system

2.4: Beam elements
( s, t) is introduced, with origin at the shear center. The ( s, t)

system is always parallel to the ( s, t ) system, and the origin of the
( s, t) is located at ( s, t ) = (cs , ct ) .
The shear center is also the center of twist.
2.4.2 Beam cross-section geometric properties
The following cross-sectional geometric properties are listed here

for reference.
cs , ct : s and t coordinates of the shear center with respect to the

centroid (terms CSOFFSET, CTOFFSET)
Assh , Atsh : effective shear cross-section areas in the s and t

directions (terms SAREA, TAREA) . Note, these terms are actual
areas, with units of length squared.
= ( s, t ) : warping function, see description below.
Governing equations for location of beam centroid:
s dA = t dA =0 (2.4-1)
Normalization equation for warping function:
dA =0 (2.4-2)
Governing equations for location of beam shear center:
s dA = t dA =0 (2.4-3)
A : Cross-sectional area (term AREA)

A = dA (2.4-4)
I ss : Inertia for bending about the s-axis (term SINERTIA):

I ss = t 2 dA (2.4-5)

Chapter 2: Elements
I tt : Inertia for bending about the t-axis (term TINERTIA):

I tt = s 2 dA (2.4-6)
I st : Product of inertia (term STINERTIA):

I st = st dA (2.4-7)
J : Saint-Venant torsional constant (term RINERTIA):

(
J = (, s t ) + (,t + s ) dA
2 2
) (2.4-8)
(do not confuse the Saint-Venant torsional constant with the polar
moment of inertia)
I : Warping constant (term WINERTIA):

I = 2 dA (2.4-9)
(this term is also referred to as C in handbooks)
A15 : Wagner effect constant (term RRINERTIA)

A15 = ( s 2 + t 2 ) dA (2.4-10)
A25 : Wagner effect constant (term SRINERTIA)

A25 = s ( s 2 + t 2 ) dA (2.4-11)
A35 : Wagner effect constant (term TRINERTIA)

A35 = t ( s 2 + t 2 ) dA (2.4-12)
A45 : Wagner effect constant (term WRINERTIA)

A45 = ( s 2 + t 2 ) dA (2.4-13)
A55 : Wagner effect constant (term DRINERTIA)

2.4: Beam elements
A55 = ( s 2 + t 2 ) dA
2
(2.4-14)
The term names are the names used in the CROSS-SECTION

PROPERTIES command. Also note that the terminology
() ()
(), s = , (),t = is used.
s t
2.4.3 Beam formulation
The formulation of the beam is a generalization of the Euler-

Bernoulli beam formulation.
The cross-section of the beam is assumed to be rigid in its own

plane so no distortion of the cross-section is considered. However,
the cross-section can warp out of its plane.
The two nodes and the origin of the local coordinate system
( s, t ) are located at the centroid of the beam cross-section. The
local coordinate system of the element may or may not be
coincident with the principal planes of inertia of the cross-section.
A brief summary of the beam formulation derivation is now given.
2.4.3.1 Beam displacements
The beam displacements in the beam local coordinate system

are u (axial r-direction displacement), v (transverse s-direction
displacement) and w (transverse t-direction displacement).
The displacements at an arbitrary point on the beam cross-

section can be written in terms of the displacements, rotations and
warping measured at the beam neutral axis:
u = uo st + t s +
v = vn t r (2.4-15)
w = wn + s r
In this expression, uo is the average longitudinal displacement of

Chapter 2: Elements
the cross-section, vn , wn are the displacements of the beam neutral

axis (corresponding to point C in Fig. 2.4-3), r , s , t are the
rotations of the section about the r, s, t axes, = ( s, t ) is the
warping function, and is the warping function multiplier. The
warping function and warping function multiplier will be discussed
in greater detail below.
Equation (2.4-15) is seen to include the assumptions that the
cross-section translates rigidly, rotates rigidly and warps (due to the
warping function). In all cases, the translations and rotations, with
respect to the beam local coordinate system, are assumed to be
small. (Later on, we will allow for large displacements by allowing
the local coordinate system to translate and rotate.)
Under the assumption of unit warping, the axial displacement

due to torsion is given by
u = ( s, t ) r ,r (2.4-16)
()
where (),r = . Hence = r ,r and (2.4-15) can be written
r
u = uo st + t s + r ,r
v = vn t r (2.4-17)
w = wn + s r
The dimension of is seen to be [L-1].
It will be convenient to express the transverse displacements in

terms of the transverse displacements at the shear center. The
transverse displacements at the shear center (corresponding to point
P in Fig. 2.4-3) are
v p = vn ct r
(2.4-18)
wp = wn + cs r
Therefore

2.4: Beam elements
v = v p t r
(2.4-19)
w = wp + s r
We now introduce the assumption that the cross-section remains

normal to the shear center axis:
s = wp , r
(2.4-20)
t = v p ,r
This assumption corresponds to the use of Euler-Bernoulli beam

theory, however here the shear center axis is used instead of the
more commonly-used neutral axis. This choice will be explained
below.
2.4.3.2 Displacement derivatives and strains
The derivatives of the displacements with respect to the r, s, t

directions are
u,r = uo ,r st , r + t s ,r + r ,rr
u, s = t + , s r , r (2.4-21)
u,t = s + ,t r ,r
v, r = t t r ,r
v, s = 0 (2.4-22)
v,t = r
w, r = s + s r ,r
w, s = r (2.4-23)
w,t = 0
in which (2.4-20) has been used.

Considering only engineering strains for now,

Chapter 2: Elements
err = u, r = uo , r st ,r + t s , r + r ,rr
rs = u,s + v,r = (, s t ) r ,r (2.4-24)
rt = u,t + w,r = (,t + s ) r ,r
Equations (2.4-24) show that shear strains arise only due to torsion.
This is the reason that we have used the shear center instead of the
neutral axis in the above developments.
2.4.3.3 Stresses
It is assumed that the only nonzero stresses are rr , rs , rt . The

stresses are computed from the strains using the material law, either
elastic or elasto-plastic. For a linear elastic material
rr = Eerr , rs = G rs , rt = G rt (2.4-25)
where E , G are the Young's and shear moduli, respectively.
2.4.3.4 Principle of virtual work
The virtual work can be written
W = ( rr err + rs rs + rt rt ) dV (2.4-26)
The force vector and stiffness matrix can be derived from this
expression, once a beam interpolation has been chosen. In general,
numerical integration must be used, however when the material is
linear elastic, the integrals are evaluated analytically.
2.4.3.5 Torsional response
In order to gain additional insight into the torsional response, we

consider a beam loaded by an external axial torque. The
displacements of the beam are

2.4: Beam elements
u = r ,r
v = t r (2.4-27)
w = s r
the corresponding engineering strains are
err = r ,rr
rs = (, s t ) r ,r (2.4-28)
rt = (,t + s ) r ,r
and the principle of virtual work can be written
( err + rs rs + rt rt ) dV = T r
L
rr 0
(2.4-29)
where T is the externally applied torque at the ends of the beam.

The left-hand-side terms can be written as
rr err dV = rr r ,rr dV (2.4-30)
( rs rs + rt rt ) dV = rs (, s t ) + rt (,t + s ) r ,r dV
(2.4-31)
Define
M b = rr dA , Fb = ( rs, s + rt,t ) dA ,
M r = ( rs t + rt s ) dA
(2.4-32a,b,c)
where M b is the bimoment (some authors insert a minus sign), Fb

is the bishear and M r is the torque calculated from the shear
stresses. Then the principle of virtual work can be written

Chapter 2: Elements
( M + ( Fb + M r ) r , r ) dr = T r
L
b r , rr 0
(2.4-33)
Using the calculus of variations, the principle of virtual work can

be rewritten as
( )
L L
Fb + M r M b ,r r dr = T Fb + M r M b ,r r M b r , r 0
,r 0
(2.4-34)
and we see the following:
1) The bimoment is energy-conjugate to the warping multiplier

(recall = r ,r )
2) If the warping multiplier is free at an end, the bimoment is zero

at that end.
3) The torque is equilibrated by the quantity .
4) If the warping multiplier is free at both ends, the bimoment is

zero throughout and the external torque is equilibrated by the sum
of the bishear and the torque calculated from the shear stresses.
2.4.3.6 Strain energy for a linear elastic material
We now return to Section 2.4.3.4 and consider a linear elastic

material. The strain energy in the beam can be written
W = 12 E err2 dV + 12G ( 2
rs )
+ rt2 dV (2.4-35)
Substituting from (2.4-24) gives
W = 12 E ( uo ,r st , r + t s ,r + r ,rr ) dV
2
(2.4-36)
(
+ 12 G (, s t ) + (,t + s ) r2,r dV
2 2
)
This can be expanded to give

2.4: Beam elements
W = 12 E { u 2
o,r dV 2 uo , rt , r s dV + 2 uo , r s ,r t dV + 2 uo ,r r ,rr dV
+ t2,r s 2 dV 2 t ,r s ,r st dV 2 t ,r r , rr s dV
+ s2, r t 2 dV + 2 s ,r r ,rr t dV
+ r2, rr 2 dV }
(
+ 12 G r2,r (, s t ) + (,t + s ) dV
2 2
)
(2.4-37)
We now recognize that dV = dAdr , where dA represents

an integral over the beam cross-section. Also since all terms such
as uo ,r , s ,r , etc. are constant over the cross-section, (2.4-37) can
be written as
W = 12 E { dA u 2
o ,r dr 2 sdA uo , rt ,r dr + 2 tdA uo ,r s , r dr + 2 dA uo ,r r ,rr dr
+ s 2 dA t2,r dr 2 stdA t , r s ,r dr 2 s dA t , r r ,rr dr

+ t 2 dA s2,r dr + 2 t dA s , r r ,rr dr
+ 2 dA r2,rr dr }
(
+ 12 G (, s t ) + (,t + s ) dA r2,r dr
2 2
)
(2.4-38)
By definition of the neutral axis, s dA = t dA = 0 , and by
definition of the warping function, dA = s dA = t dA = 0 .
Therefore, in terms of the geometric properties given in Section
2.4.2, the strain energy of the beam can be written
{
W = 12 E A uo2,r dr + I tt t2,r dr 2 I st t , r s ,r dr + I ss s2,r dr + I r2,rr dr }
+ 12 GJ r2,r dr

Chapter 2: Elements
(2.4-39)
This expression can be integrated analytically once beam

interpolations are introduced (see Section 2.4.4). And the force
vector and stiffness matrix are obtained by taking variations of the
strain energy.
2.4.3.7 Wagner effect
H. Wagner discussed the mechanics of thin-walled open sections in

the following reference:
ref. Wagner, H., Verdrehung und Knickung von offenen

Profilen (Torsion and Buckling of Open Sections), 25th
Anniversary Publication, Technische Hochschule, Danzig,
1904-1929, translated as N.A.C.A. Technical
Memorandum No. 807, National Committee for
Aeronautics, 1936.
The following observation in this paper is of significance: Consider

a beam carrying axial stress, and consider a small twist of the
beam. The axial stress now has a transverse component due to the
twist. The resultant of all of the transverse components of axial
stress acting on a cross-section has a non-zero twisting moment.
For brevity, we term the above observation, the "Wagner
effect".
Fig. 2.4-4 shows a schematic. The beam carries an initial axial
stress (this stress can vary from point to point in the cross-section).
For the material fiber shown in the figure, this stress has magnitude
rr = T and is oriented in direction (1, 0, 0) . Now suppose that the
cross-section undergoes a small twist, of magnitude per unit
length. After deformation, the material fiber has the orientation
(1, t , s ) and the Cauchy stress tensor corresponding to T is
1 t s

= t 0 0 T (2.4-40)
s 0 0

2.4: Beam elements
^t
After twist
r
^t
r+dr s^ (-t^a, s^a)T
Before twist
a dr
^s
(a) Isometric view (b) In plane of cross-section
Figure 2.4-4: Wagner effect
The twisting moment in the plane of the cross-section due to the

stress at the point ( s, t) is M r = ( s 2 + t 2 )T , and the virtual work
done by a virtual twist is
M r = ( s 2 + t 2 )T (2.4-41)
This result can also be obtained by using the proper continuum

mechanics stress and strain measures. Following standard
continuum mechanics procedures, the Green-Lagrange strain
component in the axial direction, due to the twist , is
rr = 12 ( s 2 + t 2 ) 2 (2.4-42)
and the 2nd Piola-Kirchhoff stress tensor is
1 0 0
S = 0 0 0 T (2.4-43)
0 0 0
Thus the work done by the axial stress as a result of the twist is
simply
S rr rr = ( s 2 + t 2 )T (2.4-44)

Chapter 2: Elements
Since (2.4-41) and (2.4-44) are equal, the Wagner effect can be
included simply by using the Green-Lagrange strain tensor instead
of the small strain tensor, and interpreting the stresses as 2nd Piola-
Kirchhoff stresses instead of Cauchy stresses.
It is clear that the Wagner effect is a geometrically nonlinear

effect. The Wagner effect is included only in conjunction with the
large displacement formulation described in Section 2.4.5.3.
The procedure of using the Green-Lagrange strain tensor instead

of the small strain tensor, and interpreting the stresses as 2nd Piola-
Kirchhoff stresses instead of Cauchy stresses, works extremely
well when used in the large displacement/small strain formulations
for continuum finite elements. However, it should be remembered
that the geometry of continuum elements is completely specified by
the nodes, whereas the geometry of beam elements is specified in
terms of cross-sectional properties such as A , I ss , etc. Including
all terms in the Green-Lagrange strain tensor leads to a very large
number of cross-sectional properties.
Therefore we focus on including only the minimum number of

additional terms in the strain tensor so that the Wagner effect is
properly included. Recalling that the components of the Green-
Lagrange strain tensor are
rr = err + 12 ( u,2r + v,2r + w,2r )

2 rs = rs + ( u, r u, s + v,r v, s + w,r w, s ) (2.4-45)
2 rt = rt + ( u, r u,t + v,r v,t + w,r w,t )
we will assume that all products in rs , rt can be neglected

compared to rs , rt , and that u,r2 can be neglected compared to
err .
In order to neglect additional terms, we need to use the fact that

the Wagner effect is implemented along with the large
displacement formulation described in Section 2.4.5.3. In the large

2.4: Beam elements
displacement formulation, all angles are measured with respect to a

coordinate system attached to the beam. In the limit of mesh
refinement, these angles can be made arbitrarily small (again,
measured with respect to the coordinate system attached to the
beam).
Now consider the term v,2r = t2 2tt r , r + t 2 r2,r . In the limit
of mesh refinement, t becomes arbitrarily small, however r ,r
cannot be made arbitrarily small by mesh refinement. A similar
statement can be used for term w,r2 .
So we use the approximations
v,2r t 2 r2,r
(2.4-46)
w,2r s 2 r2,r
The result is that the components of the Green-Lagrange strain

tensor are approximated by
rr err + 12 ( s 2 + t 2 ) r2,r = uo,r st ,r + t s ,r + r ,rr + 12 ( s 2 + t 2 ) r2,r

2 rs rs = (, s t ) r ,r
2 rt rt = (,t + s ) r , r
(2.4-47)
and the virtual work becomes
W = Srr rr + Srs rs + Srt rt dV (2.4-48)
Constructing the strain energy (for a linear elastic material)

using the procedure given above leads to additional terms in the
strain energy:

Chapter 2: Elements
W = ... + 12 E { ( s 2
+ t 2 ) dA uo , r r2,r dr
s ( s 2 + t 2 ) dA t ,r r2, r dr
+ t ( s 2 + t 2 ) dA s ,r r2,r dr (2.4-49)
+ ( s 2 + t 2 ) dA r , rr r2, r dr
+ 14 ( s 2 + t 2 ) dA r4,r dr
2
}
where the ... are the terms from the small strain theory (2.4-38).
Using the geometric properties given in Section 2.4.2, the strain
energy for a linear elastic material can be written
{
W = 12 E A uo2,r dr + I tt t2,r dr 2 I st t , r s ,r dr + I ss s2,r dr + I r2,rr dr }
+ 12 E {A u
15 dr A25 t ,r r2,r dr + A35 s ,r r2,r dr + A45 r ,rr r2,r dr
2
o ,r r ,r
+ 14 A55 r4,r dr }
+ 12 GJ r2,r dr
(2.4-50)
This expression can be integrated analytically once beam

interpolations are introduced (see Section 2.4.4). And the force
vector and stiffness matrix are obtained by taking variations of the
strain energy.
The Wagner effect is implemented for inelastic materials simply

by computing the strains using (2.4-47), and by including the
proper geometrically nonlinear terms in the stiffness matrix.
As a very simple example in which the Wagner effect is

important, consider the axial loading of an elastic beam, in which
bending is suppressed but in which torsion is allowed. For
simplicity the angle of twist per unit length r ,r is assumed to be
constant. (2.4-50) becomes

2.4: Beam elements
W = 12 EA uo2, r dr + 12 EA15 uo ,r r2, r dr + 12 GJ r2,r dr + 18 EA55 r4,r dr

(2.4-51)
Taking the variation of W with respect to r ,r and using

rr Srr = Euo,r gives
W = ( ( A
15 rr )
+ GJ ) r ,r dr + 12 EA55 r3,r dr r ,r (2.4-52)
For small twist, the r3,r term can be neglected. Equation (2.4-52)
J
shows that for stress rr G , the equation of motion
A15
becomes unstable, that is, torsional buckling can occur. The ratio
J
influences how important the Wagner effect is for a given
A15
cross-section. For a pipe section, both J and A15 are equal to the
polar moment of inertia of the section, so torsional buckling can
only occur for compressive stresses that are much larger than can
reasonably be expected. On the other hand, for a thin-walled open
J
section, the ratio can be much smaller than unity, then
A15
torsional buckling can occur for much smaller compressive
stresses.
Of course, the simplifications that we made in obtaining (2.4-
J
52) mean that rr = G is not the torsional buckling load in
A15
general. For actual formulas involving torsional and flexural
buckling loads, see references such as
ref. A. Gjelsvik, The Theory of Thin Walled Bars, John Wiley &
Sons, 1981
2.4.3.8 Kinetic energy and the mass matrix
The kinetic energy of the beam can be written

Chapter 2: Elements
K = 12 ( u 2 + v 2 + w2 ) dV (2.4-53)
Substituting from (2.4-17) gives
{ ( )
K = 12 A ( uo2 + vn2 + wn2 ) dr + I tt r2 + t2 dr 2 I st t s dr
( )
+ I ss r2 + s2 dr + I r2,r dr }
(2.4-54)
The mass matrix is obtained from (2.4-54), once a beam

interpolation has been introduced.
The rotary inertias and the warping constant can have a

significant effect on the torsional response of the beam, particularly
for thin-walled open sections.
2.4.3.9 The warping function
The torsional response of beams significantly depends on the

type of their cross-section (e.g., solid section, thin-walled open
section, thin-walled closed section).
The torsional response is closely related to the displacements of

the beam cross-section that are out of the plane of the cross-section
(see Fig. 2.4-5). These displacements are termed "warping
displacements", since torsion causes warping of the beam (see Fig.
2.4-5).
The warping displacements are included in the beam

formulation though the use of the warping function ( s, t ) . The
warping function must be determined in advance for each cross-
section.
For circular cross-sections, torsion does not cause warping

displacements, thus the warping function is zero for circular cross-
sections.
For the solid rectangular section undergoing plasticity, warping

is accounted for as discussed in Section 2.4.6.2. Note that it is not

2.4: Beam elements
allowed to fix the warping displacements in the solid rectangular

section.
q
Front view Side view

Figure 2.4-5: Torsional warping of an I-beam subjected to pure torsion
In the following discussion, we focus on warping functions for

thin-walled sections. The theory of the warping function for thin-
walled sections can be found in several references, including
ref. A. Gjelsvik, The Theory of Thin Walled Bars, John Wiley &
Sons, 1981
A brief summary of the theory follows, in which we have adjusted

the notations so as not to conflict with the notations used in the rest
of this section.
Open thin-walled sections
Consider a generic open thin-walled section (Fig 2.4-6). The

centerline of the section is called the contour. The coordinate
through the section thickness is denoted p and the coordinate along
the contour is denoted q.

Chapter 2: Elements
We introduce an additional point, not necessarily in the cross-

section, called a pole (Fig 2.4-7). Corresponding to a point in the
cross-section and the pole P, there is a coordinate system p , q , in
which p is parallel to dp and q is parallel to dq.
Contour branch
dq Contour
Contour origin, q=0

p
p=0 on contour
(a) Contour coordinate q (b) Thickness coordinate p
Figure 2.4-6: Cross-section coordinate system

2.4: Beam elements
t dq dp
Cross-section
Contour, p = 0
t
p
q
ct
Pole P s
C cs s
Figure 2.4-7: Notation used in construction of warping function
Now consider a rigid body rotation of the cross-section about

pole P (Fig 2.4-7). Clearly
us = t r ,
(2.4-55)
ut = s r
where us , ut are the displacements in the s, t directions. Also
u p = q r ,
(2.4-56)
uq = p r
where u p , uq are the displacements in the p , q directions. Since

p , q are parallel to dp, dq ,
u p = q r ,
(2.4-57)
uq = p r
are the displacements perpendicular to and parallel to the contour.

From this information, we can construct an approximate
warping function ( s, t ) . First consider the shear strain

Chapter 2: Elements
rq = ur ,q + uq ,r (2.4-58)
where ur is the displacement out of the plane of the cross-section

(the warping displacement). The choice
ur ,q = uq ,r = p r ,r (2.4-59)
sets rq = 0 on the contour. Now consider the shear strain
rp = ur , p + u p ,r (2.4-60)
The choice
ur , p = u p ,r = q r ,r (2.4-61)
sets rp = 0 along lines perpendicular to the contour. Since the

cross-section is thin, q is constant along these lines, so
ur ( p ,q )
= ur ( 0, q )
+ p q r , r (2.4-62)
and therefore
ur ( p ,q ) ( q
= p dq + p q r , r
0 ) (2.4-63)
From this expression, we observe that the warping function is
q
( s, t ) = p dq + p q (2.4-64)
0
Thus the total warping is composed of two contributions, the first

contribution along the centerline, called contour warping, and the
second contribution through the thickness, called thickness
warping.
From the above conditions on the shear strain, the shear strain
caused by torsion is aligned with the q direction and proportional to
p.

2.4: Beam elements
If the cross-section of the beam only consists of thin rectangular

plates all intersecting at a common point, then the contour warping
is zero and the warping function only contains thickness warping
(see Fig. 2.4-8).
(a) (b)
Figure 2.4-8: Sections with negligible contour warping
At a sharp corner or branch in the cross-section, the warping

function is continuous along the contour ( p = 0) , but is not
continuous for other values of p.
The warping function is not exact in the sense of the St. Venant
warping function. The St. Venant warping function satisfies
( + ,2t t, s + s,t ) dA = 0
2
,s (2.4-65)
but the thin-walled warping function used here has
( + ,2t t, s + s,t ) dA = 2 p 2 dA
2
,s (2.4-66)
For additional insight into the torsional response described in

Section 2.4.3.5, consider the torsional response under elastic
conditions. The bishear can then be written
(
Fb = G r ,r (, s t ) , s + (,t + s ) ,t dA )
(2.4-67)
= G r ,r (,2s + ,2t t, s + s,t ) dA
If we had used the exact St Venant warping function, this term

Chapter 2: Elements
would be equal to zero, implying that the external torque in a beam

free to warp is equilibrated entirely by the torque calculated from
the shear stresses.
But because we use the approximate warping function given
above, the bishear is nonzero. It turns out that, in an open thin-
walled section, the bishear and torque calculated from the shear
stresses can be written
Fb = G r .r 2 p 2 dA , M r = G r ,r 2 p 2 dA (2.4-68a,b)
so that each term equilibrates one-half of the external torque. The

torsional rigidity can then be observed from the expression
T = Fb + M r = G r ,r 4 p 2 dA = GJ r ,r (2.4-69)

implying J = 4 p 2 dA . For a cross-section with constant thickness
1 3
a and length b, this formula gives J = a b , which is the textbook
3
solution for the torsional rigidity of a thin plate.
Closed thin-walled sections
We consider here only closed sections with just one closed circuit
in the contour. It is assumed that the shear strain flow rq b is equal
to K r , r on the contour, where b is the cross-section thickness
and K is a constant to be determined. Thus
K
ur , q + uq , r = r ,r (2.4-70)
b
As in the open cross-section theory, uq = p r , thus

qK
ur (0, q )
= p dq r ,r (2.4-71)
0 b
The constant K is determined by noting that the change in

2.4: Beam elements
warping displacement resulting from one complete circuit must be

zero, therefore
K=
p dq = 2 A (2.4-72)
1 1
b dq b dq
where A is the area enclosed by the contour.
Using rp = 0 along lines perpendicular to the contour, the
warping displacement is
q K
ur ( p ,q )
= p dq + p q r , r (2.4-73)
b
0

and the warping function is
q K
( s, t ) = p dq + p q (2.4-74)
0
b
Again the total warping is composed of two contributions, the

contour warping and the thickness warping.
Open and closed thin-walled sections
Up to now, we have not considered the location of pole P

relative to the origin. And we have also not considered the origin of
the p,q coordinate system. Different choices of contour origin and
pole origin give different warping functions.
However, we fix the contour and pole origin using the
conditions (2.4-2) and (2.4-3) and it turns out that the pole origin
fixed using these conditions is the shear center of the cross-section.
It may be of interest to observe that the choice of contour and
pole origins does not affect the torsional rigidity
( )
J = (, s t ) + (,t + s ) dA ; however the warping constant
2 2
I = 2 dA is minimized by fixing the contour and pole origin

using conditions (2.4-2) and (2.4-3).

Chapter 2: Elements
2.4.4 Beam interpolations
2.4.4.1 Standard beam
The displacements and rotations are interpolated from the nodal

displacements and rotations using
u0 = L1u1 + L2u 2
v p = H1v1p + H 2 t1 + H 3v 2p + H 4t2 (2.4-75)
wp = H1w1p H 2 s1 + H 3 w2p H 4 s2
and
r = L1 r1 + L2 r2
s = H1,r w1p + H 2,r s1 H 3,r w2p + H 4,r s2 (2.4-76)
t = H1,r v1p + H 2,rt1 + H 3,r v 2p + H 4,rt2
where the nodal displacements and rotations are

( u , v , w , , , ) for node 1 and ( u , v , w ,
1 1
p
1
p
1
r
1
s
1
t
2 2
p
2
p r
2
, s2 , t2 ) for
node 2, and in which
r r
L1 = 1 , L2 = (2.4-77)
L L
are the linear interpolation functions and
r2 r3 r2 r3
H1 = 1 3 + 2 , H 2 = r 2 + ,
L2 L3 L L2 (2.4-78)
r2 r3 r2 r3
H3 = 3 2 2 3 , H4 = + 2
L L L L
are the cubic interpolation functions (Hermitian displacement

functions).

2.4: Beam elements
Notice that the Euler-Bernoulli condition (2.4-20) is satisfied by

the choice (2.4-75) and (2.4-76).
It is seen that the transverse shear center axis displacements vp

and w p are cubic and that the axial displacement uo and torsional
rotation r are linear.
We refer the transverse displacements to the neutral axis using

(2.4-18):
v1p = v1n ct r1 v 2p = vn2 ct r2

, (2.4-79)
w1p = w1n + cs r1 w2p = wn2 + cs r2
and we can then write the nodal degrees of freedom as

( u , v , w , , , ) for node 1 and ( u , v , w ,
1 1
n
1
n
1
r
1
s
1
t
2 2
n
2
n r
2
, s2 , t2 ) for
node 2. In this way, all nodal quantities are referred to the neutral
axis. The forces and moments at the beam element nodes
(energetically conjugate to the nodal displacements) are also all
referred to the neutral axis.
It should be clear from the above that warping effects are

included in the standard beam. It should also be clear that pure
torsion will cause the transverse displacements at the neutral axis to
be non-zero, due to the shear center offsets (cs , ct ). And,
conversely, transverse forces will cause torsion.
2.4.4.2 Warping beam
The axial rotation is interpolated using a cubic interpolation

function and the axial rotation and warping degrees of freedom:
r = H1 r1 + H 2 1 + H 3 r2 + H 4 2 (2.4-80)
Evaluating r ,r at nodes 1 and 2 gives the result r ,r = 1 and

node 1
similar for node 2. Hence 1 , are the warping function multipliers

evaluated at nodes 1, 2.

Chapter 2: Elements
The remaining displacements and rotations are interpolated as in

(2.4-75) to (2.4-78). Thus the nodal degrees of freedom
( )
are u1 , v1n , w1n , r1 , s1 , t1 , 1 for node 1 and
( u , v , w ,
2 2
n
2
n r
2
, s2 , t2 , 2 ) for node 2. Again, the nodal quantities
are referred to the neutral axis using (2.4-79).
The force conjugate to the warping degree of freedom is the

bimoment, see the discussion in Section 2.4.3.5. The dimension of
the bimoment is [FL2]. The bimoment is included in the global
force vector.
The warping degree of freedom can be fixed or left free, and

depending upon the physical situation, it may or may not be
appropriate to fix the warping degree of freedom, see discussion in
Section 2.4.9.1.
The standard beam element is not recovered by fixing the

warping degrees of freedom at all nodes in the beam. This is
because the interpolation of axial rotations in (2.4-80) remains
cubic when the warping degrees of freedom are fixed.
2.4.4.3 Comparison of standard beam and warping beam
The motivation for introducing the warping beam element is as

follows. As seen above, warping is assumed in the standard beam
element through the terms J , I . However the term r ,rr is always
zero in the standard beam element, by construction, so warping
through term I does not contribute to the strain energy (but
warping through term I contributes to the kinetic energy). More
importantly, the warping displacements between adjacent elements
are not compatible in the standard beam element. This is because
the term r ,r , which expresses the longitudinal displacement due
to warping, is constant in each element and in general different
between adjacent elements.
In the warping beam, term r ,r is continuous between adjacent

elements, thus this element could also be called a "compatible

2.4: Beam elements
warping beam" element.
Due to the incompatibility of the warping displacements in the

standard beam element, the results obtained from the standard
beam element will be incorrect in the general case.
However, it should be noticed that there are some important

special cases in which the standard beam gives correct results:
1) Warping function is zero or negligible. This occurs for

circular cross-sections.
2) Angle of twist per unit length is constant along the length of

the beam assemblage. This occurs in many static problems, e.g.
bending and torsion of a straight beam. This situation also occurs in
problems in which torsion is not present, for example, in-plane
deformations.
3) Warping, although present, does not affect the solution (other

than the torsional response) significantly. This occurs for solid
sections and possibly also for closed thin-walled sections.
In general, for the solid and closed thin-walled sections, the

standard beam can be used without significant error. However for
the open thin-walled sections, the warping beam should be used. It
is allowed, however, to use the open thin-walled sections with the
standard beam, and, as long as the angle of twist per unit length is
constant along the beam, the warping beam and standard beam will
give exactly the same results.
2.4.5 Beam element implementation
The stiffness matrix, mass matrix and force vector are

formulated in the local degrees of freedom (in the r, s, t axes).
These matrices and vectors are then transformed to the nodal
degrees of freedom (either global or skew) and assembled into the
global system matrices.

Chapter 2: Elements
2.4.5.1 Linear formulation
It is assumed that the displacements, rotations, and strains are

infinitesimally small, and the elastic-isotropic material is used. The
Wagner effect is not included.
2.4.5.2 Materially-nonlinear-only formulation

infinitesimally small. Either the elastic-isotropic material is used
(for example, in conjunction with element birth-death), the plastic
material models are used, or the moment-curvature material model
is used. The Wagner effect is not included.
2.4.5.3 Large displacement formulation
It is assumed that large displacements/rotations can occur, but

only small strains. Any beam material model, linear or nonlinear,
can be used. The input of beam cross-sections and material data is
exactly the same as when using the linear or materially-nonlinear-
only beam elements.
If a stressed large displacement beam element is subjected to a

rigid body rotation, the stresses/forces (expressed in the local
coordinate system) do not change during the rigid body rotation.
The geometry of the large displacement beam element is shown

in Fig 2.4-9. The shape of the beam neutral axis is completely
specified by the positions of the end-nodes, and by the orientation
of the end-node triads. Each end-node triad consists of three
orthogonal unit vectors Vr , Vs , Vt . Initially the triads are identical
to the element coordinate axes unit vectors, e.g, Vr is a unit vector
in the r direction, etc.
During the deformations, the positions of the end-nodes are
updated, as usual, by the nodal displacements. The orientations of
the end-node triads are updated incrementally by the increments in
nodal rotations. The end-node triads can rotate independently of
each other, however it is assumed that the relative rotation of the
end-node triads remains small.

2.4: Beam elements
z
Vs2
Vr2
x y
2 Vt2
Vs1
Vr1
1 Vt1
a) Before deformation
Vs1
Vr1 Vs2
1
Vt1 Vr2
2 Vt2
Beam neutral axis

b) After deformation
Figure 2.4-9: Geometry of large displacement beam element
In the large displacement formulation, the coordinate system in

which the beam local displacements and rotations are measured is
updated during the solution. In ADINA 8.8 and higher, this
coordinate system is taken from the orientations of the end-node
triads. In ADINA 8.7 and lower, this coordinate system is taken
from the coordinates of the end-nodes. Hence the solution output in
ADINA 8.8 and higher will in general be different than the solution
output in ADINA 8.7 and lower.
The advantage of using a coordinate system defined from the
orientations of the end-node triads is that this coordinate system is
uniquely defined even for general 3-D deformations including
torsion. A coordinate system defined from the end-nodes
coordinates is not uniquely defined for 3-D deformations including

Chapter 2: Elements
torsion.
The shape of the beam neutral axis becomes a curved space

curve. The displacements and rotations of the beam nodes, used in
the construction of the force vector and stiffness matrix, are
measured with respect to the beam local coordinate system.
Fig 2.4-10 shows an example for in-plane bending. A single

beam element is clamped at node 1 and subjected to a prescribed
rotation at node 2. The coordinate system x, y used to measure
beam local displacements and rotations is located halfway between
the nodes, and is rotated an angle / 2 with respect to the global
coordinate system x, y . Notice that the x direction does not
coincide with the line between the end-nodes.
The rotation at local node 2 with respect to the local coordinate
system is / 2 , and the rotation at local node 1 with respect to the
local coordinate system is / 2 . Also, the transverse
displacement at local node 2 with respect to the local coordinate
system is v and the transverse displacement at local node 1 with
respect to the local coordinate system is v .

2.4: Beam elements
2 q
y
y x
q /2
v
Beam neutral axis

1 x
a) Viewed in global coordinate system
y
v x q /2
-q /2
-v
b) Viewed in local coordinate system
Figure 2.4-10: Coordinate system update for large displacement beam

element, in-plane bending example
The element stiffness matrix is non-symmetric in the rotational

degrees of freedom. Therefore the option of using the non-
symmetric sparse solver (MASTER SOLVER=NONSYM-
SPARSE) can be useful in obtaining convergence.
The Wagner effect is included by default for the thin-walled

sections.
For frequency analysis and linearized buckling analysis, it is

allowed to include certain 2nd order stiffness effects into the
stiffness matrix. This choice is made using the command
KINEMATICS KBEAM-EIGENVALUE=YES
By default, this feature is not used. Note that these 2nd order

Chapter 2: Elements
stiffness effects are not introduced into the force vector, so this
feature should be used with caution.
This feature is made available only for backwards compatibility
with version 8.7 and earlier.
The following effects are not included within this feature:
warping, shear center offsets, Wagner effect.
2.4.5.4 Mass matrices
The beam element can be used with a lumped or a consistent

mass matrix, except for explicit dynamic analysis which always
uses a lumped mass matrix.
The consistent mass matrix of the beam element is evaluated in

closed form, and does not include the effect of shear deformations.
The matrix is defined in the local coordinate system using the
expression for the kinetic energy derived above (equation 2.4-54)
and the beam interpolations (either standard or warping). Therefore
the consistent mass matrix includes warping effects.
For the standard beam element, the consistent mass matrix is the
same as the one given in the following reference:
ref. J.S. Przemieniecki, Theory of Matrix Structural

Analysis, McGraw-Hill Book Co., 1968.
with the exception that there are additional terms arising from the
warping constant I .
The lumped mass for translational degrees of freedom is M / 2

where M is the total mass of the element.
The rotational lumped mass for static analysis is 0.

The rotational lumped mass for implicit dynamic analysis is
2 M I ss + I tt
M rr = . This lumped mass is applied to all
3 2 A
rotational degrees of freedom in order to obtain a diagonal mass
matrix in any coordinate system.
The rotational lumped mass for explicit dynamic analysis is
M Im
M rr = 3 where I m = max ( I ss , I tt ) is the maximum
2 A

2.4: Beam elements
bending moment of inertia of the beam. This lumped mass is

applied to all rotational degrees of freedom. Note that this scaling
of rotational masses ensures that the rotational degrees of freedom
do not affect the critical stable time step of the element.
The rotational lumped masses can be multiplied by a user-

specified multiplier ETA (for implicit dynamic analysis only). The
multiplier factor used in explicit analysis is always 1.0.
For the warping element, lumped masses for the warping

degrees of freedom are defined. However the process of lumping
implies that all coupling between warping and rotational degrees of
freedom is lost. Since there is a high degree of coupling between
these degrees of freedom, the lumped mass assumption must be
considered extremely approximate. Therefore the use of a lumped
mass matrix is not recommended for the warping beam element.
The lumped warping masses have been improved between
versions 8.9 and 9.0, so the results of analyses in which lumped
warping masses are used will change between versions 8.9 and 9.0.
In theory the consistent mass matrix, expressed in the global

coordinate system, changes as a result of large rotations in large
displacement analysis. However, the mass matrix is not updated
(except in the case of element birth/death). Therefore, when using a
consistent mass matrix in large displacement analysis, the results
can be in error if the rotations are significant.
2.4.5.5 Elastic beam element
The beam element stiffness matrix is evaluated in closed form.

The stiffness matrix used is derived from the strain energy
expression in equation (2.4-39) and the beam element
interpolations (either standard or warping).
For the standard beam element, the stiffness matrix is the same
as the one given in the following reference:
ref. J.S. Przemieniecki, Theory of Matrix Structural

Analysis, McGraw-Hill Book Co., 1968.
Certain cross-sections allow entry of the shear areas. When

shear areas are entered, the force vector and stiffness matrix are

Chapter 2: Elements
modified in order to approximate the effects of shear deformations,

as discussed in Przemieniecki.
For example, considering a cantilever beam with in-plane
deformations only, the stiffness matrix for the free transverse
displacement and rotation, including the shear area modifications is
12 EI tt 6 EI tt
L3 (1 + ) L (1 + 2 )
2
K=
2
EI tt (4 + 2 )
sym L(1 + 2 )

where
12 EI tt
2 = , Assh > 0
L2GAssh
= 0, Assh = 0
The coefficient of thermal expansion can be specified as a

material property. The coefficient of thermal expansion is constant
(independent of the temperature). In addition, the beam temperature
is taken as the average of the temperatures of the beam end-nodes.
The temperature gradients at beam nodes are ignored.
2.4.5.6 Elastic-plastic beam element
The element is used with the plastic-cyclic material model

described in Section 3.4.6. The element can also be used with a
plastic-bilinear material with isotropic hardening, but then the
program automatically converts the material input into the
equivalent plastic-cyclic material input.
The stress-strain law used incorporates the assumptions that the
stresses ss , tt , st are zero.
The material model can be used either with the materially-

nonlinear-only formulation or with the large displacement
formulation (in which case the displacements/rotations can be
large). In all cases, the strains are assumed to be small.
The Wagner effect is included by default when the large

2.4: Beam elements
displacement formulation is selected.
ref. KJB All element matrices in elasto-plastic analysis are calculated

Section 6.6.3 using numerical integration.
In the r-direction, Newton-Cotes integration is used (Fig 2.4-
11). The number of integration points is controlled with the
EGROUP BEAM ... RINT parameter. RINT can be 3, 5, 7 or 9.
The default is 5, which corresponds to an exact integration of all
system matrices under elastic small strain conditions (no Wagner
effect). When the Wagner effect is included, 9 stations can be
selected in order to exactly integrate the system matrices under
elastic conditions. However, 9 stations should not be used in
general as 9-point Newton-Cotes integration is possibly unreliable
for the numerical integration of arbitrary functions.
The numerical integration in the cross-section depends upon the
cross-section, therefore further details are given along with the
cross-sections (Section 2.4.6).
s Integration points
equally spaced
1 2 3 ...

r
r=0 r=L
t
Figure 2.4-11: Integration point locations in r-direction
Note that the system matrices are identical to the system

matrices used in elastic analysis if
The warping function assumed in elasto-plastic analysis is the

same as the one used in elastic analysis. For the pipe cross-
section, this condition is automatically satisfied since there is no
warping function in the pipe section. For the rectangular cross-
section, this condition is reached in the limit for a very thin
section (because the exact warping function is employed in the
limit of a very thin section). For the thin-walled sections, this
condition is satisfied as long as the torsional rigidity formula
used in elastic analysis is formula FW (see Section 2.4.6 for
details about the formulas FW).

Chapter 2: Elements
The numerical integration is of high enough order.
The material is (still) elastic.
For the rectangular and pipe cross-sections, the elastic-plastic

beam element can either be used using a 2-D action option or a 3-D
action option. This choice is made using the command EGROUP
BEAM SUBTYPE={TWO-D/THREE-D}; the default is THREE-
D.
2-D action assumes that the element deformations occur in the
r-s plane. In this case, the element has no stiffness corresponding
to deformations out of the r-s plane. Also, the element does not
have any torsional stiffness.
3-D action assumes that element deformations occur in any
direction.
For both 2-D action and 3-D action, the element can have an
arbitrary orientation with respect to the global coordinate axes.
For the box, I, U, L cross-sections, the elastic-plastic beam

element does not include shear deformation effects.
2.4.5.7 Element output
There are three options for calculation of beam element output:
element nodal point forces/moments (EGROUP BEAM

RESULTS=FORCES) (the default for elastic analysis)
stresses at the element integration points (EGROUP BEAM

RESULTS=STRESSES) (the default for elastic-plastic
analysis)
element section forces/moments at discrete locations along

the axial direction of the element (EGROUP BEAM
RESULTS=SFORCES)
In the description below, we use the term "forces" to describe both

forces and moments.

2.4: Beam elements
Nodal point forces: The nodal point forces are computed at the
element local nodes. These forces are equivalent, in the virtual
work sense, to the internal element stresses. The nodal point forces
are used to calculate reaction forces.
Because the local nodes are located at the neutral axis (and not
the shear center), the nodal point forces are also computed with
reference to the neutral axis (and not the shear center). This issue is
discussed in further detail below.
The element nodal point forces are accessible in the AUI using
the variable names NODAL_FORCE-R, NODAL_FORCE-S,
NODAL_FORCE-T, NODAL_MOMENT-R, NODAL_MOMENT-S
and NODAL_MOMENT-T. (If the warping beam is used, the
variable name NODAL_BIMOMENT is also available.) In the AUI,
element local nodes are defined as element points of type label. For
example, to access the result computed at element 5, local node 2,
define an element point of type label with element number 5, label
number 2.
If nodal point forces are requested in an elastic-plastic analysis,

the state of stress in the element (whether elastic or plastic) is not
indicated.
Stresses: Stress output is not available for the elastic beam

element. If stresses are requested for the elastic element, the nodal
point forces are computed instead.
Stress output for the elastic-plastic beam element is described in
Section 3.4.5.3.
Section forces: The element section forces can be computed at

equally spaced points along the neutral axis of the beam. The
number of points is controlled by EGROUP BEAM ... SPOINT;
the default number of points is 2. Station 1 is at local node 1 and
station SPOINT is at local node 2. The sign convention for section
forces is given in Figure 2.4-12. Notice that the vector of positive
moment in the s-direction points in the negative s-direction.
If section forces are requested in an elastic-plastic analysis, the
state of stress in the element (whether elastic or plastic) is not
indicated.

Chapter 2: Elements
Mt
Ft
t Ms
Fs
s
r
Fr
Mr
Figure 2.4-12: Sign convention for section forces/moments
There are two options for computation of section forces/moments.
Integration of stresses over the cross-section
The option of integration of stresses over the cross-section can be

used for the elastic and elastic-plastic beam elements, for
geometrically linear and large displacement formulations. The
section forces are computed using
Fr = rr dA
Mr = ( ( s 2
)
+ t 2 ) r ,r rr + (, s t ) rs + (,t + s ) rt dA
M s = t rr dA (2.4-81)
M t = s rr dA
M b = rr dA
These expressions are developed from the principle of virtual work:

2.4: Beam elements
Fr , M r , M s , M t , M b are conjugate to uo ,r , r ,r , s ,r ,
t ,r , r ,rr respectively. In M r , the term involving rr arises
from the Wagner effect and is included only when the Wagner
effect is included. M b is computed only for the warping beam.
The shear forces are estimated using the expressions
Fs = M t , r , Ft = M s , r . (It is not possible to compute shear
forces by integration of shear stresses because in Euler-Bernoulli
beam theory, there are no shear stresses generated except in
torsion.)
For the elastic beam, the section forces are evaluated
analytically. For the elastic-plastic beam, the axial force and
moments are evaluated using numerical integration over the cross-
section, and the shear forces are evaluated by the approximations
Mt Mt Ms Ms
Fs = r =1 r =0
, Ft = r =1 r =0
.
L L
For the elastic-plastic beam, it is necessary that each station at
which section forces are evaluated correspond to an integration
RINT 1
point along the r direction. Thus the expression must
SPOINT 1
be an integer for the elastic-plastic beam.
Fixed-end force correction
As an alternative for computing section forces by integration, it is

allowed to compute section forces using the theory of fixed-end
force correction.
In order to motivate the fixed-end force correction, consider the
problem shown in Fig. 2.4-13. Only one element is required to
obtain the analytical displacement and rotation solution. However,
the shear forces and bending moments, computed without using the
fixed-end force correction, are incorrect, and only approach the
analytical solutions as more elements are used.

Chapter 2: Elements
Elastic beam under

distributed loading
(linear analysis)
L
x
Shear
force (V)
qL/2 Convention for

positive V
1 element
0 2 elements
L x 4 elements
Analytical solution
-qL/2
Bending
moment (Mb)
Convention for
qL2 /8 positive Mb
0
L x
Figure 2.4-13: Schematic beam element forces/moments when

subjected to distributed loading

2.4: Beam elements
The basic ideas of fixed-end force correction are shown in Fig. 2.4-
14. As seen, the original distributed loading in Fig. 2.4-14(a) is
replaced by the equivalent nodal point forces and moments in Fig.
2.4-14(b). (Of course, this procedure is the one always used in
finite element analysis when distributed loads are applied.) The
solution without any fixed-end force correction is shown in Fig.
2.4-14(c) and it is clear that the bending moment distribution does
not agree with the results from beam theory. This bending moment
distribution arises from the internal force vector in the beam
element, namely the beam element is subjected to the end-node
forces and moments and equilibrium is used to determine the
intermediate section forces and moments. In the fixed-end force
correction, the beam element is subjected, instead, to: 1) the end-
node forces and moments, from which the nodal point equivalent
forces of the distributed load have been removed, and 2) the
original distributed load. Then equilibrium is used to determine the
intermediate section forces and moments. The result is shown in
Fig. 2.4-14(d), and the bending moment diagram corresponds to
beam theory.
Similarly, when the fixed-end force correction is used for the
problem in Fig. 2.4-13, the exact shear forces and bending
moments are obtained for each finite element mesh.
The fixed-end force correction feature does not account for
body force loads acting on the element, for example, from mass-
proportional loads.

Chapter 2: Elements
A B A B
l l l l
a) Problem considered, 3 beam elements used b) Nodal point equivalent forces applied to model
Bending Bending
moment moment
MA MA
MB MB
c) Bending moment diagram obtained d) Bending moment diagram obtained

without fixed-end force correction with fixed-end force correction
Figure 2.4-14: Fixed-end force correction example
For more information on the fixed end force correction, see the
following references:
ref: W. McGuire and R.H. Gallagher, Matrix Structural

Analysis, John Wiley & Sons, New York, 1979.
ref: M.L. Bucalem and K.J. Bathe, The Mechanics of Solids and
Structures Hierarchical Modeling and the Finite
Element Solution, Springer, 2011.
The option of fixed-end force correction is selected using

MASTER FEFCORR=YES (not the default).
The option of fixed-end force correction is applicable only to
linear static analysis. In addition the fixed-end force correction
cannot be used when the beam element uses the rigid end feature
described in Section 2.4.8.
If the fixed-end force correction is requested for an element

2.4: Beam elements
group in which the fixed-end force correction cannot be performed,

the section forces are computed by integration of stresses over the
cross-section.
Section forces in the AUI
The section forces are accessible in the AUI using the variable
names AXIAL_FORCE, BENDING_MOMENT-S,
BENDING_MOMENT-T, SHEAR_FORCE-S, SHEAR_FORCE-T
and TORSIONAL_MOMENT. When the warping beam element is
used, the variable BIMOMENT is also available.
Comparison of nodal point forces and section forces
Consider the cantilever beam shown in Fig. 2.4-15. The section

twists under the transverse load, because the transverse load is
applied at the neutral axis, and not at the shear center.
Figure 2.4-15: Twisting of beam due to transverse force on neutral axis
The nodal point moment in the r-direction is zero for this

problem. This result is consistent with the reaction at the fixed end,
in which the moment reaction along the beam direction is also zero.
However the section torsional moment computed using (2.4-81)
is non-zero. Near the built-in end, where the warping is fixed, the
torque produced by the load about the shear center is equilibrated

Chapter 2: Elements
by the change in bimoment and the section torsional moment is

nearly zero (see equation 2.4-34). Far from the built-in end, the
section torsional moment equals the torque produced by the load
P.
Thus for problems in which the neutral axis and shear center do
not coincide, the section forces can be more "physically realistic"
than the nodal point forces.
When the neutral axis and shear center coincide, then, in linear
elastic analysis, the nodal point forces and moments are equal to
the section forces and moments computed by integration over the
cross-section. The nodal point transverse shear forces are also equal
to the section shear forces.
When the neutral axis and shear center coincide, then, when the
element undergoes plasticity in elastic-plastic analysis, the nodal
point forces and moments are not in general equal to the section
forces and moments computed by integration over the cross-
section. Then, for example, the reaction forces might not exactly
agree with the corresponding section forces. However, as the mesh
is refined, the nodal point forces and moments will more closely
match the section forces and moments.
The shear forces are generally less accurate than the bending
moments, because numerical differentiation is used in the
computation of the shear forces.
2.4.6 Cross-sections
The following cross-section types are available:
General (properties)
Rectangular
Pipe
Box
I
U
L
2.4.6.1 General cross-section
The section area, moments of inertia, and other properties can

2.4: Beam elements
be directly specified using the CROSS-SECTION PROPERTIES

command. In this case, the formulas given in Section 2.4.2 should
be used to calculate these properties.
The general cross-section can be used in both linear elastic and

large displacement elastic analysis.
The general cross-section can be used with either the standard

beam element or the warping beam element. The choice of beam
element depends on the values of the cross-section geometric
properties, for example, if the torsional rigidity is much lower than
the polar moment of inertia, the section behaves similarly to a thin-
walled open section and the warping beam element should be used.
In version 8.9 and lower, when the general cross-section is used

with the standard beam, the offsets to the shear center are ignored
and the principal directions are aligned with the s-t axes (that is,
parameters CTOFFSET, CSOFFSET, STINERTIA in the CROSS-
SECTION PROPERTIES command are ignored). This restriction is
removed in version 9.0 and higher. Therefore when running models
created in version 8.9 and lower, it may be necessary to set
CTOFFSET, CSOFFSET, STINERTIA to zero in order to obtain
the same results as in version 8.9.
To include the Wagner effect of torsional coupling in large

displacements, set RRINERTIA, SRINERTIA, TRINERTIA,
WRINERTIA, DRINERTIA 0 in the CROSS-SECTION
PROPERTIES command.
In version 8.9 and lower, terms SRINERTIA, TRINERTIA,

WRINERTIA, DRINERTIA are defined slightly differently than in
version 9.0 and higher. In addition, term RRINERTIA is not
present in version 8.9 and lower. Therefore, when converting data
used in version 8.9 and lower, in which the Wagner effect is
included using these terms, it is necessary to recalculate these
terms, and add term RRINERTIA.
2.4.6.2 Rectangular cross-section
Figure 2.4-16 shows the rectangular cross-section and its

dimensions. ADINA evaluates the section properties from the

Chapter 2: Elements
given dimensions.
t
HEIGHT C.G. s
(H2)
WIDTH
(H1)
Figure 2.4-16: Rectangular cross-section
The rectangular cross-section can be used in both linear elastic

and large displacement elastic analysis. In addition, the rectangular
cross-section can be used in MNO and large displacement elastic-
plastic analysis.
General comments
The shear center of the cross-section coincides with the neutral

axis.
The warping constant I is assumed equal to zero.
The Wagner effect cannot be included when using a rectangular

cross-section.
Comments for elastic analysis
The formula
1 H1 H14 3
J = 1 0.63 1 H1 H2, H1 H2
3 H2 12 H24
1 H2 H24 3
= 1 0.63 1 H2 H1, H2 H1
3 H1 12 H14

2.4: Beam elements
is used to calculate the St. Venant torsional constant. J can be

multiplied by the user-specified parameter TORFAC.
Parameters SC, TC are used to adjust the bending moment

inertias Iss , Itt of the cross-section using the parallel axis theorem:
I ss I ss + A SC 2
I tt I tt + A TC2
The centroid of the beam cross-section remains at the origin of the

(s,t) system, so SC, TC cannot be interpreted as cross-section
offsets. If cross-section offsets need to be modeled, use rigid links
as described in Section 2.4.9.
Parameters SSHEARF, TSHEARF can be used to calculate the

effective shear cross-section areas in the s and t directions:
Assh = SSHEARF A, Atsh = TSHEARF A
Comments for elastic-plastic analysis
Shear deformation effects can be included in an approximate

manner as follows. Additional terms are added to the beam element
displacements:
r r2 r r2
u = ... + 1 + 6 6 2 t 1 + 1 + 6 6 2 s 2
L L L L
r2 r3
v = ... + 3 2 2 3 2
L L
r2 r3
w = ... 3 2 2 3 1
L L
where 1 , 2 are additional degrees of freedom. These additional

terms result in constant shear distortions rs and rt along the
length of the beam, as shown in Fig. 2.4-17. The additional degrees
of freedom are condensed out at the element level.

Chapter 2: Elements
t
grt
s
grs
Figure 2.4-17: Assumptions of shear deformations through element

thickness for elasto-plastic beam element, rectangular
section
The shear effects are included using the command CROSS-

SECTION RECTANGULAR ... ISHEAR=YES. The default is to
not include shear effects.
For beams in 3D action, warping effects are included by

modifying the longitudinal displacement as follows:
u = ... + st1 + ( s 3t st 3 ) 2
where 1 , 2 are additional degrees of freedom. The additional

degrees of freedom are condensed out at the element level.
Warping effects are always included for beams in 3D action.
This procedure is described in the following reference:
ref. K.J. Bathe and A. Chaudhary, "On the Displacement

Formulation of Torsion of Shafts with Rectangular
Cross-Sections", Int. Num. Meth. in Eng., Vol. 18, pp.
1565-1568, 1982.
Table 2.4-2 and Figure 2.4-18 give the integration orders used
in the cross-section.

2.4: Beam elements
Table 2.4-2:Integration orders in elasto-plastic beam analysis,

rectangular cross-section
Coordinate Action Integration orders

Default Maximum
2-D action 3 7
s
3-D action 7 7
2-D action 1 1
t
3-D action 7 7
1) Newton-Cotes integration is used in all coordinate directions
2) One point integration is used in the t-direction for the

rectangular section in 2-D action regardless of user input.
3) Seven point integration is used in the s- and t- directions for the

rectangular section in 3-D action, regardless of user input.

Chapter 2: Elements
Integration points t
equally spaced
HEIGHT s

1 2 3
WIDTH
a) Integration point locations in s-direction
t Integration points t

equally spaced
s s

1 3
2
1
b) Integration point locations in t-direction, c) Integration point locations in t-direction,

2-D action 3-D action
Figure 2.4-18: Integration point locations in elasto-plastic

beam analysis, rectangular cross-section
In explicit dynamic analysis, the use of condensed degrees of

freedom (shear deformations and warpings) will cause the program
to run considerably more slowly than if this option were not used.
2.4.6.3 Pipe cross-section
Figure 2.4-19 shows the pipe cross-section and its dimensions.

ADINA evaluates the section properties from the given
dimensions.

2.4: Beam elements
t
THICKNESS
C.G. s
H2
DIAMETER
(H1)
Figure 2.4-19: Pipe cross-section
The pipe cross-section can be used in both linear elastic and

large displacement elastic analysis. In addition, the pipe cross-
section can be used in MNO and large displacement elastic-plastic
analysis.
General comments

axis.
The warping constant I is set to zero (as the pipe section

cannot warp).
The Wagner effect cannot be included when using a pipe cross-

section.
is equal to the polar

The St. Venant torsional constant J
moment of inertia of the section, multiplied by the user-
specified parameter TORFAC.

inertias Iss , Itt of the cross-section, as in the rectangular cross-
section.

Chapter 2: Elements

effective shear cross-section areas in the s and t directions, as in the
rectangular cross-section.
Comments for plastic analysis
Table 2.4-3 and Figure 2.4-20 gives the integration orders used in
the cross-section. For the pipe section, polar coordinates are used:
t
R = s 2 + t 2 , = tan 1 .
s
Table 2.4-3:Integration orders in elasto-plastic beam analysis, pipe

section
Coordinate Action Integration Integration orders

scheme Default Maximum
radius R Any Newton- 3 7
Cotes.
tangential 2-D action Composite 5 5
angle trapezoidal
3-D action rule 8 8

2.4: Beam elements
t
Integration points
equally spaced
3
...
R
2 y
1
s
a) Integration point locations in

radial direction
t t
Integration points Integration points
equally spaced equally spaced
... 3 2 ... 3 2
R y R y
1 1

s s

b) Integration point locations in c) Integration point locations in

tangential direction, 2-D action tangential direction, 3-D action
Figure 2.4-20: Integration point locations in elasto-plastic

beam analysis, pipe section
2.4.6.4 Box cross-section
Figure 2.4-21 shows the box cross-section and its dimensions.

dimensions.

Chapter 2: Elements
THICK2
(H4)
HEIGHT s
H8 H6
(H2)
THICK1
(H3)
H5
H7
WIDTH
(H1)
Figure 2.4-21: Box cross-section
The box cross-section can be used in both linear elastic and

large displacement elastic analysis. In addition, the box cross-
section can be used in MNO and large displacement elastic-plastic
analysis.
Table 2.4-4 summarizes the choices available for the box cross-
section.

2.4: Beam elements
Table 2.4-4: Options for the box cross-section

Standard Standard Warping beam
beam, V89 beam
assumptions1
Formula for F1, FW, default F1, FW, default FW
calculation of F1 FW
torsional rigidity
(elastic material)
Warping constant No Yes Yes
I included
Wagner effect No Yes, by default Yes, by default
allowed under
large
displacement
conditions
Plastic material No Yes Yes
model allowed
1) CROSS-SECTION BOX ASSUMPTIONS=V89
The table shows that either the standard beam or the warping beam
can be used in conjunction with the box cross-section.
When the standard beam is used, there are two choices:
Standard beam, ASSUMPTIONS=V89:

Warping effects ignored except for the torsional rigidity; mass
matrix effects due to the warping constant I are ignored.
Torsional rigidity J calculated by formula F1 by default.
Wagner effect not included in large displacement analysis.
Standard beam, ASSUMPTIONS=CURRENT:

Warping effects included; the torsional rigidity includes
warping effects and mass matrix effects due to the warping
constant I are included.
Torsional rigidity J calculated by formula FW by default.
Wagner effect included by default in large displacement
analysis.

Chapter 2: Elements
General comments

axis.
The calculation of I is made to be consistent with the

calculations done in the plastic material (in which all integrals are
evaluated numerically from the warping function). Hence the value
of I might not be the same as the handbook values, particularly if
the walls of the cross-section are thick.
The Wagner effect in large displacements is controlled with

CROSS-SECTION BOX ... WAGNER={DEFAULT, YES, NO}.
DEFAULT=YES unless the standard beam element is used and
ASSUMPTIONS=V89
The torsional rigidity is controlled with CROSS-SECTION

BOX ... TORFORMULA={DEFAULT, F1, FW}.
TORFORMULA=F1 (textbook formula):
J = K H7 H8
TORFORMULA=FW (formula consistent with assumed warping

function):
H6 H4
J = 13 H33 H6+ 13 H43 H1+K 2 +
H3 H1
TORFORMULA=DEFAULT: Use formula FW unless the standard

beam is used with ASSUMPTIONS=V89, otherwise use formula
F1.
H3 H4 H7 H8
In these formulas, K = (the shear flow
H3 H7 + H4 H8
constant). Both of these formulas assume that the walls of the box
section are thin, and then these formulas are numerically almost

2.4: Beam elements
equal to each other.

When the warping beam is used, formula FW is always used.
J can be multiplied by the user-specified parameter TORFAC.

section.

The locations and the labeling of the integration points are given
in Fig. 2.4-22. For each direction in the cross-section, Newton-
Cotes integration is used with 1, 3, 5, 7 integration points in that
direction (default 3). The default integration order is sufficient to
exactly integrate all quantities in elastic analysis (not considering
the Wagner effect). However, the integration order can be
increased in order to capture the spread of plasticity throughout the
cross-section.
Each integration point is labeled with a four digit number, the
first digit giving the r integration point and the last three digits
giving the location within the cross-section. For example,
integration point 2312 denotes integration point 2 in the r direction
and integration point 312 in the cross-section. This integration
point is in the left web of the cross-section.
It can happen that there are two integration points at the same
physical location, yet the stresses are different at these points. This
happens because the warping function is not continuous at sharp
corners or branches in the cross-section.

Chapter 2: Elements
...
212 Top flange
211

221 ...
NTHICK1
Flanges and webs

separated for clarity
Integration points
evenly spaced
Left Right NHEIGHT

web web
...
...
321 121

311 312 ... 111 112 ...
...
NTHICK2 412 Bottom flange

411 ...
421
NWIDTH
Figure 2.4-22: Numerical integration in the box cross-section
2.4.6.5 I cross-section
Figure 2.4-23 shows the I cross-section and its dimensions.

dimensions.
The I cross-section can be used in both linear elastic and large

2.4: Beam elements
displacement elastic analysis. In addition, the I cross-section can be

used in MNO and large displacement elastic-plastic analysis.
Table 2.4-5 summarizes the choices available for the I cross-

section.
WIDTH 2
(H3)
THICK3 THICK 2
(H6) (H5)
t
H8 H7
HEIGHT
(H2) s
THICK1 t -ct
(H4)
s H10
H16
WIDTH1
(H1)
Figure 2.4-23: I cross-section

Chapter 2: Elements
Table 2.4-5: Options for the I cross-section

beam, V89 beam
assumptions1
Shear center At neutral axis, At shear center At shear center
shear center
effects are
ignored
Formula for F1, F2, FW, F1, F2, FW, FW
calculation of default F1 default FW
torsional rigidity
(elastic material)
I included
allowed under
large
displacement
conditions
model allowed
1) CROSS-SECTION I ASSUMPTIONS=V89
can be used in conjunction with the I cross-section. However, it is
not recommended that the standard beam be used with the I cross-
section. In order to use the standard beam with the I cross-section,
set CROSS-SECTION I ... STANDARD=YES (default is NO).
Offsets to the shear center are not included. Therefore

transverse forces applied at the nodes are assumed to be acting
through the beams shear center and hence cause no twisting.

2.4: Beam elements


Offsets to the shear center are included.
analysis.
General comments
The calculations of I and ct are made to be consistent with

the calculations done in the plastic material (in which all integrals
are evaluated numerically from the warping function). Hence the
values of I , ct might not be the same as the handbook values of
these quantities, particularly if the walls of the cross-section are
thick.

CROSS-SECTION I ... WAGNER={DEFAULT, YES, NO}.
ASSUMPTIONS=V89
The torsional rigidity is controlled with CROSS-SECTION I ...

TORFORMULA={DEFAULT, F1, F2, FW}.

1
J=
3
( H43 H1 + H63 H3 + H53 H7 )
TORFORMULA=F2 (alternate textbook formula):
1
J=
3
( H43 H1 + H63 H3 + H53 H8 )

Chapter 2: Elements

function):
1
J=
3
( H43 H1 + H63 H3 + H53 H7 ) (same formula as F1)
beam is used with ASSUMPTIONS=V89, use formula F1 if the
standard beam is used with ASSUMPTIONS=V89.
These formulas assume that the walls of the I section are thin,
and then these formulas are numerically almost equal to each other.

section.

cross-section.
point is in the top flange of the cross-section.

2.4: Beam elements
NFLANGE2,
...
WIDTH2
...
312 NTHICK3,
Top flange
THICK3
311 ...
321
NTHICK2, THICK2
Flanges and web Integration points

separated for clarity Web evenly spaced
NWEB
...
221

211 212 ...
...
NTHICK1,
112 Bottom flange
THICK1
111 ...
121
NFLANGE1,
WIDTH1
Figure 2.4-24: Numerical integration in the I cross-section
2.4.6.6 U cross-section
Figure 2.4-25 shows the U cross-section and its dimensions.

dimensions.
The U cross-section can be used in both linear elastic and large

displacement elastic analysis. In addition, the U cross-section can
be used in MNO and large displacement elastic-plastic analysis.
Table 2.4-6 summarizes the choices available for the U cross-

section.

Chapter 2: Elements
THICK 2 H9 H13
(H4)
t cs t
s s
HEIGHT H5 H6
(H2)
THICK 1
(H3)
H7
H8
WIDTH
(H1)
Figure 2.4-25: U cross-section

2.4: Beam elements
Table 2.4-6: Options for the U cross-section

beam, V89 beam
assumptions1
shear center
effects are
ignored
torsional rigidity
(elastic material)
I included
allowed under
large
displacement
conditions
model allowed
1) CROSS-SECTION U ASSUMPTIONS=V89
can be used in conjunction with the U cross-section. However, it is
not recommended that the standard beam be used with the U cross-
section. In order to use the standard beam with the U cross-section,
set CROSS-SECTION U ... STANDARD=YES (default is NO).


Chapter 2: Elements


analysis.
General comments
The calculations of I and cs are made to be consistent with

values of I , cs used in the elastic material might not be the same
as the handbook values of these quantities, particularly if the walls
of the cross-section are thick.

CROSS-SECTION U ... WAGNER={DEFAULT, YES, NO}.
ASSUMPTIONS=V89.
The torsional rigidity is controlled with CROSS-SECTION U ...


1
J=
3
( 2 H43 H1 + H33 H6 )
1
J=
3
( 2 H43 H8 + H33 H5 )

2.4: Beam elements

function):
1
J=
3
( 2 H43 H7 + H33 H2 )
These formulas assume that the walls of the U section are thin,

section.

cross-section.
point is in the top flange of the cross-section.

Chapter 2: Elements
...
NFLANGE NTHICK1
...
NTHICK2 312 Top flange

311
321 ...
Flanges and web

separated for clarity
Integration points Web NHEIGHT

evenly spaced
...
221
...
NTHICK2 112 Bottom flange

111 ...
121 211 212 ...
NFLANGE
Figure 2.4-26: Numerical integration in the U cross-section
2.4.6.7 L cross-section
Figure 2.4-27 shows the L cross-section and its dimensions.

dimensions.
The L cross-section can be used in both linear elastic and large

displacement elastic analysis. In addition, the L cross-section can
be used in MNO and large displacement elastic-plastic analysis.

2.4: Beam elements
Table 2.4-7 summarizes the choices available for the L cross-

section.
Table 2.4-7: Options for the L cross-section

beam, V89 beam
assumptions1
Orientation of Principal Horizontal leg Horizontal leg
cross-section directions aligned with s aligned with s axis
aligned with s-t axis
axes
shear center
effects are
ignored
torsional rigidity
(elastic material)
I included
allowed under
large
displacement
conditions
model allowed
1) CROSS-SECTION L ASSUMPTIONS=V89
can be used in conjunction with the L cross-section. However, it is
not recommended that the standard beam be used with the L cross-
section. In order to use the standard beam with the L cross-section,
set CROSS-SECTION L ... STANDARD=YES (default is NO).

Chapter 2: Elements
THICK 1
(H3)
HEIGHT H8 H6
(H2)
Cx
Cy
THICK 2
(H4)
H5
H7
WIDTH
(H1)
a) Section dimensions
H19
H15
t y
t
s t -c
q s
x s
-ct
H20
s
H16
a) Principal axes aligned with s, t axes c) Legs aligned with s, t axes
Figure 2.4-27: L cross-section

2.4: Beam elements
The principal axes are aligned with the s-t axes.


The horizontal leg is aligned with the s axis; the vertical leg is
aligned with the t axis.
Torsional rigidity J calculated by formula FW by default
analysis.
General comments
The calculations of I and cs , ct are made to be consistent with

values of I , cs , ct used in the elastic material might not be the
same as the handbook values of these quantities, particularly if the
walls of the cross-section are thick.

CROSS-SECTION L ... WAGNER={DEFAULT, YES, NO}.
ASSUMPTIONS=V89.

Chapter 2: Elements
The torsional rigidity is controlled with CROSS-SECTION L ...


1
J=
3
( H33H2 + H43H5)
1
J=
3
( H33H8 + H43H7 )
function):
1
J=
3
( H33H2 + H43H5) (same formula as F1)
These formulas assume that the walls of the L section are thin,
When the principal axes are aligned with the s and t axes, the
maximum and minimum principal bending moments of inertia
(about the s and t axes) are obtained as follows:
2
I xx + I yy I xx I yy
I ss = + + I xy
2
2 2
2
I xx + I yy I xx I yy
I tt = + I xy
2
2 2
where the moments and products of inertia with respect to the local
axes x and y are:

2.4: Beam elements
2 2
H4 H13 H1 H6 H3
3
H3
I xx = + H1H4 Cx + + H6 H3 Cx
12 2 12 2
2 2
H1H43 H4 H3H6
3
H6
I yy = + H1H4 Cy + + H5 H3 H4 + Cy
12 2 12 2
H1 H4 H3 H6
I xy = H1H4 Cx C y + H6 H3 Cx H4 + Cy
2 2 2 2
and where (Cx, Cy) are the centroid coordinates, measured from the
bottom of the section. In this case, the angle between the s principal
axis and the horizontal is determined to maximize the following
expression: I ss = I xx cos 2 + I yy sin 2 I xy sin 2 . The result is
( (
= 12 atan2 I xy , 12 ( I yy I xx ) ))
This angle is shown in Fig 2.4-27.
When the legs are aligned with the s and t axes, then = 0 and
I ss = I xx , I tt = I yy , I st = I xy

section. When the horizontal leg is aligned with the s axis, Ist is
also adjusted.


Chapter 2: Elements
cross-section.
NTHICK1
Horizontal and vertical

leg separated for clarity
NLEG1 Vertical
leg Integration points
evenly spaced
...
221
...
112 Horizontal leg NTHICK2

111 ...
211 212 ... 121
NLEG2
Figure 2.4-28: Numerical integration in the L cross-section


2.4: Beam elements

point is in the vertical leg of the cross-section.
2.4.7 Moment-curvature beam element
In practical engineering analysis, the data available for the

description of the behavior of beam members may be given only in
the form of relationships between bending moment and curvature,
and between torsional moment and angle of twist. ADINA offers
the capability of directly using these data without having to define
an "equivalent" stress-strain law and the exact beam cross-sectional
shape.
This element is suitable for modeling nonlinear elastic and
elasto-plastic beam problems involving arbitrary cross-sections,
especially cross-sections that are neither rectangular nor circular.
The element can be used either with the small displacement

formulation (in which case the formulation is materially-nonlinear-
only) or with the large displacement formulation (in which case the
displacements/rotations can be large). In all cases, the strains are
assumed to be small.
In the moment-curvature input, it is assumed that both the

centroid and the shear center of the beam cross-section lie on the r-
axis of the element: transverse forces applied to the element cannot
generate twisting (as with a shear center offset) and axial forces
applied to the element cannot generate bending (as with a centroid
offset).
The flexural and the torsional behavior of the beam element are
respectively described by "bending moment vs. curvature" and
"torsional moment vs. angle of twist" relationships. These
relationships are input in ADINA in the form of multilinear
functions (see Figure 2.4-29).

Chapter 2: Elements
Typical beam data set: ADINA input:

Moment Moment
Fi+1
Fi
Fi-1
Fi axial force
(can be positive
or negative)
Curvature Fi-1
(or twist angle per unit length)
Fi Curvature
Fi+1 (or twist angle
per unit length)
Axial force
Figure 2.4-29: Input curves for the moment-curvature models (curves are
shown only for positive forces, moments and curvatures/twists)
The "bending moment vs. curvature" and "torsional moment vs.

angle of twist" relationships are functions of the axial force. Note
that the axial force is positive when the element is in tension and is
negative when the element is in compression. The sign
conventions used for moments and torsion is that used for local
node 2 in Figure 2.4-2.
The flexural behavior of the beam element is defined by two

"bending vs. curvature" relationships, one for each principal plane
1
of inertia. Curvature is defined as = where is the radius

of curvature. For a linear element (linear elastic material, small
displacements/small strains), the relationship between bending
M
moment and curvature is = EI .

The input for the torsional behavior of the element is similar to

that for bending: A multilinear function is used to express the
torsional moment in terms of the angle of twist per unit length.
Note that the beam element does not include any shear

2.4: Beam elements
deformation when the moment-curvature description is used.
Note also that warping effects must be taken into account in the
definition of the "torsional moment vs. twist" input curve if
necessary.
For computation of the beam element stiffness matrix and

internal force vector, numerical integration (Newton-Cotes) is only
used for integration along the length of the element (r-axis) and no
integration is needed over the cross-section (s-axis and t-axis).
Thermal effects can be included in moment-curvature relations.
Nonlinear elastic model: A nonlinear elastic model can be used

(see Figure 2.4-30). In this case, the behavior for negative
curvatures (resp. twist angles) may be different than the behavior
for positive curvatures (resp. twist angles). The axial force versus
axial deformation relationship is always linear elastic.
Note that the last segments of the moment-curvature curve are
extrapolated if necessary, in order to calculate the moment when
the curvature or twist angle per unit length is out of the input curve
range. The end points of the curve do not represent rupture points.

Chapter 2: Elements
ADINA input for a given axial force:

Moment Extrapolation used in ADINA
Last input
data point
Curvature
(or twist angle per
unit length)
First and last data points

First input are not rupture points
data point
Figure 2.4-30: Nonlinear elastic moment-curvature beam input
Elastic-plastic model: For the analysis of beam members

undergoing plastic deformations, ADINA offers bilinear and
multilinear plasticity. Hardening can be linear isotropic, linear
kinematic or linear mixed, as shown in Figure 2.4-31.

2.4: Beam elements
y
yyield
0
yyield
2y yield
0
x
2y yield
Kinematic hardening
Mixed hardening
Isotropic hardening
(x, y): (axial strain, axial force), or

(curvature, bending moment) for a given axial force, or
(angle of twist/unit length, torsional moment) for a given
axial force
Figure 2.4-31: Hardening models for moment-curvature beams
The moment-curvature plasticity model consists of uniaxial

plasticity laws respectively applied to the axial strain, each bending
curvature, and the twist angle per unit length:
Axial force/axial strain relationship: The relationship can either

be symmetric or non-symmetric with respect to the sign of the
axial strain.
Bending moment/curvature relationship: The relationship can

either be symmetric or non-symmetric with respect to the sign
of the curvature. The relationship can depend on the axial force
and can be different in axial tension and axial compression.
Torsional moment/twist angle per unit length relationship: The

relationship can either be symmetric or non-symmetric with
respect to the sign of the twist angle per unit length. The

Chapter 2: Elements
relationship can depend on the axial force and can be different

in axial tension and axial compression.
In the symmetric case (Figure 2.4-32), enter only the positive

section of the axial force/axial strain, bending moment/curvature or
torsional moment/twist angle curve. The first data point always
corresponds to yielding and the last data point always corresponds
to rupture.
y
yrupture
Rupture:
last input
Slope must data point
decrease as
x increases
yyield
Yielding:
First input
data point
x
x yield xrupture
See Figure 2.4-31 for definition of (x, y)
Figure 2.4-32: Symmetric elasto-plastic moment-curvature beam input
In the non-symmetric case (Figure 2.4-33), enter the entire axial

force/axial strain, bending moment/curvature or torsional
moment/twist angle curve. The first data point always corresponds
to negative rupture and the last data point always corresponds to
positive rupture. One data point - the zero point - must be at the
origin. The data point prior to the zero point corresponds to the
negative yield and the data point after the zero point corresponds to
the positive yield. A different number of data points can be used
for the positive and negative sections of the curve.

2.4: Beam elements
y
ypositive
rupture
Curve cannot Slope must

pass through decrease as
y positive x increases
this quadrant
yield
One point must

be at the origin
xnegative x negative x
rupture yield x positive
x positive
yield rupture
ynegative
yield
Curve cannot
pass through
this quadrant
ynegative
rupture
First point: negative rupture point

Point before origin: negative yield point
Point after origin: positive yield point
Last point: positive rupture point
Figure 2.4-33: Non-symmetric elasto-plastic moment-curvature beam input
To obtain bilinear plasticity, define a multilinear plasticity curve

with only two segments.
Bending moment and torsion relationships can depend on the

axial force and this dependence can be different in tension and in
compression.
! The input for the bending moment/curvature and torsional

moment/twist curves consists of bending moment/curvature curves
for different levels of axial force. In the following, we discuss the
calculation of the bending moment/curvature curve for a level of
axial force that is not input (the calculation of the torsional
moment/twist curve is similar).

Chapter 2: Elements
To obtain the bending moment/curvature curve for a level of

axial force not input, interpolation is used. This interpolation is
performed, not on the bending moment/curvature curves, but on the
bending moment/plastic curvature curves; the bending
moment/plastic curvature curves are automatically calculated from
the bending moment/curvature curves.
The idea is exactly the same as is used in thermo-plasticity. In
thermo-plasticity, the input is a set of stress-strain curves for
different temperatures. Interpolation is used to obtain the stress-
strain curve for a given temperature. And this interpolation is
performed on the stress - plastic strain curves, and not on the stress-
strain curves. (See Fig 3.6-3 in Section 3.6.2.)
Because interpolation is performed on the bending
moment/plastic curvature curves instead of the bending
moment/curvature curves, care should be taken to enter enough
bending moment/curvature curves for different axial force levels,
especially when the plastic curvature is small for one of the
bending moment/curvature curves.
When the response is elastic, the interpolation is performed on
the bending moduli.
The same element section can be plastic with respect to the axial
deformation, but still elastic with respect to bending or torsion.
The same remark applies for rupture.
Bending about the two axes is treated independently. Also, there

is no interaction between bending and twist. As such, the same
element can be plastic in bending about one axis while still elastic
with respect to bending about the other axis, or with respect to
torsion.
Rupture depends on the value of the accumulated effective

plastic axial strain, accumulated effective plastic curvature or
accumulated effective plastic angle of twist per unit length.
To trigger the rupture calculation of the moment curvature

element, the following conditions must be simultaneously met:
1. number of force-strain points > 2 (asymmetric)

2. number of twist-moment points > 2 (asymmetric)
3. number of curvature-moment points > 2 (asymmetric)

2.4: Beam elements
Typical data curve: y
yi : initial elastic rigidity yu

yi
yu : cyclic elastic rigidity

ADINA result (shown here for the bilinear case):
y
y
yu
yi
yu
>1
yu yi
yi
x
Plastic strain
Figure 2.4-34: Typical material curve with a cyclic rigidity different than
the initial rigidity
The elastic-plastic model includes a special option for cyclic

behavior, in which the elastic rigidities when unloading first starts
differ from the initial rigidities (see Figure 2.4-34). With this
option, the elastic rigidity (in bending, torsion or axial deformation)

Chapter 2: Elements
changes as soon as the yield stress is reached. The ratio of the new
cyclic elastic rigidity to the initial elastic rigidity is defined by user
input. Once updated, the elastic rigidity remains constant for the
rest of the analysis, for loading and unloading.
In moment-curvature models, the cross-sectional area and the

cross-sectional moments of inertia need to be input for the mass
matrix computation (in case of mass proportional loading or
dynamic analysis). They are used only for mass matrix
computation.
The results can be obtained in the form of element nodal forces

and moments, or in the form of stress resultants given at the
integration point locations on the element.
Stress resultants: Each element outputs, at the integration point

locations on the element r-axis, the following information to the
porthole file, based on the material model. This information is
accessible in the AUI using the given variable names.
The integration point locations on the element r-axis are shown in
Figure 2.4-35. In the AUI, these integration point locations are
considered to be section points, not element result points (see Section
13.1.1).
Force-r
Node 2 r
Int pt 2
Int pt 1
s
Node 1
Moment-r
t Moment-t
Moment-s
Force-r is positive in tension.

Force-s, Force-t are not computed.
Figure 2.4-35: Stress resultant output option for

the moment-curvature models
The results are always given in the element local coordinate system

2.4: Beam elements
(r,s,t). Note that a positive value of the Moment-s at a section point

location points in the negative s-direction.
Nonlinear elastic: AXIAL_FORCE, TORSIONAL_MOMENT,

BENDING_MOMENT-S, BENDING_MOMENT-T,
AXIAL_STRAIN, TWIST, CURVATURE-S, CURVATURE-T
Plastic-multilinear: PLASTIC_FLAG_AXIAL,
PLASTIC_FLAG_TORSION, PLASTIC_FLAG_BENDING-S,
PLASTIC_FLAG_BENDING-T, AXIAL_FORCE,
TORSIONAL_MOMENT, BENDING_MOMENT-S,
BENDING_MOMENT-T, AXIAL_STRAIN, TWIST,
CURVATURE-S, CURVATURE-T,
PLASTIC_AXIAL_STRAIN, PLASTIC_TWIST,
PLASTIC_CURVATURE-S, PLASTIC_CURVATURE-T,
ACCUM_PLASTIC_AXIAL_STRAIN,
ACCUM_PLASTIC_TWIST,
ACCUM_PLASTIC_CURVATURE-S,
ACCUM_PLASTIC_CURVATURE-T, YIELD_AXIAL_FORCE,
YIELD_TORSIONAL_MOMENT,
YIELD_BENDING_MOMENT-S,
YIELD_BENDING_MOMENT-T
Note that in a materially nonlinear analysis, the element nodal

moments are in general not equal to the moment stress resultants at
the integration points located at the nodes. The reason for this is
that in such cases the beam element does not exactly satisfy the
element internal equilibrium on the differential level; that is, the
element is a "true finite element". The differential equilibrium is, of
course, satisfied more accurately as the finite element mesh is
refined.
Transverse forces are not computed at integration points since

the ADINA beam element formulation uses the equilibrium
equations of the complete element to directly determine the
transverse forces at the element end nodes.
In the case of beam elements with rigid ends, the element stress
resultants are computed at the integration point locations
corresponding to the flexible part of the element, excluding the
rigid ends.

Chapter 2: Elements
Nodal point forces: Same as for the elastic beam element.
2.4.8 Additional features
End-release option: To model beam internal hinges, a moment and

force release option can be used (see Fig. 2.4-36). Twisting
moments, axial forces and bimoments (for warping beams) can also
be released.
Shear force
Internal hinge released
l l
r r r r
Element 1 Element 2 Element 1 Element 2
Specify moment release at Specify shear force release at

element 1, local node 2, or element 1, local node 2, or
element 2, local node 1 element 2, local node 1
(a) Moment to be released at (b) Shear force to be released

internal hinge
Figure 2.4-36: Use of moment and shear force release options
A brief description of the theory of end-releases follows:
For static linear elastic analysis without warping degrees of

freedom, the stiffness matrix and internal force vector for a beam
element can be written as
Ku = F
In this expression, u contains the displacements and rotations of

the beam nodes, in the beam local coordinate system:
uT = u1r u1s ut1 r1 s1 t1 ur2 us2 ut2 r2 s2 t2 ,
F contains the forces and moments of the beam nodes,

2.4: Beam elements
FT = [ S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12 ]
and K is the stiffness matrix. These expressions use the same

notations as Figures 2.4-1 and 2.4-2.
Now suppose that one or more of the local displacements or

rotations is to be released. Releasing means that the
corresponding force is set equal to zero. For example, if the axial
displacement at local node 2 is to be released, then force S7 is set
to zero.
In order to release the selected local displacements or rotations, the

following procedure is used:
1) u is rearranged and partitioned: uT = uTA uTB , so that u A

contains all of the local displacements / rotations to be retained, and
u B contains all of the local displacements / rotations to be released.
For example, if the axial displacement at local node 2 is to be
released, then
uTA = u1r u1s ut1 r1 s1 t1 us2 ut2 r2 s2 t2

uTB = ur2
2) F and K are similarly rearranged and partitioned. The

resulting system of equations is
K AA K AB u A FA
KT =
AB K BB u B FB
3) The end release condition is now expressed as FB = 0 , and the

resulting system of equations becomes

Chapter 2: Elements
K AA K AB u A FA
KT =
AB K BB u B 0
4) This system of equations is satisfied by choosing
(K AA K AB K BB1 K TAB ) u A = FA
u B = K BB1 K TAB u A
Therefore the stiffness matrix that is assembled into the global

system of equations is K AA K AB K BB1 K TAB .
Any external forces acting onto the released degrees of freedom

are ignored.
When the warping beam is used the warping degrees of freedom

can also optionally be released, using the same theory as described
above, and the result is that the corresponding bimoments are equal
to zero.
The end release procedure is also implemented for nonlinear

analysis, both for materially nonlinear elements and also for
geometrically nonlinear elements.
The end release procedure is available for static, frequency,

implicit dynamic and explicit dynamic analysis. However the end
release procedure cannot be used in conjunction with modal
superposition.
End releases only affect the stiffness matrix and force vector,
not the mass matrix. Therefore inertial forces and moments (forces
and moments due to mass matrix effects) are not released in the
above procedure.
If end releases are used in frequency or dynamic analysis, the

elements in which end releases are specified should be very short,
in order to minimize the inertial forces and moments.
End releases are applied to the element local nodes (not to the

2.4: Beam elements
global nodes). Therefore, to model the hinge shown in Fig. 2.4-

36(a), a moment end release can be applied to local node 2 of
element 1, or to local node 1 of element 2 (but not to both local
nodes).
The command used to specify end releases is ENDRELEASE.
For example, the input corresponding to Fig 2.4-36(a) might be
ENDRELEASE 1 12
EDATA
ENTRIES EL ENDRELEASE
1 1
DATAEND
This command releases the S12 force (t bending moment at local

node 2) of element 1.
In large displacement / large rotation analysis, the global

directions corresponding to the released degrees of freedom change
as the model deforms. By default, this effect is fully included in the
end release calculations. However, this effect can occasionally slow
down convergence. For this reason, this effect can be controlled
using the options EGROUP BEAM
ENDRELEASE=(ACCURATE, APPROX1, APPROX2).
If ENDRELEASE=ACCURATE (the default), the end releases
are evaluated in a local coordinate system corresponding to the
configuration of the beam in the previous equilibrium iteration.
Therefore, at convergence, the end releases are evaluated in
(almost) the current configuration of the beam. (At convergence,
the difference between the current equilibrium iteration and the
previous equilibrium iteration is very small.)
If ENDRELEASE=APPROX1 or APPROX2, the end releases
are evaluated in a local coordinate system corresponding to the
configuration of the beam in the previous time step. Notice that the
local coordinate system does not change directions during the
equilibrium iterations. In this case, the released forces/moments
are not exactly zero at convergence, but these released
forces/moments should be small, especially if the time step size is
small. Compared to APPROX1, APPROX2 includes an additional
approximation that reduces accuracy, but improves convergence.
Thus ACCURATE is more accurate than APPROX1 and
APPROX1 is more accurate than APPROX2; however APPROX2

Chapter 2: Elements
converges better than APPROX1 and APPROX1 converges better

than ACCURATE.
In explicit dynamic analysis, the use of end releases will cause

the program to run considerably more slowly than if this option
were not used.
Rigid-end option: For the modeling of assemblages of braces and

struts, it is sometimes of interest that the joints of the structure be
considered as infinitely rigid (see Figure 2.4-37) . This condition
can be modeled using the rigid-end option: with this option, a
fraction of each end of the element can be considered as rigid.
End 1 is the end at node 1, end 2 is the end at node 2. It is
allowed for end 1 only to be rigid, end 2 only to be rigid, or both
ends to be rigid.
The lengths of the beam rigid ends are specified in the units of
the model.
One possible command input for the beam shown in Fig 2.4-38
is
EDATA
ENTRIES EL RIGID1 RIGID2
1 0.2 0.4
DATAEND
In this example, the length of rigid end 1 is 0.2 units, and the length
of rigid end 2 is 0.4 units.
Physical problem: Finite element model:
Beam element
Rigid end
Figure 2.4-37: Modeling of a beam intersection using

elements with rigid ends

2.4: Beam elements
0.2 0.4
Local node 1 Local node 2
Figure 2.4-38: Example of beam element rigid ends input
The assumptions of the rigid-end option are as follows:
In ADINA 8.8 and later, the rigid ends are assumed to be

perfectly (infinitely) rigid. In ADINA 8.7 and earlier, the rigid
ends are either assumed to be perfectly rigid, or only very stiff
(with rigidity calculated by multiplying the rigidity of the
element by a large number, such as 106).
The element, including its rigid ends, is formulated as one
single element. Hence, as for all beam elements, the internal
deformations and element internal forces are all referred to a
local coordinate system based on the end-nodes, and using the
assumptions of small internal relative deformations.
The rigid ends will never undergo plasticity. Hence, although

the stress in the rigid ends may be very high (they are not
printed by the program), plasticity is not reached. This may be
quite unrealistic in a model using the rigid-end capability.
The mass of the rigid ends is taken into account in the mass
matrix calculations.
Rigid ends can be used in all beam analyses.
When using warping beams with rigid ends, the warping degree
of freedom should be fixed at the rigid end.
2.4.9 Beam element modeling hints
For modeling tapered beams or curved beams, the user needs to

divide such members into several elements. For the tapered beams
the user needs to divide the beam into several elements and use an
appropriate constant cross-section for each of these elements.

Chapter 2: Elements
Off-centered beam elements can be created using rigid links

(see Fig. 2.4-39 and Section 5.15.2).
Physical problem: Finite element model:

Rigid panel
I-beam
I-beam Beam elements
Beam element
Rigid links
Hollow square section
Figure 2.4-39: Use of rigid links for modeling of off-centered beams
In order to model the bending due to an off-centroidal axial

force or transverse load, either apply the resulting moments directly
or apply the forces at an offset location using rigid links (Figure
2.4-40).
Beam
C.G. C.G.
Rigid link
A
A
P P
Figure 2.4-40: Using rigid links to transfer loads to the centroid of the beam
When solving problems involving flexural buckling, it may be

necessary to use a refined mesh. This is because the only second-
order strain terms included in the formulation are the terms needed

2.4: Beam elements
to model the Wagner effect. Second-order strain terms

corresponding to flexural buckling effects are not included in the
formulation. However, by refining the mesh, it is possible to obtain
accurate flexural buckling effects.
In the moment-curvature beam element, when using

nonsymmetric relationships, such as a nonsymmetric axial force -
axial strain relationship, it is helpful to apply a very small load in
the first step.
Consider a scenario in which the axial force - axial strain curve
has different moduli for tensile and compressive strains. At the
beginning of the first step, the strain is zero, and the program uses
the tensile modulus in this case. However, if the loading is
compressive, then the "wrong" modulus is used, and if the load is
large enough, the subsequent iterations do not converge. On the
other hand, if the load is very small, then even though the "wrong"
modulus is used in the first iteration, the response remains elastic,
the correct modulus is used thereafter and the iterations converge.
After the first step, the correct modulus is used, so the load
increment can be large after the first step.
2.4.9.1 Hints concerning the warping degrees of freedom
The effects of fixing the warping can be significant for thin

open section beams.
When defining the fixity for the warping beams, it is important

to know what type of physical condition leads to fixed warping in a
support and when the warping at a support is free. As an example,
when only the web of an I-beam is connected to the column while
its flanges are free to move out of their original plane, warping is
free. However, if the flanges of the beam are welded to the column,
they cannot move out of plane and warping is fixed.
For a structural connection where several beams and columns

are connected at a point it is important to know how to assign the
warping degree of freedom. Fig. 2.4-41 shows one such example.
As can be seen, the warping degree of freedom is continuous for
the columns. However, warping for the beams at the point of
connection to column should be set to zero as their flanges are
welded to the column and cannot move out of plane. As such, the

Chapter 2: Elements
user must define multiple nodes at the point of connection, assign

the warping degree of freedom for each node separately and
establish proper constraint equations between these multiple nodes
to ensure that they move and rotate together.
Also, in modeling curved members using straight warping

beams special care should be taken regarding the continuity of the
warping degrees of freedom. The angle between any two
consecutive elements should be small to ensure the continuity of
warping displacements (Fig. 2.4-42).
Column
Weld
Weld
Beam Beam
(a) Actual beam-column connection
Multiple nodes
Warping is continuous
Warping fixed Warping fixed
(b) Warping beam representation
Figure 2.4-41: Modeling of several warping beams connected at a sharp angle

2.4: Beam elements
q2 q3
q1
q i small
Warping displacements
Figure 2.4-42: Discretization of a curved member using

straight warping beams
In general, continuity of the warping deformation at a point

where several members meet not only depends on the angle at
which these members meet but also on the details of the
arrangement of the connector elements (bolts, welds, etc.)
Warping degrees of freedom cannot be prescribed. In other

words, warping degrees of freedom can only be fixed or free.
Warping degrees of freedom cannot be used in constraint

equations or rigid links.
Bimoments cannot be specified as nodal loads.

Chapter 2: Elements
2.5 Iso-beam elements; axisymmetric shell elements

The iso-beam elements (which include the axisymmetric shell

elements) can be employed in the following forms, see Fig. 2.5-1:
plane stress 2-D beam with three degrees of freedom per node,
plane strain 2-D beam with three degrees of freedom per node,
axisymmetric shell with three degrees of freedom per node,
general 3-D beam with six degrees of freedom per node.
The plane stress 2-D beam element is identical to the 3-D beam
element but constrained to act only in the YZ plane. Hence, all
motion of the 2-D plane stress beam element must occur in the YZ
plane (see Fig. 2.5-2).
The difference between the plane strain and the plane stress 2-D
beam elements is that for the plane strain element, it is assumed
that the out-of-plane strain xx is equal to zero whereas the out-of-
plane stress xx is equal to zero for the plane stress element.
Note that it can be significantly more effective to use the 2-D

beam option instead of the general 3-D beam option, since then the
numerical integration is only performed in two dimensions and the
number of degrees of freedom is also reduced.

2.5: Iso-beam elements; axisymmetric shell elements
Examples of general 3-D iso-beam elements:
Axisymmetric shell:
Z
1 radian
Plane stress iso-beam: Y

Plane strain iso-beam:
Z l
Z l
l
sxx=0 l exx=0
l
l
l Y
X l Y
X
Figure 2.5-1: 2-D iso-beam, axisymmetric shell, and 3-D iso-beam elements

Chapter 2: Elements
s
r 2
l
1l
Z Element has constant
rectangular cross-section
Y Element can have 2, 3

or 4 nodes
Auxiliary node is not used,
set the auxiliary node to 0.
a) 2-node plane stress or plane strain beam element
1
l
3 s
Axis of l
symmetry r
Z 4
Element has varying l
thickness cross-section
l
Y Element can have 2, 3
or 4 nodes 2
Auxiliary node is not used,

set the auxiliary node to 0.
b) 4-node axisymmetric shell element
Figure 2.5-2: Local node numbering; natural coordinate system
The axisymmetric shell element formulation is an extension of

the 2-D beam element formulation in that the axisymmetric hoop
stress/strain components are included in the model. The
axisymmetric shell element must be defined in the YZ plane, and
lie in the +Y half plane.
The cross-sectional areas of each of these elements are assumed

to be rectangular. The 2-D and 3-D beam elements can only be
assigned a constant area cross-section. The axisymmetric shell
element can be assigned a varying thickness.

Auxiliary node r-s plane
s 2 2-node iso-beam
1 r
Elements have constant

3 2
1
r-s plane
3-node iso-beam
Auxiliary node
4 2
3
r-s plane
1
Auxiliary node
4-node iso-beam
c) General 3-D isoparametric beam elements.
The elements can be employed with 2, 3 or 4 nodes. The 3 and

4-node elements can be curved, but it should be noted that the
element nodes must initially lie in one plane (which defines the r-s
plane).
To model the torsional stiffness of the 3-D beam element

accurately, the warping behavior is represented as described in the
following reference:
ref. K.J. Bathe and A. Chaudhary, "On the Displacement

Formulation of Torsion of Shafts with Rectangular
Cross-Sections", Int. J. Num. Meth. in Eng., Vol. 18, pp.
1565-1568, 1982.

Chapter 2: Elements
Note that the formulation directly models shear deformations in

an approximate manner; namely, the shear deformations are
assumed to be constant over the cross-section (see ref. KJB, Fig.
5.18, p. 398). This corresponds to using shear factors equal to 1.0.
The elements can be employed to model thin and thick beams and
shells.
Some applications using the element are published in the

following reference.
ref. K.J. Bathe and P.M. Wiener, "On Elastic-Plastic

Analysis of I-Beams in Bending and Torsion",
Computers and Structures, Vol. 17, pp. 711-718, 1983.
The isoparametric beam element is available in ADINA

primarily to model
Curved beams
Stiffeners to shells, when the shell element (described in

Section 2.7) is used
Beams in large displacements
Axisymmetric shells under axisymmetric loading
Considering the analysis of thick curved beams, in order to

represent the hyperbolic stress distribution through the thickness,
the higher order elements (3 or 4-node elements) should be used.
This is because the nodal director vectors are constructed from the
geometry of the elements.
ref. KJB The element matrices and vectors are formulated using the
Sections 5.4.1 isoparametric interpolation, and Gauss or Newton-Cotes numerical
and 6.5.1 integration is used to evaluate these matrices in all analyses. For the
2-D beam and axisymmetric shell elements, numerical integration
is only performed in the r-s plane, hence these elements are
considerably less expensive in terms of computer time than the
general 3-D element. The locations and labeling of the integration

points are illustrated in Fig. 2.5-3. INR is the index giving the
location of the integration point in the r direction, with INR=1
corresponding to the lowest value of r, and INR=(integration point
order) corresponding to the highest value of r. INS and INT are
analogous for the s and t directions.
The elements are based upon a mixed interpolation of

displacements and stresses.
31
32 31
r 21
r
33 32
22
23 21
s
22
s
12 11

13 12 11
label = 10(INR) + INS label = 10(INR) + INS
a) Newton-Cotes integration (3 x 3) b) Gauss integration (3 x 2)
Figure 2.5-3: Examples of integration point numbering

for 2-D beam, axisymmetric shell, and
3-D beam elements
Only the default integration order along the r-direction should

be used for the 2-, 3- and 4-node elements, i.e.:
2-node iso-beam: 1-point integration
These elements do not contain any spurious zero energy modes, do

not lock and are efficient in general nonlinear analysis.
If, however, the default r-integration order is not used, the

results of an analysis can change drastically, for example, if 2-point
integration along the r direction is specified for the 2-node element
model shown in Fig. 2.5-4, the resulting displacement () is 0.0126

Chapter 2: Elements
(i.e., locking is observed).
For the 3-D beam element, you can choose either 44 Gauss or
77 Newton-Cotes integration over the beam cross-section (along s
and t).
113

133
213 t 132

111 121

313 323
s
r
322
231

label = 100 (INR) + 10 (INS) + INT

311
331
c) Newton-Cotes integration (3 3 3)
t 112
122
212 111 121

312 s

r 221

321 label = 100(INR) + 10(INS) + INT
d) Gauss integration (3 2 2)
2.5.3 Linear iso-beam elements
ref. KJB The formulation of the element is presented in Chapter 5 of ref.

Section 5.4.1 KJB.

infinitesimally small and the elastic-isotropic material is used.

Physical problem: EI=105

P L=24
P=20
3
D th= PL = 0.9216
D 3EI
L
(theoretical solution,
no shear)
2-node Hermitian beam (no shear):
P
Element 1
l l
D = 0.9216
L D
2-node iso-beams (with shear):

P
Element 1 Element 2
l l l
D = 0.8648
L/2 L/2 D
4-node iso-beam (with shear):

P
Element 1
l l l l
D = 0.9224
L/3 L/3 L/3 D
Figure 2.5-4: Results of the analysis of a cantilever
Since the iso-beam element stiffness matrix is calculated using

numerical integration, it is clear that the 2-node Hermitian beam
element (see Section 2.4.1) is more effective when straight beam
members in linear elastic analysis are considered. Hence, for the
linear analysis of many structures (such as frames, buildings,
shafts) only the 2-node Hermitian beam element should be used.
Even when considering the linear analysis of a curved beam

member, it is frequently more cost-effective to model this beam as
an assemblage of straight 2-node Hermitian beam elements than to
use the curved iso-beam element.
Note that the iso-beam element can only be employed with a
rectangular cross-section, whereas the linear 2-node Hermitian

Chapter 2: Elements
beam element can be used for any cross-section which can be

defined by the principal moments of inertia, cross-sectional area
and the shear areas input to ADINA.
In linear analysis, the iso-beam element is primarily useful for

modeling stiffeners to shells when the 4-node or 8-node or 16-node
shell element (see Section 2.7) is employed. In this case the 2-, 3-
or 4-node iso-beam elements, respectively, together with the shell
elements can provide an effective finite element discretization of
the stiffened shell structure (see Fig. 2.5-5).
l l
l
l l
l
l Shell node
l Iso-beam node
a) Stiffened plate
l
l
l l
l l
l l l l
l
l l l
ll l
l l
l l
l l
l
b) Stiffened shell
Note: Select master nodes and use rigid links to tie

the shell nodes and the iso-beam nodes together.
Figure 2.5-5: Models of stiffened plate and stiffened shell
The solution results obtained using the isoparametric beam

elements are compared with those obtained using the 2-node

Hermitian beam elements in Fig. 2.5-4. Considering the tip loaded

cantilever shown, a 2-node Hermitian beam element representation
gives the "exact" analytical value of displacements and section
forces/moments (compared with Bernoulli-Euler beam theory).
The 2-node iso-beam element yields approximate results, while the
4-node iso-beam element is very accurate. However, as is typical
in finite element analyses, the discrepancy in the solution becomes
negligibly small when a sufficiently fine mesh is used.
The forces/moments at the element nodes are (in linear analysis)
F = KU
where is the element stiffness matrix and U is the vector of

nodal point displacements. Hence, at internal element nodes the
forces/moments in F will be equal to the forces/moments, applied
externally or by adjoining elements, and are not the internal section
forces/moments.
2.5.4 Nonlinear iso-beam elements
ref. KJB The formulation of the element is presented in Chapter 6 of ref.

Section 6.5.1 KJB.
The element can be used with the following material models:

elastic, plastic-bilinear, plastic-multilinear, thermo-isotropic,
thermo-plastic, creep, plastic-creep, multilinear-plastic-creep,
creep-variable, plastic-creep-variable, multilinear-plastic-
creep-variable, shape-memory alloy.
The element can be used either with a small displacement or a

large displacement formulation. In the large displacement
formulation, large displacements and rotations are allowed. In all
cases, only small strains are assumed.
All of the material models in the above list can be used with
either formulation. The use of a linear material with the small
formulation.

Chapter 2: Elements
The iso-beam element can be particularly effective in geometric

nonlinear analysis of straight and curved members, because the
change in geometry due to the large displacements is modeled
accurately.
ref. KJB The element matrices are evaluated using Gauss or Newton-
Section 6.6.3 Cotes numerical integration. In elastic-plastic analysis, the stress-
strain matrix is modified to include the effects of plasticity. This
stress-strain matrix is based on the classical flow theory with the
von Mises yield condition and is derived from the three-
dimensional stress-strain law with the appropriate stresses and
strains set to zero.
For the elasto-plastic and the thermo-elastic-plastic and creep

material models, the effective-stress-function algorithm of the
references given below is used.
ref. M. Koji and K.J. Bathe, "The Effective-Stress-Function

Algorithm for Thermo-Elasto-Plasticity and Creep", Int.
J. Num. Meth. Engng., Vol. 24, No. 8, pp. 509-532,
1987.
ref. M. Koji and K.J. Bathe, "Thermo-Elastic-Plastic and

Creep Analysis of Shell Structures", Computers &
Structures, Vol. 26, No 1/2, pp. 135-143, 1987.
2.5.5 Axisymmetric shell element
The axisymmetric shell element can be thought of as a special

case of the iso-beam formulation (in the same way as an
axisymmetric element is a special case of the 2-D solid element).
For the axisymmetric shell element, the hoop (circumferential)

stress and strain components are included in the iso-beam
formulation.
The use of this element can be significantly more effective than

using axisymmetric 2-D solid elements when the shell structure to
be modeled is thin.

One radian of the axisymmetric shell is modeled when using

this element (as with the 2-D axisymmetric solid element).
2.5.6 Element mass matrices
The iso-beam element can be used with a lumped or a consistent

uses a lumped mass.
ref. KJB The consistent mass matrix is calculated using the isoparametric
Section 5.4.1 formulation with the displacement interpolations given on p. 408 of
ref. KJB.

The lumped mass for degree of freedom i is M i , where M
L
is the total mass, L is the element length and i is a fraction of the
length associated with node i. The rotational mass is
1 1
M i ( b 2 + d 2 ) where b, d are the width and depth of the
3 L 6
L
cross-section. For a 2-node iso-beam element 1 = 2 = , while
2
L L
for a 4-node iso-beam element 1 = 2 = and 3 = 4 = (see
6 3
Fig. 2.5-6).
The rotational lumped mass can be multiplied by a user-

specified scalar ETA (except in explicit dynamic analysis).
You can request that ADINA either print or save stresses or

forces.


Chapter 2: Elements
variable names.
M M
2 2
l l
a) 2-node element
M M M
4 2 4 Note: M = total mass
l l l of element (2-D or 3-D
beam elements)
b) 3-node element
M M M M
6 3 3 6
l l l l
c) 4-node element
Figure 2.5-6: Construction of element lumped translational mass

matrix for the beam option of the iso-beam element
General 3-D iso-beam, elastic-isotropic material: STRESS-RR,

STRESS-RS, STRESS-RT, STRESS-SS, STRAIN-RR,
STRAIN-RS, STRAIN-RT, STRAIN-SS,
FE_EFFECTIVE_STRESS
Plane stress iso-beam, elastic-isotropic material: STRESS-RR,

STRESS-RS, STRESS-SS, STRAIN-RR, STRAIN-RS,
STRAIN-SS, FE_EFFECTIVE_STRESS
Plane strain iso-beam or axisymmetric shell, elastic-isotropic

material: STRESS-RR, STRESS-RS, STRESS-TT,
STRESS-SS, STRAIN-RR, STRAIN-RS, STRAIN-TT,
STRAIN-SS, FE_EFFECTIVE_STRESS
General 3-D iso-beam, elastic-isotropic material, thermal effects:

STRESS-RR, STRESS-RS, STRESS-RT, STRESS-SS,
STRAIN-RR, STRAIN-RS, STRAIN-RT, STRAIN-SS,
FE_EFFECTIVE_STRESS, THERMAL_STRAIN,
ELEMENT_TEMPERATURE

Plane stress iso-beam, elastic-isotropic material, thermal effects:

STRESS-RR, STRESS-RS, STRESS-SS, STRAIN-RR,
STRAIN-RS, STRAIN-SS, FE_EFFECTIVE_STRESS,
THERMAL_STRAIN, ELEMENT_TEMPERATURE
Plane strain iso-beam or axisymmetric shell, elastic-isotropic

material, thermal effects: STRESS-RR, STRESS-RS,
STRESS-TT, STRESS-SS, STRAIN-RR, STRAIN-RS,
STRAIN-TT, STRAIN-SS, FE_EFFECTIVE_STRESS,
Thermo-isotropic material: STRESS(RST),

FE_EFFECTIVE_STRESS, STRAIN(RST),
THERMAL_STRAIN(RST), ELEMENT_TEMPERATURE
Plastic-bilinear, plastic-multilinear materials: PLASTIC_FLAG,

STRESS(RST), STRAIN(RST),
PLASTIC_STRAIN(RST), THERMAL_STRAIN(RST),
ACCUM_EFF_PLASTIC_STRAIN, ELEMENT_TEMPERATURE
SMA: EFFECTIVE_STRESS, STRESS-RR, STRESS-RS,

STRESS-RT, STRESS-SS, STRESS-TT,
DETWINNED_MARTENSITE_FRACTION, STRAIN-RR,
STRAIN-RS, STRAIN-RT, STRAIN-SS, STRAIN-TT,
THERMAL_STRAIN-RR, THERMAL_STRAIN-SS,
THERMAL_STRAIN-TT, TRANSFORMATION_STRAIN-RR,
TRANSFORMATION_STRAIN-RS,
TRANSFORMATION_STRAIN-RT,
TRANSFORMATION_STRAIN-SS,
TRANSFORMATION_STRAIN-TT,
Thermo-plastic, creep, plastic-creep, multilinear-plastic-creep,

creep-variable, plastic-creep-variable, multilinear-plastic-creep-
variable materials: PLASTIC_FLAG,
NUMBER_OF_SUBINCREMENTS, STRESS(RST),
STRAIN(RST),YIELD_STRESS,
PLASTIC_STRAIN(RST),
ACCUM_EFF_PLASTIC_STRAIN, CREEP_STRAIN(RST),

Chapter 2: Elements
THERMAL_STRAIN(RST), ELEMENT_TEMPERATURE,
In the above lists,
STRESS(RST) = STRESS-RR, STRESS-RS,

STRESS-RT, STRESS-TT, STRESS-SS
with similar definitions for the other abbreviations used above.

ref. KJB Nodal forces: Nodal point forces are obtained using the relation
Section 6.3

t +t t +t t +t
F= BT dV
t +t
V
where t +t B is the strain-displacement matrix, the stresses are

stored in t +t , t +tV represents the volume, and the superscript
t + t refers to the conditions at time (load step) t + t .
These forces are accessible in the AUI using the variable names
NODAL_FORCE-R, NODAL_FORCE-S, NODAL_FORCE-T,
NODAL_MOMENT-R, NODAL_MOMENT-S,
NODAL_MOMENT-T.
The end forces/moments are computed at the element local
nodes. In the AUI, element local nodes are defined as element
points of type label. For example, to access the result computed at
element 5, local node 2, define an element point of type label with
element number 5, label number 2.
2.6 Plate elements

It is recommended to use shell elements (see Section 2.7) for
analysis of flat shells, i.e., for analysis of plates.
The plate element is kept in ADINA for backward compatibility

only. For more information on the plate element, please see the
ADINA Theory and Modeling Guides preceding ADINA 9.0.

2.7: Shell elements
2.7 Shell elements

The shell element is a 4- to 32-node (degenerate) isoparametric
element that can be employed to model thick and thin general shell
structures (see Fig. 2.7-1). However, depending on the application,
the appropriate number of nodes on the element must be employed
(see Fig. 2.7-2 and Section 2.7.10).
The shell element used in ADINA is also applicable for analysis

of flat shells, i.e., for analysis of plates.
2.7.1 Basic assumptions in element formulation
ref. KJB The basic equations used in the formulation of the element are
Sections 5.4.2 given in ref. KJB.
and 6.5.2
The shell element is formulated treating the shell as a three-
dimensional continuum with the following two assumptions used in
the Timoshenko beam theory and the Reissner/Mindlin plate
theory:
Analysis of a turbine blade: Analysis of a shell roof:
Figure 2.7-1: Some possible applications of shell elements

Chapter 2: Elements
Assumption 1: Material particles that originally lie on a

straight line "normal" to the midsurface of the structure remain
on that straight line during the deformations.
Assumption 2: The stress in the direction normal to the

midsurface of the structure is zero.
For the Timoshenko beam theory, the structure is the beam, and
for the Reissner/Mindlin plate theory, the structure is the plate
under consideration. In shell analysis, these assumptions
correspond to a very general shell theory.
In the calculations of the shell element matrices the following

geometric quantities are used:
The coordinates of the node k that lies on the shell element

midsurface at t xk , t yk , t zk (see Fig. 2.7-3); (the left
superscript denotes the configuration at time t)
The director vectors t Vnk pointing in the direction "normal"

to the shell midsurface at node k

2.7: Shell elements
a) Midsurface description
c) Top-bottom description
b) Triangle
d) Transition shell to solid
e) Multilayered shell
Figure 2.7-2: Examples of shell elements
ref. KJB The shell thickness, ak , at the nodal points measured in the
Fig. 5.33
page 437
direction of the director vectors t Vnk (see Fig. 2.7-4 and
Section 2.7.2).

Chapter 2: Elements
t
For the 9-node shell, the
2 midsurface center node
is local node number 13.
6 9 s
10 14 5
15 1
3 13
12
16
7 r
8
11
a) Midsurface nodes 4 For the 9-node shell, the

t
top-bottom center nodes
2 are local node numbers
6 13 and 29.
18 9
s
10 22 14 25 5
26 15 30 21 1
13
3 31 17
29 12
16
19 7
32 28
8
23 11
r
24
27 4
b) Top-bottom nodes 20
Figure 2.7-3: Some conventions for the shell element; local

x node numbering; natural coordinate system
Fig. 2.7-4 shows a 4-node shell element with the shell

midsurface nodal points and the nodal point director vectors.
Using the midsurface nodal point coordinates, the shell midsurface
is interpolated using the interpolation functions hk ( r , s ) given in
ref. KJB. Similarly, using the vectors t Vn , the director vector t Vn
at any point P on the midsurface is obtained by interpolation using

2.7: Shell elements
the functions hk ( r , s ) . Hence, with the shell thickness at nodal

point k equal to ak and the isoparametric coordinate t measured in
the direction of t Vn , the geometry of the shell at any time t is
defined by
q
t q
x = hk xk +
t t t
ak hk Vnxk
k =1 2 k =1
q
t q
y = hy yk +
t t t
ak hk Vnyk
k =1 2 k =1
q
t q
z = hk zk + ak hk Vnzk
t t t
k =1 2 k =1
shell midsurface interpolation for material

interpolation points not on the midsurface
where q is the number of element nodes and t Vnxk , tVnyk , tVnzk are
the direction cosines of the shell director vector t Vnk .
The direction vectors at nodes can be either directly input or

automatically generated by the program. When they are generated
by the program, they can be created as the normal vectors of a
geometrical surface, or as the averaged vectors of the surrounding
elements, which may not be exactly normal to the corresponding
geometric surface.
The assumption 1 on the kinematic behavior of the shell enters

the finite element solution in that the particles along the director
vector t Vn (interpolated from the nodal point director vectors
t
Vnk ) remain on a straight line during deformation.
Note that in the finite element solution, the vector t Vn is not
necessarily exactly normal to the shell midsurface. Figure 2.7-5(a)
demonstrates this observation for a very simple case, considering
the shell initial configuration. Furthermore, even if t Vn is
originally normal to the shell midsurface, after deformations have

Chapter 2: Elements
taken place this vector will in general not be exactly perpendicular

to the midsurface because of shear deformations (see Fig. 2.7-5(b)).
Program-calculated midsurface Program-calculated midsurface

normal vector at node k using normal vector at node k using
the geometry of element 1 the geometry of element 2
l k
element 1 l
element 2
The configuration shown
corresponds to time t=0.
0 k 0 k
Vn Vn
elt.1 elt.2
l
k kl
l
element 1 l
element 2
(a) Program-calculated normal vectors (two such vectors)

at node k. These vectors are used as director vectors
for the respective elements. Node k has 6 DOFs.
Figure 2.7-4: Convention for shell element thickness

ref. KJB The assumption 2 on the stress situation enters the finite
Section 5.4.2 element solution in a manner that is dependent on the formulation
page 440
employed:
All formulations except for the large displacement/large strain

formulations: The stress in the t-direction (i.e., in the direction of
t
Vn ) is imposed to be zero. This is achieved by using the stress-
strain relationship in the r , s , t coordinate system, shown in Fig.
2.7-6(a), with the condition that the stress in the direction t is zero.
Large displacement/large strain formulations: The stress in the

t -direction (not necessarily in the direction of t Vn ) is imposed to
be zero. This is achieved by using the stress-strain relationship in

2.7: Shell elements
the r, s, t coordinate system, shown in Fig. 2.7-6(b), with the

condition that the stress in the direction t is zero.
Note that if the nodal point midsurface director vectors are input
or generated (see Section 2.7.3) to be initially "exactly" normal to
the midsurface of the shell element (the coordinates of this
midsurface are obtained by interpolation from the nodal points on
the midsurface), then assumption 1 is also exactly satisfied in the
finite element solution, but assumption 2 is only approximately
fulfilled in geometric nonlinear analysis, because the directions of
the initially normal vectors are updated.
Program calculates director
vector by taking average of
all normal vectors at node k.
l k
element 1 l
element 2
0 k 0 k
Vn Vn
l
k kl
l
element 1 l
element 2
b) Program-calculated director vector (single vector)

at node k. Node k has 5 DOFs.
ref. KJB The transverse shear deformations are assumed by default to be

pp. 399, 440 constant across the shell thickness. The use of the correction factor
of 5/6 can be specified to improve the prediction of the
displacement response, for the linear elastic and linear orthotropic
models.

Chapter 2: Elements
The interpolation of the geometry of the shell element is always

as described above, but for a specific solution time the current
coordinates of the midsurface nodal points are used, and the current
director vectors are employed. The midsurface nodal point
coordinates are updated by the translational displacements of the
nodes and the director vectors are updated using the rotations at the
nodes (rotation increments in large displacement analysis).
Incompatible modes can be used in conjunction with the 4-node

shell element. The theory used is analogous to the theory used for
the 2-D solid element, see Section 2.2.1 and ref KJB, Section 4.4.1.
The incompatible modes are added to the midsurface displacement
interpolations; only the membrane action of the 4-node shell
element is affected.
A typical use of the incompatible modes feature would be to
improve the in-plane bending response of the 4-node element.
The incompatible modes feature is only available for 4-node
single layer shell elements in which all nodes are on the midsurface
of the element. The incompatible modes feature is available in
linear and nonlinear analysis, for all formulations.
The thermal strain calculations in shell elements depend upon

the shell element type. For the MITC3, MITC4, MITC6, MITC9,
and MITC16 single layer shell elements, the thermal strain
calculations give zero stresses for a uniform unrestrained thermal
expansion. For the other shell elements, the thermal strain
calculations can give non-zero stresses for a uniform unrestrained
thermal expansion of a curved shell structure. As the mesh is
refined, the stress calculations are improved.

2.7: Shell elements
User-defined mid-surface director vector at node k
l k
element 1 l
element 2
0 k 0 k
Vn Vn
l
k kl
l
element 1 l
element 2
c) User input director vector (single vector)

at node k. Node k has 5 DOFs.
The shell element can be used with the following material

models: elastic-isotropic, elastic-orthotropic, nonlinear-elastic,
plastic-bilinear, plastic-multilinear, plastic-orthotropic,
thermo-isotropic, thermo-orthotropic, thermo-plastic, creep,
plastic-creep, multilinear-plastic-creep, creep-variable, plastic-
creep-variable, multilinear-plastic-creep-variable, viscoelastic,
plastic-cyclic.
The shell element can be used with a small displacement/small

strain, large displacement/small strain, large displacement/large
strain ULJ formulation or a large displacement/large strain ULH
formulation. The small displacement/small strain and the large
displacement/small strain formulations can be used with any of the
above material models. The large displacement/large strain ULJ
formulation can be used with the plastic-bilinear, plastic-
multilinear, plastic-orthotropic or plastic-cyclic material models.
The large displacement/large strain ULH formulation can be used

Chapter 2: Elements
with the plastic-bilinear, plastic-multilinear or plastic-cyclic

material models.
( L1 L+2L2 ) a
0 k+1
a Vn
0
0 k
Vn Vn
l l
k k+1
L1 L2
a) Interpolation of director vector inside element

0
Vn
Initial configuration
Angle = 90
t
l
Vn
l
l
Z Final configuration
Angle 90
Y
X
b) Change in direction of director vector due to
displacements and deformations (with shear)
Figure 2.7-5: Example of change in direction of director vector

due to deformations. At time t the vector tVn
is not normal to the shell midsurface.
In the small displacement/small strain formulation, the

displacements and rotations are assumed to be infinitesimally
small. In the large displacement/small strain formulation, the
displacements and rotations can be large, but the strains are
assumed to be small. In the large displacement/large strain ULJ
formulation, the total strains can be large, but the incremental strain
for each time step should be small (< 1%). In the large
displacement/large strain ULH formulation, the total strains and

2.7: Shell elements
also the incremental strains can be large.

t
s
t
s
s
r
r
r= s t s= t r
(t = 0)
st 2 tr 2 r
Figure 2.7-6a: Definition of the local Cartesian system (r, s, t)

at an integration point
The large displacement/large strain formulation can only be

used in conjunction with 3-node, 4-node, 6-node, 9-node or 16-
node single layer shell elements, in which the shell geometry is
described in terms of midsurface nodes.
t t
s
t
s
s
r
r
t= rs , r=
s t , s=tr
(t = 0)
rs st r
Figure 2.7-6b: Definition of the midsurface Cartesian system ( r, s, t )

at an integration point

Chapter 2: Elements
The use of a linear material with the small displacement/small

strain formulation corresponds to a linear formulation, and the use
of a nonlinear material with the small displacement/small strain
formulation.
The effects of displacement-dependent pressure loading are

taken into account in the large displacement formulation.
In frequency analysis, the stiffness matrix is optionally updated
by deformation-dependent pressure loads acting onto shell
elements. The update can improve the obtained frequencies and
mode shapes when the shell structure is relatively flexible and
deformation-dependent pressure loads are applied.
2.7.3 Shell nodal point degrees of freedom
Either 5 or 6 degrees of freedom can be assigned at a shell

element midsurface node.
5 degrees of freedom node: A node "k" that is assigned 5 degrees

of freedom incorporates the following assumptions:
The translations uk, vk, wk are referred to the global Cartesian

system (or to the skew system if such a system is defined at the
node)
Only one director vector (denoted at time = 0 as 0 Vnk ) is

associated with the node. You can directly enter the director
vector or let the program calculate it (see Fig. 2.7-4). In the
latter case, the program calculates the director vector by taking
the average of all normal vectors (one normal vector is
generated per shell element attached to node k) at the node.
If two (or more) elements attached to the node have
oppositely directed normals, ADINA reverses the oppositely
directed normals, so that all normals attached to the node have
(nearly) the same direction.
The rotations k , k are referred to the local midsurface

system (see Fig. 2.7-7) defined at time = 0 by

2.7: Shell elements
Y 0 Vnk
0
V1k =
Y 0 Vnk
2
0
V2k = 0 Vnk 0 V1k
For the special case when the 0 Vnk vector is parallel to the Y
axis, the program uses the following conventions:
0
V1k Z 0
V2k X when 0
Vnk + Y
and
0
V1k Z 0
V2k X when 0
Vnk Y
No rotational
stiffness for k
Vn
this DOF
wk
l Z
l
l
vk bk
l
Y
X
uk ak
Figure 2.7-7: Shell degrees of freedom at node k
When using the large displacement formulation, the definitions

of 0 V1k and 0 V2k are only used at time=0 (in the initial
configuration) after which the vectors t Vnk and t V1k are
updated using incremental rotations at the nodal points, and
t
V2k is calculated by the cross-product t V2k = t Vnk t V1k .
The rotational degree of freedom along 0 Vnk is automatically

deleted by ADINA.

Chapter 2: Elements
6 degrees of freedom node: A node "k" that is assigned 6 degrees

of freedom incorporates the following assumptions:
The translations uk, vk, wk are referred to the global Cartesian

system or any skew system assigned to node k.
A director vector would in general not be input for this node.

(Note that any director vector input is ignored for a shell node
with six degrees of freedom.)
The program generates as many normal vectors at node k as

there are shell elements attached to the node (see Fig. 2.7-4(a)).
Hence each individual shell element establishes at node k a
vector normal to its midsurface. The components of the shell
element matrices corresponding to the rotational degrees of
freedom at this node are first formulated in the local midsurface
system defined by the normal vector and then rotated to the
global Cartesian system (or to the skew system, if defined at the
node).
The three rotational degrees of freedom at node k referred to

the global Cartesian system (or to the skew system, if defined at
the node) can be free or be deleted.
Some modeling recommendations are given below for the use of

the shell elements:
Always let all director vectors be established by the

program; director vectors should only be defined via input if
specific director vectors are required in the modeling.
Always specify 5 degrees of freedom at all shell midsurface

nodes except for the following cases in which 6 degrees of
freedom should be used:
(i) shell elements intersecting at an angle,

(ii) coupling of shell elements with other types of
structural elements such as isoparametric beams (e.g.,
in the modeling of stiffened shells),
(iii) coupling of rigid links (see Section 5.15.2) to the shell
midsurface nodes,

2.7: Shell elements
(iv) coupling of constraints or generalized constraints to

the shell midsurface nodes,
(v) imposing rotational moments or boundary conditions
at the node.
(vi) coupling of alignment elements that have align-
rotation properties to the shell midsurface nodes
All of the above considerations are illustrated in the schematic

example shown in Fig. 2.7-8. Note that in the example, the use of
6 degrees of freedom at node 1 together with the deletion of the X-
rotation degree of freedom is only applicable in small displacement
analysis.
If a large displacement formulation is used in this example,
node 1 must be assigned 5 degrees of freedom and the applied
concentrated moments must then refer to the local midsurface
system at that node, which consists of the global directions for uk,
vk, wk and the local directions t V1k and t V2k for the incremental
rotations k and k . (Alternatively, 6 degrees of freedom could
be used if a beam of very small stiffness is defined along the edge
of the plate.)
There are a number of implications regarding the use of 6 DOF

nodes on flat or nearly flat shells. These implications arise because
there is no rotational stiffness in the t Vn direction.
1) No attached structural element/rigid link (Fig. 2.7-9(a))
a) There will be a zero pivot in the stiffness matrix. If a moment is

applied into the t Vn direction, the rotation in the t Vn direction will
be infinite. In addition, the moment applied into the t Vn direction
will not cause equilibrating reactions at the fixities.
2) Structural element/rigid link attached to the node

(Fig 2.7-9(b)(-d)).
a) If the structural element/rigid link is unsupported, there will be a

zero pivot in the global stiffness matrix. Even if the structural
element/rigid link is supported, there may be a zero pivot in the
global stiffness matrix, depending upon how the structural

Chapter 2: Elements
element/rigid link is supported.
b) Moments in the t Vn direction will not be transmitted from the

structural element or rigid link into/from the shell element.
4-node shell elements

No X,Y,Z translations
No X,Y rotations
4 3 2 1l
l l l
5 MZ FZ
6
l l
MY FX
9l 8 l
7 l l
X Rigid link Concentrated

(Node 4 as loads at node 1
Z 2-node iso-beam master node)
Y elements
Number Degree of freedom Reference

Node system
of DOF 1 2 3 4 5 6
1 6 Global
2 5 * Midsurface
3 5 * Midsurface
4 6 Global
5 5 * Midsurface
6 6 Global
7 6 Global
8 6 Global
9 6 Global
= Deleted because of no stiffness at this DOF

= Free DOF
* = The AUI automatically deletes the 6th DOF
when a midsurface system is specified for the node
Figure 2.7-8: Example on the recommended use of shell elements

2.7: Shell elements
Moment causes
z infinite rotation.
y Mz
x t
Vn
Reactions at fixities
are zero.
a) Moment applied in shell normal direction.

Zero pivot in
stiffness matrix.
F t
Vn
b) Structural element/rigid link attached to node,

structural element/rigid link is unsupported.
Zero pivots in
stiffness matrix.
c) Structural element/rigid link attached to node,

structural element/rigid link is perpendicular
to shell elements
Figure 2.7-9: Flat shell with 6 DOFs at a node

Chapter 2: Elements
Moment in z direction not

transmitted into/from shell
t
Vn
F
d) Structural element/rigid link attached to node,
structural element/rigid link is supported.
ADINA has the option of applying a drilling stiffness to 6 DOF

nodes on flat shells.
The AUI command is MASTER SINGULARITY-
STIFFNESS=YES (on by default). When the drilling stiffness
option is used, any moment applied to the node into the t Vn
direction is taken by the drilling stiffness rotational spring (which
is grounded). The drilling stiffness spring eliminates the zero pivot
in the stiffness matrix, hence in some cases, a solution can be
obtained. But any moment applied into the t Vn direction will still
not cause equilibrating reactions at the fixities. And using a
drilling stiffness value that is too large may lead to an incorrect
solution.
An alternative to using the drilling stiffness option is to connect the

6 DOF nodes on flat shells to neighboring shell nodes using soft
beam elements (so-called weld elements). This idea is shown in
Fig 2.7-9(e). Then moments applied into the t Vn direction will be
taken by the weld elements, and these moments will cause
equilibrating reactions at the fixities. The weld elements also
provide stiffness in the t Vn direction, so that there will be no zero
pivots.

2.7: Shell elements
Soft beam element.

Mz t
Vn
Reactions at fixities
are nonzero.
e) Soft beam element takes applied moment.
2.7.4 Transition elements
The shell elements can also be employed as transition elements.

A transition element is obtained by using, instead of a shell
midsurface node, one node on the top surface and one node on the
bottom surface of the shell element. Each of these nodes has then
only 3 degrees of freedom corresponding to translations in the
global X, Y, Z coordinate directions. Fig. 2.7-10(a) shows a cubic
transition element.
Note that although the degrees of freedom at a shell transition

node are those of a three-dimensional isoparametric element, the
assumption that the stress "normal" to the midsurface of the shell
(in the t-direction) is zero distinguishes a transition element from a
3-D solid element even when a shell element has only transition
nodes.

Chapter 2: Elements
l l
l l
l l
l l l
l
l
l
l
l l
l
l l
l
Transition nodes
l
Z wk vk
uk
Y
X
a) Example of a transition element
t DL uniform
over element
DL s
L
r
b) Zero energy mode for flat element

with only transition nodes
Figure 2.7-10: Transition elements
In particular, if only transition nodes are used on the element,

then the element has a zero energy mode corresponding to a
uniform strain in the t-direction. Fig. 2.7-10(b) shows this zero-
energy mode. For a 3-D solid element, this mode corresponds, of
course, to a uniform strain in the t-direction.
If not all nodes on a transition element are transition nodes, i.e.,
there is at least one midsurface node, then the zero-energy mode
shown in Fig. 2.7-10(b) is not present.
The transition element is particularly useful in modeling shell to

shell and shell to solid intersections, see Fig. 2.7-11 and the

2.7: Shell elements
l l
l
l
l
l l
Shell to shell intersection
l l l
l
l l l
l l
l l l l l
l
Shell to solid intersection
Figure 2.7-11: Examples of use of transition element
ref. K.J. Bathe and L.W. Ho, "Some Results in the Analysis
of Thin Shell Structures," in Nonlinear Finite Element
Analysis in Structural Mechanics, W. Wunderlich et al
(eds.), Springer-Verlag, 1981.
ref. KJB Numerical integration is used for the evaluation of the element
Section 5.5 matrices, and the default integration or a higher order should
always be used; then no spurious zero energy modes are present.
Gauss numerical integration is used in the r-s plane. For the 4-

node element, the default order is 22 integration. For the 8- and
9-node elements, the default order is 33 integration, and for the
16-node elements, the default order is 44 integration.
Either Gauss, Newton-Cotes or trapezoidal rule numerical

ref. KJB
Section 6.8.4 integration is used through the shell thickness. For Gauss
integration, 2 to 6 points through the thickness can be selected. For
Newton-Cotes integration, 3, 5 or 7 points through the thickness
can be selected. For trapezoidal rule integration, 2 to 20 points
through the thickness can be selected. Fig 2.7-12 shows some
examples.

Chapter 2: Elements
Usually, 2-point Gauss or 3-point Newton-Cotes integration is

appropriate for an elastic material, but a higher integration order
may be more effective for elastic-plastic analysis.
t t t
3
4
2 3
2

1 2
1 1

a) 2 pt Gauss b) 3 pt Newton-Cotes c) 4 pt trapezoidal rule

integration integration integration
Figure 2.7-12: Examples of numerical integration through the

shell element thickness
The labeling of the integration points for quadrilateral and

triangular elements is given in Fig. 2.7-13. The locations of the
integration points in the r-s plane are the same as for the 2-D solid
elements (see Section 2.2.3).

2.7: Shell elements
3 3 2 Gauss integration: t s
INR=1,INS=1,INT=2

INR=1,INS=1, r
INT=1
INR=3,INS=1,INT=1
2 2 3 Newton-Cotes integration:
t s
INR=1,INS=2,INT=3

INR=2,INS=1,INT=1
a) Examples of integration point labeling for quadrilateral shell elements
Figure 2.7-13: Examples of integration point labeling

Chapter 2: Elements
s s s
3 2
1
1 1
3 4
2 r r
r
1-point 3-point 4-point
s s
3 3 4
2 6 7 8
5 2 10 12
6
1 7 1 13
4 11
r 5 9 r
7-point 13-point
b. Integration points position on r-s plane for the triangular shell elements
The default number of Gauss integration points in the r-s plane

for the MITC3 element is 3; for the MITC6 element is 7; for the
collapsed MITC4 triangular element is 4; for the collapsed DISP7
element is 7; and for the collapsed DISP10 element is 13 (see also
Fig. 2.7-18).
2.7.6 Composite shell elements
The composite shell elements are kinematically formulated in

the same way as the single layer shell elements, but
An arbitrary number of layers can be used to make up the

total thickness of the shell.
Each layer can be assigned one of the different material
models available. The element is nonlinear if any of the
material models is nonlinear, or if the large displacement
formulation is used.
The computation of accurate transverse shear stresses based

on a three-dimensional theory can be requested. This way the

2.7: Shell elements
condition of zero transverse shear stresses at the top and bottom

surfaces of the shell are satisfied, and the transverse shear
stresses are continuous at layer interfaces.
Layers are numbered in sequential order starting from 1 at the

bottom of the shell.
The layer thicknesses can be assigned using one of two general

approaches:
Specify the total element thickness and the percentage of

thickness for each layer.
Specify the data of each layer (or ply) in terms of the weight
per unit surface of the fiber Wf, the density of the fiber f and
the fiber volume fraction of the fiber-matrix compound f .
The AUI then computes the thickness of each layer
Wf
automatically using the formula h = , in which it is
f f
assumed that the ply is made of uniaxial fibers, so that the fiber
thickness fraction perpendicular to the plate midsurface is
proportional to the fiber volume fraction. The total thickness of
the multilayered shell is then the sum of the ply thicknesses.
This approach is especially useful when the layers are fiber-
matrix composites.
In order to take into account the change of material properties

from one layer to another, numerical integration of the mass and
stiffness matrices is performed layer by layer using reduced natural
coordinates through the thickness of the element (see Fig. 2.7-14
and 2.7-15). The relation between the element natural coordinate t
and the reduced natural coordinate tn of layer n is:
1 n i
t = 1 + 2
a i =1
n
(1 t )
n
(2.7-1)


Chapter 2: Elements
with
t = element natural coordinate through the thickness

tn = layer n natural coordinate through the thickness
i
= thickness of layer i
a = total element thickness
i
a and are functions of r and s.
e3
l
3
Y t l
e2
e1 l
X l 2
s
r l
a4 4 l l 1
Vn
l
l 1
e2 lV kn 1l V2
l k
V1 =
|| e2 lVnk || 2 l 1 b1
V1 a1
l k
V 2 = lVnk lV 1k
Figure 2.7-14: 8-node composite shell element
z,w
y,v
x,u
ln E nxx G nxz nnxy Layer n

a
Layer 1
Figure 2.7-15: Multilayered plate

2.7: Shell elements
The geometry of layer n is given by:
N N
n

xi = hk xik + mkn + t n k hk Vnik (2.7-2)
k =1 k =1 2
with
xi = coordinates of a point inside layer n

N = number of nodes
hk = interpolation functions
xik = Cartesian coordinates of node k
Vnik = components of normal vector Vnk
ak = total element thickness at node k
i
k = thickness of layer i at node k
[ ] = 0 in the initial configuration, 1 in the deformed

configuration
In the above formula, the expression
n n
ak
mkn = + i
k k
(2.7-3)
2 i =1 2
corresponds to the distance at node k between the element

midsurface and the midsurface of layer n.
Accordingly, the displacements in layer n are:
N N
n t n nk 0 k
ui = hk u + mk + ( V2i k + V1i k )
k 0 k
i
k =1 k =1 2
(2.7-4)
with
uik = components of nodal displacements at node k

k , k = rotations of 0 Vnk about 0 V1k and 0 V2k (see Fig. 2.7-14)
Using the expressions of the coordinates and displacements

Chapter 2: Elements
defined in Eqs. 2.7-2 and 2.7-4, the contribution of each layer to

the element stiffness and mass matrices can be evaluated.
In the analysis of thick plates and shells or in the analysis of

multilayered structures such as composite sandwiches with low
shear rigidity, the transverse shear deformation energy may not be
negligible compared to the flexural energy and high transverse
shear stresses may be encountered. In these cases, the shell
kinematic formulation must be amended to obtain more accurate
expressions of the shear deformation energy and transverse shear
stresses.
By combining for a plate in cylindrical bending, the plate
equilibrium equations and the 3-dimensional equilibrium equations,
an expression of the shear stress which satisfies stress continuity at
the layer interfaces and zero boundary values at the top and bottom
surfaces of the shell can be obtained. Simultaneously, a shear
reduction factor is obtained which can be used to improve the shear
deformation energy approximation.
The expression of the shear factor is given by
I xx2
k xz = (2.7-5)
a a
g xz2

2 2
Gxz dz dz
a 2 a 2 Gxz
and the transverse shear stress in terms of the shear strain value
xz is
I xx g xz ( z )
xz ( z ) = xz (2.7-6)
a
g xz2
a 2 Gxz dz
2
where
a
I xx = a Dxx ( z z0 ) dz
2 2
2
Exxn
D =
n
1 xyn yxn
xx
Exx = Young's modulus

Gxz = shear modulus

2.7: Shell elements
xyn , yxn = Poisson's ratios

a = total thickness
z
g xz ( z ) = Dxxn ( z0 ) d
a 2
Equations 2.7-5 and 2.7-6 are used in ADINA to calculate shear

correction factors for each element and shear stress profiles at each
integration point.
The shear correction factor calculation is only available for the
linear isotropic elastic and linear orthotropic material models.
ref. O. Guillermin, M. Koji, K.J. Bathe, "Linear and

Nonlinear Analysis of Composite Shells," Proceedings,
STRUCOME 90, DATAID AS & I, Paris, France,
November 1990.
2.7.7 Mass matrices
The shell element can be employed with a lumped or a

consistent mass matrix, except for explicit dynamic analysis which
always uses a lumped mass.
The consistent mass matrix is calculated using the isoparametric

formulation with the shell element interpolation functions.
The lumped mass for translational degrees of freedom of

midsurface nodes is M / n where M is the total element mass and n
is the number of nodes. No special distributory concepts are
employed to distinguish between corner and midside nodes, or to
account for element distortion.
The rotational lumped mass for all except explicit dynamic analysis
M 1
is, ( tav2 ) where tav is the average shell thickness. The same
n 12
rotational mass matrix is assumed for 5- and 6-degree of freedom
nodes, and is applied to all rotational degrees of freedom.
The rotational lumped mass for explicit dynamic analysis is

Chapter 2: Elements
M 1 2
( t + A ) , where tav is the average shell thickness and A is
n 12 av
the cross-sectional area. The rotational masses are scaled up to
ensure that the rotational degrees of freedom do not reduce the
critical time step for shell elements. The same rotational mass
matrix is assumed for 5- and 6-degree of freedom nodes and is
applied to all rotational degrees of freedom.

specified scalar ETA (except in explicit analysis).
2.7.8 Anisotropic failure criteria
The study of material failure in composite structures has shown

that various criteria can be used to assess failure. The following
failure criteria are provided in ADINA for the analysis of shell
structures using the elastic-isotropic, elastic-orthotropic, thermo-
isotropic or thermo-orthotropic material models:
Maximum stress failure criterion

Maximum strain failure criterion
Tsai-Hill failure criterion
Tensor polynomial failure criterion
Hashin failure criterion
User-supplied failure criterion
These criteria can be evaluated during the analysis to determine

whether the material has failed. However, the stresses and material
properties of the model are not changed as a consequence of this
calculation. It is therefore intended only for use in linear elastic or
thermo-elastic analysis, or as a stress state indicator in nonlinear
analysis.
The material failure is investigated at each integration point of

each element during the analysis. The failure criteria values (and
possibly modes) are printed and/or saved according to the ADINA
stress printing and saving flags.
The maximum stress failure criterion compares each

component of the stress tensor (resolved in the material (abc)

2.7: Shell elements
directions) to maximum stress values. The input constants for this

criterion are:
Xt = maximum allowable tensile stress in the material a-

direction
Xc = maximum allowable compressive stress in the
material a-direction
Yt = maximum allowable tensile stress in the material b-
direction
Yc = maximum allowable compressive stress in the
material b-direction
Zt = maximum allowable tensile stress in the material c-
direction
Zc = maximum allowable compressive stress in the
material c-direction
Sab = maximum allowable absolute shear stress in the a-b
plane
Sac = maximum allowable absolute shear stress in the a-c
plane
Sbc = maximum allowable absolute shear stress in the b-c
plane
The failure of the material occurs when any of the following

inequalities is not satisfied anymore:
Xc < a < Xt
Yc < b < Yt
Zc < c < Zt
Sab < ab < Sab
Sac < ac < Sac
Sbc < bc < Sbc
where (a,b,c,ab,ac,bc) is the stress vector referred to the principal

material directions.
The above conditions are applicable to 3-D analysis. In plane
stress analysis, the parameters Zt, Zc, Sac and Sbc are not used.
Note that the maximum stress criterion indicates the mode of

failure. Note that this criterion does not take into account any

Chapter 2: Elements
interaction between failure modes.
The maximum strain failure criterion compares each

component of the strain tensor (resolved in the material (abc)
directions) to maximum strain values. The input constants for this
criterion are:
Xt = maximum allowable tensile strain in the material a-

direction
Xc = maximum allowable compressive strain in the
Yt = maximum allowable tensile strain in the material b-
direction
Yc = maximum allowable compressive strain in the
Zt = maximum allowable tensile strain in the material
third c-direction
Zc = maximum allowable compressive strain in the
Sab = maximum allowable absolute shear strain in the a-b
plane
Sac = maximum allowable absolute shear strain in the a-c
plane
Sbc = maximum allowable absolute shear strain in the b-c
plane
The failure of the material occurs when any one of the following
inequalities is not satisfied anymore:
Xc < a < X t
Y c < b < Y t
Z c < c < Z t
S ab < ab < S ab
S ac < ac < S ac
S bc < bc < S bc
where (a, b, c, ab, ac, bc) is the strain vector referred to the
principal material directions.
The above conditions are applicable to 3-D analysis. In plane

2.7: Shell elements
stress analysis, however, the parameters Zt, Zc, Sac and Sbc are not
used.
Note that the maximum strain criterion indicates the mode of

failure. Note that this criterion does not take into account any
interaction between failure modes.
The Tsai-Hill failure criterion provides a yield criterion

applicable to materials with anisotropic failure strengths. This is an
extension of the von Mises yield criterion. Failure of the material
occurs when the following inequality is not satisfied anymore:
(G + H ) a2 + ( F + H ) b2 + ( F + G ) c2 2 H a b 2G a c 2 F b c
+ 2 L bc2 + 2 M ac2 + 2 N ab2 < 1
with
1 1 1 1
F = 2 + 2 + 2
2 X Y Z
1 1 1 1
G= 2
2+ 2
2 X Y Z
1 1 1 1
H= 2
+ 2 2
2 X Y Z
1
L=
2 Sbc2
1
M=
2 Sac2
1
N=
2 S ab2
where
X = maximum allowable absolute stress in the material a-

direction
Y = maximum allowable absolute stress in the material
b-direction
Z = maximum allowable absolute stress in the material c-
direction

Chapter 2: Elements

plane
Sac = maximum allowable absolute shear stress in the a-c
plane
Sbc = maximum allowable absolute shear stress in the b-c
plane
In the case when plane stress conditions are used, the material
behavior with respect to failure is considered as transversely
isotropic (Y=Z), and the failure criterion reduces to
a2 a b b2 ab2
+ + <1
X2 X2 Y2 Sab2
In this case, the maximum allowable stresses Z, Sbc and Sac are not
used.
Note that this criterion includes the effect of interactions

between failure modes, but does not indicate a specific mode of
failure.
The tensor polynomial failure criterion assumes the existence

of a failure surface in the stress space of the form:
Fi i + Fij i j = 1 with i, j = 1,..., 6
with the convention
1 = a ; 2 = b ; 3 = c
4 = ab ; 5 = ac ; 6 = bc
and with

2.7: Shell elements
1 1 1
F1 = + ; F11 =
X at X ac X at X ac
1 1 1
F2 = + ; F22 =
X bt X bc X bt X bc
1 1 1
F3 = + ; F33 =
X ct X cc X ct X cc
F4 , F5 , F6 = 0
1 1 1
F44 = 2 ; F55 = 2 ; F66 = 2
Sab Sac Sbc
where the following strength components need to be defined:
Xat = maximum allowable tensile stress in the material a-

direction
Xac = maximum allowable compressive stress in the
Xbt = maximum allowable tensile stress in the material b-
direction
Xbc = maximum allowable compressive stress in the
Xct = maximum allowable tensile stress in the material c-
direction
Xcc = maximum allowable compressive stress in the
Sab = maximum allowable absolute shear stress in the
material a-b plane
Sac = maximum allowable absolute shear stress in the
material a-c plane
Sbc = maximum allowable absolute shear stress in the
material b-c plane
For plane stress conditions, the parameters Xct, Xcc, Sbc, Sac, Fbc
and Fac are not used.
The interaction strengths Fij with i j can be input, or calculated

by ADINA according to Hoffman's convention:

Chapter 2: Elements
1 1
Fij = +
X it X ic X jt X jc

with i j and i, j = a, b, c
Note that this failure criterion includes interactions between the

failure modes, but that the mode of failure is not given by the
analysis.
The Hashin failure criterion applies especially to fibrous

composites where the fiber and the matrix failure mechanisms are
distinct. In ADINA, we assume that the fibers are predominantly
aligned with the first material principal direction (a-direction), thus
making the material transversely isotropic about the a-direction.
Failure will occur when any of the following inequalities is not
satisfied anymore:
a2
+
( 2
ab + ac2 )
<1 (if a > 0) (2.7-7)
X t2 Sab2
a
<1 (if a < 0) (2.7-8)
Xc
( b + c )
2
( 2
b c ) ( 2
+ ac2 )
Yt 2
+
bc
Str2
+
ab
Sab2
<1 ( if ( b + c ) > 0)
(2.7-9)
Yc
2
( b + c ) 1
2 Str ( b + c ) ( b b c ) ( ab + ac )
2 2 2 2
+ + + <1
Yc 4 Str2 Str2 S ab2
(if ( b + c ) < 0)
(2.7-10)
with
Xt = maximum allowable tensile stress in the material a-

direction
Xc = maximum allowable compressive stress in the

2.7: Shell elements
Yt = maximum allowable tensile stress in the material b-
direction
Yc = maximum allowable compressive stress in the
plane (note that Sab = Sac)
Str = maximum allowable absolute shear stress in the b-c
plane
The above inequalities are usually referred to as: tensile fiber

failure (2.7-7), compressive fiber failure (2.7-8), tensile matrix
failure (2.7-9), compressive matrix failure (2.7-10).
The user-supplied failure criterion allows for the description

of a failure envelope using up to four quadratic failure relations of
the form:
6
Fi i + Fij i j = 1 if
i =1
i i >0
with
1 = a ; 2 = b ; 3 = c
4 = bc ; 5 = ac ; 6 = ab
The material failure occurs when any one of the failure surfaces is
reached and the corresponding stress condition is satisfied at the
same time.
This flexible and very general user-supplied failure model

allows you to input any quadratic failure criteria with failure mode
interactions.
The following references contain additional information about

composite failure criteria:
ref. Jones R.M., Mechanics of Composite Materials,

McGraw Hill, 1975.
ref. Tsai S.W., "A Survey of Macroscopic Failure Criteria

Chapter 2: Elements
for Composite Materials," Journal of Reinforced Plastics

and Composites, Vol.3, pp.40-62, January 1984.
ref. Hashin Z., "Analysis of Composite Materials A

Survey," Journal of Applied Mechanics, Transactions of
the ASME, Vol. 50, pp. 481-505, September 1983.
You can request that ADINA either print and save stresses or
forces.

variable names.
The results can be requested to be referred to the global (X,Y,Z)

axes, or to the local ( r , s , t ) axes (see Fig. 2.7-6a), or to the
( )
midsurface r, s, t axes (see Fig. 2.7-6b). For the large
displacement/large strain formulations, both the local and
( )
midsurface axes are the r, s, t axes, so the results in both the
local and midsurface systems are identical. In addition, for
orthotropic material models, the results referred to the material axes
can be requested.
Elastic-isotropic, elastic-orthotropic: STRESS(XYZ),

STRAIN(XYZ), FE_EFFECTIVE_STRESS

STRAIN(XYZ), FE_EFFECTIVE_STRESS,
Thermo-isotropic: STRESS(XYZ), STRAIN(XYZ),

THERMAL_STRAIN(XYZ), FE_EFFECTIVE_STRESS,
ELEMENT_TEMPERATURE

2.7: Shell elements
Plastic-bilinear, plastic-multilinear, without thermal effects:

YIELD_STRESS, ACCUM_EFF_PLASTIC_STRAIN
Plastic-bilinear, plastic-multilinear with thermal effects:

YIELD_STRESS, ACCUM_EFF_PLASTIC_STRAIN,

variable: PLASTIC_FLAG, NUMBER_OF_SUBINCREMENTS,
PLASTIC_STRAIN(XYZ), CREEP_STRAIN(XYZ),
ACCUM_EFF_PLASTIC_STRAIN,
Plastic-orthotropic without thermal effects: PLASTIC_FLAG,

YIELD_STRESS, ACCUM_EFF_PLASTIC_STRAIN
Plastic-orthotropic with thermal effects: PLASTIC_FLAG,
YIELD_STRESS, ACCUM_EFF_PLASTIC_STRAIN,
Viscoelastic: STRESS(XYZ), STRAIN(XYZ),

In the above lists,
when results are referred to the global (X,Y,Z) axes,

STRESS(XYZ) = STRESS-XX, STRESS-YY,
STRESS-ZZ, STRESS-XY, STRESS-XZ,
STRESS-YZ
( )
when results are referred to the ( r , s , t ) or r, s, t axes,

Chapter 2: Elements
STRESS(XYZ) = STRESS-RR, STRESS-SS,

STRESS-TT, STRESS-RS, STRESS-RT,
STRESS-ST
when results are referred to the material axes,

STRESS(XYZ) = STRESS-AA, STRESS-BB,
STRESS-CC, STRESS-AB, STRESS-AC,
STRESS-BC
with similar definitions for the other abbreviations used above.

In addition, when a failure model is used, each element outputs,

at each integration point, the following information to the porthole
file, based on the failure criterion. This information is accessible in
the AUI using the following variable names:
Maximum stress, maximum strain:

FAILURE_FLAG_TENSION-A,
FAILURE_FLAG_COMPRESSION-A,
FAILURE_FLAG_TENSION-B,
FAILURE_FLAG_COMPRESSION-B,
FAILURE_FLAG_TENSION-C,
FAILURE_FLAG_COMPRESSION-C,
FAILURE_FLAG_SHEAR-AB, FAILURE_FLAG_SHEAR-AC,
FAILURE_FLAG_SHEAR-BC, FAILURE_FLAG
Tsai-Hill, tensor polynomial: FAILURE_FLAG,

FAILURE_CRITERION
Hashin: FAILURE_FLAG_TENS-FIBER,
FAILURE_FLAG_COMP-FIBER,
FAILURE_FLAG_TENS-MATRIX,
FAILURE_FLAG_COMP-MATRIX,
FAILURE_CRITERION_TENS-FIBER,
FAILURE_CRITERION_COMP-FIBER,
FAILURE_CRITERION_TENS-MATRIX,
FAILURE_CRITERION_COMP-MATRIX
User-supplied: FAILURE_FLAG_SURFACE-1,
FAILURE_FLAG_SURFACE-2,
FAILURE_FLAG_SURFACE-3,

2.7: Shell elements
FAILURE_FLAG_SURFACE-4, FAILURE_FLAG,
FAILURE_CRITERION_SURFACE-1,
FAILURE_CRITERION_SURFACE-4
Section results: The computation of stress resultants, forces and

moments, the computation of membrane strains and curvatures, and
the position of the neutral axes can be requested in ADINA. The
latter is especially useful for multilayered shell elements.
These results are given at the locations corresponding to the
projections of the integration points on the shell element
midsurface (see Fig. 2.7-16).
The position of the neutral axes is available only for multilayer
shell elements.
Fig. 2.7-17 shows the convention used for positive forces and
moments. Table 2.7-1 contains the formulas used to evaluate the
stress and strain resultants and the neutral axes positions. They are
always referred to the local Cartesian system ( r , s , t ) (see Fig.
2.7-16).

Chapter 2: Elements

B
D
Point labels INR INS
A 1 1
B 1 2
r C 2 1

A

C D 2 2

h/2
a) Point labeling for section results
(2 2 2 Gauss integration) E r zdz
-h/2
Z 0r =
Layer number h/2
+ _h
5 2 E r dz
4 -h/2
3 Z 0r
2 h/2
1
- h_ Es zdz
2
-h/2
Z 0s =
h/2
b) Definition of the neutral axes Es dz
-h/2
Figure 2.7-16: Conventions and definitions for section results
output

2.7: Shell elements
Table 2.7-1: Section results output for the shell element

Stress resultants (forces and moments)
h/2 h/2
Rr r = r r dz Mr r = s s z dz
h / 2 h / 2
h/2 h/2
Rs s = s s dz M s s = r r z dz
h / 2 h / 2
h/2
Rt t = t t dz
h / 2
h/2 h/2
Rr s = r s dz Mr s = r s z dz
h / 2 h / 2
h/2
Rr t = r t dz
h / 2
h/2
Rs t = s t dz
h / 2
Membrane strains and curvatures
2r r 1r r 2r r 1r r
m r r = 1r r z1 ss =
( z2 z1 ) ( z2 z1 )
2s s 1s s 2s s 1s s
m s s = 1s s z1 rr =
( z2 z1 ) ( z2 z1 )
2r s 1r s 2r s 1r s
m r s = 1r s z1 rs =
( z2 z1 ) ( z2 z1 )
Indices 1 and 2 refer respectively to the lowest and highest integration points on a
normal to the element midsurface.
Note that the section results are given with respect to the local Cartesian system of
the shell elements.
The stress resultants do not satisfy equilibrium except in the limit as the mesh is
refined.

Chapter 2: Elements
t
s
Rss
Rrs
Rrs
r Rrr
t
s
Rst Mrr
Mss Rrt Mrs
r Mrs
Figure 2.7-17: Nomenclature for shell stress resultants (positive)
The corresponding AUI variables are (for all material models):

MEMBRANE_FORCE-RB, MEMBRANE_FORCE-SB,
MEMBRANE_FORCE-RBSB, SHEAR_FORCE-RB,
SHEAR_FORCE-SB, BENDING_MOMENT-RB,
BENDING_MOMENT-SB, BENDING_MOMENT-RBSB,
MEMBRANE_STRAIN-RB, MEMBRANE_STRAIN-SB,
MEMBRANE_STRAIN-RBSB, CURVATURE-RB,
CURVATURE-SB, CURVATURE-RBSB,
SHEAR_STRAIN-RB, SHEAR_STRAIN-SB,
NEUTRAL_AXIS_POSITION-RB,
NEUTRAL_AXIS_POSITION-SB.
ref. KJB Nodal forces: The nodal point force vector which corresponds to
Example 5.11 the element internal stresses can also be requested in ADINA. The
pp. 358-359 procedure for the calculation of the force vector is given in ref.
KJB.
names NODAL_FORCE-X, NODAL_FORCE-Y,
NODAL_FORCE-Z, NODAL_MOMENT-X, NODAL_MOMENT-
Y, NODAL_MOMENT-Z.
2.7.10 Selection of elements for analysis of thin and thick shells
The following types of shell elements are available:

2.7: Shell elements
Number of Single layer with Multilayer

nodes mid-surface nodes or top & bottom
only
Quadrilateral 4-node MITC4 MITC4
elements 8-node MITC8 MITC8
9-node MITC9 DISP9
16-node MITC16 DISP16
Triangular 3-node MITC3 MITC4 Collapsed
elements 6-node MITC6 DISP7 Collapsed
9-node DISP10 Collapsed DISP10 Collapsed
The types of triangular elements are illustrated in Fig. 2.7-18.
s s
3 3
6 5
r r
1 2 1 4 2
MITC3 MITC6
s s s
2 2 2
6 9
6 5
10 5
r 1,4 r 1,4 r
1,4
3 3 7 11
3 7
MITC4 collapsed DISP7 collapsed DISP10 collapsed
Figure 2.7-18. Types of triangular elements
The most effective element for analysis of general shells is

usually the 4-node element, shown in Fig. 2.7-19(a). This element
does not lock and has a high predictive capability and hence can be

Chapter 2: Elements
used for effective analysis of thin and thick shells. See references:
ref. Dvorkin, E. and Bathe, K.J., "A Continuum Mechanics

Based Four-Node Shell Element for General Nonlinear
Analysis," Engineering Computations, Vol. 1, pp. 77-88,
1984.
Recommended element: MITC4 t

s
l
r
l
l
(a) 4-node shell element for thick and thin shells.
Sometimes also useful elements: MITC16 and MITC9
t t
l s
l
l
l
s
l
l l
l l
l
l l l
l l
l
l l l
l
l
l
l l
r
l r
MITC16 MITC9
(b) 16-node cubic shell element for thick and thin shells.
9-node shell element for thick and thin shells.
Figure 2.7-19: Recommended elements for analysis of any shell
ref. Bathe, K.J. and Dvorkin, E., "A Four-Node Plate

Bending Element Based on Reissner/Mindlin Plate
Theory and a Mixed Interpolation," Int. J. Num. Meth. in
Eng., Vol. 21, pp. 367-383, 1985.
ref. Bathe, K.J. and Dvorkin, E., "A Formulation of General

2.7: Shell elements
Shell Elements The Use of Mixed Interpolation of

Tensorial Components," Int. J. Num. Meth. in Eng., Vol.
22 pp. 697-722, 1986.
ref. Bucalem, M.L and Bathe, K.J., "Higher-Order MITC

General Shell Elements," Int. J. for Num. Meth in Engng,
Vol. 36, pp. 3729-3754, 1993.
ref. Bathe, K.J., Iosilevich, A. and Chapelle, D., "An

Evaluation of the MITC Shell Elements," Comp. Struct,
Vol. 75, pp.1-30, 2000.
ref. Lee, P.H. and Bathe, K.J., Development of MITC

isotropic triangular shell finite elements, Comp. Struct,
Vol.82, 945-962, 2004.
In linear analysis it can be effective to use the 9-node shell

element (the MITC9 element) or the 16-node shell element (the
MITC16 element) (see Fig. 2.7-19(b)). However, the use of these
elements can be costly.
ref. KJB The phenomenon of an element being much too stiff is in the
pp. 403-408 literature referred to as element locking. In essence, the
phenomenon arises because the interpolation functions used for an
element are not able to represent zero (or very small) shearing or
membrane strains. If the element cannot represent zero shearing
strains, but the physical situation corresponds to zero (or very
small) shearing strains, then the element becomes very stiff as its
thickness over length ratio decreases.
The MITC3, MITC4, MITC6, MITC8, MITC9 and MITC16
elements are implemented to overcome the locking problem in the
shell element. When you choose to generate the 4-node shell
element (through the AUI), the MITC4 elements are created for the
quadrilateral elements and MITC3 elements are used for the
triangular elements. When you choose to generate the 8-node shell
element, the MITC8 elements are created for the quadrilateral
elements and the MITC6 elements are created for the triangular
elements. When you choose to generate the 9-node shell element,
the MITC9 elements are created for the quadrilateral elements and
the MITC6 elements are used for the triangular elements. When
you choose to generate the 16-node shell element, the MITC16

Chapter 2: Elements
elements are created for the quadrilateral elements and DISP10

collapsed triangular elements are used for the triangular elements.
The MITC8 element is not as effective as the other MITC elements,
and its use is not recommended in general.
In order to arrive at an appropriate element idealization of a thin

shell, it may be effective to consider the behavior of a single
element in modeling a typical part of the shell. As an example, if a
shell of thickness h and principal radii of curvatures R1 and R2 is to
be analyzed, a single element of this thickness and these radii and
supported as a cantilever could be subjected to different simple
stress states. The behavior of the single element when subjected to
the simple stress states (e.g., constant bending moments) tells what
size of element, and hence element idealization, can be used to
solve the actual shell problem.
For the analysis of thick shells, the elements depicted in Fig.

2.7-20 can be used as well. If necessary, these elements can also be
employed together with the elements of Fig. 2.7-19 in the analysis
of a thin shell, but then only a few elements should be employed to
model a special region, where necessary, such as a transition
region, a cut-out, and a triangular region. Also, the elements
should be small enough in size.
3
l l
l
l
2 l
2
l
l
l l
l l
l
4
l
3 l
l
l
l l
l
l
l 1 and 4
1
a) Quadrilateral elements b) Triangular elements
l Nodes 1 to 4 (corner nodes) must be input

l Optional nodes
Figure 2.7-20: Elements for analysis of thick shells
For cases where the ratio thickness/radius of curvature is large

(in the original configuration or in the deformed configuration in
large deformation analysis) it is best to use 5 dofs at each shell

2.7: Shell elements
node. Then all elements will represent the hyperbolic stress

distributions through the element thickness. If 6 dofs per node are
used, then only the higher order elements will represent the
hyperbolic stress distribution because the nodal director vectors are
constructed from the geometry of the elements.
It is recommended that, as often as possible, only the 4-node, 9-

node and 16-node elements are employed for the quadrilateral
elements, 3-node and 6-node elements are employed for the
triangular elements.
Note that the elements in Fig. 2.7-20 can also be used as

transition elements by assigning top and bottom surface nodes
instead of a midsurface node.
The shell elements can be used in conjunction with the beam-

bolt elements (Section 11.17) to model plates connected by bolts.

2.7.11 3D-shell element
Overview
One characteristic of the shell elements described earlier is that

the change in thickness of the element is not explicitly calculated
from the element degrees of freedom. This is because the zero
stress through the shell thickness assumption is used in the material
descriptions.
However, in large strain analysis, the change in thickness can

become important. For example, during out-of-plane bending, the
material in compression thickens, and the material in tension thins
(for nonzero Poisson's ratio). Hence the midsurface is no longer
exactly halfway between the top and bottom surfaces. This effect
is shown in Fig 2.7-21.

Chapter 2: Elements
Midsurface
a) Undeformed configuration
Material particles
initially on midsurface
b) Deformed configuration
Figure 2.7-21: Kinematics of pure bending
This change in thickness is naturally modeled when quadratic

3D solid elements are used, because the nodes on the top and
bottom surfaces can move relative to the nodes on the midsurface,
as shown in Fig 2.7-22. However 3D solid elements tend to lock
when they are very thin, so that they are unsuitable for bending
analysis of thin structures.

2.7: Shell elements
Figure 2.7-22: Pure bending modeled with quadratic 3D elements
In the 3D-shell element, the change in thickness of the element

is modeled using control vectors, as shown for the case of pure
bending in Fig 2.7-23. The motion of the control vectors is
controlled by element degrees of freedom at the shell midsurface,
as discussed in more detail below.

Chapter 2: Elements
Top control vector
Bottom control vector
Top control vector shrinks
Bottom control vector elongates
Figure 2.7-23: Pure bending modeled with 3D-shell elements
Because the change in thickness is explicitly calculated from

element degrees of freedom, the assumption of zero stress through
the shell thickness is not used in the 3D-shell element.
In addition, the 3D-shell elements use MITC tying rules to

relieve shear locking. Therefore these elements are suitable for
out-of-plane bending analysis of thin structures, even for large
bending strains.
The 3D-shell elements can be used with 3 or 4 nodes. The 4-

node element is recommended for general use.
The 3D-shell elements can be used in static, implicit dynamic,

explicit dynamic or frequency analysis.

2.7: Shell elements
The 3D-shell elements cannot be used in the following types of

analysis: cyclic symmetry or periodic symmetry, thermo-
mechanical-coupling analysis, in which the structural and thermal
models are specified by two input files.
The following features available for other shell elements are not
available for the 3D-shell element: 6-, 8-, 9-, 16-node elements,
composite (multilayer) shells, transition elements, anisotropic
failure criteria. When section results are requested, the membrane
strains and curvatures are not available.
When using contact groups with true offsets (CGROUP ...

OFFSET-TYPE=TRUE), the 3D-shell elements contribute to the
offsets. However, the offset at a 3D-shell element node is
computed as half the current shell thickness at the node, and this
offset is used regardless of whether the contact occurs on the shell
top surface or on the shell bottom surface.
The 3D-shell is more fully described in the following reference:
ref. T. Sussman & K.J. Bathe, 3D-shell elements for

structures in large strains, Computers & Structures, 122,
2-12, 2013.
Kinematics and degrees of freedom of the 3D-shell element
The kinematics of the 3D-shell element are similar to the

kinematics of the quadratic 3D element. The correspondence
between the quadratic 3D element and the 3D-shell element is
shown in Fig 2.7-24. Fig 2.7-24(a) shows a cross-section of a shell
structure that has undergone large deformations. Three material
particles are labeled: A at the midsurface, B at the top surface and
C at the bottom surface. In both elements, the position of material
particle A is given by the position of a node at A. Now consider
the material particles B and C. In the quadratic 3D element, the
positions of material particles B and C are given by the positions of
nodes at B and C. In the 3D-shell element, the positions of
material particles B and C are given by the position of the node at
A, and also two control vectors. In both elements, the line of
material particles between points B and C is specified by the
positions of material particles A, B, C.

Chapter 2: Elements
a) Material particles in thin structure

Node on top surface
Node on midsurface
Node on bottom surface
b) Kinematics of quadratic 3D element
Top control vector

Node on midsurface
Bottom control vector
Line of
material
particles
c) Kinematics of 3D-shell element
Figure 2.7-24: Correspondence between quadratic 3D element and 3D-shell element
Fig. 2.7-25 shows the corner of a 3D-shell element, with its top and
bottom surfaces described by control vectors. Initially the control
vectors are equal and opposite. During deformations, the control
vectors can evolve independently. Thus in the deformed
configuration, the control vectors are in general not equal and
opposite, as shown in Fig 2.7-25(b).

2.7: Shell elements
Top surface
Bottom surface
a) Undeformed configuration b) Deformed configuration,

warping rotations used
c) Deformed configuration,
warping rotations not used
Figure 2.7-25: Control vectors at 3D-shell element node
The control vector motions are governed by element degrees of

freedom. For ease of use of the element, the element degrees of
freedom include the same degrees of freedom as for the usual shell
elements:
x, y, z translations
, rotations (5 DOF node), or x , y , z rotations (6 DOF
node)
and extra degrees of freedom:
constant and linear thickness incremental strains

, rotations (5 DOF node), or x , y , z rotations (6 DOF
node)
The elongations of the control vectors are governed by the constant

and linear thickness incremental strains. The directions of the

Chapter 2: Elements
control vectors relative to each other are governed by the additional

rotations, which are termed "warping rotations". (The term
"warping rotation" is used because the additional rotational degrees
of freedom cause the material particles initially on a straight line to
warp into a curved quadratic line.)
Typically the 3D-shell element is used without warping rotations,

then the extra degrees of freedom are the constant and linear
thickness strains. In this case the control vectors always point in
opposite directions, but in general have different lengths after
deformations. And material particles that were initially on a
straight line remain on a straight line after deformations. This
situation is shown in Fig 2.7-25(c).
All of the considerations for selection of 5 and 6 DOF nodes

discussed in Section 2.7.3 directly apply to the 3D-shell element
nodes.
It is not allowed to prescribe or fix any of the extra degrees of

freedom. If any of the usual rotations are prescribed or fixed, then
the AUI automatically fixes the corresponding warping rotations.
Material models and formulations for the 3D-shell element
Material models for the usual shell elements are developed using
the assumption of zero stress through the shell thickness. Hence
these models do not directly apply to the 3D-shell element.
The following material models are implemented for the 3D-shell

element: elastic-isotropic, plastic-orthotropic, plastic-cyclic,
Mooney-Rivlin, Ogden, Arruda-Boyce, hyper-foam, Sussman-
Bathe, eight-chain, three-network.
Small displacement/small strain, large displacement/small strain or

large displacement/large strain formulations can be used with these
material models, as described under the material models
descriptions.
Material model notes
The thermal strains are assumed to be small in all material models,

2.7: Shell elements
including the rubber-like material models.
When stresses are output in the shell local coordinate system, the
coordinate system is the r, s, t system shown in Figure 2.7-6(b).
No shear correction factor is used in the 3D-shell elements.
Mixed u/p formulation
The mixed u/p formulation should be used when the material is

almost incompressible, to avoid volumetric locking. When the u/p
formulation is used, by default the assumed pressure field is
p = p0 + p1t where p0 and p1 are the pressure degrees of freedom,
and t is the isoparametric coordinate through the shell thickness.
Note that it is necessary to allow a linear variation of pressure
through the shell thickness in order to model out-of-plane bending.
It is also allowed to select the number of pressure degrees of

freedom in the plane of the shell, and the number of pressure
degrees of freedom used through the shell thickness. The case
p = p0 + p1t corresponds to one pressure degree of freedom in the
plane of the shell and two pressure degrees of freedom through the
shell thickness.
Incompatible modes can be used in the 4-node 3D-shell element.

However incompatible modes and the u/p formulation cannot be
used together.
The tables at the end of this section show under what conditions the
incompatible modes and u/p formulations are used. Notice that
these conditions are material dependent.

Chapter 2: Elements
Element group input
3D-shell elements are placed in a shell element group (EGROUP

SHELL), with the following additional command parameters:
SHELLTYPE, WARPING_ROTATIONS, NELPRS, NELPT,
TYING. SHELLTYPE=3DSHELL activates the 3D-shell
elements.
Warping rotations are specified using parameter

WARPING_ROTATIONS. Warping rotations are not used by
default.
The u/p formulation is specified using parameters NELPRS and

NELPT. NELPRS is the number of pressure degrees of freedom in
the plane of the shell, and NELPT is the number of pressure
degrees of freedom through the shell thickness. These parameters
have the possible values 0 (no u/p), >0, (u/p) or AUTOMATIC (u/p
depending upon the material), see tables at the end of this section.
Tying is specified using parameter TYING. TYING is used by

default.
The interaction between incompatible modes, u/p formulation and

the material models is given in the following tables. The default
behavior for each material is given in the first row of each table.

2.7: Shell elements
Materials that do not allow the u/p formulation: Hyper-foam material
KINEMATICS EGROUP EGROUP Result

INCOMPATIBLE- INCOMPATIBLE- NELPRS /
MODES MODES NELPT
AUTOMATIC or DEFAULT AUTOMATIC No u/p or incompatible
NO modes
YES DEFAULT AUTOMATIC Incompatible modes used
NO AUTOMATIC No u/p or incompatible
modes
YES AUTOMATIC Incompatible modes used
AUTOMATIC or DEFAULT 0/0 No u/p or incompatible
NO modes
YES DEFAULT 0/0 Incompatible modes used
NO 0/0 No u/p or incompatible
modes
YES 0/0 Incompatible modes used
AUTOMATIC or DEFAULT A/B Error
NO
YES DEFAULT A/B Error
NO A/B Error
YES A/B Error

Chapter 2: Elements
Materials that allow the u/p formulation, but for which the u/p formulation is
not the default: Elastic material

MODES MODES NELPT
AUTOMATIC or DEFAULT AUTOMATIC No u/p or incompatible
NO modes
YES DEFAULT AUTOMATIC Incompatible modes
used
NO AUTOMATIC No u/p or incompatible
modes
YES AUTOMATIC Incompatible modes
used
NO modes
YES DEFAULT 0/0 Incompatible modes
used
modes
YES 0/0 Incompatible modes
used
AUTOMATIC or DEFAULT A/B u/p used with A*B
NO pressure dofs
NO A/B u/p used with A*B
pressure dofs
YES A/B Error

2.7: Shell elements
Materials that allow the u/p formulation, and for which the u/p formulation is
the default: Plastic-cyclic, plastic-orthotropic, Ogden, Mooney-Rivlin,
Arruda-Boyce, Sussman-Bathe materials, eight-chain, three-network model

MODES MODES NELPT
AUTOMATIC or DEFAULT AUTOMATIC Automatic u/p
NO
YES DEFAULT AUTOMATIC Incompatible modes
used
NO AUTOMATIC Automatic u/p
YES AUTOMATIC Incompatible modes
used
NO modes
YES DEFAULT 0/0 Incompatible modes
used
modes
YES 0/0 Incompatible modes
used
AUTOMATIC or DEFAULT A/B u/p used with A*B
NO pressure dofs
NO A/B u/p used with A*B
pressure dofs
YES A/B Error
Automatic u/p: u/p used with 1 / 2 pressure dofs, i.e. p = p0 + p1t
2.8 Pipe elements

Two types of pipe element are available: pipe-beam element
and pipe-shell element. The program automatically selects one of
these element types depending on the setting used for the pipe
element. If no ovalization or warping of the cross-section are
present, the pipe-beam element is used. If ovalization or warping of
the cross-section are present, the pipe-shell element is used. Only

Chapter 2: Elements
4-node elements are supported. Figure 2.8-1 shows the

conventions used.
Assemblages of pipe elements can be subjected to internal

pressure, see Section 2.8.2.
ref. KJB The pipe-beam element has 6 displacement degrees of freedom

Sections 5.4.1, per node (3 translations and 3 rotations, see Fig. 2.8-1).
6.5.1
tab
Convention for
pipe stresses Pipe global displacement
taa
degrees of freedom
tcc
Integration point
tac
b
3 4 c a
s
t r
1 2
f
q R
Z
Nodes 1 to 4
and KK must
lie in one plane
Y (r-s plane).
X
Auxiliary node KK
Element coordinate system convention:

longitudinal direction = a-direction (or h)
radial direction = b-direction (or z)
circumferential direction = c-direction (or x)
Figure 2.8-1: Pipe element configuration and displacement

degrees of freedom and stress convention
The pipe-shell element has, in addition to the 6 pipe-beam

degrees of freedom, 3 or 6 ovalization and 3 or 6 warping degrees

2.8: Pipe elements
of freedom per node allowing the cross-section to ovalize and

warp. Ovalization refers to the deformation of the pipe-skin in the
cross-sectional plane, and warping refers to the deformation of the
pipe-skin in the direction normal to the pipe cross-section. The
additional degrees of freedom are allocated as follows: 3
ovalization and 3 warping degrees of freedom for in-plane bending
deformation; 3 ovalization and 3 warping degrees of freedom for
out-of-plane bending deformation; 6 ovalization and 6 warping
degrees of freedom for combined in-plane and out-of-plane
bending. The user determines whether to activate one or both of
in-plane or out-of-plane warping/ovalization.
t Pipe element
r
s Mt
Mt
(in-plane moment)
Plane of pipe element

Auxiliary node
a) In-plane-only bending conditions

Pipe element
t r
Ms
Ms s
(out-of-plane moment)
Plane of pipe element

Auxiliary node
b) Out-of-plane bending conditions

Figure 2.8-2: Loading conditions for the pipe element
Note that the in-plane bending in this terminology corresponds

to the bending in the r-s plane and the out-of-plane bending

Chapter 2: Elements
corresponds to the bending in the r-t plane (see Fig. 2.8-1 and
2.8-2).
The ovalization degrees of freedom are based on the von

Karman ovalization modes, with the following circumferential
displacement:
Nc Nd
w = cm sin 2m + d m cos 2m
m =1 m =1
where cm and dm are generalized ovalization displacements which

are interpolated from their nodal values using polynomial shape
functions. The cm terms correspond to in-plane bending and the dm
terms correspond to out-of-plane bending. Nc and Nd are the
number of von Karman modes used in the analysis for in-plane and
out-of-plane bending actions (Nc = Nd = 3 is always used in
ADINA).
Similarly, warping of the cross-section is described by a

longitudinal displacement field analogous to the interpolation for
the ovalization displacement
Nc Nd
w = pm cos 2m + qm sin 2m
m =1 m =1
where pm and qm are generalized ovalization displacements which

are interpolated from their nodal values using polynomial shape
functions. The pm terms correspond to in-plane bending and the qm
terms correspond to out-of-plane bending.
The formulation of the pipe-shell element (ovalization/warping
included) is described in more detail in the following references:
ref. K.J. Bathe and C.A. Almeida, "A Simple and Effective
Pipe Elbow Element Linear Analysis," J. Appl. Mech.,
Transactions of the ASME, Vol. 47, No. 1, pp. 93-100,
1980.
ref. K.J. Bathe, C.A. Almeida and L.W. Ho, "A Simple and
Effective Pipe Elbow Element Some Nonlinear
Capabilities," Comp. & Structures, Vol. 17, No. 5/6, pp.

2.8: Pipe elements
659-669, 1983.
The pipe-shell formulation also takes into account the effects of

stiffening due to internal pressure, and the effects of interactions
between pipes and flanges, and pipes with different radii, see Fig.
2.8-3. The formulations are described in the following references:
Pipe Elbow Element Interaction Effects," J. Appl.
Mech., Transactions of The ASME, Vol. 49, pp. 165-171,
1982.
Pipe Elbow Element Pressure Stiffening Effects," J.
Appl. Mech., Transaction of the ASME, Vol. 49, pp. 914-
916, 1982.

Chapter 2: Elements
Rigid flange
Pipe centroidal axis
a) Pipe connected to a rigid flange
Pipe centroidal axis
Pipe elements
R1
Junction
R2
b) Pipes of different radii joined together
Figure 2.8-3: Interaction effects with the pipe element
Whereas the pipe-beam element models the usual beam strains

only, the pipe-shell element models in addition ovalization and
warping effects by including the following strains (see Fig. 2.8-1
for notation):
The ovalization is included by the von Karman ovalization

modes with
w
w =

and
w sin w cos 1 2 w
e =
ov

aa
R a cos ( R a cos )2 2


2.8: Pipe elements
abov = 0 (2.8-1)
1 w
acov =
R a cos
1 2 w
eccov = w
+
a 2 2
The warping is included using shell theory as
cos 1 w
w
eaa = 1
R a cos R a cos
abw = 0
1 w w sin
acw = (2.8-2)
a R a cos
eccw = 0
Here, a is the pipe mean radius, R is the bend radius and and
are the angles defined in Fig. 2.8-1.
In the formulation of the element the displacements w for the

ovalization in Eq. 2.8-1 and w for the warping in Eq. 2.8-2 are
interpolated by sine and cosine functions associated with the
ovalization and warping degrees of freedom.
For the pipe-shell element, continuity of ovalization and

warping generalized displacements is enforced by assigning the
same degrees of freedom to the node which the adjacent elements
are attached. However, a restriction in ADINA is that the continuity

Chapter 2: Elements
of the derivative w is only enforced between elements of the

same element group. Hence, elements which model a region of
significant longitudinal variation of the ovalization should be
placed into one element group.
A further restriction when using the pipe-shell element in

ADINA is that for continuity in the ovalization and warping
displacements, adjacent elements should lie in one plane. This is
because the ovalization and warping displacements are measured
using the angle for the s-axis of each element. In practice,
adjacent bend elements lying in different planes are usually joined
by straight sections (or a rigid flange) and these cases can be
accurately modeled with the pipe-shell elements, because all
ovalization and warping displacements are approximately zero in
the middle of the straight section and at the rigid flange.
Note that without warping/ovalization, the pipe-beam and pipe-

shell elements give the same stress response. Hence, the
ovalization and warping of the pipe-shell element can be thought of
as additional kinematic modes to the pipe-beam response which
result in an appropriate reduction of stiffness of pipe bends.
For the pipe-shell element, the following use of the

ovalization/warping degrees of freedom is recommended:
Activate the in-plane ovalization/warping degrees of

freedom if the pipe elements undergo only in-plane
deformations, see Fig. 2.8-2(a).
Activate the out-of-plane ovalization/warping degrees of

freedom if the pipe elements undergo only out-of-plane
deformations, see Fig. 2.8-2(b).
Activate all ovalization/warping degrees of freedom if the

pipe elements undergo general three-dimensional deformations.
In order to model the presence of a flange at a pipe node (of a

pipe-shell element), all ovalization/warping degrees of freedom
should be deleted at the node. The deletion of all nodal
ovalization/warping degrees of freedom implies that the pipe
section at that node remains circular. In addition, the option of

2.8: Pipe elements
enforcing a zero pipe-skin slope must be employed to model the

flange (see Fig. 2.8-4).
The symmetry condition, also shown in Fig. 2.8-4, implies zero

warping but not necessarily zero ovalization.
For a pipe elbow, a nondimensional geometric factor is

defined as
R
=
a 2
1 2
where R = pipe bend radius, = pipe wall thickness, a = pipe cross-

section mean radius, v = Poisson's ratio.

Chapter 2: Elements
Wb
=0
a
Wb a Ovalization and warping
equal to zero at node j
for element i.
Node j
Element i
Wb
=0
a
Node j
ai+1 ai Ovalization and warping
equal to zero at node j
for elements i and i + 1.
Element i + 1 Element i
Wb
(a) Flange conditions
Wb
Symmetry plane, =0
a a
Node j Nonzero ovalization and
zero warping at node j
for element i.
Element i Wb
(b) Symmetry condition
Figure 2.8-4: Zero pipe-skin slope conditions
With a decreasing value of , the ovalization/warping effects of

the cross-section become more pronounced. Note that for a straight
pipe element, R and thus the cross-section does not
ovalize/warp except if connected to another pipe element with a
finite bend radius.
The pipe elements can be used with the following material

models: elastic-isotropic, plastic-bilinear, plastic-multilinear,
thermo-isotropic, thermo-plastic, creep, plastic-creep,
multilinear-plastic-creep, creep-variable, plastic-creep-
variable, multilinear-plastic-creep-variable.

2.8: Pipe elements
The pipe elements can be used with a small displacement or a

large displacement formulation. In the small displacement
formulation, the displacements and rotations are assumed to be
infinitesimally small. In the large displacement formulation, the
displacements and rotations are assumed to be large. In all cases,
only small strains are assumed.
All of the material models listed above can be used with either
formulation. The use of a linear material with the small
formulation.
In the large displacement formulation, large displacement
effects are only included for the overall beam displacements and
not for the warping/ovalization degrees of freedom. Hence, large
displacement effects are not accounted for in the element cross-
sectional deformations.
An element is to be considered a nonlinear pipe element if a

large displacement analysis is performed or a nonlinear material is
used and/or pipe internal pressure is present.
2.8.2 Pipe internal pressures
Both types of pipe elements support internal pressure loading.

The load vector corresponding to that pressure is calculated as
follows:
( p) ( p)
First, two axial forces Fa1 and Fa2 are acting at the element
end nodes, as shown in Fig. 2.8-5. Their magnitudes are
Fa1( p ) = ai2 p1 ; Fa2( p ) = ai2 p2
where p1 and p2 are the internal pressures at nodes 1 and 2 (input as

nodal point pressures to the program), and ai is the internal radius
of the cross-section. Also, if the element is curved, transverse
forces are generated by the internal pressure. The transversal force
acting on the elementary axial length dL of the pipe axis is

Chapter 2: Elements
ai2
dFT = p s dL
R
and the nodal transversal forces are obtained as
Fk = hk d FT
L
where hk is the interpolation function corresponding to node k.
l
4
3l
(p) 2
dF T dL l
s Fa(p)
2
1l
R l Auxiliary node
(p)
Fa
1
Figure 2.8-5: Forces due to pipe internal pressure
The nodal forces due to the internal pressure for an element

represent a self-balanced system. Consequently, if a structure is
free to deform, as in the case of the cantilever in Fig. 2.8-6(a), the
support reactions are equal to zero. However, in the case of a
restrained system, the reactions at a support balance the forces and
moments which are a result of the internal stresses and the nodal
forces that are equivalent to the internal pressure, see Fig. 2.8-6(b).
Note that internal pressure acts against ovalization and hence

increases the stiffness of the ovalization degrees of freedom.

2.8: Pipe elements
RT
Ra MA Ra = RT = MA = 0
l
A
l
Internal
pressure p
a) Cantilever pipe free to deform (zero reactions)
RA RB
2a+d l l l l
A B
d/2 p
In elastic deformations:
RA =RB = paip(ai -2na) a = mean radius
ai = a - d/2 n = Poisson's ratio
= internal radius
b) Fully restrained pipe
Figure 2.8-6: Support reactions for pipe loaded by internal pressure
In the stress calculation the following assumptions are made

regarding the internal pressure effects.
For the pipe-beam element it is assumed that the hoop stress

cc is determined by the given pressure, t +t cc = t +t cc( p ) ,
t +t
where t +t
cc( p ) is the hoop stress equivalent to the pressure
t +t
p (using thin-walled cylinder theory). Then, using that the
normal stress through the pipe skin is equal to zero, the axial
stress t +t aa is obtained as

Chapter 2: Elements
t +t
aa =t +t E ( t +t eaa t +t eaaIN t +t eTH ) +
t +t t +t
cc( p )
where E and t +t are the Young's modulus and Poisson's

t +t
ratio, and t +t eTH is the thermal strain corresponding to the

temperature t +t ; t +t eaa and t +t eaa
IN
are the total mechanical and
inelastic strains, respectively.
For the pipe-shell element the basic condition regarding the

deformation of the pipe skin is that the material is free to
deform in the circumferential direction. Hence, the total hoop
strain is
1 t +t IN
t +t
ecc = t +t ( t +t eaa t +t eaa
IN
) eaa + t +t OV
e cc
2
1 2 ( p )
+ (1 + ) t +t eTH + cc
E
where t +t ecc
OV
is the strain due to ovalization of the cross-section.
This relation is based on the assumption that the inelastic
deformation is incompressible (as it is in metal plasticity and
creep).
Pressures at all nodal points of a piping model using the pipe-

beam or pipe-shell elements (and other elements) must be defined
if internal pressure effects are to be included in some of the
elements. The internal pressure loads are only used for the pipe-
beam and pipe-shell elements. See Section 5.7 for information
about specifying internal pressures.
The element stiffness and mass matrices and force vectors are in
all formulations evaluated using Gauss or Newton-Cotes numerical
integration in the axial and the thickness directions and the
composite trapezoidal rule along the circumference of the pipe
section. The locations and labeling of the integration points are
given in Fig. 2.8-7.

2.8: Pipe elements
Newton-Cotes formulas:
1 2 3
Order -3
Node 1 Node 2
1 2 3 4 5
Order -5
Node 1 Node 2
1 2 3 4 5 6 7
Order -7
Node 1 Node 2
Gauss formulas:
1
Order 1
Node 1 Node 2
1 2
Order 2
Node 1 Node 2
1 2 3
Order 3
Node 1 Node 2
1 2 3 4
Order 4
Node 1 Node 2
a) Integration point locations on pipe centroidal axis (label INR)
Figure 2.8-7: Locations and labeling of integration points

Chapter 2: Elements
s s
b
b
23 5
1 3 4
12
t t
Gauss: order 3 Newton-Cotes: order 5

b) Integration point locations along thickness direction (label INB)
t t
3
2 4 2

3 1 5 1

s s

6 8
4 7
Order 4 Order 8
t
t
5 4 3 9 8 76

10 5 4
6 11 3
2 2
1 12 1
7 13
s
14 24 s
12 23
8 15

9

11 16 22
10 17 18 19 20 21
Order 12 Order 24
c) Integration point locations along circumferential direction (label INC)
For the pipe-beam elements the following integration orders

should be used (and are default in ADINA):
1-point and 3-point Gauss integration along the axial

direction for the 2-node and 4-node elements, respectively
2-point Gauss integration through the thickness of the pipe
skin

2.8: Pipe elements
8-point integration along the circumference
For the pipe-shell element the following integration orders

should be used (and are default in ADINA):
3-point Gauss integration along the axial direction

2-point Gauss integration through the thickness of the pipe
skin
12-point integration along the circumference if only the in-
plane ovalization degrees of freedom are activated
24-point integration along the circumference if the out-of-
plane ovalization degrees of freedom are activated
The recommended numerical integration orders for nonlinear

analysis are:
When no inelastic strain effects are considered, use the same

integration orders as in linear analysis.
When material nonlinear conditions are considered, use a
higher integration order through the pipe skin if the pipe skin
progressive yielding or creep is to be represented. For the
integration along the length of the elements and the
circumference of the cross-section, an integration order higher
than that used in linear analysis may, for the same reason, also
be appropriate.
The integration order along the axial r-direction (INR)
should not be changed from the default value for the pipe-beam
element otherwise the pipe-beam element will lock. The same
locking phenomenon is also observed for iso-beam elements;
see Section 2.5.2 for more details.
2.8.4 Element mass matrices
The pipe element can be used with a lumped or a consistent

uses a lumped mass.
The consistent mass matrix associates mass only with the pipes
ref. KJB
Section translational and rotational degrees of freedom and is calculated
5.4.1 using the isoparametric formulation with the displacement
interpolations given on p. 408 of ref. KJB. Hence, no masses are

Chapter 2: Elements
associated with the ovalization/warping degrees of freedom.
The lumped mass for translational degree of freedom i is

M i where M is the total element mass, L is the element
L
length and i is a fraction of the length associated with node i. The
1 1
rotational lumped mass is M i ( d o2 + di2 ) where
3 L 4
d o and di are the outside and inside diameters of the pipe. The
lumped ovalization and warping mass for each of the ovalization
and warping degrees of freedom are set to be equal to the
i
translational lumped masses M .
L

specified scalar ETA (except in explicit analysis).
You can request that ADINA either print or save stresses or forces.

variable names.
Elastic-isotropic: STRESS(ABC), STRAIN(ABC),

EQUIV_INTERNAL_AXIAL_PRESSURE,
EQUIV_INTERNAL_HOOP_PRESSURE
Elastic-isotropic with thermal effects: STRESS(ABC),

STRAIN(ABC), FE_EFFECTIVE_STRESS,
EQUIV_INTERNAL_HOOP_PRESSURE, THERMAL_STRAIN,
ELEMENT_TEMPERATURE
Thermo-isotropic: STRESS(ABC), STRAIN(ABC),

THERMAL_STRAIN(ABC),

2.8: Pipe elements
EQUIV_INTERNAL_HOOP_PRESSURE,
ELEMENT_TEMPERATURE
Plastic-bilinear, plastic-multilinear: PLASTIC_FLAG,

PLASTIC_STRAIN(ABC), THERMAL_STRAIN(ABC),
ACCUM_EFF_PLASTIC_STRAIN, ELEMENT_TEMPERATURE

variable: PLASTIC_FLAG, NUMBER_OF_SUBINCREMENTS,
PLASTIC_STRAIN(ABC), CREEP_STRAIN(ABC),
THERMAL_STRAIN(ABC), ELEMENT_TEMPERATURE,
ACCUM_EFF_PLASTIC_STRAIN,
In the above lists,
STRESS(ABC) = STRESS-AA, STRESS-AB, STRESS-

AC, STRESS-CC, STRESS-BB
with similar definitions for the other abbreviations used above. See
Fig. 2.8-1 for the pipe stress coordinate system convention.
Nodal forces: In linear analysis, the forces/moments at the

element nodes are
F=K U
where K is the element stiffness matrix and U is the vector of

nodal point displacements which may include ovalizations and
warpings. Hence, the forces/moments in F will be equal to the
forces/moments applied externally or by adjoining elements, and

Chapter 2: Elements
are not necessarily the internal section forces/moments.

names NODAL_FORCE-R, NODAL-FORCE-S,
NODAL_FORCE-T, NODAL_MOMENT-R, NODAL_MOMENT-
S, NODAL_MOMENT-T.
The end forces/moments are computed at the element local
nodes. In the AUI, element local nodes are defined as element
points of type label. For example, to access the result computed at
element 5, local node 2, define an element point of type label with
element number 5, label number 2.
The pipe elements are available in ADINA to model general

pipe networks subjected to general boundary and loading
conditions including internal pressure and thermal loading.
When ovalization effects are insignificant (straight pipes or

pipes of large thickness) the 2- or 4-node pipe-beam element
should be used. In this case, linear or nonlinear behavior can be
modeled with large displacements but small strains.
When ovalization effects are important, the 4-node pipe-shell

element should be used. In this case, linear or nonlinear behavior
can be modeled with the assumptions that the displacements are
always relatively small and that the strains are small. The
displacements are always relatively small because in the calculation
of the flexibility due to ovalization/warping the pipe radius is
constant (and equal to the initial pipe radius) and no large
displacement strain terms are included in the ovalization/warping
calculations. The only large displacement strain terms accounted
for are those corresponding to the beam behavior.
However, if the large displacement option is employed, the
change of the pipe radius is taken into account in the calculation of
the element nodal forces due to the pipe internal pressure.
Pipe Elbow Element Linear Analysis," J. Appl. Mech.,
Transactions of the ASME, Vol. 47, No. 1, pp. 93-100,
1980.

2.8: Pipe elements
Since the pipe-beam element stiffness matrix is calculated using

numerical integration, it is clear that the linear 2-node Hermitian
beam element (see Section 2.4.1) for which the stiffness matrix is
evaluated in closed-form can be much more cost-effective in
analyzing linear piping systems with negligible ovalization,
warping and internal pressure effects.
2.9 General and spring/damper/mass elements

2.9.1 General elements
General elements are linear elements which can have from one
node to as many nodes as there are in the structure.
General elements are useful in the definition of special elements

whose formulations are not directly available in ADINA. General
elements are also useful in providing flexibility in obtaining any
desirable linear stiffness/mass/damping matrices.
You directly enter the stiffness matrix (and in dynamic analysis,

the mass and damping matrices) of the general element.
The dimension of the stiffness/mass/damping matrices cannot
exceed ND, defined as NDOFIELD, where IELD is the number of
nodes of the general element and NDOF is the number of active
degrees of freedom for each node (set in the MASTER command,
with a default of 6). The components for the
stiffness/mass/damping matrices must be defined for all ND
degrees of freedom. Note that only the upper triangular half of the
matrices needs to be input (the stiffness/mass/damping matrices
must be symmetric).
For example, for a 4-node 3D solid element with three active
master degrees of freedom, ND=3*4=12, the stiffness matrix is
12x12 symmetric and the internal force vector is calculated using

Chapter 2: Elements
Fx1 k1,1 k1,2 k1,3 k1,4 k1,5 k1,6 k1,7 k1,8 k1,9 k1,10 k1,11 k1,12 u1x
1 k2,2 k2,3 k2,4 k2,5 k2,6 k2,7 k2,8 k2,9 k2,10 k2,11 k2,12 u1y
Fy
Fz1 k3,3 k3,4 k3,5 k3,6 k3,7 k3,8 k3,9 k3,10 k3,11 k3,12 u1z
2
Fx k4,4 k4,5 k4,6 k4,7 k4,8 k4,9 k4,10 k4,11 k4,12 u x2
F 2 k5,5 k5,6 k5,7 k5,8 k5,9 k5,10 k5,11 k5,12 u y2
y2
Fz k6,6 k6,7 k6,8 k6,9 k6,10 k6,11 k6,12 u z2
3 = k7,7 k7,8 k7,9 k7,10 k7,11 k7,12 u x3
Fx
Fy3 k8,8 k8,9 k8,10 k8,11 k8,12 u 3y
3
Fz k9,9 k9,10 k9,11 k9,12 u z3
Fx4 sym k10,10 k10,11 k10,12 u x4
4
Fy k11,11 k11,12 u y4
F 4 k12,12 u z4
z
The stiffness matrix is entered using the following command:
MATRIX STIFFNESS 1 12
1 k1,1 k1,2 k1,3 k1,4 k1,5 k1,6 k1,7 k1,8 k1,9 k1,10 k1,11 k1,12
2 k2,2 k2,3 k2,4 k2,5 k2,6 k2,7 k2,8 k2,9 k2,10 k2,11 k2,12
3 k3,3 k3,4 k3,5 k3,6 k3,7 k3,8 k3,9 k3,10 k3,11 k3,12
4 k4,4 k4,5 k4,6 k4,7 k4,8 k4,9 k4,10 k4,11 k4,12
5 k5,5 k5,6 k5,7 k5,8 k5,9 k5,10 k5,11 k5,12
6 k6,6 k6,7 k6,8 k6,9 k6,10 k6,11 k6,12
7 k7,7 k7,8 k7,9 k7,10 k7,11 k7,12
8 k8,8 k8,9 k8,10 k8,11 k8,12
9 k9,9 k9,10 k9,11 k9,12
10 k10,10 k10,11 k10,12
11 k11,11 k11,12
12 k12,12
where, of course, the numerical values of the stiffness matrix

entries are substituted for ki,j. Note that in this case, all except for
the three translational degrees of freedom must be disabled from
the MASTER command.
An alternative is to define a user-supplied general element, see

Section 2.9.3.
If skew systems are used, then either all matrices must be input in
the global coordinate system and proper transformations are made
by ADINA if there are skew systems at the element group nodes, or

2.9: General and spring/damper/mass elements
all matrices must be input in the coordinate systems used at the

nodes and no transformations between global and skew systems are
made in ADINA (SKEWSYSTEMS = YES in the EGROUP
GENERAL or EGROUP SPRING command).
Note that the general element can be used to define a constant

damping matrix for other element types in ADINA. Alternative
ways to define damping are Rayleigh damping, concentrated
dampers and the damping feature in spring elements.
Forces are calculated for general elements using
F=K U
where U stores the displacements and rotation of the ND degrees of

freedom.
The forces are accessible in the AUI using the variable names
NODAL_FORCE-X, NODAL_FORCE-Y, NODAL_FORCE-Z,
NODAL_MOMENT-X, NODAL_MOMENT-Y, NODAL_MOMENT-
Z. The forces are with reference to the nodal skew systems, if skew
systems are used.
The damping force is simply calculated as
F=CU
and the inertia is calculated as
F=MU
For output purposes, stresses can also be calculated for general

elements if you define a stress-displacement matrix S. The stress
vector is obtained using
=SU
Note that the matrix S must be input as a full matrix and is of

dimension NSND, where NS equals the number of stress
components in (NS cannot exceed 600).

Chapter 2: Elements
Each element outputs the vector to the porthole file. This

vector is accessible in the AUI using the variable name
GENERAL_ELEMENT_STRESS. You select the component of
in the AUI by setting the label point number equal to the
component number in .
2.9.2 Linear and nonlinear spring/damper/mass elements
Linear spring single-degree-of-freedom element
The stiffness matrix has only 1 component and corresponds to a

grounded spring acting in the single (user-specified) degree of
freedom, see Fig. 2.9-1(a).
The mass matrix corresponds to a concentrated mass acting in

the single degree of freedom, see Fig. 2.9-1(b).
The damping matrix corresponds to a grounded damper acting

in the single degree of freedom, see Fig. 2.9-1(c).

u q
k
l
l
k m
Single Single Single Single

translational rotational translational rotational
DOF spring DOF spring DOF mass DOF mass
(a) Stiffness K = [k] (b) Mass M = [m]
.
u
c c
l l
Single Single
translational rotational
DOF damper DOF damper
(c) Damping C = [c]
Figure 2.9-1: Single-degree-of-freedom element
Linear spring two-degree-of-freedom element
The stiffness matrix is shown in Fig. 2.9-2(a) and corresponds

to a spring coupling the two (user-specified) degrees of freedom.
The lumped mass matrix is diagonal, see Fig. 2.9-2(b), and

corresponds to two concentrated masses acting in the two degrees
of freedom.
The consistent mass matrix is obtained using a linear

interpolation of the accelerations in the two degrees of freedom, see
Fig. 2.9-2(b) and ref. KJB, Example 4.5, pp. 166-170.
The damping matrix is shown in Fig. 2.9-2(c) and corresponds

to a damper coupling the two degrees of freedom.
MNO-G spring element
The element definition is the same as that of the single- or two-

Chapter 2: Elements
degree of freedom linear spring element.
Spring stiffness can be defined using a nonlinear force-

displacement curve as in the nonlinear spring element.
Spring stiffness can be in any direction defined by the command

EGROUP SPRING ... SKEWSYSTEM=YES
Damping coefficient can be dependent on the relative velocity

of the element end nodes as in the nonlinear damper element.
Element mass can vary with time as in the nonlinear mass

element.
k -k m/2 0
K = Mlumped =
-k k 0 m/2
k = spring constant
m/3 m/6
a) Stiffness matrix Mconsistent =
m/6 m/3
m = total mass
c -c of element
C =
-c c b) Mass matrix
c = damping constant
c) Damping matrix
Figure 2.9-2: Two-degrees-of-freedom element matrices
Nonlinear spring/damper/mass element
A nonlinear "spring" element is available in ADINA for static

and dynamic analysis. This element can be used to connect two
nodes or to attach a node to the ground. It can be a translational or
a rotational "spring" (see Figure 2.9-3). The stiffness, damping and
mass of the element can each be either nonlinear or linear.

Large displacements can be applied to nonlinear spring

elements.
In the case of a translational spring, the nonlinear stiffness is

defined by a multilinear elastic force versus relative-displacement
relationship. The corresponding curve can be different for negative
relative-displacement than for positive relative-displacement. The
input curve need not pass through the origin (0,0).
The nonlinear damping is given by a function of the form
N
FD = C U where F D is the damping force in the element, U
is the relative velocity between the element end nodes, and C and N
are real constants.
The mass of the element can vary with time (see Figure 2.9-4).

Chapter 2: Elements
Nonlinear translational stiffness: Nonlinear torsional stiffness:

ME
FE
U= U2 - U1 2 2
U2 U2
1 1
U1 U1
FE ME
Nonlinear translational damping: Nonlinear torsional damping:
MD
FD
U = U2 - U 1 2
2
U2 U2
1 1
FD U1 U1
MD
Nonlinear translational stiffness,

grounded spring:
FE n
U = U1 1
U1
Figure 2.9-3: Description of the nonlinear spring/damper/mass element
The relative displacement U is defined as follows: Let U1 and

U be the displacement vectors at local nodes 1 and 2, and 0x1 and
2
0 2
x be the respective initial positions of these nodes.
For initially coincident nodes (0x1 = 0x2), U = U id2 U iddx1
where id specifies a global direction in the global Cartesian
coordinate system, and is specified in the input. id = 1,2,3 for the
X, Y, Z translational directions and id = 4,5,6 for the rotation about
the X, Y, Z directions. There is also an option (using id = 0) to use
another arbitrary direction.

Force versus displacement or Force versus velocity or moment

moment versus rotation: versus torsional velocity:
FE FD
1
2 F D = CU N
u
Case 1: no rupture
2 Case 2: rupture
1 U
Mass versus time:

Total
mass
Time
Figure 2.9-4: Nonlinear stiffness, damping and mass
If the nodes are not initially coincident, then the element can be
used to define a translational spring along the direction given by
the vector from node 1 to node 2. Let U1 store only the
translational displacements, then the relative translational
displacement is given by U = (U 2
U )
1 ( 0
x 2 0 x1 )
, i.e., the
( 0
x 2 0 x1 )
relative displacement is the difference in displacements in the
direction of the line (spring) element.
In a similar manner, a torsional spring is defined between the

local nodes 1 and 2.
Similar definitions are used for the relative velocity.
Rupture based on the value of the relative displacement may or

may not be included in the nonlinear spring element (see Figure
2.9-4). If rupture is not desired, the last segments of the elastic-
force versus relative displacement curve are extrapolated when
necessary. If rupture is desired, then the elastic force in the
element is set to zero whenever the relative displacement is larger
(resp. smaller) than the last (resp. first) point on the curve. Note

Chapter 2: Elements
that this effect is reversible, as the force will be computed again

according to the curve if the displacement value comes back into
the curve displacement range.
Note that the linear spring element and the nonlinear spring
element act quite differently: the linear spring element connects
specific degrees of freedom, whereas the nonlinear spring element
connects nodes and acts in a specified direction. Likewise the input
for these elements is also different.
Element output
You can request that ADINA either print or save stress

resultants or forces.
Stress resultants: Each element outputs the following information

to the porthole file. This information is accessible in the AUI using
the given variable names.
Linear element: GENERAL_ELEMENT_STRESS (NS values per

element)
Nonlinear element: DAMPING_FORCE, ELASTIC_FORCE
Nodal forces: For linear elements, these are the elastic nodal
forces (in the global Cartesian reference system) equivalent to the
axial elastic stress. Any stresses due to damping are not included.
For nonlinear elements, these are the nodal forces (in the global
Cartesian reference system) equivalent to the axial elastic and
damping stresses.
The forces are accessible in the AUI using the variable names
NODAL_MOMENT-X, NODAL_MOMENT-Y, NODAL_MOMENT-
Z. The forces are with reference to the nodal skew systems, if
skew systems are used.
2.9.3 6DOF Spring Element
The 6DOF spring element is a generalized spring-damper

element which can be linear or materially-nonlinear only (MNO). It
can have single node, two coincident or two non-coincident nodes.
In each degree of freedom, the element stiffness can be defined as a
constant or using a force-displacement curve in the element

coordinate system. The damping coefficients are always constants

in units of force per unit velocity. Please note that mass property is
not available in the 6DOF spring element. In the case of the MNO
6DOF spring, the stiffness is defined by a multilinear elastic force
versus relative-displacement curve. The input curve can be
different for negative relative-displacement than for positive
relative-displacement. The input curve need not pass through the
origin (0.0).
The displacement (skew) system can be defined in the 6DOF

spring element to prescribe loads and constrains. Element
birth/death is also supported. Currently, the 6DOF spring element
is not supported in explicit dynamic analysis.
If a 6DOF spring element has a single node or two coincident

nodes, its element coordinate system must be defined using CID
(coordinate system used to define spring element) as shown in Fig.
2.9-5 and Fig. 2.9-6. For single node 6DOF spring element, it
corresponds to a grounded spring acting in the user-specified
degree of freedom.
Y CID
Z
X
y-element
z-element
x-element
GA
Figure 2.9-5: 6DOF spring element with single node

Chapter 2: Elements
Y CID
Z
X
y-element
z-element x-element
GA GB
Figure 2.9-6: 6DOF spring element with two coincident nodes
If a 6DOF spring element has two non-coincident nodes, its

element coordinate system can be defined using CID, orientation
vector or its axial direction as shown in Fig. 2.9-7, Fig. 2.9-8 and
Fig. 2.9-9.
In Fig. 2.9-7, the element coordinate system is defined by CID.

Note that GA and GB may or may not have displacement (skew)
coordinate system.
Y CID
Z
X
y-element
z-element x-element
GA
GB
Figure 2.9-7: 6DOF spring element with two non-coincident nodes

In Fig. 2.9-8, the element coordinate system is defined by

orientation vector using G0 or (Ax, Ay, Az). Note that (Ax, Ay,
Az) refers to the displacement (skew) coordinate system of GA.
z-element
(normal to plane)
y-element
(in plane) G0 or (Ax, Ay, Az)
plane of GA, GB, G0

x-element
GA
GB
Figure 2.9-8: 6DOF spring element with two non-noncoincident

nodes
In Fig. 2.9-9, a 6DOF spring element is defined with two non-

coincident nodes without G0, (Ax, Ay, Az) or CID. This defines a
1-D axial/torsional spring/damper. In this case, axial stiffness (or
damping) or torsional stiffness (or damping) or both must be
specified but all other stiffness (or damping) must not be specified.
x-element
GA
GB
Figure 2.9-9: 1-D spring element with two non-coincident nodes
Element output
You can request that ADINA either print or save stress resultants or
forces.
Stress resultants: Each element outputs the following information

Chapter 2: Elements
to the porthole file. This information is accessible in the AUI using

the given variable names.
STRESS_TRANS_R, STRESS_TRANS_S, STRESS_TRANS_T,

STRESS_ROT_R, STRESS_ROT_S, STRESS_ROT_T
The above results are always given in the element local coordinate
system (r,s,t).
Nodal forces: the forces (in the global Cartesian coordinate system)
are accessible in the AUI using the given variable names.

NODAL_MOMENT-X, NODAL_MOMENT-Y,
NODAL_MOMENTZ
2.9.4 User-supplied element
This feature is useful for implementing elements that have a

non-standard behavior such as an element that has different axial
and bending parameters, a nonlinear spring for which the force is
described as a nonlinear function of the displacement, among other
cases.
The non-standard element behavior must be coded in the
ADINA subroutine CUSERG. Subroutine CUSERG must calculate
the element stiffness matrix and nodal force vector. In certain types
of analysis (e.g., in a dynamic analysis), the mass and damping
matrices must also be calculated. Subroutine CUSERG can also
output the desired results into the corresponding output files.
A template and example for the user-supplied element

subroutine CUSERG is provided in file ovl160u.f. The
mentioned example will be discussed later in this section and
corresponds to a three-dimensional nonlinear translational spring
that takes into account geometric nonlinearities. Also, a
Makefile file is provided, which includes instructions for
compiling and linking the user-supplied element subroutine to
ADINA.

The following element types are allowed:
Triangular 3-node, 6-node, or 7-node 2-D solid element
Quadrilateral 4-node, 8-node, or 9-node 2-D solid element
Tetrahedral 4-node, 10-node, or 11-node 3-D solid element
Hexahedral 8-node, 20-node, 21-node, or 27-node 3-D solid

element
2-node, 3-node, or 4-node iso-beam element
4-node, 8-node, 9-node, or 16-node shell element
transition shell element with variable number of nodes up to

a maximum of 32 nodes.
The nodal connectivity of the user-supplied elements must be

compatible with the ordinary ADINA elements so that the user-
supplied elements can be generated by the AUI mesh generation
commands. This will allow the AUI to load and display the user-
supplied elements for post-processing purposes.
There are two ways within the AUI to activate the user-supplied
element feature. Note that regardless of the way chosen, the
material definition of the element can only be done with the
MATERIAL USER-SUPPLIED command (see AUI Command
Reference Manual). The material properties can be temperature
dependent or independent. However, this does not imply that a
user-supplied material subroutine is needed. In fact, any material
implementation needed for the user-supplied element must be done
within the user-supplied element subroutine CUSERG.
The first way to activate the user-supplied element feature

consists of basically two parts: material definition and element
group definition. The element group definition consists of applying
either the EGROUP TWODSOLID or EGROUP THREEDSOLID
or EGROUP ISOBEAM or EGROUP SHELL commands with the
parameter OPTION=USER-CODED (see AUI Command Reference
Manual). Clearly, the choice of element group command depends

Chapter 2: Elements
on the type of element the user-supplied element resembles. This is

the preferred way.
The second way to activate the user-supplied element feature

consists of applying the general element strategy, which has four
parts: material definition, matrix definition, matrix set definition
and element group definition.
As with any general element, the stiffness, mass and damping
matrices must be provided (of course, the mass and damping
matrices are necessary only for certain types of analysis). The
stiffness, mass and damping matrices are grouped into a matrix set
and the matrix set number is included in the element group
definition. In ordinary general elements (see Section 2.9.1), the
user directly specifies the above matrices. By implementing the
user-supplied element as a general element one can provide the
above matrices by either calculating all of them in subroutine
CUSERG or by calculating at least the stiffness matrix in
subroutine CUSERG and directly specifying the rest as it is done
for ordinary general elements.
The element subtype, material label and number of integration
points are provided to ADINA through a matrix definition. Note
that in the AUI the matrix definition is accessed as a user-supplied
matrix from the Matrix Set dialog box. The matrix definition has
the following parameters:
The element subtype to which the user-supplied element

resembles. The options are 2-D solid, 3-D solid, iso-beam or
shell.
The user-supplied material number.
The total number of user-supplied integration points NUIPT.

The default value of NUIPT is NUIT1NUIT2NUIT3.
The integration order in the first local direction NUIT1. If

NUIPT is entered, then NUIT1 is ignored.
The integration order in the second local direction NUIT2. If


The integration order in the third local direction NUIT3. If

It is important to highlight that the number of active degrees of

freedom per node in a user-supplied element is defined by the
IDOF parameter of the MASTER command (see AUI Command
Reference Manual). This fact may prevent users from creating
models that combine user-supplied elements with standard ADINA
elements which possess different active degrees of freedom per
node.
The standard number of active degrees of freedom per node are
2, 3 and 6 for the 2-D solid, 3-D solid and iso-beam elements,
respectively. Note that six degrees of freedom per node are always
assumed for the shell user-supplied element.
The nodes per individual user-supplied element can be assigned

with the ENODES command. In addition, the COORDINATES
NODE and NODESET commands can be useful when
implementing a model with user-supplied elements. Please see the
AUI Command Reference Manual for a description of these
commands.
The calls from ADINA to subroutine CUSERG are divided into

several phases, which are controlled by the integer variable KEY.
KEY = 1: This is the input phase and it is called once at the

beginning of the analysis. The working arrays
ARRAY(LGTH1,NUIPT) and IARRAY(LGTH2,NUIPT), which
must store real and integer history dependent variables (for all the
integration points of an element), respectively, must be initialized
to their proper values. Also the coordinates of the integration points
of an element XYZIPT(3,NUIPT) must be provided. These points
are used for storing, e.g., stress and strain results. The
XYZIPT(3,NUIPT) array is arranged through the sequence of 1 to
NUIPT or through the local integration orders NUIT1, NUIT2 and
NUIT3. The definitions of NUIT1, NUIT2 and NUIT3 follow the
standard ADINA (2-D solid, 3-D solid, iso-beam and shell)
elements local integration orders. Note that the array XYZ(3,IELD)
provides the element nodal coordinates and can be used to calculate
the requested integration point coordinates.

Chapter 2: Elements
KEY = 2: The element nodal forces RE(ND) must be calculated in

the global coordinate system, where ND is the total number of
degrees of freedom in an element. For this purpose, ADINA
provides the element displacements DISP2(ND) and DISP1(ND) at
the current and previous time steps, respectively. In addition,
material properties and other parameters are provided (see file
ovl160u.f), which can be useful. Note that the element nodal
forces are the equilibrating forces during the equilibrium iterations
and the resultant nodal forces when a solution converges. If the
element subtype is iso-beam, then the element nodal forces in the
local coordinate system REBM(ND) must also be provided.
In this phase, it is recommended to calculate the desired user-
calculated quantities for all the integration points of an element,
such as stresses, strains, stress state flag, etc. Depending on the
requested element group results (i.e., forces or stresses) this might
be a necessary step since ADINA writes the requested results to the
porthole file for post-processing. The real and integer user-
calculated quantities are stored in RUPLOT(100,NUIPT) and
IUPLOT(50,NUIPT), respectively. In order for ADINA to correctly
process the information stored in these arrays, the user should
follow the order specified in file ovl160u.f.
Finally, the real and integer history dependent variables (for all
the integration points of an element) stored in
ARRAY(LGTH1,NUIPT) and IARRAY(LGTH2,NUIPT),
respectively, must be updated.
KEY = 3: The element stiffness matrix AS(ND,ND) must be

calculated in this phase. It should be calculated in the global
coordinate system. As in the case of the element nodal forces, the
stiffness matrix can also be calculated via the element
displacements, material properties, and other variables and
parameters provided by the ADINA subroutine CUSERG.
KEY = 4: The user printout must be coded in this phase. In order

for ADINA to access it, use the AUI command PRINTOUT in your
input file.
KEY = 5: In this phase, the array AS(ND,ND) stores the element

mass matrix, which must be calculated in the global coordinate
system and in terms of the variables and parameters provided by
the ADINA subroutine CUSERG.

KEY = 6: In this phase, the array AS(ND,ND) stores the element

damping matrix, which must be calculated in the global coordinate
system and in terms of the variables and parameters provided by
the ADINA subroutine CUSERG.
A complete description of the argument list of subroutine

CUSERG is provided in file ovl160u.f.
The following steps provide a guide for using the user-supplied

element feature via the AUI commands:
1. Once the user-supplied element has been coded in file

ovl160u.f, compile and link it to ADINA with the
help of the provided Makefile.
2. Use the MATERIAL USER-SUPPLIED command to

define the material properties, control parameters and
working array sizes.
3. If following the preferred way, then use either the

EGROUP TWODSOLID or EGROUP
THREEDSOLID or EGROUP ISOBEAM or
EGROUP SHELL commands with the parameter
OPTION=USER-CODED to define the type of user-
supplied element.
If following the general element strategy, then
(a) use the MATRIX USER-SUPPLIED command to

define the type of user-supplied element to relate to
the stiffness, mass and damping matrices. Note that if
the mass or damping matrices are constants, then they
can be input directly with the MATRIX MASS or
MATRIX DAMPING commands, respectively;
(b) use the MATRIXSET command to define the general

element matrix set; and
(c) use the EGROUP GENERAL command with the label

number corresponding to the defined MATRIXSET

Chapter 2: Elements
and the parameter USER-SUPPLIED=YES to

conclude the user-supplied element definition.
4. Define the number of active degrees of freedom per

node in the user-supplied element with the IDOF
parameter of the MASTER command.
5. Define the rest of the input file (i.e., nodes, nodal

connectivity, boundary conditions, loads, etc.) with
the corresponding AUI commands and generate the
data file. Run ADINA to solve the problem and use
the AUI to display the results.
2.9.4.1 User-supplied element example
A three-dimensional nonlinear translational spring that takes

into account geometric nonlinearities is given as an example of the
user-supplied element subroutine CUSERG, which can be found in
file ovl160u.f.
The spring can be implemented with either the EGROUP

ISOBEAM or EGROUP GENERAL AUI commands, and will
have three translational degrees of freedom active per node.
Assume that initially the spring is undeformed and that nodes 1

and 2 are attached to the ends of the spring, then its undeformed
length is given by 0 L = 0 x 2 0 x1 .
The spring force is a function of the displacement, that is, F =

F(u). Consequently, at the next time step (say at time t) the force at
node 1 can be written as
L t 1 t 2
t
F1 = k ( L ) t
L
(x x) (2.9-1)
where t xi = 0 xi + u i and u i is the displacement vector of the i-th

node, L =t L 0 L, and, in this particular example, it will be
assumed that

k ( L ) = k0 + k1L (2.9-2)
with k0 and k1 being spring constants.
Note that the force at node 2 is

t
F 2 = t F1.
The stiffness matrix of the user-supplied element can be easily

obtained from the above equations by taking the partial derivative
of each nodal force vector with respect to all the degrees of
freedom of the user- supplied element. In this particular example,
the stiffness matrix is a 6 x 6 symmetric matrix.
2.10 Displacement-based fluid elements

The elements discussed in this section incorporate the following
assumptions:
Inviscid, irrotational medium with no heat transfer
Compressible or almost incompressible medium
Relatively small displacements
No actual fluid flow
For other types of fluid flow, use ADINA-F and ADINA-FSI

(see the ADINA-F Theory and Modeling Guide for details of the
formulations employed). In the remainder of this section, only the
ADINA fluid elements are discussed.
The types of problems for which the ADINA displacement-

based fluid elements can be employed are:
Static analyses, where the pressure distribution in the fluid

and the displacement and stress distribution in the structure is of
interest
Frequency analyses, where natural frequencies and mode

shapes of a structure/fluid medium are to be determined

Chapter 2: Elements
Transient analyses, where a pressure wave propagates

rapidly through the fluid, which does not undergo large motions
(transient acoustic problems)
However, in practice, the use of the displacement-based

elements is rather restricted to special applications in static and
dynamic analyses. The potential based element (see Section 2.11)
is much more general and is recommended for use for solving this
class of problems.
Simply stated, the displacement-based fluid elements can be

thought of as derived from the solid two- and three-dimensional
elements (see Sections 2.2 and 2.3) by using an elastic stress-strain
relation with a bulk modulus K and a zero shear modulus.
The elements can be employed in 2-D and 3-D analyses. Two-

dimensional fluid elements can be employed in planar and
axisymmetric analyses. Two-dimensional fluid elements must be
defined in the YZ plane, and axisymmetric elements must lie in the
+Y half plane (all nodal point coordinates must have positive y
values).
Although the fluid elements can be employed using a large

displacement formulation, the allowed fluid motion is relatively
small in a practical analysis, because the elements must not become
distorted. Actual flow of a fluid cannot be analyzed using the
elements. Use ADINA-F or ADINA-FSI if the allowed fluid
motion is large.
Difficulties in the use of the elements and various experiences

in solutions obtained are discussed in the following references:
ref. K.J. Bathe and W. Hahn, "On Transient Analysis of

Fluid-Structure Systems," Computers & Structures, Vol.
10, pp. 383-391, 1981.
ref. J. Sundqvist, "An Application of ADINA to the Solution

of Fluid-Structure Interaction Problems," Computers &
Structures, Vol. 17, pp. 793-808, 1983.
ref. L. Olson and K.J. Bathe, "A Study of Displacement-

2.10: Displacement-based fluid elements
Based Fluid Finite Elements for Calculating Frequencies

of Fluid and Fluid-Structure Systems," Nuclear
Engineering and Design, Vol. 76, pp. 137-151, 1983.
For example, in Olson and Bathe, it is noted that problems

involving structures moving through fluids that behave almost
incompressibly (e.g., an ellipse vibrating on a spring in water)
cannot be solved satisfactorily with the displacement-based fluid
elements.
In linear analysis, you can impose irrotational conditions in the

element formulation. This is achieved using a penalty constraint.
Note that if the penalty constraint is imposed, rigid body rotations
of the elements are no longer possible.
These elements are defined within 2-D and 3-D fluid element
groups (in ADINA). Set the formulation of the element group to
either displacement-based without rotation penalty or
displacement-based with rotation penalty (if you are using the AUI
user interfaces, set the formulation with the interpolation type
field).
Pressures are output at the integration points. The integration

point numbering is the same as the numbering convention used for
the solid elements, see Sections 2.2.3 and 2.3.3.
The pressure is evaluated using the following relation:
p = K ev
where p is the pressure, K is the bulk modulus and ev is the

volumetric strain ( volume/volume).
The pressures and strain components are accessible in the AUI
using the variable names FE_PRESSURE, STRAIN-XX,
STRAIN-YY, STRAIN-ZZ, STRAIN-XY, STRAIN-XZ,
STRAIN-YZ (STRAIN-XY and STRAIN-XZ are applicable only
for 3-D elements). See Section 13.1.1 for the definitions of those
variables that are not self-explanatory.
You can also request nodal force output. The nodal forces are
accessible in the AUI using the variable names NODAL_FORCE-X,
NODAL_FORCE-Y, NODAL_FORCE-Z (NODAL_FORCE-X is

Chapter 2: Elements
applicable only for 3-D elements).
2.11 Potential-based fluid elements

The elements discussed in this section incorporate the following
assumptions:
Inviscid, irrotational medium with no heat transfer
Compressible or almost incompressible medium
Relatively small displacements of the fluid boundary
Actual fluid flow with velocities below the speed of sound

(subsonic formulation) or no actual fluid flow (linear
infinitesimal velocity formulation)
For other types of fluid flow, use ADINA-F and ADINA-FSI

(see the ADINA-F Theory and Modeling Guide for details of the
formulations employed). In the remainder of this section, only the
ADINA fluid elements are discussed.
The types of problems for which the ADINA potential-based

fluid elements can be employed are:
Static analyses, where the pressure distribution in the fluid

and the displacement and stress distribution in the structure is of
interest
Frequency analyses, where natural frequencies and mode

shapes of a structure/fluid medium are to be determined
Transient analyses, where a pressure wave propagates

rapidly through the fluid, which does not undergo large motions
(transient acoustic problems)
Transient analyses, where fluid flows through the domain,

and the boundaries of the domain undergo only small motions.
The potential-based fluid elements can be employed in 2-D and

3-D analyses. Two-dimensional elements can be employed in

2.11: Potential-based fluid elements
planar and axisymmetric analyses. Two-dimensional elements

must be defined in the YZ plane, and axisymmetric elements must
lie in the +Y half plane (all nodal point coordinates must have non-
negative y values).
The potential-based fluid elements can be coupled with ADINA

structural elements, as described in detail below. The structural
motions cause fluid flows normal to the structural boundary, and
the fluid pressures cause additional forces to act on the structure.
The potential-based fluid elements can be coupled to a pressure

boundary condition (i.e., no structure adjacent to the potential-
based fluid element boundary). This feature can be used to model
free surfaces.
The potential-based fluid elements can be coupled directly to

ADINA-F fluid elements. The ADINA-F fluid element motions
cause potential-based fluid flows normal to the ADINA-F
boundary, and the potential-based fluid pressures cause additional
forces to act on the ADINA-F boundary.
The potential-based elements can model (approximately) an

infinite domain through the use of special infinite elements.
Two formulations are available, a subsonic velocity

formulation, which is nonlinear, and an infinitesimal velocity
formulation, which is linear. These are described in detail in the
following sections.
In some analyses, either ADINA-F/ADINA-FSI or the

potential-based fluid elements can be used in the modeling.
ADINA-F/ADINA-FSI is far more general than the potential-based
fluid elements and can model a much wider range of flow
conditions. In addition, in ADINA-FSI, the fluid mesh need not be
compatible with the structural mesh.
However, for the class of problems in which the assumptions
given above are acceptable, the potential-based fluid elements are
more efficient. This is because
The number of degrees of freedom in the fluid region is less

for the potential-based formulation. In 3-D analysis, each

Chapter 2: Elements
ADINA-F node requires a minimum of 4 degrees of freedom,

whereas each node in the interior of a potential-based mesh
requires only one degree of freedom.
If the velocities are small, then the linear infinitesimal

velocity formulation can be used. ADINA-F is always
nonlinear.
In addition, frequency analysis is not possible in ADINA-F.
The potential-based element formulation has been extensively

revised in version 8.0. Some of the features of earlier versions of
ADINA are superseded by new features of ADINA 8.0. Features
of earlier versions of ADINA that are superseded by new features
of ADINA 8.0 include:
The P0 degree of freedom (superseded in ADINA 7.4)

Volume infinite elements (superseded in ADINA 8.0)
These features are retained in ADINA 8.0 so that models developed

for earlier versions of ADINA still work in ADINA 8.0. However
these features are not described here, and are not recommended for
new models.
For another description of the potential-based fluid element,

including examples, see the following reference:
ref. T. Sussman and J. Sundqvist, "Fluid-structure interaction

analysis with a subsonic potential-based fluid
formulation," Computers & Structures, Vol. 81 (2003),
pp. 949-962.
2.11.1 Theory: Subsonic velocity formulation
Figure 2.11-1 shows a fluid region with volume and bounding

surface. In the fluid, we use the basic equations of continuity and
energy/momentum, as written in terms of the velocity potential:
+ i( ) = 0 (2.11-1)
and
h = ( x ) 12 i (2.11-2)

where is the density, is the velocity potential ( v = where

v is the fluid velocity), h is the specific enthalpy (defined as
dp
h= ), p is the pressure and ( x ) is the potential of the

(conservative) body force accelerations at position x . For
example, when the body forces are due to gravity, = g , where
g is the acceleration due to gravity.
Body force acceleration g,

body force acceleration potential W
Bounding surface S
Unknowns on S:
f = fluid potential Fluid region V
u = displacements n
Unknowns in V:
f = fluid potential
Essential boundary condition: f prescribed

Natural boundary condition: ru n prescribed
Figure 2.11-1: Fluid region
Equations (2.11-1) and (2.11-2) are valid for an inviscid

irrotational fluid with no heat transfer. In particular, (2.11-2) is
valid only when the pressure is a function of the density (and not
of, for example, the density and temperature).
For the pressure-density relationship, we use the slightly

compressible relationship
p
= 1+ (2.11-3)
0
where is the bulk modulus and 0 is the nominal density.

(2.11-3) then directly gives the density-enthalpy relationship and

Chapter 2: Elements
the pressure-enthalpy relationships:
0 h 0 h
= 0 exp , p = exp 1 (2.11-4a,b)

The continuity equation (2.11-1) is then approximated using the

standard Galerkin approach to obtain
F = ( + i( ) ) dV = 0 (2.11-5)
V
where is the variation of symbol and F is the variation in

the mass flux rate. (2.11-5) can be rewritten as
F = ( i ) dV in dS (2.11-6)
V S
where S is the boundary of V and n is the inwards normal on S .

From this equation, we observe that the natural boundary condition
is in prescribed, in other words, prescribed mass flux rate.
At this point, we notice that is a function of the fluid velocity

and position through the density-enthalpy relationship (2.11-4a),
because the enthalpy is a function of fluid velocity and position. In
V , we compute the enthalpy using (2-11.2). Hence, in V , the
density is a function only of the fluid potential and position. But
on S , we anticipate that part or all of S might be a moving
boundary with velocity u(x) . We adopt the convention that the
fluid velocity v used in the enthalpy calculation on S is computed
in terms of both the boundary velocity u and the internal fluid
velocity using
v n = (uin)n , v t = ( in)n , v = v n + v t (2.11-7a,b,c)
In other words, the fluid velocity normal to the surface is taken to

be the velocity of the moving boundary and the remainder of the
fluid velocity (which is tangential to the surface) is taken from the
fluid potential. We also include the motions of S in the body

force potential calculations in the enthalpy. Combining the above

considerations, we take
h = ( x + u ) 12 v n i v n 12 v t i v t (2.11-8)
on S .
(2.11-6) can be written as

F = h i dV + u in dS (2.11-9)
V
h S
The surface integral is obtained using (2.11-7).
On S , we assume that the boundary motion u(x) is small

enough so that the volume V and the normal n can be assumed
constant. In other words, we perform all integrations on the
undeformed fluid boundary. We discuss the implications of this
assumption later.
From (2.11-9), if no boundary conditions are applied to S, the

boundary condition u n = 0 is implied. This boundary
condition corresponds to no fluid flow through the boundary.
From (2.11-9), the natural boundary condition is
u n = prescribed . We use this boundary condition to apply the
structural motions to the fluid domain. We also use this boundary
condition to model infinite fluid regions. Both of these cases are
discussed in more detail below.
Notice that there is no boundary condition corresponding to
flow tangential to the boundary. In other words, fluid can slip
tangential to the boundary without restriction.
Also from (2.11-9), the essential boundary condition
= prescribed is possible. Through (2.11-2) this boundary
condition corresponds to a partially prescribed enthalpy and hence
a partially prescribed pressure. The enthalpy and pressure are not
fully prescribed since is not prescribed when is prescribed
only on the boundary.

Chapter 2: Elements
Modifications to equations of motion for the structure
We assume that part of the boundary S is adjacent to the

structure (Figure 2.11-2). The part of the boundary adjacent to the
structure is denoted S1 .
Surface S1 (Fluid and structure

are separated for
clarity)
Fluid Structure
n
Pressure p Traction -pn
Figure 2.11-2: Forces on structure from fluid
The fluid pressure on S1 provides additional forces on the

structure adjacent to S1 :
Fu = p ni u dS1 (2.11-10)
S1
where Fu is the variation in the applied force vector and the

minus signs are used in anticipation of Fu being considered an
internal force vector in (2-11.12) below. The pressure is evaluated
using
( )
p = p (h) = p ( x + u ) 12 v n i v n 12 v t i v t (2.11-11)
We emphasize that n points into the fluid (and out of the

structure).
Finite element equations of motion
Equations (2.11-6) and (2.11-9) are linearized using standard

procedures to obtain

0 0 u CUU CUF u
0 M + C
( CFF + ( CFF )S )
+
FF FU
K K FU
UU UF u 0

K FU

(
K
FF
( ))
+ K
FF S
= 0
F

F ( )
+ F
F S

(2.11-12)
where the increment in the vector of unknown potentials is
written and the increment in the vector of unknown
displacements u is written u . (We drop the left superscript
t + t here and below for ease of writing.) In the linearization
process, increments in both the nodal displacements and increments
in the nodal potentials (and their time derivatives) are considered.
In (2-11.12),
FU = vector from (2.11-10)

FF = vector from volume integration term in (2.11-9)
( FF )S = vector from surface integration term in (2.11-9)
and the matrices are obtained by linearization. Note that vectors
and matrices with the subscript S are integrated over the surface.
Also the matrices with displacement degrees of freedom are
integrated over the surface.
The sum of FF and ( FF )S can be interpreted as an out-of-

balance mass flux vector. (2-11.12) is satisfied only if this sum is
zero at every node in the fluid, that is, if the consistent mass fluxes
at each of the nodes from all of the attached fluid elements sum to
zero.
In (2.11-12), we do not include any of the structural system

matrices. (2.11-12) only gives the contribution of the potential-
based fluid elements to the system matrices. Of course, the rest of
the structure will make additional contributions to the first row of
the above equations.

Chapter 2: Elements
Equation (2.11-12) is a nonlinear system of equations. The

nonlinearities come from several sources:
1) The 12 i term in h . This term is sometimes called

the Bernoulli effect.
2) The nonlinear relationship between the density, pressure and

the enthalpy (2.11-4).
Because the system is nonlinear, equilibrium iterations must be

employed for an accurate solution, just as in nonlinear structural
analysis.
There is no explicit restriction on the magnitude of the fluid

velocities in equation (2.11-12). However, in practice, there are
several restrictions
1) The velocity must be smaller than the speed of sound,

otherwise, the equations (2.11-9) become hyperbolic and cannot
be solved using ordinary finite element techniques.
2) The density change should not be too great, otherwise the

change in mass flux may be quite different than the change in
volume flux when the boundary velocities are perturbed.
3) Also, the pressure-density relationship in (2.11-3) only holds

for relatively small density changes.
With these restrictions, the Mach number of the fluid flow

should not exceed about 0.3 or so.
Equation (2.11-12) is in general nonsymmetric. But in the limit

of very small velocities, equation (2.11-12) becomes symmetric
(see equation (2.11-19) below). Even for finite velocities, equation
(2.11-12) is nearly symmetric. This means that we can
symmetrize (2.11-12) and use the usual symmetric equation solvers
of ADINA for the solution of (2.11-12). The convergence rate is
very fast for small velocities and becomes slower as the velocities
become larger.

2.11.2 Theory: Infinitesimal velocity formulation
If we assume that the velocities and the density changes are

infinitesimally small, then the continuity equation (2.11-1)
becomes
0 p
+ i( ) + 0 2 + 0 2 = 0 (2.11-13)

The momentum/equilibrium equation (2.11-2) becomes
p
h ( x) (2.11-14)

from which we see
( ) (
p ( x ) 0 ( x ) ) (2.11-15)
Substituting (2.11-15) into (2.11-13) gives
0 + 2 = 0 (2.11-16)
Equation (2.11-16) is a special form of the wave equation. It is

linear in the solution variable . (2.11-16) can be written in
variational form using standard techniques. The result is
0 dV dV u n dS
V V S
= 0 dV (2.11-17)
V
The fluid pressure onto the structure becomes

Fu = pni u dS1 0 + 0 iu 0 ni u dS1 (2.11-18)
S1 S1
x
The finite element contributions to the system matrices

corresponding to (2.11-17) and (2.11-18) are

Chapter 2: Elements
0 0 U 0 CTFU U ( K UU )S 0 U
0 M + +
K FF
FF CFU 0 0
(2.11-19)
( R ) 0
= UB S +
0 R FB
where
M FF = matrix from term in (2.11-17)

K FF = matrix from term in (2.11-17)
CFU = matrix from u n term in (2.11-17)

( K UU )S = matrix from 0 iu ni u term in (2.11-18)
x
( R UB )S = loads vector from ( 0 ) ni u term in (2.11-18)
R FB = loads vector from 0 term in (2.11-17)
U = vector containing unknown nodal displacements
= vector containing unknown nodal fluid potentials.
We note that the term ( K UU )S is numerically very small

compared with the rest of the structural stiffness matrix, when there
is a structure adjacent to the fluid. But ( K UU )S is important in the
case when there is no structure adjacent to the fluid.
The left-hand-side of equation (2.11-19), with the exception of

the term ( K UU )S , is identical to the formulation presented in the
ref. L.G. Olson and K.J. Bathe, Analysis of fluid-structure

interactions. A direct symmetric coupled formulation
based on the fluid velocity potential, J. Computers and
Structures, Vol 21, No. 1/2, pp 21-32, 1985.

2.11.3 Theory: Infinite fluid regions
Both the subsonic and infinitesmal velocity formulations

include special boundary conditions for the modeling of infinite
fluid regions.
The basic approach is to replace the infinite fluid region with a

boundary condition that simulates the infinite fluid region. The
boundary condition allows outwards-going waves to be propagated
into the infinite fluid region without reflection.
There are many approaches for setting up infinite elements.

One approach that is physically intuitive is to use the results from
acoustic analysis regarding outwards-going waves.
Planar waves: To illustrate the procedure, we consider a planar

boundary, that is, a boundary on which plane acoustic waves
propagate outwards through the boundary (Figure 2.11-3). If the
waves have very small amplitude,
p = cv (2.11-20)
where p is the change in pressure applied to the boundary from

the fluid on the outside of the boundary (in other words, the change
in pressure applied to the boundary from the fluid that is not
explicitly modeled), v is the change in the outwards velocity of
the fluid and c is the speed of sound in the fluid. Hence the
pressure on the boundary is
p = p + c ( v v ) (2-11.21)
where v is the outwards velocity at the boundary, and p and v

are the pressure and outwards velocity at infinity ( p and v
must be specified as part of the model definition).

Chapter 2: Elements
At infinity
Direction of p=p
wave propagation v=v
a) Physical problem
The pressure and

Fluid velocity on the boundary
are related by
n p=p + rc(v - v )
b) Modeling with planar infinite boundary
Figure 2.11-3: Planar infinite boundary
The variation in mass flow through the boundary is
F B = in dS = v dS (2.11-22)
SB SB
We need to express the mass flow through the boundary in terms of

the potential on the boundary. This can be done as follows. First,
we assume that there is no fluid flow tangential to the boundary
and that the boundary does not move. Under those conditions,
p(h) = p( 12 v 2 ) (2.11-23)
Given a value of , v can be numerically evaluated by combining

equations (2.11-21) and (2.11-23) and hence, given a value of ,
(2.11-22) can be numerically evaluated. We use this procedure in
the implementation of the planar infinite element for the subsonic
velocity formulation.
As an illustrative example, in the special case when there are no

body forces, the velocities are small, and the velocities and pressure

at infinity are zero

p
v= = (2.11-24)
c c
so that the variation in mass flow rate is, in this special case,

F B = dS (2.11-25)
SB
c
We use (2.11-25) in the implementation of the planar infinite

element for the infinitesimal velocity formulation.
Spherical waves: The same technique can be used for spherical

waves, provided that the appropriate relationship is used for the
relationship between p and v . The relationship that we use for
spherical waves is based on the relationship between pressure and
mass flux in the frequency domain
p p
v= + (2.11-26)
c i r
where i = 1 , is the frequency of the outwards-going wave

and r is the radius of the boundary. Here we assume that both the
pressure and velocity at infinity are equal to zero and that the
velocity itself is small. The equivalent of (2.11-26) in the time
domain is
p p dt
v= + (2.11-27)
c r
We now neglect body force effects, so that p = on the

boundary. Then

v= (2.11-28)
c r
so that the variation in mass flow rate is

Chapter 2: Elements

F B = + dS (2.11-29)
SB
c r
Note that this derivation holds regardless of the wave frequency
Cylindrical waves: The same technique and derivation can be used

for cylindrical waves. For cylindrical waves, the relationship
between pressure and mass flux in the frequency domain is
p J1 (kr ) iY1 (kr )

v= (2.11-30)
ic J 0 (kr ) iY0 (kr )
where J 0 , J1 , Y0 , Y1 are the Bessels functions of the first and

r
second kinds of orders 0 and 1, and kr = . (2.11-30) cannot be
c
transformed into the time domain, but for frequencies higher than
r
about = 1 , it turns out that (2.11-30) is well approximated by
c
(2.11-26), so that the spherical wave derivation can be used,
provided that the frequency is high enough.
For planar, spherical and cylindrical infinite boundaries, only

surface integrations are required. Also, no additional degrees of
freedom are required on the boundary. For planar waves, the
infinite boundaries contribute to ( FF )S ( CFF )S and for spherical
and cylindrical waves, the infinite boundaries contribute to ( FF )S ,
( CFF )S and ( K FF )S .
It is instructive to compare the above approach with the doubly-
asymptotic approach in Olson and Bathe.
ref. L.G. Olson and K.J. Bathe, An infinite element for

analysis of transient fluid-structure interactions,
Engineering Computations, Vol 2, pp 319-329, 1985.

In the DAA, the infinite fluid region is modeled as the

superposition of two effects, an added mass approximation and a
plane wave approximation. The added mass approximation
dominates for low frequencies and the plane wave approximation
dominates for high frequencies. The plane wave approximation
gives exactly the results given above for planar waves (equation
2.11-25). But the added mass approximation in the DAA requires a
volume integral and hence volume infinite elements are required
for modeling non-planar waves. Provided that the nodes in the
volume infinite elements at infinity are placed at the correct
locations, the volume infinite elements give exactly the same
results as the spherical infinite elements (equation (2.11-29).
Since the surface integral formulation given in equations (2.11-
20) to (2.11-30) gives the same results as the DAA approach, and
since the surface integral formulation does not require volume
elements, we feel that the surface integral formulation is preferable.
2.11.4 Theory: Ground motion loadings
Ground motion loadings require special treatment. In this case,
U = Ur + Ug (2.11-31)
where U g is the vector of nodal point ground displacements and

U r is the vector of nodal point displacements relative to the
ground motion. The ground motions are expressed as
U g = u gk d k (2.11-32)
k
where u gk are the ground displacements in direction k and d k is

the vector of nodal point values in which d k ( i ) = 1 if equation i
corresponds to a translation in direction k and d k ( i ) = 0 otherwise.
Because the ground motions are known, the increment in

displacements U is equal to the increment in the relative
displacements U r Hence the left hand sides of equations (2.11-
12) and (2.11-19) are unchanged and the right-hand-sides of these

Chapter 2: Elements
equations are updated only by additional internal forces. For

example, the additional internal forces added to (2.11-19) from the
fluid elements are
( K ) d 0
UU S k u gk u gk
0 CFU d k
Note that still contains the nodal point values of and is

the potential corresponding to the absolute (not relative) velocities.
Also note that ground motions are handled differently than
physical body forces. This is quite different than in ordinary
structural analysis, in which physical body forces and ground
motions are both modeled using mass-proportional loads.
2.11.5 Theory: Static analysis
The case of static analysis also requires special attention. It is

not correct simply to set the velocities to zero in (2.11-1), because
(2.11-1) is identically satisfied for zero velocities.
We envision the process of obtaining a static solution as a

quasi-static process. In this process, all fluid velocities are
assumed negligible and therefore the fluid potential is constant (in
space). Under these conditions,
h = (2.11-32)
and the variational statement of continuity is
F = dV u in dS (2.11-33)
V S
Equation (2.11-33) is simply the integral equation of

conservation of mass because is constant in space.
We now numerically integrate (2.11-33) in time. We use the

Euler backwards approximations
t +t
= ( t +t t ) / t , t +t
u = ( t +t u t u ) / t (2.11-34a,b)

and then obtain, at time t + t ,
F = ( t +t t ) dV t +t ( t +t u t u )in dS (2.11-35)
V S
where we write = / t .
The modification to the structural equations of motion is
Fu = t +t p ni u dS1 (2.11-36)
S1
which is the same as equation (2.11-10), except that the t + t is

explicitly written.
We prefer the Euler backwards approximations because then the
stiffness matrix remains symmetric when the displacement
increments are infinitesimally small. Of course, the steps must be
small enough so that the Euler backwards approximations apply,
but in many problems, these approximations are quite good,
especially when the fluid is almost incompressible.
Note that we assume that the domain of integration and its
boundary remain (nearly) unchanged during the time integration.
All integrations are performed over the original domain of
integration.
Also note that the solution in the fluid is the single value .
This value represents the Bernoulli constant.
The linearized equations of motion for the subsonic formulation
become
(
K UU

)
S
K UF FU
u = 0

(2.11-37)
0 FF + FF
K
FU ( (
K FF + K FF ))
S ( )
S
where we use the ~ to emphasize that the vectors and matrices are
different than the corresponding dynamic vectors and matrices,
along with the constraints = constant (if there is more than one
fluid region in the problem, then = constant in each fluid

Chapter 2: Elements
region). As in the dynamic case (equation 2.11-12), equation

(2.11-37) is in general non-symmetric, but (2.11-37) becomes
symmetric when the displacement increment t +t u t u becomes
infinitesimally small. So, as in the dynamic case, we symmetrize
the system and use the usual symmetric equation solver.
( )
In (2.11-37) the sum of FF + FF can be interpreted as an
S
out-of-balance mass vector. (2.11-37) is only satisfied if the total
mass of each fluid region is conserved. Motions of the boundary
that do not change the total mass of any fluid region are associated
with zero pivots in the system matrices, unless there is structural
stiffness associated with these motions. Note that the K UU( )
S
matrix also provides structural stiffness, so if there are body forces,
motions of the boundary that do not change the total mass of any
fluid region are not associated with zero pivots.
The equations of motion for the infinitesimal velocity

formulation can be obtained either from the above derivation, or
can be formally derived from (2.11-19) by applying the Laplace
transform to both sides of (2.11-19) and applying the final value
theorem. The result is
( K UU )S CTFU U ( R UB )S
= (2.11-38)
CFU M FF R FB
together with the condition
K FF = 0 (2.11-39)
There are a number of unusual characteristics of (2.11-37),

(2.11-38) and (2.11-39):
The solution involves instead of . This makes sense as

(2.11-32) then implies that p is constant (in time) in a static
solution.
The condition (2.11-39) must be satisfied. When there are

no infinite boundaries, this condition is satisfied whenever

= constant within each separate fluid region. Hence the

number of unknown potential degrees of freedom in static
analysis is equal to the number of separate fluid regions in the
analysis.
The condition = constant within each separate fluid

region implies that p = 0 + C where C is a constant
determined from the solution. Hence the variation of pressure
within each separate fluid region is contained within C and any
choice of constant of integration within 0 (recall that is a
potential and therefore includes an arbitrary constant of
integration) is balanced by an opposite change in C .
If the degree of freedom is fixed at a node, or if infinite

boundaries are included in the fluid region, then you must set
= 0 for all the nodes in the fluid region. This condition
implies p = 0 within the fluid region. Any choice of
constant of integration within then affects the solution.
It is necessary to enter the density of the fluid in static

analysis, even when the solution does not depend on the
density.
2.11.6 Theory: Frequency analysis
Frequency analysis is possible when there is no structural

damping, when the infinitesimal velocity formulation is used, and
when there are no infinite boundaries. The eigenvalue problem to
be solved is
2 M 0 0 CTFU K + ( K UU )S 0 U ( j) 0
j + =

0 M FF
j
CFU 0 0 K FF F ( j) 0
(2.11-40)
where F (j) = i (j) , i = 1 and in which we also include the

structural stiffness matrix K and structural mass matrix M .

Chapter 2: Elements
(2.11-40) is derived from (2.11-19) by taking the Fourier transform

of the left-hand-side.
(2.11-40) is a non-standard eigenvalue problem in which all of
the eigenvalues are real and non-negative. It is solved using either
the determinant search method or the Lanczos iteration method.
The eigenvectors are scaled according to the following
orthogonality condition:
u (i)CTFU F (j) + F (i)CFU u (j) + (i + j ) ( u (i) M UU u (j) + F (i) M FF F (j) )

=2 i j ij
(2.11-41)
where ij is the Kronecker delta. Notice that when there is no
fluid, (2.11-41) reduces to the usual orthogonality condition
u (i) M UU u (j) = ij .
(2.11-40) has one rigid body mode for each separate fluid
region in which all of the degrees of freedom are free. Each
rigid body mode is of the form U (j) = 0, F (j) = constant in each
separate fluid region. For future reference, we term these rigid
body modes rigid body modes. The determinant search method
can determine all rigid body modes. The Lanczos frequency
solver does not print the rigid body modes.
The modal pressures and modal fluid particle displacements are
computed using
1
p( j ) = j f ( j ) , u( j ) = f ( j ) (2.11-42)
j
in which p ( j ) is the modal pressure at the point of interest for

mode j, f ( j ) is the fluid potential eigenvector F ( j ) interpolated to
the point of interest and u ( j ) is the modal displacement vector at
the point of interest for mode j.
2.11.7 Theory: Mode superposition
Mode superposition is possible when there is no structural

damping, the infinitesimal velocity formulation is used and when

there are no infinite boundaries. The theory used to determine the
modal equations of motion is given in Chapter 6 of the following
reference:
ref. L. Meirovitch, Computational Methods in Structural

Dynamics, Sijthoff & Noordhoff, 1980.
The derivation starts with the equations of motion (2.11-19)

being put into the form
M 0 Ur 0 CTFU U r K + ( K UU )S 0 Ur
0 M + +
K FF
FF CFU 0 0
R ( R ) 0 ( K UU )S dk 0
= + UB S +
u gk + C d u gk
0 0 R FB 0 FU k
(2.11-43)
where we include the structural stiffness matrix K and structural

mass matrix M . We also include the ground motion loading terms
in (2.11-43) in anticipation of the response spectrum derivation
below. (2.11-43) corresponds to an undamped gyroscopic system.
The modal expansion used is
n j (j) n
Ur = U , = j F (j) (2.11-44a,b)
j =1 j j =1
where j is the generalized coordinate for mode j. The modal

equation is
j + 2j j = ( j ) (2.11-45)
where
( j ) = ( F (j) ) R + j ( U (j) ) R u
T T
(2.11-46)

Chapter 2: Elements
and, from (2.11-43),
R u = R + ( R UB )S ( K UU )S d k u gk
R = CFU d k u gk (2.11-47a,b)
With modal damping (damping ratio j ), (2.11-45) becomes
j + 2 j j j + 2j j = ( j ) (2.11-48)
Then we use
n j (j) n n
Ur = U , U r = U (j) , U r = U (j)
j j
j =1 j j =1 j j =1 j
(2.11-49a,b,c)
n t n n
= j dt F (j) , = j F (j) , = j F (j)
j =1 0 j =1 j =1
(2.11-50a,b,c)
to obtain the structural and fluid response.

The initial modal coordinates are computed from the initial
conditions (displacements, velocities, fluid potentials and time
derivatives of fluid potentials). Initial accelerations and initial
second time derivatives of fluid potentials are not used.
In the first solution step, we use the Newmark method with
= 1/ 2, = 1 because this choice of Newmark parameters does
not require initial accelerations. In the successive solution steps,
we use the Newmark method with the usual Newmark parameters
= 1/ 4, = 1/ 2 .
2.11.8 Theory: Response spectrum, harmonic and random vibration

analysis
Modal participation factors are calculated for response spectrum

analysis, harmonic vibration analysis and random vibration
analysis. There are two cases: modal participation factors
corresponding to applied forces (in which case we do not consider

response spectrum analysis) and modal participation factors

corresponding to ground motions.
In the modal participation factor calculation, we only consider
the dynamic solution (due to dynamically applied loads).
To begin the derivation, we introduce
j
xj = (2.11-51)
j
as the new generalized coordinate. The modal expansion is then
n n
U r = x j U (j) , = j x j F (j) (2.11-52a,b)
j =1 j =1
and the modal equation of motion is
1
(F )
(j) T
R + ( U (j) ) R u (2.11-53)
T
x j + 2 j j x j + 2j x j =
j
Applied forces: As it is assumed that the physical body forces are

entirely static and that there are no ground motions, the modal
equation of motion becomes
x j + 2 j j x j + 2j x j = ( U (j) ) R
T
(2.11-54)
and evidently the modal participation factor is simply
( j ) = ( U (j) ) R
T
(2.11-55)
We use ( j ) rather than the ( j ) given in (2.11-45) because then

the modal expansion for Ur is then the commonly used one.
As a consequence of (2.11-55), the modal participation factor for

any rigid body mode is zero.
Ground motions: The loads vectors are

Chapter 2: Elements
R u = ( K UU )S d k u gk Md k u gk ,
R = CFU d k u gk (2.11-56a,b)
Hence
1
x j + 2 j j x j + 2j x j = ( F (j) ) CFU d k + ( U (j) ) Md k u gk
T T

j (2.11-57)
( U (j) ) ( K UU )S d k u gk
T
( ) (K )
T
We neglect the term U (j) UU S d k u gk (which is probably
small anyway) to obtain
x j + 2 j j x j + 2j x j = ( j )u gk (2.11-58)
where the ground motion modal participation factor is
1
(F )
(j) T
CFU d k + ( U (j) ) Md k
T
( j) =
j
This factor has the usual physical interpretation, provided that
each fluid region is completely surrounded by interface elements.
The modal participation factor for a rigid body mode can be
zero or non-zero. It can be shown that the modal participation
factor for a rigid body mode is zero if each fluid region is
completely surrounded by interface elements.
Note: static corrections (residual calculations) are not
implemented for the potential-based fluid element because residual
displacement calculations are based upon non-constant (in space)
body force loading. Such loading is not possible in general for the
potential-based fluid elements.
2.11.9 Modeling: Formulation choice and potential master degree of

freedom
The potential-based formulation discussed in this section is

introduced in ADINA 7.4 and extensively modified in ADINA 8.0.

You can select the potential-based formulation in the AUI

command-line input using the parameter FLUIDPOTENTIAL in
the MASTER command:
FLUIDPOTENTIAL=AUTOMATIC: The potential-based

formulation of ADINA 7.4 and 8.0 is employed. This is the
default.
FLUIDPOTENTIAL=YES: The potential-based formulation of

ADINA 7.3 and lower is employed.
FLUIDPOTENTIAL=NO: The potential-based formulation is not

employed.
Note, it is not possible to select the potential-based formulation

using the AUI user interfaces. The potential-based formulation of
ADINA 8.0 is employed when using the AUI user interfaces.
When FLUIDPOTENTIAL=AUTOMATIC, the AUI
automatically detects the presence of potential-based elements, and,
if there are any potential-based elements, activates the potential
master degree of freedom.
2.11.10 Modeling: Potential degree of freedom fixities
Subsonic formulation: Deleting the potential degree of freedom

along part of the bounding surface has the effect of partially
specifying the enthalpy along that surface, see equation (2.11-2).
This has no physical meaning.
Infinitesimal velocity formulation: Deleting the potential degree of

freedom along part of the bounding surface has the effect of setting
the pressure equal to 0 along that part of the bounding surface,
see equation (2.11-15). If there are no body forces, then = 0
and the pressure is therefore set to zero along that part of the
bounding surface. This boundary condition can be used if the
displacements of the boundary are not of interest.
It is recommended that the potential degrees of freedom not be

deleted anywhere in the fluid.

Chapter 2: Elements
2.11.11 Modeling: Elements
The volume V of the fluid domain is modeled using two-

dimensional or three-dimensional fluid elements. These elements
are analogous to the two-dimensional or three-dimensional solid
elements and the nodal point numbering of the fluid elements is the
same as the nodal point numbering of the solid elements.
The two-dimensional elements are either planar (unit thickness
of fluid assumed) or axisymmetric (1 radian of fluid assumed).
These elements are defined within 2-D and 3-D fluid element
groups (in ADINA). Set the formulation of the element group to
either Linear Potential-Based Element (for the infinitesimal
velocity formulation) or Subsonic Potential-Based Element (for
the subsonic velocity formulation).
The bounding surface S of the fluid domain is modeled with
potential-interfaces and/or interface elements, as discussed in detail
below.
It is required that each separate fluid domain be modeled with
separate fluid element groups. This is because the AUI constrains
the potential degrees of freedom of each element group together in
static analysis during phi model completion, step 7, see Section
2.11.15.
It is not permitted to have fluid regions of different densities
sharing the same potential degrees of freedom. This is because the
nodal pressure would be different as computed from the fluid
regions connected to the node.
It is recommended that the fluid and adjacent structure not share
the coincident nodes. This allows the AUI to construct appropriate
constraint equations between the fluid and structural degrees of
freedom that are most appropriate during phi model completion,
see Section 2.11.15.
2.11.12 Modeling: Potential-interfaces
For ease of modeling, you can define potential-interfaces of

various types along the surface of the fluid domain. When you
generate a data file, the AUI places interface elements along the
boundaries specified by the potential-interfaces, except as noted
below. ADINA itself does not use the potential-interfaces.
There are several types of potential-interface:

Fluid-structure: Place a fluid-structure potential interface on the

boundary between a potential-based fluid and the adjacent
structure.
In many cases, the AUI can automatically generate fluid-
structure interface elements along the boundary between the fluid
and structure during phi model completion, step 1, see Section
2.11.15. So you typically do not need to define fluid-structure
potential-interfaces.
Free surface: Place a free surface potential interface on the

boundary where the pressures are to be prescribed and the
displacements are desired, for example, on the free surface of a
fluid. Surface waves can be approximately modeled in this
manner, but note that the displacement of the waves is assumed to
be small.
ADINA-F: Place an ADINA-F potential interface on the

boundary adjacent to an ADINA-F mesh.
Usually the AUI automatically generates ADINA-F interface
elements along the boundary adjacent to an ADINA-F mesh during
phi model completion, step 1. So you typically do not need to
define ADINA-F potential-interfaces.
Infinite: Place an infinite potential-interface wherever infinite

boundary conditions are desired.
There are three types of infinite potential-interface.
Planar: In the subsonic formulation, the pressure and

velocity at infinity must be specified. In the infinitesimal
velocity formulation, the pressure and velocity are assumed to
be zero.
Spherical: The radius of the boundary must be specified.

The pressure and velocity at infinity are assumed to be zero, and
the velocities at the boundary are assumed to be small.
Cylindrical: The radius of the boundary must be specified.

The pressure and velocity at infinity are assumed to be zero, and
the velocities at the boundary are assumed to be small. This
element cannot accurately model low-frequency waves, that is,

Chapter 2: Elements
r
waves with <1.
c
Inlet-outlet: Place an inlet-outlet potential-interface wherever the

pressure of the boundary is specified, and where the displacements
of the boundary are not of interest. For example, the outlet of a
pipe on which the pressure is known can be modeled using an inlet-
outlet potential-interface.
If there is more than one inlet or outlet in the model, each inlet
and outlet must have its own inlet-outlet potential-interface.
Fluid-fluid: Place a fluid-fluid potential-interface on the boundary

between two potential-based fluid elements of two different
element groups.
Note that only one fluid-fluid potential-interface need be
defined for each boundary. The AUI generates a fluid-fluid
interface element for each of the two elements that share a common
boundary during phi model completion, step 1, see Section 2.11.15.
Rigid-wall: Place a rigid-wall potential interface wherever the fluid

is not to flow through the boundary.
Note that this boundary condition is modeled in ADINA by the
absence of any interface element. Therefore a rigid-wall potential-
interface suppresses any automatic generation of interface elements
along the boundary of the rigid-wall potential-interface. However
the AUI uses the rigid-wall potential interface during phi model
completion, step 2, in constructing structural normals, see Section
2.11.15.
2.11.13 Modeling: Interface elements
Interface elements are used on the surface of the fluid domain to

specify a boundary condition. In many cases, you do not have to
define interface elements. Rather you define potential-interfaces
and the AUI then generates the interface elements. However we
describe the interface elements here for completeness.
Interface elements are defined within the same element group as
the fluid elements themselves. In 2-D analysis, the fluid-structure
interface elements are 2 or 3 node line segments, in 3-D analysis,
the fluid-structure interface elements are 3 to 9 node area segments.
The nodal point numbering conventions are shown in Figure

2.11-4. The AUI automatically reorders all interface elements

when generating the ADINA input data file so that the interface
element normals point into the fluid.
Fluid Fluid
n n
1 1
Solid Solid 3
2 2
a) 2-D elements
Fluid Fluid
1 1
n n 8
4 5 4
9
2 7
2
3 6
Solid Solid 3
b) 3-D elements
Figure 2.11-4: Fluid-structure interface elements, showing

the local node numbering convention
There are several types of interface elements:
Fluid-structure interface element: This element connects the

potential-based fluid element with an adjacent structural element
(Figure 2.11-5). Each node of the element contains the potential
degree of freedom and displacement degrees of freedom. It is
assumed that the displacements of the nodes of the interface
element are small.
It is assumed that the structure provides stiffness to all
translational degrees of freedom, because the fluid-structure
interface element does not provide stiffness, mass or damping to
the tangential directions.
We emphasize that the potential-based fluid element, fluid-

Chapter 2: Elements
structure interface element and adjacent structural element must all

be compatible.
Structural element
Nodal unknown: u
Fluid-structure
interface element
Nodal unknowns: f, u
Fluid element
Nodal unknown: f Nodes circled by
share degrees of
Assumptions: u is small. freedom
Figure 2.11-5: Fluid-structure interface element
Free surface interface element: This element is placed onto the

boundary of a potential-based fluid element wherever the pressure
is to be prescribed, and wherever the displacements of the fluid are
required (Figure 2.11-6). For example, the free surface of a fluid in
a basin can be modeled using free surface interface elements. Each
node of the element contains the potential degree of freedom and
displacement degrees of freedom. It is assumed that displacements
and velocities of the nodes of the interface element are small.
Applied pressure
or no applied pressure
Free surface
interface element
Fluid element
Nodal unknown: f
Assumptions: u.n is small.
un is small.
Figure 2.11-6: Free surface interface element
It is necessary to fix all displacements that are tangential to the

free surface interface element, because the free surface interface
element does not provide stiffness, mass or damping to the
tangential directions.
In many cases, the AUI can generate skew systems and fixities
corresponding to the tangential directions during phi model

generation, see Section 2.11.15.
ADINA-F interface element: This element is placed onto the

boundary of a potential-based fluid element wherever the boundary
is adjacent to an ADINA-F mesh (Figure 2.11-7). Each node of the
element contains the potential degree of freedom and displacement
degrees of freedom. It is assumed that displacements and velocities
of the nodes of the interface element are small.
Note that the coupling between ADINA and ADINA-F is
through the displacements of the shared boundary. Therefore,
because the displacements of the ADINA-F interface element are
assumed to be small, the ADINA/ADINA-F coupling cannot be
used for modeling actual fluid flow between the ADINA and
ADINA-F models.
ADINA-F interface element, because the ADINA-F interface
tangential directions. In many cases, the AUI can generate these
fixities automatically during phi model completion, see Section
2.11.15.
The ADINA-F interface element need not be compatible with
the elements from the adjacent ADINA-F mesh .
It is necessary to define an ADINA-F fluid-structure boundary
in the ADINA model to connect the ADINA and ADINA-F
models, just as in ordinary ADINA-F FSI analysis.
In most cases, the AUI automatically generates the ADINA-F
interface elements during phi model completion, step 1, see Section
2.11.15.
Applied tractions
from ADINA-F
ADINA-F
interface element
Fluid element
Nodal unknown: f
un is small.
Figure 2.11-7: ADINA-F interface element
Infinite interface elements: This element is placed onto the

Chapter 2: Elements
boundary of a potential-based fluid element wherever infinite

boundary conditions are desired. Each node of the element
contains only a potential degree of freedom.
There are three types of infinite interface element:
Planar infinite element: In the subsonic formulation, the

pressure and velocity at infinity must be specified. In the
infinitesimal velocity formulation, the pressure and velocity are
assumed to be zero. A planar infinite element is shown in
Figure 2.11-8.
p , v specified
Planar infinite
interface element
Nodal unknown: f
Fluid element
Nodal unknown: f
Assumptions: f constant along boundary.

Flow velocity is close to v
Figure 2.11-8: Planar infinite interface element
Spherical infinite element: The radius of the boundary must be

specified. The pressure and velocity at infinity are assumed to
be zero, and the velocities at the boundary are assumed to be
small.
Cylindrical infinite element: The radius of the boundary must be

specified. The pressure and velocity at infinity are assumed to
be zero, and the velocities at the boundary are assumed to be
small. This element cannot accurately model low-frequency
r
waves, that is, waves with <1.
c
When considering where to place the infinite interface elements,
remember that the infinite elements are derived under the
assumption that waves travel normal to the boundary. Hence the
boundary must be placed so that any anticipated waves travel
normal to the boundary (Figure 2.11-9).

Direction of
wave propagation
indicated by
a) Correct modeling b) Incorrect modeling
Figure 2.11-9: Direction of wave propagation must be normal to the

infinite interface element
Inlet-outlet interface element: This element is placed onto the

boundary of a potential-based fluid element wherever the pressure
of the boundary is specified, and where the displacements of the
boundary are not of interest (Figure 2.11-10). For example, the
outlet of a pipe on which the pressure is known can be modeled
using inlet-outlet interface elements. Each node of the element
contains the potential degree of freedom and displacement degrees
of freedom. The displacement degrees of freedom are only used to
compute velocities and accelerations; the displacements themselves
are not used.
Applied pressure Inlet-outlet

or no applied pressure interface element
Fluid element
Nodal unknown: f
.
Assumptions: un constant along boundary.
f constant along boundary.
Figure 2.11-10: Inlet-outlet interface element

inlet-outlet interface element, because the inlet-outlet interface
tangential directions. In addition, it is necessary to set the
tangential fluid velocity to zero by constraining all of the potential

Chapter 2: Elements
degrees of freedom to be equal. Finally it is necessary to set the

normal velocity to be uniform along the element, by constraining
the normal velocities to be equal. In many cases, the AUI can
generate these fixities and constraints automatically during phi
model completion, see Section 2.11.15.
Fluid-fluid interface element: This element is placed along the

interface between two potential-based fluid elements of two
different element groups (Figure 2.11-11). For example, the
interface between air and water can be modeled using fluid-fluid
interface elements.
Fluid element of Fluid-fluid interface

element group 2 element of element
group 2
Nodal unknown: f(2)
Nodal unknowns: f(1), un
Fluid-fluid interface
Fluid element of element of element
element group 1 group 1
Nodal unknowns: f(2), un
Nodal unknown: f(1)

un is small.
Figure 2.11-11: Fluid-fluid interface element
Each node of the interface element contains the potential degree

of freedom and displacement degrees of freedom. The normal
velocities and displacements are assumed to be small. It is
necessary to fix all displacements that are tangential to the fluid-
fluid interface element, because the fluid-fluid interface element
does not provide stiffness, mass or damping to the tangential
directions.
Each of the potential-based elements that share a common
boundary requires a fluid-fluid interface element. The two fluids
are connected by constraining the normal displacements of one of
the interface elements to the corresponding normal displacements
of the other interface element.

In many cases, the AUI can generate these fixities and

constraints automatically during phi model completion, see Section
2.11.15.
2.11.14 Modeling: Loads
Concentrated forces, pressure loads, prescribed displacements
Concentrated forces, pressure loads and/or prescribed

displacements can be applied directly to any part of the fluid
boundary on which there are fluid-structure, free surface, inlet-
outlet or fluid-fluid interface elements. However, when applying
concentrated forces, remember that the AUI can apply skew
systems to certain nodes on the fluid boundary during phi model
completion, see Section 2.11.15. Therefore you should make sure
that the nodes on which you apply concentrated forces have the
degree of freedom directions that you anticipate.
Mass-proportional loads
Mass-proportional loads can be applied. However the AUI and

ADINA make a distinction between those mass-proportional loads
used to model physical body forces and those mass-proportional
loads used to enter ground accelerations. For each mass-
proportional load, you must specify its interpretation: body force or
ground acceleration.
Body force: Mass-proportional loads interpreted as physical

body forces must be constant in time when there are
infinitesimal velocity potential-based fluid elements in the
model. These loads are used in the construction of g k and
as follows:
g k = MAGNITUDE A(k) TF
where MAGNITUDE is the magnitude of the mass-proportional

loading, A(k) is the vector giving the direction of the mass-
proportional loading and TF is the value of the time function.
Then

Chapter 2: Elements
3
= g k ( xk xk0 ) (2.11-59)
k =1
where xk0 is a datum value (entered as part of the fluid material

description). Notice that the choice of the datum value affects
the solution only when infinite interface elements are present, or
if at least one potential degree of freedom is fixed in the fluid
region.
Body forces can be applied in a static analysis as the only
loads in the analysis. Then, if other time-varying loads are
present, a restart to dynamic analysis can be performed. The
body forces should be kept in the restart dynamic analysis.
Ground acceleration: Mass-proportional loads interpreted as

ground motions can be time-varying, and are used in the
construction of ugk as follows: Suppose that you specify a
mass-proportional load interpreted as ground motions of
magnitude gk(t), where gk(t) is entered as for body force loads
(but the time function need not be constant). Then ADINA
computes the ground motions using
t t
u gk = g k , u gk =
tstart
u gk dt , u gk =
tstart
u gk dt (2.11-60a,b,c)
Note that it is assumed that u gk (tstart ) = 0, u gk (tstart ) = 0 .

For example, if g k = sin t , then the above equations imply
1 1 sin t
u gk = (1 cos t ) , ugk = t
. It is seen that
the average ground velocity is non-zero, which is probably not
realistic. As an alternative, use
t
gk = sin t , t < t p ,
tp
= sin t , t tp

2
in which t p = is one period of the ground motion.

The choice of interpretation for the mass-proportional loads

does not affect the structural elements used in the model.
Centrifugal loads
Centrifugal load effects are not included in the potential-based

fluid elements.
Mass flux loads (phiflux loads)
Mass flux loads (also referred to as phiflux loads) can be

prescribed directly onto potential-based fluid elements. The mass
fluxes can be distributed or concentrated. The dimensions of
[mass]
distributed mass flux are in dynamic analysis and
[time] [area]
[mass]
are in static analysis). The dimensions of concentrated
[area]
[mass]
mass flux are in dynamic analysis and are [mass] in static
[time]
analysis.
Positive mass flux is assumed to represent mass flowing into the
fluid domain.
No interface elements or potential-interfaces should be defined
on boundaries with distributed or concentrated mass fluxes.
2.11.15 Modeling: Phi model completion
As can be seen above, there are many restrictions and conditions

that must be considered when specifying boundary conditions on
potential-based fluid elements. Many of these conditions have
been automated in the AUI in the following way. The AUI
performs phi model completion whenever generating a data file
in which potential-based fluid elements are used. The steps in phi
model completion are:

Chapter 2: Elements
1) The AUI loops over all fluid element sides on the boundary
of each fluid element region. If the fluid element side already has
an interface element, the side is skipped. If the fluid element side
has a potential-interface, an interface element of the appropriate
type is generated. Otherwise, the side is checked to see if it is
attached to a structure (shares structural degrees of freedom with
structural elements), close to a structure (nodes coincident with
nodes of a structural element) or on a ADINA-F fluid-structure
boundary; and, if any of the above conditions are met, an interface
element of the appropriate type is generated. The intent of this step
is to cover as much of the fluid boundary as possible with interface
elements.
Note that phi model completion relies on determining the nodes
that are close together. The tolerances used in this determination
are controlled by command PHI-MODEL-COMPLETION,
parameters CLOSE-TOL, XTOL, YTOL, ZTOL. By default, the
tolerances used during phi model completion are the same
tolerances used in mesh generation coincident node checking .
2) The AUI loops over all nodes attached to interface elements.

If the node is attached to structural elements, the node is skipped.
Otherwise the types of the attached interface elements are
determined. Then
a) If the node is attached only to a free surface interface,

ADINA-F interface, inlet-outlet interface or fluid-fluid interface,
then the node has a free normal direction (normal to the interface)
and zero stiffness directions that are tangential to the free normal.
The free normal and zero stiffness directions are identified, and if
they are not aligned with the global directions, a skew system is
generated that is aligned with the free normal and zero stiffness
directions.
b) If the node is attached to a free surface interface, ADINA-F

interface, inlet-outlet interface or fluid-fluid interface, and is also
attached to a fluid-structure interface or rigid-wall interface, the
AUI proceeds as follows. The node has a free normal direction
(determined from the free surface interface, ADINA-F interface,
inlet-outlet interface or fluid-fluid interface), a structural normal
direction (determined from the fluid-structure interface or rigid-
wall interface), and, in 3D, another direction orthogonal to the free

normal and structural normal directions, which may be a zero

stiffness direction or another structural normal direction. The free
normal direction is modified to be orthogonal to the structural
normal directions. The free normal, structural normal and zero
stiffness directions are identified, and, if they are not aligned with
the global directions, a skew system is generated that is aligned
with the free normal, structural normal and zero stiffness
directions.
This process relies on an angle tolerance in 3D analysis to
determine if the structure is smooth. This tolerance is parameter
PHI-ANGLE in command PHI-MODEL-COMPLETION. See
Example 2 below.
The intent of step 2 is to identify the zero stiffness and free

normal directions of the nodes.
3) The AUI loops over all nodes attached to interface elements.

If the node is attached to a structural element, the node is skipped.
If the node (node A) is attached to a fluid-structure interface
element and is close to a structural node B, node A is constrained to
node B as follows. Each displacement degree of freedom for node
A is constrained to the corresponding degrees of freedom for node
B, accounting for differences in skew systems between A and B,
accounting for B possibly being a slave node in a constraint
equation (but not accounting for B possibly being a slave in a rigid
link), accounting for B possibly being fixed. But a displacement
degree of freedom for node A is not constrained if the degree of
freedom is a free normal direction (see 2 above).
The intent of step 3) is to connect the fluid mesh with the

structural mesh, when different nodes are used for the fluid and
structure. The connection still allows the fluid nodes to slip
relative to the structural nodes on intersections between free
surfaces and the structure.
4) In static analysis, when there are no body force loads, the AUI
loops over all nodes on a free surface, ADINA-F interface, or fluid-
fluid interface. If the node is attached to a structural element, the
node is skipped. Otherwise, constraint equations are defined for all
nodes so that the displacements in the direction of the free normal
are equal.

Chapter 2: Elements
The intent is to remove the zero pivots in the system matrices

that are otherwise present (see the discussion after equation
(2.11-37)).
5) The AUI loops over all nodes on an inlet-outlet interface. If the

node is attached to a structural element, the node is skipped.
Otherwise, constraint equations are defined for all nodes so that the
displacements in the direction of the free normal are equal.
The intent is to make the displacement / velocity constant over
the inlet-outlet interface.
6) In dynamic analysis, or in static analysis when there are body

force loads, the AUI loops over all nodes on a fluid-fluid interface.
If the node is attached to a structural element, the node is skipped.
Otherwise constraint equations are defined for all pairs of nodes, so
that the displacements in the direction of the free normal are
compatible.
The intent is to enforce displacement compatibility between the
fluids.
7) The AUI then loops over all nodes with zero stiffness degrees of
freedom and defines fixities for each zero stiffness degree of
freedom.
8) In static analysis, the AUI constrains all of the potential degrees

of freedom for an element group together.
9) In dynamic analysis, the AUI constrains all of the potential

degrees of freedom on an inlet-outlet together. The intent is to
remove the tangential flows on an inlet-outlet.
Note that the model that results from the phi model completion
process is not stored in the AUI, but is immediately written to the
ADINA data file. Therefore the model that results from phi model
completion cannot be displayed during model definition. Phi
model completion does write some messages to the log file or user
interface, which you may find helpful. However, we recommend
that you also check the model by running the model through
ADINA, for example, for one load step; then use ADINA-PLOT to
display the model.

Example 1: We now present a detailed example for a 2-D fluid

filled basin with flexible walls.
Figure 2.11-12(a) shows the model before phi model
completion. The model is defined with separate nodes for the fluid
and the structure. One way to guarantee that the nodes on the fluid
and structure are separate is to generate the structural elements as
usual, but to set coincidence checking to group when generating
the fluid elements.
A potential-interface of type free surface is defined on the
geometry line corresponding to the free surface.
Potential-interface of type
free surface is defined on
this line
16 3 2 1 10 Potential-based
Fluid nodes are at the same fluid elements
coordinates as structural nodes.
6 5 4 11
The fluid nodes and structural Beam elements
nodes are separated in this figure 15
for clarity. 9 8 7
30o
14 13 12
a) Finite element model before phi model completion
Figure 2.11-12: Example 1 of phi model completion
In step 1 of phi model completion, the AUI generates fluid-

structure interface elements where the fluid is adjacent to the
structure, and free surface interface elements corresponding to the
potential-interface (Figure 2.11-12(b)).

Chapter 2: Elements
Interface elements
of type free surface
3 2 1
6 4
9 8 7
Interface elements of
type fluid-structure interface
b) Step 1 of phi model completion;

interface elements are created
In step 2 of phi model completion, the AUI classifies the

displacement directions on the free surface (Figure 2.11-12(c)).
Notice that the free normal for node 2 is taken from the free
surface, but the free normal for node 3 is modified by the presence
of the adjacent structure. A skew system is defined for node 3
because the free normal and structural normal are not aligned with
the global coordinate directions. The zero stiffness direction of
node 2 will be fixed in step 6 below.

b z z
y y
c 3 2 1
Node 1: y = structural normal direction,

z = free normal direction
Node 2: y = zero stiffness direction,
z = free normal direction
Node 3: b = free normal direction,
c = structural normal direction
c) Step 2 of phi model completion: classification

of displacement directions on free surface
In step 3 of phi model completion, the AUI constrains the fluid

displacement directions to the structure (Figure 2.11-12(d)). At
node 1, only the y displacement direction is constrained; the z
displacement is left free so that the free surface can slip along the
wall. At node 3, only the c displacement direction is constrained;
the b displacement is left free so that the free surface can slip along
the wall.
Notice that at node 4, both the y and z displacements are
constrained to the structure. The fluid still slips in the z direction
because only the normal displacement (the y displacement in this
case) is used by the fluid equations. Similar statements hold for
nodes 6 and 8.
Nodes 7 and 9 are fixed because corresponding nodes 12 and 14
are fixed.

Chapter 2: Elements
16 3 1 10
6 4 11
15
9 8 7
14 13 12
Node 1: uy=u10
y
o o
Node 3: uc=-cos30 uy16 - sin 30 uz16
Node 4: uy=u11 11
y , uz=uz Node 6: uy=u15 15
y , uz=uz
Node 7: uy=uz =fixed Node 8: uy=u13 13
y , uz=uz
Node 9: uy=uz =fixed
d) Step 3: Creation of constraint equations and fixities
If the analysis is static without body forces, then the AUI

performs step 4 of phi model completion (Figure 2.11-12(e)). The
free surface can only translate vertically as a rigid body.
b z z
y y
c 3 2 1
Node 2: uz=u1z
o
Node 3: ub= uz1 / cos30
Step 4 is only performed in static analysis when
there are no body forces.
e) Step 4 of phi model completion: defining constraint

equations to set normal displacements equal on free surface

This motion affects the total mass of the fluid region, so that
there is no zero pivot in the system matrices.
If there are body forces, then step 4 is not necessary because all
boundary motions are given stiffness by the matrix ( K UU )S .
Step 5 of phi model completion is skipped because there are no
fluid-fluid interfaces.
In step 6 of phi model completion, the zero stiffness direction at
node 2 is fixed (Figure 2.11-12(f)). Vertical motions of the nodes
attached only to free surface interface elements are allowed, but
horizontal motions of these nodes are not allowed (because the
fluid does not provide stiffness, damping or mass to horizontal
motions).
z
y
2
Node 2: uy=fixed
f) Step 6 of phi model completion: defining fixities to

eliminate zero stiffness degrees of freedom
Figure 2.11-12: (continued) .
If the analysis is static, then the AUI performs step 7 of phi

model completion (Figure 2.11-12g). Only constant (in space)
potentials are allowed in static analysis.

Chapter 2: Elements
3 2 1
6 5 4
9 8 7
f2=f1, ..., f9=f1
Step 7 is performed only in static analysis
g) Step 7: defining constraint equations to set all potential

degrees of freedom equal
Example 2: An alternative way to model the problem of Example

1 is shown in Figure 2.11-13. Here the free surface is shifted
downwards slightly so that the nodal coincidence checking feature
will not reuse any existing nodes on the free surface boundary.
Depending on how much you shift the fluid free surface, you may
need to adjust the tolerance used in the coincidence checking
algorithms.
Potential-interface of type
free surface is defined on
this line
11 2 10 Potential-based
3 1 fluid elements
6 5 4 Beam elements
9 8 7
30o
Figure 2.11-13: Alternative modeling of problem in Figure 2.11-12
During phi model completion, it is necessary to set the

coincidence tolerance to be loose enough so that node 1 is
considered close to node 10 and node 3 is considered to be close
to node 11. Then steps 1, 2, 4, 6, 7 of phi model completion
proceed as in Example 1, and step 3 only processes nodes 1 and 3.
The solution results will be almost exactly the same as in Example

1 (the solutions will be slightly different because the geometry is

slightly different).
We think that the user input for Example 2 is more difficult than
the user input for Example 1 because Example 2 requires a good
working knowledge of the nodal coincidence checking features.
Example 3: In 3-D analysis of a fluid-filled basin, there is one

additional consideration. Consider the model shown in Figure
2.11-14, in which only the free surface and the adjacent structural
nodes are shown.
y
x
13 14 15
3 Sides of 3-D potential-based

1 2
fluid elements on free surface
16
22
4 5 6 Sides of shell elements
21 17 Fluid nodes are at the same

7 8 9 coordinates as structural nodes.
The fluid nodes and structural
10 11 12 nodes are separated in this figure
for clarity.
20 19 18
a) Top view of finite element model
During step 2 of phi model completion, the AUI determines the

structural normal direction(s), zero stiffness direction(s) and free
normal direction for the nodes of the free surface (Figure 2.11-
14(b)). Notice that node 3 has two structural normals, but node 6
has only one structural normal. That is because the angle between
the two structural normals for node 3 is greater than PHI-ANGLE,
but the angle between the two structural normals for node 6 is less
than PHI-ANGLE.
Nodes 3, 6, 9 and 12 are assigned skew systems because the
structural normal directions are not aligned with the global system.

Chapter 2: Elements
n12 n23
1 2 3
n14 n36
4 5 6
n69
7 8 9
y
10 11 12 x
Node 1: Structural normal 1 = n14 Node 2: Structural normal = n12 = n23

Structural normal 2 = n12 Zero stiffness direction = x
Free normal = z Free normal = z
Node 3: Structural normal 1 = n23 Node 5: Zero stiffness direction 1 = x
Structural normal 2 = n36 Zero stiffness direction 2 = y
Free normal = z Free normal = z
Node 6: Structural normal = average of n36 and n69
Free normal = z
Zero stiffness direction = remaining orthogonal direction
b) Classification of structural normals and zero stiffness directions

for some nodes on the free surface
During step 3 of phi model completion, the AUI creates

constraint equations for the fluid nodes adjacent to the structural
nodes. For example, node 1 is constrained in both the x and y
directions to node 13, because both directions are structural normal
directions. Node 2 is also constrained in both the x and y
directions to node 14, here because the y direction is a structural
normal direction and the x direction is a zero stiffness direction. (It
is assumed that the structure provides stiffness in the x direction.)
During step 6 of phi model completion, the AUI fixes the x and
y directions for nodes 5 and 8, because these directions are zero
stiffness directions, and there is no adjacent structure.

Example 4: We now present some of the steps for phi model

completion of an enclosure with two distinct fluid regions. Figure
2.11-15(a) shows the model before phi model completion.
Fluid nodes are at the same coordinates as structural nodes.

The fluid nodes and structural nodes are separated in this figure
for clarity.
15 14 13 2-D potential-based
fluid elements, group 1
3 2 1
Beam elements
6 5 4 Enclosed nodes all lie on

17 16 the same geometry line,
potential-interface of type
9 8 7 fluid-fluid is defined on
this line
2-D potential-based
12 11 10 fluid elements, group 2
20 19 18
a) Finite element model before phi model completion
In step 1 of phi model completion, the AUI generates fluid-

structure interface elements where the fluid is adjacent to the
structure, and fluid-fluid interface elements corresponding to the
potential-interface. Four fluid-fluid interface elements are
generated, two for each shared element side.
In step 2 of phi model completion, the AUI classifies the
displacement directions on the fluid-fluid interface. Here the free
normal is always in the z direction and the zero stiffness directions
for nodes 5 and 8 are in the y direction.
In step 3 of phi model completion, the fluid nodes are
constrained to the adjacent structural nodes (Figure 2.11-15(b)).
Notice that the nodes on the fluid-fluid interface are allowed to slip
relative to the structure.

Chapter 2: Elements
15 14 13
3 2 1
6 5 4
17 16
9 8 7
12 11 10
20 19 18
Node 1: uy=u13 13
y , uz=uz Node 2: uy=u14 14
y , uz=uz
Node 3: uy=u15 15
y , uz=uz Node 4: uy=u16
y
Node 6: uy=u17
y Node 7: uy=u16
y
Node 9: uy=u17
y Node 10: uy=u18 18
y , uz=uz
Node 11: uy=u19 19 20 20

y , uz=uz Node 12: uy=uy , uz=uz
b) Step 3: constraining fluid nodes to adjacent structural nodes
If the analysis is static without body forces, then the AUI

performs step 4 of phi model completion. In this case, the z
displacements of nodes 5 to 9 are constrained to be equal to the z
displacement of node 4. The free surface can only translate
vertically as a rigid body.
In step 5 of phi model completion, the fluid nodes on the fluid-
fluid interface are constrained to each other (Figure 2.11-15(c)).
Step 5 is not performed if step 4 was performed. Notice that the
potential degrees of freedom are not constrained.

15 14 13
3 2 1
6 5 4
17 16
9 8 7
12 11 10
20 19 18
Node 4: uz=u7z Node 5: uz=u8z

Node 6: uz=u9z
c) Step 5: constraining adjacent nodes of the two fluid groups together
In step 6 of phi model completion, the zero stiffness direction of

nodes 5 and 8 are fixed. Vertical motions of the nodes attached
only to fluid-fluid interface elements are allowed, but horizontal
motions are not allowed (because the fluid does not provide
stiffness, damping or mass to horizontal motions).
If the analysis is static, then the AUI performs step 7 of phi
model completion (Figure 2.11-15(d)). Only constant (in space)
potentials are allowed in static analysis, but the potential can be
different for the two fluid element groups.

Chapter 2: Elements
15 14 13
3 2 1
6 5 4
17 16
9 8 7
12 11 10
20 19 18
f2=f1, ..., f6=f1 f8=f7, ..., f12=f7
Step 7 is performed only in static analysis
d) Step 7: defining constraint equations to set the potential

degrees of freedom of each element group together
Example 5: We demonstrate how to model an initial pressure in

a confined fluid region. This can be done in a static analysis as
follows: Define an auxiliary fluid element, physically unattached
to the fluid region but belonging to the same fluid element group.
On the auxiliary fluid element, place a potential-interface of type
inlet-outlet, and apply the specified initial pressure to the inlet-
outlet.
The constraint equations that the AUI defines in static analysis
connects the auxiliary fluid element to the confined fluid region.
The result is that the value of in the confined fluid region is the
same as the value of in the auxiliary fluid element; the value of
in the auxiliary fluid element is determined by the pressure that
you apply to the inlet-outlet.
If you restart to a dynamic or frequency analysis, fix all of the
degrees of freedom of the auxiliary fluid element. Since, in
dynamic or frequency analysis, the AUI does not define constraint
equations connecting the potential degrees of freedom, the
auxiliary fluid element is no longer connected to the confined fluid

region. The initial values of in the confined fluid region are that
from the static analysis, so the confined fluid region has the same
initial pressure as in static analysis.
The modeling process is schematically illustrated in Figure
2.11-16.
Potential-interface
of type inlet-outlet
Applied pressure
Confined
fluid region Finite fluid
with nonzero elements
fluid pressure
Auxiliary
Constraint equations fluid element,
automatically generated in same fluid
by the AUI during phi element group
model completion
a) Physical problem b) Modeling in static analysis
Pressure at
beginning of Fix all degrees of freedom
restart analysis attached to the auxiliary fluid
= pressure at element.
end of static
analysis.
c) Restart to dynamic or frequency analysis
Figure 2.11-16: Modeling an internal pressure in a confined fluid region
2.11.16 Modeling: Considerations for static analysis
In static analysis, the potential degree of freedom output should

be interpreted as , not .

Chapter 2: Elements
2.11.17 Modeling: Considerations for dynamic analysis
Any of the direct time integration methods available in ADINA,

except for the central difference method, can be used in direct time
integration. Rayleigh damping can be specified in the structure.
Please remember that there is no damping within the fluid.
Therefore the choice of initial conditions is extremely important,
because poorly chosen initial conditions will cause transient
responses that are not damped out.
2.11.18 Modeling: Considerations for frequency analysis and modal

participation factor analysis
Frequency analysis (and all analyses that depend upon

frequency data) cannot be performed if infinite interface elements
are present in the analysis.
For ground motion modal participation factor calculations, it is
recommended that each fluid region be completely surrounded by
interface elements. This sets the modal participation factors for the
rigid body modes to zero.
Figure 2.11-17 shows a one-dimensional example in which
there is a rigid body mode and the ground motion modal
participation factor is nonzero. The motion excited by the nonzero
ground motion modal participation factor corresponds to a constant
ground acceleration. Notice that this ground motion causes the
unbounded expansion of the fluid region.

Fixed to
ground
uz = constant ground motion
k k
uz is the absolute
Fluid Fluid displacement
No boundary uz = 0
condition
a) Finite element model b) Interpretation of f rigid body mode
Figure 2.11-17: Problem in which the ground motion modal

participation factor for the f rigid body mode
is nonzero
2.11.19 Modeling: Element output
Each fluid element outputs, at its integration points, the

following information to the porthole file (the integration point
numbering is the same as the numbering convention used for the
solid elements): FE_PRESSURE,
FLUID_REFERENCE_PRESSURE, ELEMENT_X-VELOCITY,
ELEMENT_Y-VELOCITY, ELEMENT_Z-VELOCITY
(ELEMENT_X-VELOCITY is applicable only for 3-D elements).
FE_PRESSURE is the value of p and
FLUID_REFERENCE_PRESSURE is the value of (see
Equation (2.11-59)).
In the output of modal quantities, ELEMENT_X-VELOCITY,
ELEMENT_Y-VELOCITY, ELEMENT_Z-VELOCITY should be
interpreted as modal particle displacements, not velocities, in
accordance with equation (2.11-42). However in all other types of
analysis (static, direct time integration, modal superposition,
response spectrum, harmonic and random), these quantities are
velocities. The velocities are absolute velocities, even when
ground motions are entered using mass-proportional loads.
Interface elements do not have any output.

Chapter 2: Elements
2.12 Alignment elements

2.12.1 Overview
The alignment element is a two-node element, as shown in Fig

2.12-1. The element is used to prescribe the relative translations,
distances and/or rotations of the two nodes.
Possible applications for the element include:
Bringing two bodies together

Robot arms and other parts in which the relative rotations are
prescribed
Simulation of body-body contact
There are three types of alignments
Translation alignments
Distance alignments
Rotation alignments
Any or all of these alignments can be used in the alignment element

definition.
Each alignment can be prescribed as a function of time. Thus

the alignment element has some of the characteristics of a
prescribed load or prescribed constraint boundary condition.

2.12: Alignment elements
Local
node 2
offset
Local
node 1 Local
offset node 2
Local
node 1
Figure 2.12-1: Alignment element
In order to introduce the concepts used in the alignment

element, we present some illustrative examples. Details are given
later on.
Example 1) Translational alignment: As a simple example,

consider an alignment element that aligns translations only. In Fig
2.12-2, there is a misalignment between the two nodes. This
misalignment is measured by the vector connecting the two nodes.
Also shown in this figure are equal and opposite forces generated
by the alignment element. These forces are proportional to the
misalignment, and these forces act to reduce the misalignment. The
element provides stiffness in all three translational directions.
This element can be used to keep the two nodes nearly coincident.
Notice that, although the nodes are nearly coincident, their relative
rotations are not affected by the element, so the nodal rotations can
be very different.

Chapter 2: Elements
F
Local
node 2
F
Local Misalignment vector
node 1
Figure 2.12-2: Example 1, translational alignment
Example 2) Translational alignment with prescribed translational

alignment: This example considers the same alignment element as
in Example 1, but now a prescribed alignment is included (Fig
2.12-3). The prescribed alignment is a vector measured from local
node 1. The misalignment is the vector difference of the total
alignment and the prescribed alignment. The forces generated by
the alignment element are equal and opposite, and these forces act
to reduce the misalignment. Notice that the forces do not in
general act along the line connecting the two nodes, in particular a
moment is generated by the alignment element force acting on local
node 1.
The prescribed alignment is a function of time, entered as part of

the definition of the translation alignment.
As a common special case, the prescribed alignment can be

specified as a fraction of the initial alignment. This case is useful
when the two nodes are initially misaligned, and the nodes are to be
brought into alignment through several solution steps. In each
solution step, the fraction is reduced, until the fraction is zero; then
the prescribed alignment is zero.
In Fig 2.12-3(b), the prescribed alignment is equal to the initial

misalignment. Therefore the misalignment is zero and no forces are
generated by the alignment element. In Fig 2.12-3(c), the
prescribed alignment is reduced, therefore the misalignment is non-
zero and forces are generated. The forces will cause the nodes to
approach each other.

F
Local
node 2
F
Misalignment vector
Prescribed translational alignment

Local
node 1
(a): General case
Local
node 2
Prescribed translational alignment

Local
node 1
(b): Special case, prescribed alignment equals initial misalignment
F
Local
node 2
Misalignment vector
F
Local
node 1 Prescribed translational alignment
(c): Special case, prescribed alignment equals a fraction of the initial misalignment
Figure 2.12-3: Example 2, translational alignment with prescribed alignment
Example 3) Translational alignment with prescribed translational

alignment and nodal offsets: This example considers the same
alignment element as in Example 2, but now nodal offsets are

Chapter 2: Elements
included (Fig 2.12-4). Thus the prescribed alignment, total

alignment and misalignments are measured from positions 1b and
2b. As in Example 2, the forces generated by the alignment
element are equal and opposite, and these forces act to reduce the
misalignment. Notice that the forces do not in general act along the
line connecting the two nodes, and in this example moments act on
both local nodes 1 and 2.
F F
Prescribed alignment 2b
Misalignment
Offset for
local node 2 Local
1b Actual alignment node 2
Offset for
local node 1
Local
node 1
Figure 2.12-4: Example 3: Translational alignment with
prescribed alignment and nodal offsets
Example 4) Distance alignment with prescribed distance alignment

and nodal offsets: This example considers an alignment element
that aligns distance, instead of aligning the translational
components (Fig 2.12-5). Here the prescribed distance is a scalar
quantity, measured from local node 1. The distance misalignment
is the scalar difference of the total distance and the prescribed
distance. The forces generated by the alignment element are equal
and opposite, and these forces act to reduce the misalignment.
Of course, it is also possible to not use nodal offsets, then the

distance is measured directly between the two nodes.
For the distance alignment, the force always acts along the line
connecting points 1b and 2b. The element provides stiffness only in
the direction of the line connecting points 1b and 2b.

Misalignment F Offset for

F 2b local node 2
Prescribed distance
Local
Total distance node 2
1b
Offset for
local node 1
Local
node 1
Figure 2.12-5: Example 4: Distance alignment with prescribed
alignment and nodal offsets
Example 5) Rotation alignment: The discussion of rotational

alignments is more difficult than the discussion of translational or
distance alignments. In order to provide simple examples, we
consider only planar rotations here. The general case of 3D
rotations is discussed below.
Fig 2.12-6 shows an example of rotational alignments in the plane.

Here each local node is seen to have a local coordinate system,
called a triad, attached to it. The directions of the triad axes for
local node 1 are denoted a11 , a21 and the directions of the triad
axes for local node 2 are denoted a12 , a22 . The angle giving the
orientation of the directions a12 , a22 in terms of the directions
a11 , a21 is the rotational misalignment angle . This
misalignment creates equal and opposite moments acting onto the
two nodes.
As the nodes rotate, the triads also rigidly rotate.
The alignment element in this example can be used to keep the

rotations of the two nodes nearly coincident. Notice that, although
the rotations of the nodes are nearly coincident, their translations
are not affected by the element, so the nodal positions need not be
coincident.

Chapter 2: Elements
2
a2
1
a2 2
a1
Local q
node 2
Local M M
node 1 1
a1
Figure 2.12-6: Example 5: Rotation alignment
Example 6) Rotation alignment with prescribed rotational

alignment: Fig 2.12-7 shows another example of a rotational
alignment element, here including a prescribed rotational alignment
. Angle is used to construct another triad, with directions
denoted d11 , d12 , relative to the directions a11 , a21 . The rotational
misalignment angle is measured from directions d11 , d12 to
directions a12 , a22 . Again, the misalignment creates equal and
opposite moments acting onto the two nodes.

2
a2
2
d1
1 2
a2 a1
q
Local
node 2
M
Local 1
node 1 g d1
1
M a1
Figure 2.12-7: Example 6: Rotation alignment with prescribed

rotational alignment
Example 7) Rotation alignment with prescribed rotational

alignment and offset triads: Figure 2.12-8 shows another example
of a rotational alignment element, here including both a prescribed
rotational alignment and offset triads. The offset triad for node 1 is
given by directions b11 , b21 and the offset triad for node 2 is given
by directions b12 , b22 . The prescribed rotational alignment is
measured from directions b11 , b21 and the rotational misalignment
angle is measured from directions d11 , d12 to directions
b12 , b22 .
Although in the figure, the origins of the b triads is different than

the origins of the a triads, it is certainly possible to have the origins
of the b triads coincident with the origins of the a triads, and to
have the b triad directions different than the a triad directions.

Chapter 2: Elements
2
b2
2
d1 1
b2 2
b1
q
2b 2
a2
Offset for
M local node 2
1b 1
g d1
1 1
a2 M b1
Local
Offset for node 2
local node 1
2
a1
1
a1
Local
node 1
Figure 2.12-8: Example 7: Rotation alignment with prescribed

rotational alignment and offset triads
A single alignment element can align translations and/or

distances and/or rotations. For example, a single element that aligns
both translations and rotations can be used to keep two nodes
nearly coincident, and also the relative rotations of the two nodes
nearly equal.
2.12.2 Triads
In this section the nodal offsets and triads are discussed in more
detail.
Figure 2.12-9 shows the triads at a node.

d3
d1
b1 c3
c1
b2
c2 xdb d2
a3
a2
b3 xba
Node
a1
Figure 2.12-9: Triads at a node
A triad
Each node of the alignment element has a triad attached to the

node. This triad is called the "a" triad, and the coordinate axes of
the a triad are denoted a1, a2, a3. This triad is an orthonormal
right-handed coordinate system, with origin at the node. As the
node translates and rotates, the triad rigidly translates and rotates
with the node.
B triad
Each node of the alignment element also has a triad which is in

general offset from the a triad. This offset triad is called the "b"
triad, and the coordinate axes of the b triad are denoted b1, b2, b3.
The b triad is an orthonormal right-handed coordinate system, with
the origin of the b triads given by vector xba.
The origin and coordinate axes of the b triad are defined with
reference to the a triad. Thus the b triad is rigidly attached to the a
triad. In particular, as the node rotates, vectors xba, b1, b2, b3 also
rotate.

Chapter 2: Elements
C triad
Local node 1 of the alignment element also has a triad that provides
a coordinate system in which the prescribed alignments, total
alignments and misalignments are measured. This triad is called
the "c" triad and the coordinate axes of the c triad are denoted c1,
c2, c3. The c triad is an orthonormal right-handed coordinate
system, with origin at the origin of the b triad. The coordinate axes
of the c triad are defined with reference to the b triad. Thus the c
triad is rigidly attached to both the b triad and the a triad.
Thus, for example, the components of the prescribed translational

misalignments are specified in the c triad coordinate system.
D triad
Local node 1 of the alignment element also has a triad that

represents the prescribed translation or rotation. This triad is called
the "d" triad and the coordinate axes of the d triad are denoted d1,
d2, d3. The d triad is an orthonormal right-handed coordinate
system, with origin offset from the origin of the b triad.
The prescribed translational alignment is used to determine the

origin of the d triad relative to the b triad of local node 1 (in other
words, the vector xdb), and the prescribed rotational alignment is
used to determine the orientation of the d triad relative to the
orientation of the b triad of local node 1. This will be discussed in
more detail later.
2.12.3 Notations
The notations used in the discussion of the theory are based on the
triad names (a, b, c, d). Although the notations are cumbersome,
the notations clearly show the meanings of the quantities described.
Vectors
xba : vector from a triad origin to b triad origin

xdb : vector from b triad origin to d triad origin

When it is necessary to denote a particular local node, a superscript

is used:
xb1a1 : vector from a triad origin of local node 1 to b triad origin of

local node 1
xb 2 a 2 : vector from a triad origin of local node 2 to b triad origin of
local node 2
xd1b1 : vector from b triad origin of local node 1 to d triad origin of
local node 1
xb 2d1 : vector from d triad origin of local node 1 to b triad origin of
local node 2.
Obviously the laws of vector addition hold, for example

xd1a1 = xd1b1 + xb1a1 .
Since the above quantities are vectors, they can be expressed using
components in any convenient coordinate system. In order to
denote which coordinate system the components are expressed in,
we use a letter in the subscripts, for example ( xd 1bc11 , xd 1bc12 , xd 1bc13 )
are the components of vector xd1b1 , expressed in the c triad
coordinate system.
Rotation matrices
For rotations, we need to express the directions of one triad in

terms of the directions of another triad. For example
b1b211 b1b2 21 b1b231

VB 2 B1 = b 2b211 b 2b2 21 b2b231
b32 1 b3b2 21 b3b231
b1
The first row of the matrix gives the direction b1 for local node 2,
the second row gives the direction b2 for local node 2, and the third
row gives the direction b3 for local node 2. Each of these direction
vectors is expressed using the coordinate system of the b triad of
local node 1.
Clearly if the directions of the b triad axes are the same for both

Chapter 2: Elements
local nodes, then there is no relative rotation between the b triads of

local nodes 1 and 2, and this matrix reduces to the identity matrix.
We also use
b12d 11 b1d2 21 b1d2 31

VB 2 D1 = b2d2 11 b2d2 21 b 2d2 31 : the b triad directions for local
b32 1 b3d2 21 b3d2 31
d1
node 2 in terms of the d triad directions for local node 1
d11b11 d11b 21 d11b 31

VD1B1 = d 21b11 d 21b 21 d 21b 31 : the d triad directions for local
d 31 1 d 31b 21 d 31b 31
b1
node 1 in terms of the b triad directions for local node 1
The rules of matrix multiplication apply, for example

VB 2 B1 = VB 2 D1 VD1B1 . Also the inverse of the matrix is the
same as the transpose, for example (VD1B1 ) 1 = (VD1B1 )T .
2.12.4 Translation alignment theory
Figure 2.12-10 shows an alignment element that uses a translation

alignment.

c3 1 1 xb2d1 a3
xd b
c2 xb2 a2
Local
xb2b1 node 2
a3
1 1
xb a
a2 a2 a1
c1
Local
node 1 a1
Figure 2.12-10: Translation vectors used in translation alignment
Translation alignment calculation
When the translation alignment is used, the vector between the

origin of the b triad of local node 1 and the origin of the b triad of
local node 2 is calculated ( xb 2b1 ).
Notice that the directions of the b triad coordinate axes are not used
in the translation calculation.
Prescribed translation
The user enters a table giving the "prescribed translation" for

various times. This prescribed translation can be specified using
the following options:
Aligned: The prescribed translation is set to zero.
Components: The prescribed translation is directly specified as

the vector xd1b1 , given in terms of components in the c triad
coordinate system ( xd 1bc11 , xd 1bc12 , xd 1bc13 ) .
Tbfactor: The prescribed translation is specified relative to the

translations at the time of the birth of the alignment. The vector
xb 2b1 is evaluated at the time of birth of the alignment, and this
vector is used to construct a prescribed translation with the same

Chapter 2: Elements
direction, but with a scaled value of the magnitude, that is

xd1b1 = factor xb 2 b1 .
birth
Same: The prescribed translation is the same as in the previous

row of the align-translation table. (This option cannot be used in
the first row of the align-translation table.)
Notice that the directions of the d triad coordinate system are not
used in this calculation.
The ALIGN-TRANSLATION command is used to enter this table.
Translation misalignment
The difference between the actual translation and the prescribed

translation is calculated using xb 2d1 = xb 2 b1 xd1b1 . This
difference is termed the translation misalignment. Then vector
xb 2d1 is expressed using its components in the c triad coordinate
system: ( xb 2 d c11 , xb 2 d c12 , xb 2 d c13 )
Forces
The alignment element calculates a force from the translation

misalignment in each coordinate direction:
Fc1 = kc1 xb 2 d c11 , Fc 2 = kc 2 xb 2 d c12 , Fc 3 = kc 3 xb 2 d c13 . Notice that
the stiffness can be different in each coordinate direction. Hence,
for example, the user might specify that the stiffness be zero for
any misalignment in the c1 coordinate direction, but nonzero for
misalignments in the other coordinate directions. Figure 2.12-11
shows one possibility, drawn in two dimensions for simplicity.

xb2b1c2
xd1b1c2
Fc2 Fc1
xb2d1 a2
xd1b1 xb2a2
Local
Fc2 node 2
xb2b1 Fc1
a2 a1
xb1 a1
Local xd1b1c1
node 1
a1
xb2b1c1
Figure 2.12-11: Prescribed translations and forces for translation alignment,

two-dimensional case
The stiffness coefficients are entered in the ALIGN-

TRANSLATION command.
All forces act to reduce the translation misalignments.
The principle of virtual work is used to obtain the corresponding

forces and moments acting at the nodes.
2.12.5 Distance alignment theory
Fig. 2.11-12 shows an alignment element that uses a distance

alignment.

Chapter 2: Elements
a3
xb2 a2
Local
a3 xb2b1 node 2
1 1 a2
xb a
a2 a1
Local
node 1 a1
Figure 2-12.12: Distance vector used in distance alignment
Distance calculation
When the distance alignment is used, the distance D is measured

using the length of vector xb 2b1 . Vector xb 2b1 connects the origin
of the b triad of node 2 with the origin of the b triad of node 1.
Notice that the directions of the b triad axes are not used.
Also notice that the distance changes if the nodes rotate without
translation, since the offset vectors xb1a1 , xb 2 a 2 also rotate as the
nodes rotate.
Prescribed distance
The user enters a table giving the "prescribed distance" for various
times. The prescribed distance can be specified using the following
options:
Aligned: The prescribed distance is set to zero.
Distance: The prescribed distance is directly specified as the

number D P .
Tbfactor: The prescribed distance is specified relative to the

distance at the time of the birth of the alignment. D is evaluated
at the time of birth of the alignment, and this vector is used to
construct a prescribed distance with a scaled value of the

magnitude, that is D P = factor D birth .
Same: The prescribed distance is the same as in the previous

row of the align-distance table. (This option cannot be used in
the first row of the align-distance table.)
The ALIGN-DISTANCE command is used to enter this table.
Distance misalignment
D D P is termed the distance misalignment.
Forces
The alignment element calculates a force from the difference

between the distance and the prescribed distance Fd = kd ( D D P ) .
This force acts along the line connecting the origins of the b triads,
and acts to reduce the difference between the distance and the
prescribed distance. Fig. 2.12-13 shows some possibilities, drawn
in two dimensions for simplicity.

Chapter 2: Elements
a2
2
xb a2
D F Local
node 2
a2 a1
F
DP
1 1
xb a
Local a1
node 1 Figure 2.12-13(a): Distance greater than prescribed distance
a2
D xb2 a2
Local
node 2
a2
a1
DP
1 1
xb a
Local a1
node 1
Figure 2.12-13(b): Distance equal to prescribed distance

F
a2
xb2 a2
D Local
node 2
a2 a1
1 1
xb a
DP
F
Local
a1
node 1
Figure 2.12-13(c): Distance less than prescribed distance
The stiffness coefficient is entered in the ALIGN-DISTANCE

command.
The principle of virtual work is used to obtain the corresponding

forces and moments acting at the nodes.
2.12.6 Rotation alignment theory
Fig. 2.12-14 shows an alignment element that uses a rotation

alignment. Only the directions of the triads are used, and not their
origins, as shown in Fig. 2.12-14(b).

Chapter 2: Elements
b3
b1
c1 c3
b1 a3
xb2 a2
d1
d2
b2 Local
c2 node 2
a3
b2
a1
b3 xb1 a1 a2
a2 d3
c1
Local
node 1
a1
Figure 2.12-14(a): Rotation alignment
b11
b32
b21
2
b1 d1
d2
b31
b22 d3
Figure 2.12-14(b): Relative orientation of b and d triads,
corresponding to (a)
Rotation calculation
When the rotation alignment is used, the relative rotation between

the b triad of local node 2 and the b triad of local node 1 is
b1b211 b1b2 21 b1b231

calculated: VB 2 B1 = b 2b211 b 2b2 21 b2b231 .
b32 1 b32 1 b32 1
b1 b2 b3
It is easier to visualize the relative rotation for the special case in

which the b3 axes of the two triads have the same direction. This
case is shown in Fig 2.12-15(a). In this special case, the matrix has

the form
b1b211 b1b2 21 0

VB 2 B1 = b 2b211 b 2b2 21 0
0 0 1

b21 2
b1
b22
2 b11
b1b21
2
b1b11
(a): Relative orientation of b triads
b21 2
b1 q
d2 d3d1
P1
gb3
b22 b11
(b): Prescribed rotations and rotation misalignment
Figure 2.12-15: Rotations in the plane, b3 axes have the same direction
Prescribed rotation
The user specifies the "prescribed rotation" for various times.

Internally, the prescribed rotation is expressed as a coordinate triad
d1, d2, d3, defined relative to the b triad of node 1. In matrix
form, the prescribed rotation can be written as

Chapter 2: Elements
d11b11 d11b 21 d11b 31

VD1B1 = d 21b11 d 21b 21 d 21b 31
d 31 1 d 31b 21 d 31b 31
b1
Again, this situation is easier to visualize for the special case in

which the b3 axes of the two triads, and also the d3 axis, have the
same directions. This case is shown in Fig 2.12-15(b).
The prescribed rotation can be specified using the following

options:
Aligned: VD1B1 = I
Angle: The user enters an angle P and axis of rotation

(n P
c11
, ncP21 , ncP31 ) . The program normalizes the axis of rotation,
then the program calculates the axis of rotation in the b triad
system of node 1 ( nbP11 , nbP21 , nbP31 ) . Next the program calculates
( P
b11
, bP21 , bP31 ) = P ( nbP11 , nbP21 , nbP31 ) and then calculates the
prescribed rotation matrix using the formula
d1b11 d1b 21 d1b 31 P

sin 2
sin P
1 2
VD B = d 2b11
1 1
d 2b 21 d 2b 31 = I + S+ S ,
P
2 P
d3 1 d 3b 21 d 3b 31
b1 2
0 bP31 b 21
P

where S = bP31 0 bP11
P 1 bP1 0
b2 1
Notice that the components of the axis of rotation are specified

in the coordinate system of the c triad of local node 1. Also
notice that angle P and axis of rotation ( ncP11 , ncP21 , ncP31 )
represents the total angle and axis corresponding to the
prescribed rotation matrix. If the prescribed rotation is
expressed as a series of incremental rotations, the incremental
rotations must be combined, then converted back into the form

P , ( ncP1 , ncP2 , ncP3 ) .

1 1 1
Relative to rotations at time of birth: The matrix

b1b211 b1b2 21 b1b231

VB 2 B1 = b 2b211 b 2b2 21 b2b231 is evaluated at the time of birth
b32 1 b32 1 b32 1
b1 b2 b3
of the alignment, and this matrix is used to construct a

prescribed rotation VD1B1 with the same axis, but with a scaled
value of the angle.
Incremental: The matrix VD1B1 is computed by incremental

update of the matrix VD1B1 for the previous row of the table.
Specifically

sin sin
1 2 S 2 VD1B1
VD B = I +
1 1
S+
2
previous row

2
where S , P (n P
c11
, ncP21 , ncP31 ) are defined above. If this option is
used for the first row of the table, then the effect is the same as
if option=Angle.
Same: The prescribed rotation is the same as in the previous row

of the align-rotation table. (This option cannot be used in the
first row of the align-rotation table.)
The ALIGN-ROTATION command is used to enter this table.
When angles are specified, the unit is either radians or degrees,

depending on the value of ALIGN-ROTATION UNITANGLE
(the default is radians).

Chapter 2: Elements
Rotation misalignment
The rotation misalignment can be calculated as the product of the

actual rotation and the inverse of the prescribed rotation:
VB 2 D1 = VB 2 B1 (VD1B1 )T
The rotation misalignment matrix VB 2 D1 is then represented as a

rotation misalignment angle and axis: angle m and axis of rotation
( ndm11 , ndm21 , ndm31 ) , where the superscript m stands for "misalignment".
The axis of rotation is then transformed from the d1 triad system to
the c1 triad system to obtain ( ncm11 , ncm21 , ncm31 ) .
Moments
The rotation misalignment angle and axis can also be written in

component form:
(c1 ,c 2 , c3 ) = m ( ncm1 , ncm2 , ncm3 )

1 1 1
The moments are then calculated using the rotation misalignment

vector:
M c1 = kc1 c1 , M c 2 = kc 2 c 2 , M c 3 = kc 3 c 3 .
Notice that the rotational stiffness can be different in each

coordinate direction. Hence, for example, the user might specify
that the stiffness be zero for any rotational misalignment in the c1
coordinate direction, but nonzero for misalignments in the other
coordinate directions.
The unit of kci is always moment/radian.
It is also allowed to enter a sawtooth moment / angle relationship

for any of the coordinate directions. The sawtooth moment / angle
relationship is shown in Fig 2-12.16.

Moment
-360 -270 -180 -90 0 90 180 270 360

Rotational
misalignment
(degrees)
Figure 2.12-16: Sawtooth moment / angle relationship
Using this option, the moment is zero for misalignments of 0

degrees, 180 degrees, 360 degrees, etc.
The stiffness coefficients, and choices of constant or sawtooth

stiffnesses, are entered in the ALIGN-ROTATION command.
All moments act to reduce the rotation misalignments.
2.12.7 Triadsets and node-triadset pairs
Triadsets
The combination of the a triad, b triad and c triad is called a

triadset. The TRIADSET command is used to define the triads of
the triad set.
The triadset consists of three parts:
Specification of the a triad: The choices are:
ELEMENTS: The a triad corresponds to the local axes of the

element to which the node is attached. Currently, only

Chapter 2: Elements
Hermitian beam elements contribute local axes to the a triad.

Thus, for example, if the node is connected to a single
Hermitian beam element, the a1 direction is the r direction of
the beam element, the a2 direction is the s direction of the beam
element, etc.
If the node is not attached to any element with local axes, the
a triad directions are the global coordinate directions.
If the node is connected to two or more elements with local
axes, then the program averages the local axes to obtain the a
triad axes.
VECTORS: Components (a1x, a1y, a1z) and (a2x, a2y, a2z)

are input. Then the a triad directions are computed using
a1 = (a1x, a1y, a1z), normalized to unit length

a3 = a1 (a2x, a2y, a2z), normalized to unit length
a2 = a3 a1
Specification of the b triad: The choices are
ATRIAD: The b triad coincides with the a triad.
VECTORS: The b triad origin is offset from the a triad origin

using components (b0a1, b0a2, b0a3) . Using the notation of
2.12.3, (b0a1, b0a 2, b0a3) = ( xb1a1a1 , xb1a1a 2 , xb1a1a 3 ) .
The b triad directions are computed as follows: Components
(b1a1, b1a2, b1a3) and (b2a1, b2a2, b2a3) are input. Then the
b triad directions are computed using
b1 = (b1a1, b1a2, b1a3), normalized to unit length

b3 = b1 (b2a1, b2a2, b2a3), normalized to unit length
b2 = b3 b1
Specification of the c triad: The choices are:
BTRIAD: The c triad coincides with the b triad.
VECTORS: The c triad directions are computed as follows:

Components (c1b1, c1b2, c1b3) and (c2b1, c2b2, c2b3) are

input. Then the c triad directions are computed using
c1 = (c1b1, c1b2, c1b3), normalized to unit length

c3 = c1 (c2b1, c2b2, c2b3), normalized to unit length
c2 = c3 c1
Node-triadset pair
A triadset is assigned to a node using the SET-TRIADSET NODES

command. There can be more than one triadset assigned to a node.
Thus, for example, if a node is attached to more than one alignment
element, one alignment element can use one triadset and the other
alignment element can use a different triadset.
It is also allowed for no triadset to be assigned to a node. Then, if

the node is attached to an alignment element, the program
calculates a default triadset. The calculation of the default is
discussed below.
2.12.8 Align-translation specification
The ALIGN-TRANSLATION command is used to define the

characteristics of the translation alignment of an alignment element.
These characteristics include the translational stiffnesses and the
prescribed translational alignments.
The prescribed translational alignments are input in the form of a

table. Each row of the table gives the prescribed translational
alignment at a time. This time need not be a solution time.
For example, consider the command

ALIGN-TRANSLATION 1
ENTRIES TIME OPTION XCP YCP ZCP TBFACTOR
CLEAR
3.0 TBFACTOR 0.0 0.0 0.0 1.0
4.0 COMPONENTS 1.0 2.0 3.0 0.0
5.0 ALIGNED
DATAEND

Chapter 2: Elements
The lowest time in the table is time 3.0. For solution times before
time 3.0, the translational alignment is inactive (not born yet).
For the first solution time at or after time 3.0, the translational
alignment becomes active. The total alignment for the preceding
solution time is measured and is used to construct xb 2b1 . Since
birth
OPTION=TBFACTOR and TBFACTOR=1.0, the prescribed
translational alignment for time 3.0 is set to xd1b1 = xb 2b1 .
birth
For solution time 4.0, the prescribed translational alignment is

( xd 1bc11 , xd 1bc12 , xd 1bc13 ) = (1.0, 2.0,3.0) .
For solution time 5.0, the prescribed translational alignment is

( xd 1bc11 , xd 1bc12 , xd 1bc13 ) = (0.0,0.0,0.0) .
For solution times after time 5.0, the prescribed translational

alignment is the same as the prescribed translational alignment at
time 5.0.
For solution times between times given in the align-translation

table, linear interpolation is used to determine the prescribed
translational alignment.
2.12.9 Align-distance specification
The ALIGN-DISTANCE command is used to define the

characteristics of the distance alignment of an alignment element.
These characteristics include the distance stiffness and the
prescribed distance alignments.
The prescribed distance alignments are input in the form of a table,

analogous to the table used in the align-translation definition. The
options are given in Section 2.12.5.
2.12.10 Align-rotation specification
The ALIGN-ROTATION command is used to define the

characteristics of the rotation alignment of an alignment element.
These characteristics include the rotational stiffnesses and the

prescribed rotational alignments.
The prescribed rotational alignments are input in the form of a

table, analogous to the table used in the align-translation definition.
The options are given in Section 2.12.6.
2.12.11 Element group specification
The EGROUP ALIGNMENT command is used to define the

characteristics of an alignment element group. The default values
of the align-translation, align-distance and align-rotation definitions
can be entered in the EGROUP ALIGNMENT command.
2.12.12 Element specification
The nodes of the alignment element are specified using the

ENODES command, just as for other element types. Notice that
local node 1 is different than local node 2, because local node 1 is
used to measure prescribed alignments. Thus two elements
differing only in the order of the local nodes will in general give
different results.
Data for the alignment element is specified using the EDATA
command, just as for other element types. Note that it is allowed to
enter the align-translation, align-distance and align-rotation
definitions separately for each element. Therefore one element
might align translations only and another element in the same
group might align rotations only.
The data for the alignment element also includes the triadset to
be used for each local node. This data is entered using parameters
TRIADST1 and TRIADST2.
The assignment of a triadset to a local node is now discussed in
detail for local node 1 (local node 2 is similar):
Case 1: The global node is not in a node-triadset pair:
The a triad is determined as follows:
Case 1a: Global node is attached to at least one element

with local axes. Then the program averages the triads of
all attached elements with local axes to obtain the a triad
at the node.

Chapter 2: Elements
Case 1b: Global node is not attached to any element with

local axes. Then the program sets the a triad to the global
The b triad is coincident with the a triad (with no offset).
The c triad is coincident with the b triad.
The TRIADST1 parameter must be set to zero.
Case 2: The global node is in a single node-triadset pair
The TRIADST1 parameter can either be set to zero, or to the

triadset for the global node. The triadset is obtained from the
node-triadset pair.
Case 3: The global node is in more than one node-triadset pair
The TRIADST1 parameter must be nonzero. This parameter

is used to choose the triadset for the node. The triadset
specified by parameter TRIADST1 must be in a node-
triadset pair with the global node corresponding to local
node 1.
2.12.13 Input examples
Example 1: In this example, nodes 1 and 2 are aligned using a

distance alignment only. The nodes are to be aligned at time 1.0,
with the two nodes coinciding.
TIMESTEP
CLEAR
1 1.0
DATAEND
*
ALIGN-DISTANCE 1
ENTRIES TIME OPTION
CLEAR
1.0 ALIGNED
DATAEND
*

EGROUP ALIGNMENT 1 ALIGN-DISTANCE=1

EDATA
ENTRIES N1 N2
CLEAR
1 2
DATAEND
Example 2: This example is similar to example 1. However, the

alignment starts at time 1.0 with a prescribed alignment distance
equal to the initial alignment distance. In the absence of other
external loads, the solution should immediately converge at time
1.0. The alignment ends at time 2.0, with the two nodes
overlapping.
TIMESTEP
CLEAR
2 1.0
DATAEND
*
ALIGN-DISTANCE 1
ENTRIES TIME OPTION
CLEAR
1.0 TBFACTOR
2.0 ALIGNED
DATAEND
*
EDATA
ENTRIES N1 N2
CLEAR
1 2
DATAEND
Example 3: This example is similar to example 2. However the

alignment takes place over two solution steps, with the prescribed
alignment distance at time 2 equal to the initial alignment distance,
multiplied by 0.1.
TIMESTEP
CLEAR
3 1.0
DATAEND
*
ALIGN-DISTANCE 1
ENTRIES TIME OPTION FACTOR
CLEAR

Chapter 2: Elements
1.0 TBFACTOR
2.0 TBFACTOR 0.1
3.0 ALIGNED
DATAEND
*
EDATA
ENTRIES N1 N2
CLEAR
1 2
DATAEND

alignment begins at solution time 2.0. The prescribed alignment
distance at time 2.0 is equal to the alignment distance at time 1.0.
Note that time 1.0 is the solution time that immediately precedes
time 2.0. The prescribed alignment distance at time 3.0 is equal to
the alignment distance at time 1.0, multiplied by 0.1.
TIMESTEP
CLEAR
4 1.0
DATAEND
*
ALIGN-DISTANCE 1
CLEAR
2.0 TBFACTOR
3.0 TBFACTOR 0.1
4.0 ALIGNED
DATAEND
*
EDATA
ENTRIES N1 N2
CLEAR
1 2
DATAEND

times specified in ALIGN-DISTANCE do not correspond to
solution times. The alignment begins at solution time 3.0. The
prescribed alignment distance at time 3.0 is obtained as follows:
Dummy alignment distance at time 2.1 = alignment distance at

time 2.0 (2.0 is the solution time immediately preceding time

2.1)

time 2.0, multiplied by 0.1.
Prescribed alignment distance at time 3.0 = Linear interpolation

of dummy alignment distances at times 2.1, 3.1. In this case, the
result is equal to the alignment distance at time 2.0, multiplied
by 0.19, since
0.19 = 1.0 + (0.1 - 1.0)*(3.0 - 2.1)/(3.1 - 2.1)
The prescribed alignment distance at time 4.0 is obtained as

follows:

time 2.0, multiplied by 0.1
Dummy alignment distance at time 4.1 = zero
Prescribed alignment distance at time 4.0 = Linear interpolation

of dummy alignment distance at times 3.1, 4.1.
The prescribed alignment distance at time 5.0 is equal to zero.

TIMESTEP
CLEAR
5 1.0
DATAEND
*
ALIGN-DISTANCE 1
CLEAR
2.1 TBFACTOR
3.1 TBFACTOR 0.1
4.1 ALIGNED
DATAEND
*
EDATA
ENTRIES N1 N2
CLEAR
1 2
DATAEND

Chapter 2: Elements

rotational alignment only. The alignment starts at time 3 and the
rotations are aligned at time 4. Here we suppose that the two nodes
are each connected to Hermitian beam elements. Also, it is not
necessary to set the triads in the EDATA command since there is
only one triad per node.
The nodes are aligned when the b triad axes of the two nodes are
coincident. Since no triadset is defined, the b triad axes correspond
to the r-s-t axes of the attached beam elements.
ALIGN-ROTATION 1
ENTRIES TIME OPTION
CLEAR
3.0 TBFACTOR
4.0 ALIGNED
DATAEND
*
EGROUP ALIGNMENT 1 ALIGN-ROTATION=1
EDATA
ENTRIES N1 N2
CLEAR
1 2
DATAEND

rotational alignment only. The alignment starts at time 3 and the
rotations are aligned at time 4.
Here we explicitly define the a and b triads at the nodes. Triadset 1

gives the triads for node 1 and triadset 2 gives the triads for node 2.
Triadset 1: The a triad initially has directions a1=(0,-1,0), a2=(-

1,0,0) (components in the global system). a3 is computed to create
a right-handed orthonormal system.
The b triad has offset (1, 2, 3). This offset is measured in the a1, a2,
a3 directions. The b triad directions are b1=(0,1,0), b2=(0,0,1). b3
is computed to create a right-handed orthonormal system. Again,
the components of the b1, b2, b3 axes are expressed in the a triad
coordinate system.
Thus for demonstration, the initial b triad offset and directions in

the global system can be calculated as follows:

a3=(0,0,-1) in global system

b3=(1,0,0) in a triad system
offset=1*(0,-1,0) + 2*(-1,0,0) + 3*(0,0,1) = (-2,-1,3) in global

system
b1=0*(0,-1,0) + 1*(-1,0,0) + 0*(0,0,-1) = (-1,0,0) in global system

b2=0*(0,-1,0) + 0*(-1,0,0) + 1*(0,0,-1) = (0,0,-1) in global system
b3=1*(0,-1,0) + 0*(-1,0,0) + 0*(0,0,-1) = (0,-1,0) in global system
Triadset 2: The a triad initially has directions a1=(1,0,0),

a2=(0,0,1). a3 is computed to create a right-handed orthonormal
system.
The b triad has offset (-1, -2, -3). The b triad directions are b1=
(-1,0,0), b2=(0,1,0) (again, the components of b1, b2 are expressed
in the a triad coordinate system). b3 is computed to create a right-
handed orthonormal system.
The nodes are aligned when the b triad axes of the two nodes are
coincident. The triad offsets are not used in this example.
TRIADSETS
ENTRIES,
TRIADSET,
AOPTION A1X A1Y A1Z A2X A2Y A2Z,
BOPTION B0A1 B0A2 B0A3,
B1A1 B1A2 B1A3 B2A1 B2A2 B2A3
CLEAR
1,
VECTORS 0.0 -1.0 0.0 -1.0 0.0 0.0,
VECTORS 1.0 2.0 3.0,
0.0 1.0 0.0 0.0 0.0 1.0
2,
VECTORS 1.0 0.0 0.0 0.0 0.0 1.0,
VECTORS -1.0 -2.0 -3.0,
-1.0 0.0 0.0 0.0 1.0 0.0
DATAEND
SET TRIADSET NODES
CLEAR
1 1
2 2
DATAEND

Chapter 2: Elements
ALIGN-ROTATION 1
ENTRIES TIME OPTION
CLEAR
3.0 TBFACTOR
4.0 ALIGNED
DATAEND
*
EGROUP ALIGNMENT 1 ALIGN-ROTATION=1
EDATA
ENTRIES N1 N2
CLEAR
1 2
DATAEND
Example 8: In this example, nodes 1 and 2 are aligned using both a

translational and a rotational alignment. The same triadsets are used
as in the previous example. However different triadset numbers are
assigned to the nodes, just to show that the triadset number need
not be the node number.
The translational alignment starts at time 1 and the translations are

aligned at time 2. The nodes are translationally aligned when the
offsets of the b triad axes of the two nodes are coincident.
The rotational alignment starts at time 3 and the rotations are

aligned at time 4. The nodes are rotationally aligned when the b
triad axes of the two nodes are coincident.
TRIADSETS
ENTRIES,
TRIADSET,
AOPTION A1X A1Y A1Z A2X A2Y A2Z,
B1A1 B1A2 B1A3 B2A1 B2A2 B2A3
CLEAR
10,
VECTORS 0.0 -1.0 0.0 -1.0 0.0 0.0,
VECTORS 1.0 2.0 3.0,
0.0 1.0 0.0 0.0 0.0 1.0
20,
VECTORS 1.0 0.0 0.0 0.0 0.0 1.0,
VECTORS -1.0 -2.0 -3.0,
-1.0 0.0 0.0 0.0 1.0 0.0
DATAEND
SET TRIADSET NODES
CLEAR
1 10

2 20
DATAEND
ALIGN-TRANSLATION 1
ENTRIES TIME OPTION
CLEAR
1.0 TBFACTOR
2.0 ALIGNED
DATAEND
*
ALIGN-ROTATION 1
ENTRIES TIME OPTION
CLEAR
3.0 TBFACTOR
4.0 ALIGNED
DATAEND
*
EGROUP ALIGNMENT 1 ALIGN-TRANSLATION=1,
ALIGN-ROTATION=1
EDATA
ENTRIES N1 N2
CLEAR
1 2
DATAEND
Example 9: In this example, nodes 1 and 2 are assumed to be at the

end of a straight line of beam nodes. It is desired to bend the
straight line so that the line becomes a circle.
The rotational alignment starts at time 1. At time 2, the prescribed

rotational alignment is approximately 120 degrees with alignment
axis (0,0,1). The components of the axis are in the c triad system
of node 1. (For this problem, since no triadset is used, the a, b and c
triads of node 1 all coincide, therefore the components of the axis
are in the a triad system of node 1, in other words, the direction of
the t axis at node 1.)
At time 3, the prescribed rotational alignment is approximately 240

degrees with alignment axis (0,0,1). At time 4, the prescribed
alignment is set to 0. Since the rotational interpolation always takes
the shortest angular distance, the prescribed alignment between
times 3 and 4 increases from 240 degrees to 360 degrees.
The nodes are rotationally aligned when the b triads at node 1 and 2
have the same directions.

Chapter 2: Elements
The change in prescribed rotational alignment is always less than

about 120 degrees. This change must be less than 180 degrees per
time step.
At time 5, the translational alignment starts and at time 6, the

translational alignment ends. The nodes are translationally aligned
when the nodes coincide.
ALIGN-TRANSLATION 1
ENTRIES TIME OPTION
CLEAR
5.0 TBFACTOR
6.0 ALIGNED
DATAEND
*
ALIGN-ROTATION 1
ENTRIES TIME OPTION ANGLE AXISC1 AXISC2 AXISC3
CLEAR
1.0 TBFACTOR
2.0 ANGLE 2.1 0.0 0.0 1.0
3.0 ANGLE 4.2 0.0 0.0 1.0
4.0 ALIGNED
DATAEND
*
ALIGN-ROTATION=1
EDATA
ENTRIES N1 N2
CLEAR
1 2
DATAEND
Example 10: In this example, we want to simulate body-body

contact.
Fig 2.12-17 shows the bodies in the aligned configuration.

Radius = 3
s
r
Node 8 t
s Radius = 2
r
t
Node 5
(a): At alignment
a2=s
a1=r
Triadset 8 a3=t
b2
b1
b3
b2
b1
b3
a2=s
a1
a1=r b1
Triadset 5
a3=t
a3
30 o
b3
(b): Definition of triadsets (c): Top view of triadset 8
Figure 2.12-17: Simulation of body-body contact

Chapter 2: Elements
Triadset 5 is attached to node 5 and triadset 8 is attached to node 8.
The triadsets are defined so that the b triads are aligned in the
aligned configuration:
a triads: The a triads are the same as the triads from the attached
elements.
b triads, triadset 5: The b triads are offset from the node in the a2
direction:
(b0a1, b0a2, b0a3) = (0,2,0)
The b triad directions are the same as the a triad directions (the
default)
(b1a1, b1a2, b1a3) = (1,0,0)

(b2a1, b2a2, b2a3) = (0,1,0)
(b3a1, b3a2, b3a3) = (0,0,1)
b triads, triadset 8: The b triads are offset from the node in the
negative a2 direction:
(b0a1, b0a2, b0a3) = (0,-3,0)
The b triad directions are rotated relative to the a triad directions:
(b1a1, b1a2, b1a3) = (cos 30o,0, sin 30o)

(b2a1, b2a2, b2a3) = (0,1,0)
(b3a1, b3a2, b3a3) = (-sin 30o,0, cos 30o)
c triads: The c triads are the same as the b triads.
The alignment element is defined using node 5 as local node 1 and

node 8 as local node 2.
At alignment, we would like to have the b triad origins coincide.

This means that we need high translational stiffnesses in all three

We would also like to have the axis of rotation correspond to the c2

= b2 direction of node 5. Therefore we need high rotational
stiffnesses in the c1 and c3 coordinate directions, but a low
rotational stiffness in the c2 coordinate direction. In addition, we
would like the stiffness in the c2 coordinate direction to be
periodic. Thus, if node 5 is rotated 360 degrees relative to node 8,
the alignment element forces are unchanged.
We will reach alignment by simultaneously applying the

translational and rotational alignments in five time steps.
TRIADSETS
ENTRIES TRIADSET,
AOPTION,
B1A1 B1A2 B1A3 B2A1 B2A2 B2A3,
CLEAR
5,
ELEMENTS,
VECTORS 0.0 2.0 0.0,
1.0 0.0 0.0 0.0 1.0 0.0
8,
ELEMENTS,
VECTORS 0.0 -3.0 0.0,
0.86603 0.0 0.5 0.0 1.0 0.0
DATAEND
SET TRIADSET NODES
CLEAR
5 5
8 8
DATAEND
ALIGN-TRANSLATION 1 KTC1=1E15 KTC2=1E15 KTC3=1E15

ENTRIES TIME OPTION
CLEAR
1.0 TBFACTOR
6.0 ALIGNED
DATAEND
*
ALIGN-ROTATION 1 KRC1=1E15 KRC2=2 KRC3=1E15,
KRC2TYPE=SAWTOOTH
ENTRIES TIME OPTION
CLEAR
1.0 TBFACTOR
6.0 ALIGNED
DATAEND
*
ALIGN-ROTATION=1
ENODES

Chapter 2: Elements
ENTRIES EL N1 N2
CLEAR
1 5 8
DATAEND
The element can either output nodal point forces or alignment data
(the choice is made in the EGROUP ALIGNMENT command).
When the element outputs alignment data, two levels of output can
be selected:
Usual output: The actual (total) alignments are output. In the

notation of Section 2.12.4 to 2.12.6:
Translation: xb 2b1 in component form:

( xb 2 b1c1 , xb 2 b1c 2 , xb 2b1c 3 ) . The postprocessing variables are
ALIGN_TRANS_ACTUAL-C{123}.
Distance: D = length of vector xb 2b1 . The postprocessing

variable is ALIGN_DIST_ACTUAL.
Rotation: VB 2 B1 , transformed into an equivalent angle-axis

representation, with the axis given in the c triad system:
( nc11 , nc 21 , nc 31 ) . The angle is output in degrees or radians,
depending upon the setting of ALIGN-ROTATION
UNITANGLE. The postprocessing variables are
ALIGN_ROT_ACTUAL_MAGNITUDE,
ALIGN_ROT_ACTUAL_AXIS-C{123}
Verbose output: In addition to the usual output, verbose output can

be selected.
For printing of results, verbose output is selected either using
PRINTOUT PRINTDEFAULT=STRAINS
EGROUP ALIGNMENT PRINT=DEFAULT
EDATA ... printi=DEFAULT

or
EGROUP ALIGNMENT PRINT=VERBOSE

EDATA ... printi=DEFAULT
or
EDATA ... printi=VERBOSE
For saving of results, verbose output is selected using EGROUP

ALIGNMENT SAVE=VERBOSE.
Verbose output includes the usual output described above, and also
the following addtional output:
Translation:
xd1b1 in component form: ( xd 1b1c1 , xd 1b1c 2 , xd 1b1c 3 ) . The
postprocessing variables are ALIGN_TRANS_PRESC-
C{123}.
xb 2d1 in component form: ( xb 2 d 1c1 , xb 2 d 1c 2 , xb 2 d 1c 3 ) The

postprocessing variables are ALIGN_TRANS_MISALIGN-
C{123}.
The alignment forces ( Fc1 , Fc 2 , Fc 3 ) . The postprocessing

variables are ALIGN_TRANS_FORCE-C{123}.
Distance:
D P . The postprocessing variable is ALIGN_DIST_PRESC.
D D P . The postprocessing variable is

ALIGN_DIST_MISALIGN.
Fd .The postprocessing variable is ALIGN_DIST_FORCE.
Rotation:
VD1B1 , transformed into an equivalent angle-axis


Chapter 2: Elements
P ( ncP1 , ncP2 , ncP3 ) . The angle p is output in degrees or

1 1 1
radians, see above. The postprocessing variables are

ALIGN_ROT_PRESC_MAGNITUDE,
ALIGN_ROT_PRESC_AXIS-C{123}
VB 2 D1 , transformed into an equivalent angle-axis

m ( ncm11 , ncm21 , ncm31 ) . The angle m is output in degrees or
radians, see above. The postprocessing variables are
ALIGN_ROT_MISALIGN_MAGNITUDE,
ALIGN_ROT_MISALIGN_AXIS-C{123}
The alignment moments ( M c1 , M c 2 , M c 3 ) , output as a

resultant moment about an axis M ( ncM11 , ncM21 , ncM31 ) . The
postprocessing variables are
ALIGN_ROT_MOMENT_MAGNITUDE,
ALIGN_ROT_MOMENT_AXIS-C{123}
2.12.15 Recommendations
1) When using the rotation alignment, it is best to define the b

triads so that they have the same orientation at alignment.
2) When using the distance alignment, the directions of the b triads,

and also the c triads, are not used.
3) When using the translation alignment, the directions of the b

triads are not used.
4) If it is necessary to keep the translational misalignment small for

all coordinate directions, then the stiffness in each translational
direction should be set to a large value. In this case the directions
of the c triads are not important and therefore the c triads can be set
to the b triads (the default).
5) If it is necessary to keep the rotational misalignment small for all

coordinate directions, then the stiffness in each rotational direction
should be set to a large value. In this case the directions of the c

triads are not important and therefore the c triads can be set to the b
triads (the default).
6) Hence the directions of the c triads only become important when

the stiffnesses are different in the different coordinate directions, or
when the translational components are explicitly specified, or when
the rotation angle-axis is explicitly prescribed.
7) When defining the triadsets, it is easiest to consider the nodes in

the aligned configuration, then define the triadsets in the aligned
configuration.
8) The nonsymmetric solver can be useful when the direction of the

prescribed rotational alignment changes during the analysis.
2.13 Cohesive elements

The cohesive element is a 4-node (for 2-D) or 8-node (for 3-D)

element that can be employed to model the separation at zero-
thickness interfaces, such as laminated composites, bonded
interfaces, etc.
2.13.2 Element formulation
The cohesive element is composed of top and bottom surfaces

with initially zero thickness. As shown in Figure 2.13-1, the 2-D
cohesive element has 4 nodes that connect 2-D 4-node solid
elements (axisymmetric, plane stress or plane strain elements). The
3-D cohesive element has 8 nodes that only connect 3-D 8-node
solid elements.
The mid-surface of the cohesive elements is tracked by

averaging the coordinates of nodal pairs from the top and bottom
surfaces so that the effect of geometrical nonlinearity can be taken
into account.

Chapter 2: Elements
2-D Solid Elements

2-D Solid Elements
3 4
Thickness = 0 2-D Cohesive Element 1 2
1 2
2-D Cohesive Element
2-D Cohesive Element in ADINA 2-D Cohesive Element in AUI-PLOT
(a) 4-node 2-D cohesive element
3-D Solid Elements
3 2 3-D Solid Elements
3 2
4 1
Thickness = 0
4 1
7 6
8 5 3-D Cohesive Element

3-D Cohesive Element
3-D Cohesive Element in ADINA 3-D Cohesive Element in AUI-PLOT
(b) 8-node 3-D cohesive element
Figure 2.13-1: 4-node and 8-node cohesive elements
2.13.3 Constitutive law of cohesive element
The following is a quick summary of the basic concepts used in

the constitutive law of the cohesive element. For further
information, please see the following references:
ref. Turon, A., Camanho, P.P., Costa, J., and Davila, C.G.,
"A Damage Model for the Simulation of Delamination in
Advanced Composites under Variable-Mode Loading,"

2.13: Cohesive elements
Mechanics of Materials, Vol. 38, No. 11, pp. 1072-1089,

2006.
ref. Camanho, P., Dvila, C., "Mixed-mode decohesion finite

elements for the simulation of delamination in composite
materials," NASA/TM-2002-211737, pp. 1-37, 2002.
ref. Alfano, G., Crisfield, M.A., Finite element interface

models for the delamination analysis of laminated
composites: mechanical and computational issues,
International Journal for Numerical Methods in
Engineering, 50 (7), pp. 17011736, 2001.
The cohesive element uses a bilinear constitutive law that

relates the traction, , to the relative displacement, , at the
element mid-surface as shown in Figure 2.13-2. Initially linear
elastic behavior followed by the initiation and evolution of damage
is assumed. After the onset of damage, the stiffnesses of the
cohesive element are gradually reduced to zero in its softening
envelope. Unloading and reloading in the softening envelope are
indicated by the arrows in Figure 2.13-2.
K reloading reloading
unloading

Figure 2.13-2 Bilinear constitutive law of cohesive element
For pure mode I, II or III loading, after the interfacial normal or

shear tractions reach their respective tensile or shear strengths, the
stiffnesses are gradually reduced to zero. The area under the curve

Chapter 2: Elements
is the respective (mode I, II or III) fracture toughness. The onset of

damage at the interface can be determined by comparing the
tractions with their respective allowable.
Under mixed-mode loading, the onset of damage may occur

before any of the traction components involved reach their
respective allowable. In this case, the following criteria are
provided in ADINA to predict delamination growth under mixed-
mode loadings.
Power law criterion

BenzeggaghKenane (B-K) criterion
The power law criterion is established in terms of a linear or

quadratic interaction between the energy release rates using

GI GII
+ =1
GIC GIIC
where GI and GII are the mode I and mode II energy release rate;
GIC and GIIC are the mode I and mode II fracture toughness; is
mixed-mode interaction parameter which is usually between 1 and
2.
With BenzeggaghKenane (B-K) criterion, the critical energy
release rate GC is expressed as a function of the mode I and mode
II fracture toughness and a mixed-mode interaction parameter ,

G
GC = GIC + ( GIIC GIC ) S
GT
where GT is total energy release rate under mixed-mode loading;

GS is the energy release rate for shear loading.
The properties required to define the bilinear constitutive law

are the mode I and mode II fracture toughness ( GIC , GIIC ), the
penalty stiffness ( K ), the normal and shear interfacial strengths,

2.13: Cohesive elements
and the mixed-mode interaction criterion and its parameter. Please

note that the same penalty stiffness is assumed for all three modes
and the shear interfacial strengths in the two orthogonal directions
are assumed to be the same.
For the calculation of all element matrices and vectors, numerical

Newton-Cotes integration is used. As shown in Figure 2.13-3, 21
Newton-Cotes integration points are used for 2-D 4-node cohesive
element, and 22 Newton-Cotes integration points for 3-D 8-node
cohesive element.
2 4
r r
1 2
2-D cohesive element
1 3
3-D cohesive element
Figure 2.13-3: Newton-Cotes integration points layout
You can request that ADINA either print or save stresses/strains.
Stresses/strains: Each element outputs, at its nodal points, the

following information to the porthole file. This information is
accessible in the AUI using the given variable names.
COHESIVE_NORMAL_STRESS,
COHESIVE_TANGENTIAL_STRESS,
COHESIVE_NORMAL_STRAIN,
COHESIVE_TANGENTIAL_STRAIN, COHESIVE_ DAMAGE

Chapter 2: Elements
This page intentionally left blank

3.1: Stress and strain measures in ADINA
3. Material models and formulations

The objective of this chapter is to summarize the theoretical basis
and practical use of the material models and formulations available
in ADINA.
3.1 Stress and strain measures in ADINA

3.1.1 Summary
It is important to recognize which stress and strain measures are

employed in the use of a material model:
In the preparation of the input data in which the material

parameters are defined with respect to these stress and strain
measures
In the interpretation of the analysis results in which the type

of stresses and strains output must be considered
The practical use of the material models available in ADINA

regarding the stress and strain measures used for the input data and
analysis results is summarized in the following. These stress/strain
measures are described in detail in ref. KJB, Section 6.2.
In the following discussion, small strains are strains less than

about 2%.
Small displacement/small strain formulation
Input of material parameters: All elements and material models use

the engineering stress-engineering strain relationship.
Output: All elements and material models output engineering

stresses and engineering strains.
Note that under small displacement, small strain conditions,

engineering stresses and Cauchy stresses are nearly identical.

Chapter 3: Material models and formulations
Large displacement/small strain formulation
Input of material parameters: 2nd Piola-Kirchhoff stresses and

Green-Lagrange strains. Note that under small strain conditions,
2nd Piola-Kirchhoff stresses are nearly equal to engineering
stresses, and Green-Lagrange strains are nearly equal to
engineering strains.
Output:
(1) 2-D, 3-D elements: all material models output Cauchy

stresses and Green-Lagrange strains.
(2) Beam, iso-beam, pipe elements: all material models output

2nd Piola-Kirchhoff stresses and Green-Lagrange strains.
(3) Shell elements, results output in local coordinate system: all

material models output 2nd Piola-Kirchhoff stresses and Green-
Lagrange strains.
(4) Shell elements, results output in global coordinate system:

all material models output Cauchy stresses and rotated small
strains.
Large displacement/large strain formulation, 2D and 3D

elements
For the two- and three-dimensional solid elements, the following

material models can be used:
(1) Plastic-bilinear, plastic-multilinear, thermo-plastic. In these

cases either the ULH formulation or the ULJ formulation is
used. The default is the ULH formulation, except when explicit
time integration is employed.
Input of material parameters: Cauchy (true) stresses and

logarithmic (true) strains. For the multilinear stress-strain
curves, it is also possible to enter engineering stress-strain
data along with the input MASTER CONVERT-
SSVAL=YES, see detailed description of the CONVERT-
SSVAL feature in Section 3.4.1.

Output: ULH formulation: Cauchy stresses and deformation

gradients.
ULJ formulation: Cauchy stresses and Jaumann strains.
(2) Drucker-Prager, Mroz bilinear, creep, plastic-creep,

multilinear-plastic-creep, plastic-creep-variable, multilinear-
plastic-creep-variable, viscoelastic, user-supplied. In these
cases the ULH formulation is always used.

logarithmic (true) strains
Output: Cauchy stresses and deformation gradients.
(3) Plastic-orthotropic. In this case the updated ULJ

formulation is always used.

logarithmic (true) strains
Output: Cauchy stresses and Jaumann strains.
(4) Mooney-Rivlin, Ogden, Arruda-Boyce, hyper-foam, and

Sussman-Bathe. In this case the TL (total Lagrangian)
formulation is used.
Input of material parameters: Mooney-Rivlin, Ogden,

Arruda-Boyce or hyper-foam constants. For Sussman-Bathe,
the uniaxial stress-strain curve (see Section 3.8.1.5 for
details).
Output: Cauchy stresses and deformation gradients.
Large displacement/large strain formulation, shell elements
For the shell elements, the following material models can be used,
provided that the elements are single layer elements described
using 3, 4, 6, 9 or 16 midsurface nodes:

(1) Plastic-bilinear, plastic-multilinear. Either the ULJ

formulation or the ULH formulation can be used. The default is
the ULH formulation, except when explicit time integration is
employed.
(2) Plastic-orthotropic. The ULJ formulation is always

employed.
Input of material parameters: Kirchhoff stresses and

logarithmic (true) strains. For the multilinear stress-strain
curves, it is also possible to enter engineering stress-strain
data along with the input MASTER CONVERT-
SSVAL=YES, see detailed description of the CONVERT-
SSVAL feature in Section 3.4.1.
Output: Kirchhoff stresses and left Hencky strains (ULH) or

Jaumann strains (ULJ).
Strain measures: The strain measures used in ADINA are

illustrated hereafter in the simplified case of a rod under uniaxial
tension (see Fig. 3.1-1).
0
Engineering strain: e0 =
0
1 2
2
Green-Lagrange strain: = 2
0
2 0
Logarithmic strain, Hencky strain, Jaumann strain:

d
e = ln =
0 0
Stretch: =
0
ref. KJB Green-Lagrange strains are used in the large displacement/small

Sec. 6.2.2 strain formulations. This is because large rotations do not affect
the Green-Lagrange strains (Green-Lagrange strains are invariant
with respect to rigid-body rotations), and for small strains, small
rotations, Green-Lagrange strains and engineering strains are
equivalent.

In the small strain formulations, the current area is always

assumed to be equal to the initial undeformed area.
F
A

A0
A0 = initial section area
A = final section area
F = applied force
0 = 0 +
F
Figure 3.1-1: Rod under uniaxial tension
Engineering strains are also called nominal strains in the

literature.
Logarithmic strains are also known as true strains.
Stress measures: The stress measures used in ADINA are

illustrated hereafter in the simplified case of a rod under uniaxial
tension (see Fig. 3.1-1).
F
Engineering stress: =
A0
F A0
Cauchy stress: = =
A A
F 0
2nd Piola-Kirchhoff stress: S = = 0
A0
A F
Kirchhoff stress: J = = =
A0 0 A0 0 0

Cauchy stresses are also known as true stresses.
For the case in which the material is incompressible,

= J = can be used to compute the Cauchy stress and the
0
Kirchhoff stress from the engineering stress.
When the strains are small, the 2nd Piola-Kirchhoff stresses are
nearly equal to the Cauchy stresses from which the rigid body
rotations have been removed.
2nd Piola-Kirchhoff stresses are input only for the large

displacement/small strain formulation. Because the strains are
assumed to be small, the engineering stresses can be entered as the
stress input quantities.
2nd Piola-Kirchhoff stresses are output only for the large

displacement/small strain formulation and for element types in
which the stresses expressed in the element coordinate system of
the deformed element are of physical significance. If the element
does not undergo rigid body rotations, the 2nd Piola-Kirchhoff
stresses are nearly equal to the Cauchy stresses, because the strains
are assumed to be small. If the element undergoes rigid body
rotations, the Cauchy stresses expressed in the element coordinate
system do not change (because the element coordinate system
rotates). Since the 2nd Piola-Kirchhoff stresses, as expressed in the
undeformed element coordinate system, do not change during a
rigid body rotation, these 2nd Piola-Kirchhoff stresses remain nearly
equal to the Cauchy stresses. The 2nd Piola-Kirchhoff stresses here
provide a convenient way of removing the rigid body rotations
from the output stresses.
When the large displacement/small strain formulation is used,

and when stresses and strains are output in an element local
coordinate system, 2nd Piola-Kirchhoff stresses and Green-
Lagrange strains are output. The justification for using 2nd Piola-
Kirchhoff stresses is given in the preceding paragraph, and the
justification for using Green-Lagrange strains is similar.

When the shell element is used with the large

displacement/small strain formulation, and when stresses and
strains are output in the global coordinate system, the stresses and
strains are calculated in two steps. First the stresses and strains are
calculated in the element local coordinate system, according to the
preceding paragraph. Then the stresses and strains are rotated from
the element local system into the global system. The resulting
stresses can be interpreted as Cauchy stresses, because the strains
are assumed small. The resulting strains can be interpreted as
rotated small strains.
When the material is nearly incompressible, the Kirchhoff

stresses are nearly equal to the Cauchy stresses.
Since Kirchhoff stresses are input/output only for materials that

are nearly incompressible, practically speaking, the differences
between Kirchhoff and Cauchy stresses are negligible.
3.1.2 Large strain thermo-plasticity analysis with the ULH formulation
This section discusses the ULH formulation for large strain

analysis. ULH stands for updated Lagrangian Hencky.
The following is a quick summary of the theory of large strain

inelastic analysis with the ULH formulation. For further
information, see ref KJB, Section 6.6.4 and also the following
references:
ref. F.J. Montns and K.J. Bathe, "Computational issues in

large strain elasto-plasticity: an algorithm for mixed
hardening and plastic spin", Int. J. Numer. Meth. Engng,
2005; 63;159-196.
ref. M. Koji and K.J. Bathe, Inelastic Analysis of Solids and

Structures, Springer-Verlag, 2003.
Total deformation gradient tensor: Let X be the total

deformation gradient tensor at time t with respect to an initial
configuration taken at time 0. For ease of writing, we do not
include the usual left superscripts and subscripts.

Polar decomposition into rotation and right stretch tensor: The

total deformation gradient tensor X can be decomposed into a
material rigid-body rotation tensor R and a symmetric positive-
definite (right) stretch tensor U (polar decomposition):
X = RU (3.1-1)
Principal directions of right stretch tensor: The right stretch

tensor U can be represented in its principal directions by a diagonal
tensor , such that
U = R L R TL (3.1-2)
where R L is a rotation tensor with respect to the fixed global axes

(see Figure 3.1-2).
y R
(t) RL
(0)
x
(0) Initial configuration at time 0

Configuration at time t not including rigid body rotation
(t) Configuration at time t
Directions of maximum/minimum total stretches
and strains
R Material rigid-body rotation between time 0 and time t
RL Directions of initial configuration fibers with

maximum/minimum total stretches and strains
Figure 3.1-2: Directions of maximum/minimum

total stretches and strains

(Note that the rotation R L does not correspond to a material

rigid-body rotation, but to a rotation of the coordinate system: U
and are two representations of the same deformed state,
respectively in the global coordinate system and in the U principal
directions coordinate system.)
Right Hencky strain tensor: The Hencky strain tensor (computed

in the right basis) is given by
E R = ln U = R L ln R TL (3.1-3)
The superscript R symbolizes the right basis.
Polar decomposition into rotation and left stretch tensor: The

total deformation gradient tensor X can also be decomposed into a
material rigid-body rotation R and a symmetric positive-definite
(left) stretch tensor V (polar decomposition):
X= VR (3.1-4)
R in (3.1-4) is the same as R in (3.1-1).
Principal directions of left stretch tensor: The left stretch tensor

V can be represented in its principal directions by a diagonal tensor
, such that
V = R E RTE (3.1-5)
where R E is a rotation tensor with respect to the fixed global axes.

Note that R E = R R L .
Left Hencky strain tensor: The Hencky strain tensor (computed

in the left basis) is given by
E L = ln V = R E ln RTE (3.1-6)
The superscript L symbolizes the left basis.
Comparison of left and right Hencky strain tensors: The

principal values of the left and right Hencky strain tensors are

identical, and equal to the logarithms of the principal stretches.

Hence both of these strain tensors can be considered to be
logarithmic strain tensors. However, the principal directions of the
left and right Hencky strain tensors are different. The principal
directions of the right Hencky strain tensor do not contain the rigid
body rotations of the material, but the principal directions of the
left Hencky strain tensor contain the rigid body rotations of the
material.
Therefore, for a material undergoing rigid body rotations, the

principal directions of the right Hencky strain tensor do not rotate,
however the principal directions of the left Hencky strain tensor
rotate with the material. Hence, the left Hencky strain tensor is
preferred for output and visualization of the strain state.
Multiplicative decomposition of deformation gradient in

inelastic analysis: In inelastic analysis, the following
multiplicative decomposition of the total deformation gradient into
an elastic deformation gradient X E and an inelastic deformation
gradient X P is assumed:
X = XE XP (3.1-7)
To understand (3.1-7), consider a small region of material under a

given stress state with deformation gradient X. If this region of
material is separated from the rest of the model and subjected to the
same stress state, the deformation gradient is still X. Now if the
stress state is removed, (3.1-7) implies that the deformation
gradient of the unloaded material is X P . The stresses are due
entirely to the strains associated with the elastic deformation
gradient X E .
It can be shown (see Montns and Bathe), that (3.1-7) is equivalent

to the additive decomposition of the displacements into elastic
displacements and plastic displacements.
For the materials considered here, det X P = 1 .
Polar decomposition of elastic deformation gradient: The elastic

deformation gradient can be decomposed into an elastic rotation

tensor R E and elastic right and left stretch tensors U E , V E :
XE = R E UE = V E R E (3.18-a,b)
Elastic Hencky strain tensors: The elastic Hencky strain tensors

in the right and left bases are given by
E ER = ln U E , E EL = ln V E (3.1-9a,b)
Stress-strain relationships: The stresses are computed from the

elastic Hencky strain tensors using the usual stress-strain law of
isotropic elasticity. However, the stress measures used depend
upon the strain measures used. When the right Hencky strain
measure is used, the stress measure used is the rotated Kirchhoff
stress
= ( R E ) J R E
T
(3.1-10)
and when the left Hencky strain measure is used, the stress measure
is the (unrotated) Kirchhoff stress J . J = det X is the volume
change of the material, and, using det X P = 1, J = det X E .
With these choices of stress and strain measures, the stresses and
strains are work-conjugate.
The choice of right Hencky strain and rotated Kirchhoff stresses

gives the same numerical results as the choice of left Hencky strain
and (unrotated) Kirchhoff stresses.
Implementation notes: For 2-D and 3-D solid elements, the

difference between the Cauchy and Kirchhoff stresses is neglected.
The stress measure used with the right Hencky strains is
= ( R E ) R E . The input of material properties is assumed to be
T
in terms of Cauchy stresses, and the output of stresses is in terms of

Cauchy stresses.
For shell elements, Kirchhoff stresses are used throughout. The

input of material properties is assumed to be in terms of Kirchhoff

stresses, and the output of stresses is in terms of Kirchhoff stresses.
These assumptions are justified because they are used with material
models in which the plastic deformations are incompressible and
the plastic deformations are generally much larger than the elastic
deformations.
3.1.3 Large strain thermo-plasticity analysis with the ULJ formulation
This section discusses the ULJ formulation for large strain

inelastic analysis (ULJ formulation). ULJ stands for updated
Lagrangian Jaumann.
The following is a quick summary of the theory of large strain

inelastic analysis with the ULJ formulation:
For further information, see ref KJB, Section 6.2.2 and also the

Velocity gradient tensor: The velocity gradient tensor is defined

as
tu
L = t i = XX 1 (3.1-11)
x j
Notice that the derivative is taken with respect to the current

coordinates.
Rate of deformation tensor, spin tensor: The rate of deformation

tensor is defined as
1
D=
2
( L + LT ) (3.1-12)
and the spin tensor is defined as

1
W=
2
( L LT ) (3.1-13)
D is the symmetric part of L and W is the skew-symmetric part of

L.
Rate of change of Jaumann strain tensor: The rate of change of

the Jaumann strain is defined as
J = D +W J J W (3.1-14)
The quantity J is termed the Jaumann strain in analogy with the

more often-used Jaumann stress. But we do not use the Jaumann
stress in the ULJ formulation.
Jaumann strain tensor: In practice, increments are used in

computing the Jaumann strain tensor, i.e.
t +t
J = t J + ( Dt ) +( Wt ) t J t J ( Wt ) (3.1-15)
Comparison of Jaumann strain with left Hencky strain: When

the rate of change of the principal directions of the left stretch
tensor V is zero, the rate of change of the left Hencky strain is the
same as the rate of change of the Jaumann strain. Hence the
Jaumann strain can be used as an approximate replacement for the
left Hencky strain. The Jaumann strain can be computed more
efficiently than the left Hencky strain, because it is not necessary to
take the square root or logarithm of a tensor when computing the
Jaumann strain. On the other hand, the time step size affects the
Jaumann strain, so that finite time step sizes lead to an error in the
calculation of the Jaumann strain.
For a uniaxial deformation, the Jaumann strain approaches the

logarithmic strain as the step size is reduced. For a rigid-body
rotation, the Jaumann strain also rotates, with the rotation of the
Jaumann strain approaching the expected rotation as the step size is
reduced.
It can also be shown that the Jaumann strain is path-dependent in

general, so that a deformation history in which the final

deformations equal the initial deformations can produce (non-

physical) non-zero Jaumann strains, even in the limit of
infinitesimally small time steps.
Stress-strain relationships: In elasto-plasticity, the same

algorithms are used as in small-strain elasto-plasticity. The
mechanical strains are computed as the total strains minus the
plastic strains (and also any thermal strains), in which the total
strains are the Jaumann strains.
As in the ULH formulation, the stresses are Cauchy stresses for 2-

D / 3-D elements, and are Kirchhoff stresses for shell elements.
3.1.4 Thermal strains
Calculation of thermal strains is needed for temperature-

dependent material models, as well as temperature-invariant
material models with non-zero thermal expansion coefficients.
The temperature at an integration point is evaluated based on

the nodal temperatures and the element shape functions, and then
used to calculate the thermal strains.
For isotropic temperature independent materials, the following

expression is used for thermal expansion.
t TH
eij = ( t 0 ) ij (3.1-16)
where ij is the Kronecker delta ( ij = 1 for i = j and ij = 0 for

i j ).
If the thermal expansion is temperature dependent and isotropic,

the thermal strains are calculated as follows:
t TH
eij = t ( t 0 ) ij (3.1-17)
where

t
=
(
1
0 )
t (
( t )( t REF ) ( 0 )( 0 REF ) )
(3.1-18)
and REF is the material reference temperature and is input as a
material property, see below for more details.
For temperature independent orthotropic materials Eq. (3.1-16)

is replaced by a thermal expansion coefficient vector,
t TH
e ij = i ( t 0 ) ij (no summation over i ) (3.1-19)
For temperature dependent orthotopic materials Eq. (3.1-17) and

Eq. (3.1-18) are modified for each direction similar to Eq. (3.1-19).
Equations (3.1-17) and (3.1-18) are derived as follows: Suppose

that, from experimental data, the dependence of the length of a bar
as a function of temperature is obtained, as shown in Fig. 3.1-3.
L
Secant to curve
Length, L
LREF
REF
Temperature,
Figure 3.1-3: Length of bar vs. temperature
The thermal strain with respect to the reference length may be

calculated as
L LREF
TH =
LREF
Then we define the mean coefficient of thermal expansion for a
given temperature as follows:

TH ( )
( ) =
REF
With this definition, the secant slope in Fig. 3.1-3 is LREF ( ) .

Now, in ADINA, we assume that the thermal strains are initially
zero. To do this, we subtract the thermal strain corresponding to
0
to obtain
t
TH = ( t )( t REF ) ( 0 )( 0 REF )
which leads to Equations (3.1-17) and (3.1-18).
Notice that if the mean coefficient of thermal expansion is constant,

REF no longer enters into the definition of t and the equations
reduce to Eq. 3.1-16. In general, when the mean coefficient of
thermal expansion is not constant, REF must be chosen based on
knowledge of the experiment used to determine ( ) since for the
same material curve, different choices of REF yield different
values of ( ) .
3.2 Linear elastic material models

The following material models are discussed in this section:
Elastic-isotropic: isotropic linear elastic

Elastic-orthotropic: orthotropic linear elastic
In each model, the total stress is uniquely determined by the total

strain.
These models can be employed using the small displacement

or large displacement formulations. In all cases, the strains are
assumed small.
When the elastic-isotropic and elastic-orthotropic materials are

3.2: Linear elastic material models
used with the small displacement formulation, the formulation is

linear.
If the material models are employed with a large

displacement/small strain or large displacement/large strain
formulations, the total Lagrangian formulation is used.
In the small displacement formulation, the stress-strain

relationship is
t
0 = C t0e
in which t0 = engineering stresses and t0 e = engineering strains.
In the total Lagrangian formulation, the stress-strain relationship

is
ref. KJB
Section 6.6.1
t
0 S = C t0
in which t0 S = second Piola-Kirchhoff stresses and t0 = Green-

Lagrange strains.
The same matrix C is employed in all of these formulations.
If the strains are large, it is recommended that the linear elastic

material model not be used (see ref KJB, pp 589-590 for a
discussion in which a linear elastic material is used under large
strain conditions).
.
3.2.1 Elastic-isotropic material model
This material model is available for the truss, 2-D solid, 3-D
solid, beam, iso-beam, plate, shell and pipe elements.
The two material constants used to define the constitutive

relation (the matrix C) are
E = Young's modulus, v = Poisson's ratio

The same constants are employed in the small and large

ref. KJB
Table 4.3, displacement formulations, and hence the matrices C are identically
p. 194 the same in all formulations.
You can specify the coefficient of thermal expansion as part of

the elastic-isotropic material description. The coefficient of
thermal expansion is assumed to be temperature-independent.
3.2.2 Elastic-orthotropic material model
The elastic-orthotropic material model is available for the 2-D

solid, 3-D solid, plate and shell elements.
2-D solid elements: Figure 3.2-1 shows a typical two-dimensional

element, for which the in-plane orthogonal material axes are "a"
and "b". The third orthogonal material direction is "c" and is
perpendicular to the plane defined by "a" and "b". The material
constants are defined in the principal material directions (a,b,c).
1l b
Z
l 8
5l
Y a l4
l
2l l 7
l
6 3
Figure 3.2-1: Principal in-plane material axes orientation for the
orthotropic material model for 2-D solid elements
The stress-strain matrices are as follows for the various 2-D

element types:

Plane stress:
ea 1 ab ac a
0
Ea Eb Ec

eb ba 1 bc
0 b
= Ea Eb Ec

ca cb 1
ec 0 c = 0
Ea Eb Ec
1
ab 0 0 0 ab
Gab
Plane strain:
ea 1 ab ac a
0
Ea Eb Ec

eb ba 1 bc
0 b
= Ea Eb Ec

cb 1
ec = 0 ca 0 c
Ea Eb Ec
1
ab 0 0 0
Gab ab
Axisymmetric:
ea 1 ab ac a
0
Ea Eb Ec

eb ba 1 bc
0 b
= Ea Eb Ec

ca cb 1
ec 0 c
Ea Eb Ec
1
ab 0 0 0
Gab ab

The seven material constants (Ea, Eb, Ec, vab, vac, vbc and Gab)
define the symmetric compliance matrix C1 .
1 ac
ab 0
Ea Eb Ec

1 bc
0
Eb Ec
C1 =
1
0
Gab
1
symmetric
Ec
where the subscript " " in C1 indicates that the material law is
given in the local system of the material axes. Note that the
determinant of C1 must not be zero in order to be able to calculate
the inverse C . This imposes the following restrictions on the
constants. The material constants must be defined so that the
stress-strain constitutive matrix is positive-definite; i.e.,:
1
E 2
ji < i , i, j = a, b, c
E
j
Ea E E
ab bc ca < 0.5 1 ab2 bc2 b ca2 c 0.5
Eb Ec Ea
or
Ea 2 Ea E E
ab bc ac < 0.5 1 ab bc2 b ac2 a 0.5
Ec Eb Ec Ec
using ADINA input values.
Based on the input values for ij , ADINA calculates ji so as to

have a symmetric constitutive matrix; i.e.,

ji ij
=
Ei Ej
See also for example the following reference:
ref. Jones, R.M., Mechanics of Composite Materials,

McGraw-Hill p. 42, 1975.
In ADINA, the Poissons ratio ij is defined differently from

Jones, although it is consistent with Jones nomenclature. In Figure
3.2-2, a uniaxial tensile test is illustrated in which the material
under test is loaded in the a-direction. The Poissons ratio is
defined in ADINA by
b
ba =
a
F eb F
b
a ea
Figure 3.2-2: Definition of nba
In general,
i
ij = (i, j = a, b, c)
j
In Jones, however, the same Poissons ratio is notated
j
J
ij = .
i
Using the correspondence between the Jones and ADINA

nomenclature for the Poissons ratio,

ji = J ij
and the relationship
ji ij
=
Ei Ej
it is possible to calculate the value of ji given a value for J ij

from
Ej J
ij = ij
E
i
When material data is available from sources that follow Jones

notation of Poissons ratio, the above conversion allows the
equivalent ADINA Poissons ratio to be calculated from the Jones
Poissons ratio for input to ADINA.
To obtain the stress-strain matrix C corresponding to the global

material axes, first C is calculated and then
C = QT C Q
where
2
1 m12 1 2 0
2 2
m m1m2 0
Q= 2 2
2m1m2 2 1 2 1 m2 + 2 m1 0
0 0 0 1
and 1 , 2 and m1 , m2 are the direction cosines of the a, b material

axes to the global axes.
Note that in a large displacement analysis it is assumed that the
angle remains constant throughout the incremental solution.
Hence, the use of the total Lagrangian formulation is usually most
appropriate.

3-D solid elements: In three-dimensional analysis we have
1/ Ea ab / Eb ac / Ec 0 0 0
1/ Eb bc / Ec 0 0 0

1/ Ec 0 0 0
C1 =
1/ Gab 0 0
1/ Gac 0

symmetric 1/ Gbc
and, as in two-dimensional analysis, this matrix is inverted and then

transformed to the global coordinate axes.
Plate elements: Figure 3.2-3 shows a typical plate element for

which the in-plane orthogonal material axes are "a" and "b". The
material constants are defined in the principal material directions
(a,b), for which we have
ea 1/ Ea ab / Eb ac / Ec 0 a
e / E 1/ Eb bc / Ec 0 b
b = ba a
ec ca / Ea cb / Eb 1/ Ec 0 c = 0

ab 0 0 0 1/ Gab ab
Note that the transverse normal strain ec is not calculated for the
plate elements.

z,c
b
The a-b and x-y planes 1
l
are coplanar. Nodes
1, 2 and 3 belong to
b y
the x-y plane.
a
Z
l 3
2 l
Y
X
x
Figure 3.2-3: Principal in-plane material axes orientation for the

orthotropic material model for plate elements
In order to obtain the stress-strain matrix C corresponding to the

local x y z directions of the element, we use
C = QT C QT
cos cos sin sin cos sin

where Q = sin sin
T
cos cos cos sin

2cos sin 2cos sin cos 2 sin 2
is the angle between the a and x axes, see Fig. 3.2-2.
Shell elements: Figure 3.2-4 shows typical shell elements for

which the orthogonal material axes are "a", "b" and "c".
The constitutive relation defined in the a,b,c system is

ea 1/ Ea ab / Eb ac / Ec 0 0 0 a
e / E 1/ Eb bc / Ec 0 0 0 b
b ba a
ec ca / Ea cb / Eb 1/ Ec 0 0 0 c = 0
=
ab 0 0 0 1/ Gab 0 0 ab
ac 0 0 0 0 1/ Gac 0 ac

bc 0 0 0 0 0 1/ Gbc bc
Table 3.2-1 summarizes which material constants are necessary,

depending on the type of element used in the analysis.
Table 3.2-1: Required material constants for the element types

available in ADINA
Required material constants
(these constants must be nonzero except for
Element type Poisson's ratios)
2-D solid Ea, Eb, Ec, ab, ac, bc, Gab
3-D solid Ea, Eb, Ec, ab, ac, bc, Gab, Gac, Gbc
Plate Ea, Eb, ab, Gab
Shell Ea, Eb, Ec, ab, ac, bc, Gab, Gac, Gbc

t ,c
b
a
3
r
b
1
s d
r12
2
(a) 4-node element
t, c
l l
b l
l
l
a
rl l
b 3
1
l l
l
r12 s d
l l
2
(b) 16-node element
Note: (r, s, t) is the local Cartesian system.

The a-b and r-s planes are coplanar.
d is the unit projection of r12 onto the r-s plane.
b is the input material angle.
Figure 3.2-4: Definition of axes of orthotropy for shell elements

3.3: Nonlinear elastic material model
3.3 Nonlinear elastic material model

The nonlinear elastic material model is available for truss, 2-D
solid (plane stress, plane strain and axisymmetric), 3-D solid and
shell elements. The formulations used are slightly different (and
simpler) for the truss element, as detailed in Section 3.3.1.
This material uses a nonlinear elastic uniaxial stress-strain data

input in tabular form and shown in Fig. 3.3-1. Note that the input
curve must pass through the origin (0,0). This material is not based
on the classical theory of finite elasticity, and is not intended for
large strain analysis. However, it is a useful material model when
used appropriately, and with awareness of its limitations. This
material model is also available as a user-coded material for 2-D
and 3-D solids only.
Note that the material unloads along the same curve, so that no
permanent inelastic strains are introduced.
The material can have different stress-strain curves in tension

and compression. Under predominantly uniaxial tension or
compression, the material response will follow the input curve
exactly. Under shear dominated loading, the stress is interpolated
from both tension and compression parts of the material stress-
strain curve.
e e
Figure 3.3-1: Stress-strain behavior of nonlinear elastic material

model

In order to use the unaxial stress-stain data ( ) of Fig. 3.3-1,

the effective stress and strain ( and ) must be calculated based
on the 2-D or 3-D total stress and strain tensors ( and ). The
von-Mises stress is used as the effective stress, while the effective
strain is based on
d =
T
d (3.3-1)
which equates the deformation work per unit volume in unaxial

loading to the multi-dimensional state. This results in a unique
equation for as a function of , and the stress-strain state that
depends on the element type.
The effective strain, , is defined by
1 1
E0 2 = T C0 (3.3-2)
2 2
where E0 is Youngs modulus which is determined by the most

stiff region of the input stress-strain curve, C0 is the elastic stress-
strain matrix obtained using E0 and . ( E0 cancels out from both
sides of Eq. (3.3-2))
Differentiating Eq. (3.3-2) with respect to the total strain, we have
1 T
d = C0 d (3.3-3)
E0
Substituting Eq. (3.3-3) into Eq. (3.3-1), the stresses can be

expressed in terms of total strains, i.e.,

= C0 (3.3-4)
E0
or


= e (3.3-5)
E0
where e = C0 which is the elastic trial stress.
The effective stress is taken from the tensile part of the

stress-strain curve for predominantly tensile loading, from the
compression part of the stress-strain curve for predominantly
compression loading and is interpolated between the two curves
otherwise.
The consistent tangent stress-strain matrix is obtained by

differentiating Eq. (3.3-4)or (3.3-5) with respect to the total strain
tensor. The stress-strain matrix is symmetric in predominant tensile
or compression loading (when only one of the two material curves
is used), and is non-symmetric otherwise (when interpolation
between the curves is required). If a symmetric solver is used, the
constitutive matrix is symmetrized and in most cases reasonable
convergence is still obtained.
Note that discontinuities are not allowed in user-supplied stress-

strain curve. The table look-up is performed using linear
interpolation within the table and linear extrapolation outside the
table using the two starting or ending points.
Stess update algorithm
For an iteration i , given t , t , t +t

u (i ) , E0 , , update t +t
(i ) ,
t +t
(i )
t +t
Step 1. Calculate the new total strain state (i ) based on
displacements t +t u ( i )
Step 2. Calculate the elastic trial stress,
t +t
e = C0 t +t (3.3-6)
Step 3. Compute the magnitude of the effective strain, .

Step 4. Calculate the ratio
I
1e
r =C (3.3-7)
t +t
e
where C is a constant that biases the general stress state towards

the pure tension or compression curves and is internally set to 3/2,
I1e is the first elastic stress invariant, and t +t e is the effective
elastic stress which is calculated as follows,
t +t
e = E0 t +t (3.3-8)
Restrict r to be between -1 and 1.
Step 5. Calculate the effective stress in tension t and in

compression c , based on the user-supplied stress-strain curve and
, as follows:
t +t
t =t +t (t +t ) (3.3-9)
t +t
c = t +t (t +t ) (3.3-10)
Step 6. Calculate the actual effective stress, t +t , as
1 + r t +t 1 r t +t
t +t
= t + c (3.3-11)
2 2
Step 7. Evaluate the new stress state by
t +t
t +t
t +t
= e (3.3-12)
E0 t +t

Step 8. Evaluate the tangential stress-strain matrix. Symmetrize it if

necessary.

Step 9. A secant stress-strain matrix can be used if the tangential

stress-strain matrix exhibits a softening behavior.
3.3.1 Nonlinear elastic material for truss element
For the truss element, an elastic material model is available for

which the stress-strain relationship is defined as piecewise linear.
Fig. 3.3-2 illustrates the definition of the stress-strain law.
Stress
6 l
5 l
4 l
e1 e2 e3
3 e4
l
e5 e6 Strain
l 2
l 1
Figure 3.3-2: Nonlinear elastic material for truss element
Note that the stress is uniquely defined as a function of the

strain only; hence for a specific strain te, reached in loading or
unloading, a unique stress is obtained from the curve in Fig. 3.3-2.
A typical example of the truss nonlinear elastic model is shown

in Fig. 3.3-3. Note that the stress-strain relation shown corresponds
to a cable-like behavior in which the truss supports tensile but no
compressive loading.

Stress
Point 3
4.32104
Point 1 Point 2

-1.0 1.0 Strain
Figure 3.3-3: Example of nonlinear elastic stress-strain material

model for truss
A sufficient range (in terms of the strain) must be used in the

definition of the stress-strain relation so that the element strain
evaluated in the solution lies within that range; i.e., referring to Fig.
3.3-2, we must have e1 t e e6 for all t.
The truss element with this material model is particularly useful

in modeling gaps between structures. This modeling feature is
illustrated in Fig. 3.3-4. Note that to use the gap element, it is
necessary to know which node of one body will come into contact
with which node of the other body.
ref. S.M. Ma and K.J. Bathe, "On Finite Element Analysis of

Pipe Whip Problems," Nuclear Engineering and Design,
Vol. 37, pp. 413-430, 1976.
ref. K.J. Bathe and S. Gracewski, "On Nonlinear Dynamic

Analysis using Substructuring and Mode Superposition,"
Computers and Structures, Vol. 13, pp. 699-707, 1981.
Note that an alternate gap element can be obtained by the use of

the 2-node truss element employing the plastic-bilinear material
model, see Section 3.4.1.
A more general way of modeling contact between bodies is the
use of contact surfaces, see Chapter 4.
Other modeling features available with the truss element and

this material model are shown in Figure 3.3-5.

Pipe (beam)
n
l
Gap
Gap element L
m
l
Stress

_
L
Strain
Figure 3.3-4: Modeling of gaps
The model shown in Fig. 3.3-5(a) corresponds to a

compression-only behavior.
The model shown in Fig. 3.3-5(b) corresponds to a tension

cut-off behavior, the snapping of a cable, for example.
The model shown in Fig. 3.3-5(c) corresponds to a behavior

exhibiting both tension and compression cut-offs.

Stress Stress
e2

e1 e2 e3 e4 e1 Strain

Strain e3 e4

(a) Compression only (b) Tension cut-off
Compression Stress
cut-off limit
e6

e1 e2

e7 e8 Strain
e5
e4
e3 Tension
cut-off limit
(c) Tension and compression cut-offs
Figure 3.3-5: Various modeling features available with the

nonlinear elastic truss model
3.4 Isothermal plasticity material models

ref. KJB This section describes the following material models:
Section 6.6.3
Plastic-bilinear, plastic-multilinear: von Mises model with
isotropic, kinematic hardening or mixed hardening
Mroz-bilinear: von Mises model with Mroz hardening
Plastic-orthotropic: Hill yielding with bilinear proportional

hardening
Gurson: Gurson plastic model for fracture/damage analysis
Plastic-cyclic: von Mises model with hardening rules suitable

for modeling cyclic plasticity.

3.4: Isothermal plasticity material models
All elastic-plasticity models use the flow theory to describe the

elastic-plastic response; the basic formulations for the von Mises
models are summarized on pp. 596-604, ref. KJB, and for the Mroz
model in:
ref. Y.F. Dafalias and E.P. Popov, "Plastic Internal Variables

Formalism of Cyclic Plasticity," J. Appl. Mech., Trans.
ASME, Vol. 43, pp. 645-651, 1976.
The formulation for the Ilyushin model is given in the following

reference:
ref. K.J. Bathe, E. Dvorkin and L.W. Ho, "Our Discrete-

Kirchhoff and Isoparametric Shell Elements An
Assessment," J. Computers and Structures, Vol. 16, pp.
89-98, 1983.
The Drucker-Prager material model is described in Section

3.9.2.
3.4.1 Plastic-bilinear and plastic-multilinear material models
These material models are based on
The von Mises yield condition (see p. 597, ref. KJB)
An associated flow rule using the von Mises yield function
An isotropic or kinematic, bilinear or multilinear, hardening

rule
Figs. 3.4-1 to 3.4-3 summarize some important features of these

material models.
These models can be used with the truss, 2-D solid, 3-D solid,
beam (plastic-bilinear only), iso-beam, shell and pipe elements.

Stress
Multilinear stress-strain curve
0
y
Bilinear stress-strain curve
Strain
Figure 3.4-1: von Mises model
These models can be used with the small displacement/small

strain, large displacement/small strain and large
displacement/large strain formulations. The large
displacement/large strain formulation can only be used with the
2-D solid, 3-D solid and shell elements (the shell elements must be
3-node, 4-node, 6-node, 9-node or 16-node single layer shell
elements described entirely by midsurface nodes).
3
(1,1,1) s3
23
0
y
23
0
y
s
s
Elastic region
a) Principal stress space b) Deviatoric stress space
Figure 3.4-2: von Mises yield surface

Stress Stress
ET ET
0 t 0
y y y
E E 2 0 y
Strain E Strain
E t
y E
E ET
ET
a) Bilinear isotropic hardening b) Bilinear kinematic hardening
Stress Stress
t
0 y
0 2 0 y
y y
Strain Strain
t
y
c) Multilinear isotropic hardening d) Multilinear kinematic hardening
Figure 3.4-3: Isotropic and kinematic hardening
When used with the small displacement/small strain

formulation, a materially-nonlinear-only formulation is employed,
when used with the large displacement/small strain formulation,
either a TL or a UL formulation is employed (depending on
element type), and when used with the large displacement/large
strain formulation, either a ULH or ULJ formulation is employed.
If geometrically nonlinear effects are to be included, the large

displacements/large strain kinematics are preferred to the large
displacement/small strain kinematics, even when the strains are
numerically small. The large displacement/small strain kinematics
should be used only when the large displacement/large strain

kinematics are not supported by the element.
For multilinear plasticity, there is no restriction on the number

of stress-strain points in the stress-strain curve.
Mixed hardening is available only in bilinear plasticity.
Plane strain, axisymmetric or 3-D solid elements that reference

these material models should also employ the mixed displacement-
pressure (u/p) element formulation. The u/p formulation is used by
default for these elements.
In the von Mises model with isotropic hardening, the following

yield surface equation is used:
1 t t 1
t
fy =
2
( s s ) t y2 = 0
3
where s is the deviatoric stress tensor and t y the updated yield

t
stress at time t.
In the von Mises model with kinematic hardening, the following
1 t 1
t
fy =
2
( s t ) ( t s t ) 0 y2 = 0
3
where t is the shift of the center of the yield surface (back stress
tensor) and 0 y2 is the virgin, or initial, yield stress.
In the von Mises model with mixed hardening, the following
1 t 1
t
fy =
2
( s t ) ( t s t ) t y2 = 0
3
where
t
y = 0 y + ME p e p

The back stress t is evolved by
d = C p (1 M ) de p
Cp is Pragers hardening parameter, related to the plastic

modulus Ep by
2
Cp = Ep
3
and M is the factor used in general mixed hardening (0 < M < 1)
which can be a variable, expressed as
M = M + ( M 0 M ) exp ( e p )
The formulation for the von Mises model with mixed hardening
is given in the following reference:
ref K.J. Bathe and F.J. Montns, On Modeling Mixed

Hardening in Computational Plasticity, Computers and
Structures, Vol. 82, No. 6, pp. 535 - 539, 2004.
Note that the convergence might not be good when the mixed
hardening parameter 0 . Therefore, = 0 is preferred.
The yield stress is a function of the effective plastic strain,
which defines the hardening of the material. The effective plastic
strain is defined as
t
P 2 p
t
e = de de p
0
3
in which de p is the tensor of differential plastic strain increments

and in which de p de p is calculated as deijp deijp (see ref. KJB, p.
P
599) . In finite element analysis, we approximate t e as the sum
of all of the plastic strain increments up to the current solution
time:

P P
t
e = e
all solution steps

P 2 p
where e = e e p and e p is the tensor of plastic strain
3
increments in a solution step. Because of the summation over the
P
solution steps, we refer to the calculated value of t e as the
accumulated effective plastic strain.
If a thermal load is applied to the structure, the thermal strains
are taken into account but the material characteristics are
considered to be temperature independent.
Stress-strain input data for multilinear plasticity in large strain

analysis
In large strain analysis, ADINA works internally with true

(Cauchy) stresses and true (logarithmic) strains. However, typical
tension tests used to determine experimental data return forces and
displacements. These forces and displacements are used to compute
engineering stresses (force per unit original area) and engineering
strain (change in length per unit length). Therefore it is necessary
to convert engineering stress-strain data to true stress-strain data.
This conversion is either done by the user, or is done automatically
by the AUI (using the MASTER CONVERT-SSVAL=YES
option). We now discuss this conversion process in detail.
Consider an experiment in which a fully incompressible
material is put into uniaxial tension. The force-displacement curve
is determined, and the following information extracted from the
force-displacement curve:
The engineering stress is computed as the force / original area.

The engineering strain is computed as the displacement /
original length.
This data is determined for a number of points on the force-

displacement curve, starting at point 1, which is considered to be
the elastic limit. And the Youngs modulus is also determined as
the ratio of engineering stress / engineering strain at the elastic
limit.
Now consider duplicating the experimental results with a finite
element model that uses a large strain formulation. The tension test
finite element model should return the correct force for each

displacement point on the force-displacement curve.

Because the stress and strain measures used in the large strain
finite element formulation are true stress and true strain, it is
necessary to convert the engineering stress / engineering strain data
to true stress / true strain data.
The conversion can be done using an algorithm similar to
{
For (each stress-strain point i, i=1, 2, ... ) {
ei = engineering strain
i = engineering stress
i = i (1 + ei ) = true stress
i = ln(1 + ei ) = true strain
}
E = 1 / 1
}
Notice that E also needs to be converted. The reason is as

follows. If E is not converted, ADINA computes 1 = 1 / E but
now 1 is no longer the true strain at the elastic limit.
Here is a numerical example:
e1 = 0.1 , 1 = 30 MPa , E = 1 / e1 = 300 MPa
The conversion given above produces
1 = 0.09531 , 1 = 33 MPa
This point is assumed to be at the elastic limit. If E is kept at 300,

then ADINA computes
1 = 33 / 300 = 0.11
and this 1 is no longer at the elastic limit. Therefore E must be

recalculated as part of the conversion process:
E = 33 / 0.09531 = 346.2 MPa

Assuming that the above conversion is performed, then the tension

test finite element model will return the correct force for each of
the displacements corresponding to the points on the original force-
displacement curve.
Assumptions in the conversion formulas
There are a number of assumptions in the above formulas, as

follows:
Elastic response
It is assumed that the stress-strain behavior is linearly elastic up

to the elastic limit. However, since ADINA uses true stress / true
strain data internally, the linear elastic behavior is also based on
true stress / true strain data. In the above example, the linear elastic
response computed by the tension test finite element model is based
on a Youngs modulus of 346.2 MPa. This response is, of course,
quite different than a linear elastic response based on a Youngs
modulus of 300 MPa.
The reason that the Youngs modulus is so different is because
the strain at the elastic limit is large. When the strain at the
elastic limit is small, then the change in Youngs modulus is also
small. The change in Youngs modulus caused by the conversion is
3
approximately equal to E = e1 E .
2
Compressible elastic material
The above conversion assumes that the material response is

fully incompressible, under both elastic and plastic conditions.
However, in most cases, the material is compressible under elastic
conditions.
Nevertheless, the above formulas are frequently used anyway.
The error thus incurred will be largest for point 1, and will diminish
for larger values of strain.
Response in uniaxial compression
The above formulas use tension data to convert from

engineering to true values. However, it should be recognized that

ADINA uses the same stress-strain curve in both tension and
compression. Therefore, if a large strain finite element model is
put into uniaxial compression, the observed (compressive) force
will not be equal in magnitude to the expected (tensile) force, at a
given (compressive) displacement level.
Here is an example. Suppose that in a tensile test, the following
data is obtained:
Point Engineering Engineering True strain True

strain stress (MPa) stress
...
3 0.05 50 0.04879 52.5
4 0.05263 51 0.05129 53.7
...
First consider a materially-nonlinear-only analysis. Enter the

engineering stress and strain values for the two points as part of the
stress-strain input data. When the model is run in compression to
an engineering strain of -0.05, the engineering stress is -50 MPa.
The force-displacement response is symmetric in tension and
compression.
Next consider a large strain analysis. Enter the true stress and
strain values for the two points as part of the stress-strain input
data. Now run the model in compression to an engineering strain of
-0.05. An engineering strain of -0.05 corresponds to a true strain of
-0.05129, therefore the true stress is -53.7 and the engineering
stress is -56.5. The force-displacement response is not symmetric
in tension and compression.
Homogeneous deformation
The above conversion assumes that the specimen is uniform and

that the specimen deforms homogeneously under load. Therefore
the conversion is only valid up to the ultimate tensile strength of
the material, because beyond that point the deformation of the
specimen might be no longer homogeneous due to localized
necking.
MASTER CONVERT-SSVAL feature

When CONVERT-SSVAL=NO in the MASTER command (the

default), the AUI does not perform any conversion. True stress-
strain data should be input, and the user could compute this true
stress-strain data using the formulas given above.
When CONVERT-SSVAL=YES, then the AUI performs this
conversion using the formulas given above. Therefore engineering
stress-strain data should be input.
The CONVERT-SSVAL=YES feature should only be used
when all of the elements that use multilinear plastic materials also
use large strain formulations.
The CONVERT-SSVAL=YES feature does not perform any
conversions for the bilinear plastic material models, or for the
plastic-cyclic material models.
Material behavior beyond the last point of the stress-strain

curve in multilinear plasticity
The material behavior beyond the last point of the stress-strain

curve in multilinear plasticity can be considered ruptured, or the
curve can be extended indefinitely with the slope of its final
segment. This depends on the global setting of the EXTEND-
SSCURVE paramter in the MASTER command with indefinite
extension as the default.
Modeling of rupture: Rupture conditions can also be modeled

with these material models except for the beam element. For the
bilinear stress-strain curve, a maximum allowable effective plastic
strain eAP can be specified for the rupture condition. For the
multilinear stress-strain curve, the rupture plastic strain corresponds
to the effective plastic strain at the last point input for the stress-
strain curve.
When rupture is reached at a given element integration point,
the corresponding element is removed from the model (see Section
11.5).
There is also the option of user-supplied rupture. You code the
rupture condition into one of the CURUP subroutines: CURUP2
for 2-D solid elements, CURUP3 for 3-D solid elements, CURUP4
for beam elements, CURUP5 for iso-beam elements, CURUP7 for
shell elements and CURUP8 for pipe elements. These subroutines
are in file ovlusr.f. ADINA provides the calculated latest stresses,
total strains, thermal strains, plastic strains, creep strains, yield

stress and accumulated effective plastic strain to these subroutines

and these subroutines return the current rupture state.
In subroutines CURUP2, CURUP3 and CURUP7, ADINA also
provides solution time, accumulated effective creep_strain
(AECST2) of the creep material model, and user-input real and
integer arrays RRUPTR and IRUPTR. Viscoplastic strains and
accumulated effective viscoplastic strains of the Anand material
model are also provided through EPSP2 and EPSTR2.
Rate-dependent plasticity for truss, 2-D solid, 3-D solid,

isobeam, pipe and shell elements: The rate dependent model in
ADINA is used to simulate the increase in the yield stress with an
increase in strain rate.
The rate-dependent model only applies to the isotropic plasticity
models with isotropic hardening (bilinear or multilinear).
The rate-dependent model is implemented for truss, 2-D solid,
3-D solid, isobeam, pipe and shell (either single-layered or
multilayered) elements.
The effective yield stress including strain rate effects is
P
y = 1 + b ln 1 +
0

0
y

where y0 is the static yield stress, 0 is the transition strain rate

and b is the strain rate hardening parameter. Here P is calculated
e P
using P = .
t
For more information on this formula, see the following
reference:
ref. W.H. Drysdale and A.R. Zak, Mechanics of Materials

and Structural Theories A Theory for Rate Dependent
Plasticity, Computers and Structures, Vol. 20, pp. 259-
264, 1985.
The AUI can determine constants 0 and b using curve-fitting.

The input for the curve-fitting is one or more plastic strain rates and
associated user-input stress-strain curves. For each plastic strain

y
rate P , the average overstress ratio is determined from the
y0
associated user-input stress-strain curve; then the curve-fitting is
performed using the strain rates and associated overstress ratios. If
only one strain rate is specified, then 0 must be specified and b is
determined using curve-fitting. If two or more strain rates are
specified, both 0 and b are determined.
Modeling of gaps using the truss element: When used in

conjunction with the 2-node truss element the elastic-plastic
material models can be used to model gaps in the elements (the
nonlinear elastic model can also be used to model gaps with the
truss elements; see Section 3.3.1). In this case the gap elements can
only resist compressive loads, i.e., gap elements have no tensile
stiffness. The gap width input for each element is used to
determine a strain egap (compressive strain is defined as negative).
The compressive stiffness of a gap element is zero if the strain in
the element is greater than or equal to egap and is nonzero when the
strain is less than egap.
3.4.2 Mroz-bilinear material model
This material model is based on

Isotropic elasticity
An associated flow rule
The Mroz kinematic hardening rule for the yield surface
The Prager kinematic hardening rule for the bounding
surface
The model is described in the following references:
ref. Z. Mrz, On the description of anisotropic work

hardening, J. Mech. Phys. Solids, 15 (1967) 163-175.
ref. F.J. Montns, Implicit algorithms for multilayer J2-

plasticity, Comput. Meth. Appl. Engrg., 189 (2000)
673-700.
The Mroz-bilinear model can be used with the 2-D solid and

3-D solid elements.
The Mroz-bilinear model can be used with the small

displacement/small strain, large displacement/small strain and
large displacement/large strain formulations.
when used with the large displacement/small strain formulation, the
TL formulation is used, and when the large displacement/large
strain formulation is used, the ULH formulation is used.

this material model should employ the mixed displacement-
pressure (u/p) formulation.
The material behavior is characterized by a uniaxial bilinear

stress-strain curve plus a bounding line. A representation of the
stress-strain curve is shown in Fig. 3.4-4. Under uniaxial loading,
plastic deformation initiates at = y , and continues along the
line AB until the bounding stress yB is reached. With further
plastic deformation, the stress follows the bounding line along BC.
The material response along ABC is the same as in the case of the
plastic-bilinear model with isotropic or kinematic hardening. If the
tangent modulus ETB is equal to zero, the deformation along BC
would correspond to perfect plasticity.
The uniaxial loading is represented in the principal deviatoric
stress space in Fig. 3.4-5. The stress path ABC corresponds to
the stress-strain path ABC. When the yield surface has translated so
that it comes in contact with the bounding surface at point B (B),
the two surfaces translate together along their common contact
point normal along BC (BC).
When the loading is reversed, the material undergoes elastic

unloading along CD, and deforms plastically once D is reached, see
Fig. 3.4-4. From this point, the material response differs from that
of the traditional plastic-bilinear model with kinematic or isotropic
hardening. For example, in the case of kinematic hardening,
continued plastic deformation follows DF with slope ETB , whereas
in the case of the Mroz-bilinear model, continued plastic

deformation follows DF with slope ET . In Fig. 3.4-5, this

corresponds to translation of the yield surface along DF. During
this translation, the bounding surface does not translate. When the
yield surface again comes in contact with the bounding surface at
point F (F), the two surfaces translate together along their
common normal along FG (FG). If the loading is again reversed,
the elastic unloading of the material follows GH in the stress-strain
curve.
Stress
C ETB
yB B
Bounding line ET 2 y
A E
y
E
D
ET
O Strain
H F
ET
E Bounding line
2 y
F
G ETB
Figure 3.4-4: Mroz-bilinear model; stress-strain curve

Bounding surface Bounding surface

at stress state C at stress state G
Yield surface Yield surface
s3 at stress state C at stress state G s3
yB 2
3
C C
B
y 2 A
3
D
s1 s2 s1 G s2
(a) Loading path O-A-B-C (b) Reverse loading C-D-F-G
Figure 3.4-5: Mroz-bilinear model behavior in

deviatoric stress space
In order to describe the Mroz-bilinear material model under

general, non-radial loading (see Fig. 3.4-6), two conditions are
defined for plastic deformation during a time step:
Case I:
The yield surface translates in the deviatoric stress space
during the time step, but does not come into contact with the
bounding surface.
Case II:
- The yield surface translates in the deviatoric stress space
until it reaches the bounding surface, after which both
surfaces translate together along their common contact
point normal.
- The two surfaces are already in contact at the beginning

of the time step, and the state of the loading causes both
surfaces to translate together along their common contact
point normal.
The implicit Mroz translation rule for the yield surface is based
on a target normal tensor defined as

t +t SE t
t= E t
S
where S E is the trial stress for the time step and t is the back
stress of the bounding surface at the end of the previous time step.
The target normal tensor is explicit in the sense that no iterations
are required to determine it, but implicit in the sense that it depends
on the trial stress for the time step. This target normal tensor is the
same for both Case I and Case II. In order for the target normal
tensor to be uniquely defined, the following condition on the
choice of the yield stress and the bounding stress must be satisfied:
y < yB < 2 y
Case I:
The translation direction for the yield surface, according to the

implicit Mroz translation rule, is given by
2
t +t
m = t t +
3
( yB y ) t +t t
with the increment in the back stress of the yield surface given by
t +t
m
= t +t
t = C p t +t
.
m
In the above equation, the effective plastic modulus C p is given by
2 E ET
C p = .
3 E ET
The plastic strain increment for the step is given by the associated

flow rule
t +t
S t +t
e p = t +t n = t +t
S t +t
where is a consistency parameter which has to be determined

iteratively. The yield condition f y = 0 is solved for the
consistency parameter :
1 t +t 1
fy =
2
( S t +t ) ( t +t S t +t ) y2 = 0 .
3
e S
E
t+t
n
t+t
t
S S s3
t+t

t t+t
t
t
t+t
m
t+t
m
s1 s2
(a) Case I; Only the yield surface translates during the time step
Figure 3.4-6: Mroz model under general, non-radial loading

E
S
e
t+t
S
t+t

t
S s3
t+t

t t+t
t
t

s1 s2
(b) Case II; Both surfaces translate during the time step
Case II:
The back stress of the bounding surface evolves according to
= t +t
t = C p t +t t
where is a consistency parameter associated with translation

of the bounding surface, and C p is an effective plastic modulus
given by
2 ET ETB
C p = .
3 ET ETB

By geometry, the position of the yield surface at the end of the time
step is related to that of the bounding surface through
2
t +t
= t +t
+
3
( yB y ) t +t t .
The increment in the back stress of the yield surface becomes
2
= t +t
t = t t + C p +
3
( yB y ) t +t t

from which the consistency parameter associated with translation

of the yield surface is calculated as

= .
C p
Finally, the plastic strain increment for the step is given by the
associated flow rule
e p = ( + ) t +t t .
As the consistency parameter is a function of the consistency

parameter , the yield condition for the bounding
surface, f yB = 0 , can be solved for the consistency parameter
alone:
1 t +t 1 2
f yB =
2
( S t +t ) ( t +t S t +t ) yB
3
=0
Note that if a zero tangent modulus is prescribed for the

bounding surface, ETB = 0 , the bounding surface does not
translate. In this case, the effective stress remains smaller than or
3 t +t t +t
equal to the bounding stress, t +t
= S S yB .
2

If a thermal loading is applied to the structure, thermal strains

are taken into account, but the material characteristics are taken to
be temperature independent.
Rupture conditions can be modeled: When a positive value eAp

is input for the maximum allowable effective plastic strain, the
program compares the accumulated effective plastic strain, t +t e p ,
with eAp for each material point used in the analysis and over the
whole history of deformation. It is considered that the rupture of a
material point occurs when t +t e p is greater than eAp . When
rupture is reached at a material point, the corresponding element is
removed from the model (see Section 11.5). The user-supplied
rupture option is also available (see Section 3.4.1).
3.4.3 Plastic-orthotropic material model
The plastic-orthotropic model is based on:

The Hill yield condition
An associated flow rule
A proportional hardening rule
The elastic constants can be either isotropic or orthotropic.
This model can be used with the 2-D solid, 3-D solid and shell
elements.
The plastic-orthotropic model can be used with the small

large displacement/large strain formulations. Shell elements
used in conjunction with the large displacement/large strain
formulation must be 3-node, 4-node, 6-node, 9-node or 16-node
single layer elements described entirely by midsurface nodes.
formulation, a materially-nonlinear-only formulation is employed;
TL formulation is employed; and when used with the large
displacement/large strain formulation, the ULJ formulation is
employed.


The mixed formulation can be used (and is the default) for 2-D
axisymmetric, 2-D plane strain and 3-D solid elements. But when
the mixed formulation is used, the elastic constants must be
isotropic.
The following is a quick summary of the basic concepts used in

the plastic-orthotropic material model. For further information, see
the following reference:

ref. R. Hill, The Mathematical Theory of Plasticity, Oxford
University Press, 1998.
The Hill yield condition is given by:
F ( bb cc ) + G ( cc aa ) + H ( aa bb )
2 2 2
+ 2L ab2 + 2M ac2 + 2N bc2 1 = 0
where (a, b, c) are the material principal axes, and F, G, H, L, M, N

are material constants. These constants are given by
1 1 1 1
F= 2+ 2 2
2Y Z X
1 1 1 1
G= 2 + 2 2
2 Z X Y

1 1 1 1
H= 2+ 2 2
2 X Y Z
1 1 1
L= 2 ; M = 2 ; N = 2
2Yab 2Yac 2Ybc
where X, Y, Z, are the yield stresses in the material directions a, b,

c, and Yab, Yac, Ybc are the yield stresses for pure shear in the
planes (a,b), (a,c), and (b,c) (see Figure 3.4-7).
cc Initial yield surface

Yield surface
at time t
C Z
0
A B bb
aa
X
a, b, c material axes X, Y, Z initial yield stresses
Figure 3.4-7: Orthotropic yield surface
The Hill yield condition can also be written as
f ( bb cc ) + g ( cc aa ) + h ( aa bb )
2 2 2
+ 2 ab
2
+ 2m ac2 + 2n bc2 y2 = 0
where
f = y2 F , g = y2G , h = y2 H , = y2 L , m = y2 M , n = y2 N

and
11
y2 = ( X 2 + Y 2 + Z 2 ) + Yab2 + Yac2 + Ybc2
23
The Hill effective stress is defined as
a2 = f ( bb cc ) + g ( cc aa ) + h ( aa bb )
2 2 2
+ 2 ab
2
+ 2m ac2 + 2n bc2
so that the yield condition can be written a2 y2 = 0 . Notice that

under uniaxial tension, the Hill effective stress is not equal to the
uniaxial stress, in general.
The accumulated effective plastic strain eaP is defined using the
principle of the equivalence of plastic work dW P = a deaP . The
formula for the increment of effective plastic strain is different than
the corresponding formula used for isotropic plasticity. Again,
under uniaxial tension, the accumulated effective plastic strain is
not equal to the uniaxial plastic strain, in general.
The hardening rule is defined in terms of the relationship
between the yield stress and the accumulated effective plastic
d y
strain. The universal plastic modulus is defined as Eup = ,
deaP
and can be either constant or varying, as discussed in more detail
below. During hardening, the shape coefficients f, g, h, l, m, n are
considered constant.
Note that orthotropic proportional hardening reduces to
isotropic hardening for the appropriate values of the elastic and
tangent moduli.
The effective-stress-function algorithm is used to calculate
stresses and plastic strains when plasticity occurs.
There are two options for the input of the universal plastic
modulus and the initial yield stress:
Bilinear universal plastic modulus Eup . In this case the plastic

moduli for directions a, b, c and planes (a,b), (a,c), (b,c) are

Eup
given by E = E = E = E and E = E = E =
a
p
b
p
c
p
u
p p
ab . p
ac
p
bc
3
Either enter Eup directly, or enter the independent moduli
EaT , EbT , EcT for the a, b, c directions and for the (a,b), (a,c),
(b,c) planes. Then Eup is equal to
(
1 1 p2 p2 p2 p2 p2
)
p2
Ea + Eb + Ec + Eab + Eac + Ebc
23
where
Ei EiT
Eip = i = a ,b ,c
Ei EiT
and
Eij EijT
E =p
ij ij = ab,ac,bc
Eij EijT
The initial yield stress y is entered by direct specification of

the yield stresses X , Y , Z , Yab , Yac , Ybc , then by the formula
11
y2 = ( X 2 + Y 2 + Z 2 ) + Yab2 + Yac2 + Ybc2 .
23
Analytical uniaxial plastic stress-strain curve: The plastic

stress-strain curve is assumed to have the following analytical
form:
y = C ( e0 + eaP )
n
in which e0 is the initial yield strain, n is the strain hardening

coefficient and C is a constant. The initial yield stress y is
computed using y = Ce0n , and the universal plastic modulus
Eup is computed using Eup = nC ( e0 + eaP )

n 1
.

There are three ways to specify material constants f , g , h , ,

m , n:
Enter the yield stresses X , Y , Z , Yab , Yac , Ybc then

ADINA calculates the material constants using the equations
given above. When this option is used in conjunction with an
analytical plastic stress-strain curve, the initial yield stress
11
computed from y2 = ( X 2 + Y 2 + Z 2 ) + Yab2 + Ybc2 + Yac2 must
23
equal the initial yield stress computed from y = Ce0 . n
Directly specify f , g , h , , m , n . Note that f , g , h ,

, m , n must all be positive. When this option is used in
conjunction with a bilinear universal plastic modulus, all of the
yield stresses X , Y , Z , Yab , Yac , Ybc must be entered, but only
the initial yield stress computed from
11
y2 = ( X 2 + Y 2 + Z 2 ) + Yab2 + Yac2 + Ybc2 is used.
23
Enter the Lankford coefficients r0 , r45 , r90 , then

r0 1 r
f = ,g = , h= 0 ,
r90 ( r0 + 1) r0 + 1 r0 + 1
( r0 + r90 )( 2r45 + 1)
2 = , m = 1.5 , n = 1.5
r90 ( r0 + 1)
in which r0 is the Lankford coefficient for the rolling direction,

r45 is the Lankford coefficient for the direction 45o to the rolling
direction and r90 is the Lankford coefficient for the direction 90o
to the rolling direction (the transverse direction). When this
option is used in conjunction with a bilinear universal plastic
modulus, all of the yield stresses X , Y , Z , Yab , Yac , Ybc must
be entered, but only the initial yield stress computed from

11
y2 = ( X 2 + Y 2 + Z 2 ) + Yab2 + Yac2 + Ybc2 is used.
23
A typical combination of input options is the analytical

universal plastic stress-strain curve option together with Lankford
coefficients. When this combination of options is used, the
material a-direction is the rolling direction, and, under uniaxial
tension in the rolling direction, the Hill effective stress is equal to
the uniaxial stress and the effective plastic strain is equal to the
uniaxial plastic strain. Therefore the analytical plastic stress-strain
constants are taken from the plastic stress-strain curve in the rolling
direction. In particular, y = Ce0n gives the initial yield stress in
the rolling direction.
When the 2-D solid element or shell element is used, material

axes a and b must lie in the plane of the element.
If a thermal load is applied to the structure, the thermal strains

are taken into account but the material characteristics are
considered to be temperature independent.
Rupture conditions can also be modeled: a maximum allowable

effective plastic strain eaP can be specified for the rupture
condition. When rupture is reached at a given element integration
point, the corresponding element is removed from the model (see
Section 11.5).
The user-supplied rupture option can also be used (see Section
3.4.1).
3.4.4 Gurson material model
The Gurson model is intended for use in fracture/damage

analysis. The purpose of the Gurson plastic model is to predict
ductile crack growth by void growth and coalescence. This is a
micromechanical model based approach.
The Gurson model is available for the 2-D solid and 3-D solid
elements.

The Gurson model can be used with the small

large displacement/large strain formulations.
TL formulation is employed and when used with the large
displacement/large strain formulation, the ULH formulation is
employed.
In the Gurson model, the yield function has the form
2
q
0
3 q p
2 0
2
= + 2q1 f * cosh 2 1 + q3 f * = 0 ( )
in which is the yield function, q is the von Mises stress, 0 is
the equivalent tensile flow stress, p is the pressure (positive in
compression), f * is based on the void volume fraction and q1,q2,
q3 are the Tvergaard constants. Note that this reduces to the von
Mises yield function when f * equals zero. f * is defined as
f for f f c

f =
*
fu f c
*
fc + f f ( f fc ) for f > f c
F c
in which f is the void volume fraction, f c is the critical void

volume fraction, f F is the void volume fraction at final failure and
fu* = 1/ q1 .
Initially, the void volume fraction is f 0 , which must be input.

The increase of void volume fraction is controlled by
df = (1 f ) d kkp + Ad p

in which d kkp is the sum of the normal plastic strain components,

p is the equivalent plastic strain and A is the void nucleation
intensity. A is defined as
fN 1 p 2
A= exp N

S N 2 2
S N
in which f N is the volume fraction of void nucleating particles,

N is the mean void nucleation burst strain and S N is the
corresponding standard deviation.
The uniaxial stress-strain curve is

N
0 0 3G
= + p

y y y
where y is the tensile yield stress of the matrix, G is the shear

modulus and N is a material constant (typically on the order of 0.1).
For more information, see the following reference:
ref. N. Aravas, On the numerical integration of a class of

pressure-dependent plasticity models, Int. J. Num.
Meth. in Engng., Vol. 24, 1395-1416 (1987).
3.4.5 Plastic-cyclic material model
The von Mises yield condition (see p. 597, ref. KJB)
A flow rule using the von Mises yield function
An isotropic and/or kinematic hardening rule. The isotropic

and kinematic hardening rules used in the plastic-cyclic model
are suitable for use in modeling cyclic plasticity.

The plastic-cyclic material model differs from the plastic-

bilinear and plastic-multilinear material models because the
isotropic and kinematic hardening rules are different. However, the
plastic-cyclic material model can reproduce the plastic-bilinear and
plastic-multilinear models when suitable material constants are
chosen.
This material model can be used with the truss, 2-D solid, 3-D
solid, Hermitian beam, shell and 3-D shell elements. The shell
elements must be MITC3, MITC4, MITC6, MITC9 or MITC16
single-layer shell elements.
This material model can be used with the small

large displacement/large strain formulations. Large
displacement/large strain kinematics can only be used with the 2-D
solid, 3-D solid, shell and 3D-shell elements.

3.4.5.1 Fundamental concepts
Many of the ideas used in the plastic-cyclic material model are

given in the following reference:
ref J. Lemaitre and J.-L. Chaboche, Mechanics of Solid

Materials, Cambridge University Press, 1990.
We abbreviate this reference as ref LC in the discussion below. As

an aid to understanding the model, whenever our notation differs
from the notation used in ref LC, we give the equivalent Lemaitre
and Chaboche notation.
The motivation for the plastic-cyclic material model is

illustrated in Figs. 3.4-8 and 3.4-9. Fig. 3.4-8 shows a bar subjected

to uniaxial cycling, with the strain prescribed. Fig. 3.4-9 shows

response predictions from the plastic-bilinear and plastic-
multilinear material models of Section 3.4.1. On repeated cyclic
loading, perfect plasticity and multilinear hardening plasticity
produce stabilized plastic cycles with no additional hardening.
Bilinear isotropic hardening does not produce a stable plastic cycle
and bilinear kinematic hardening produces a very rough
approximation to a stable plastic cycle.
Response predictions from the plastic-cyclic material model are

illustrated in Fig. 3.4-10. When nonlinear kinematic hardening is
used without isotropic hardening, a stable plastic cycle is reached
after one cycle. In this stable plastic cycle, the transition from
elastic to plastic conditions occurs more gradually than the
corresponding transition from bilinear kinematic hardening. Cyclic
hardening and cyclic softening are obtained by combining the
nonlinear kinematic hardening with isotropic hardening or
softening.
Strain
Time
a) Uniaxial cycling of a bar b) Prescribed strain time history
Figure 3.4-8: Uniaxial cycling example
The plastic-cyclic material model can optionally contain a strain

memory surface. The motivation for using the strain memory
surface is shown in Fig. 3.4-11. Considering an increase in the

prescribed strain amplitude, if no strain memory surface is used,

then no additional cyclic hardening takes place, whereas if a strain
memory surface is used, additional cyclic hardening takes place.
Stress
Stress
Strain Strain
a) Perfect plasticity b) Bilinear isotropic hardening
Stress
Stress
Strain Strain
c) Bilinear kinematic hardening d) Multilinear isotropic hardening
Stress
Strain
e) Multilinear kinematic hardening
Figure 3.4-9: Response predictions using the plastic-bilinear and plastic-multilinear models

Stress Stress
Strain Strain
a) Nonlinear kinematic hardening, b) Nonlinear kinematic hardening,

no isotropic hardening or softening isotropic hardening
Stress
Strain
c) Nonlinear kinematic hardening,

isotropic softening
Figure 3.4-10: Response predictions using the plastic-cyclic material model

Strain
Time
a) Prescribed strain time history

Additional
Stress Stress cyclic
hardening
Strain Strain
b) Stress-strain curve without c) Stress-strain curve with

strain memory surface strain memory surface
Figure 3.4-11: Response predictions with and without

strain memory surface
Ratchetting occurs when prescribed stresses with non-zero mean

stress are applied to the bar, as shown in Fig. 3.4-12.

Stress Stress
Strain
Time
a) Prescribed stress time history b) Stress-strain curve
Figure 3.4-12: Ratchetting using the plastic-cyclic material model
Stresses, strains, stress-strain law:
t +t
=t +t s + t +t m I
where t +t is the stress tensor ( in ref LC), t +t

s is the deviatoric
stress tensor ( in ref LC) and t +t
m is the mean stress
(hydrostatic stress H in ref LC).
t +t
e= t +t
e + t +t em I
where t +t e is the strain tensor ( in ref LC), t +t

e is the
deviatoric strain tensor ( in ref LC) and t +t
em is the mean strain
(hydrostatic strain H in ref LC).
t +t
m = 3 t +t em
t +t
s = 2G ( t +t e t +t e P )
where and G are the bulk modulus and shear modulus, and
t +t P
e is the plastic strain ( P in ref LC). Thermal strains are not
included in the above equations, but are included in ADINA when
there are thermal effects.

Yield condition: The von Mises yield condition is
1 2 1
t +t
fy = t +t
s t +t t +t
y2 = 0
2 3
where t +t
is the back stress tensor ( X and X in ref LC, note
that back stress is always deviatoric) and t +t
y is the yield stress
( Y + R or k in ref LC). The norm of a symmetric tensor a is
defined as a = a : a . The yield condition is always evaluated
at time t + t .
Figure 3.4-13 shows the yield condition.
s3
t+t
fy=0
t+t
s
t+t

2 t+t
y
3
s1 s2
Figure 3.4-13: von Mises yield surface in principal

deviatoric stress space

Flow rule: The flow rule states that the direction of plastic strain
increments is normal to the yield surface.
directions of e P = directions of ( t + t
s t + t
)
where e P is the increment in plastic strain, and is a constant

used to choose the yield surface configuration ( = 0 corresponds
to the configuration at time t , = 1 corresponds to the
configuration at time t + t , other values of correspond to
intermediate configurations). (The concept is similar to that used
in alpha-integration; we would have used instead of , except
that is used for the back stresses.) The directions of a
a
symmetric tensor a are defined as directions of a = .
a
Figure 3.4-14 shows the evolution of the yield surface using the
stress-strain law, yield condition and flow rule.

s3
s t+t
ts s
t+t

t
2 t+t
y
2 t 3
3 y
s1 s2
a) Overview
eP parallel to n
2Ge 2Ge
ts -2GeP ts
n -2GeP
t+t t+t
s s
Yield surface Yield surface Yield surface Yield surface

at time t at time t + t at time t at time t + t
Yield surface
at time t + t/2
b) =1, eP normal to yield c) =1/2, eP normal to yield

surface at time t + t surface at time t + t/2
Figure 3.4-14: Incremental update of yield surface and plastic strains
Strain memory surface: The memory-exponential isotropic

hardening rule uses the concept of a strain memory surface with
additional internal variables t +t and t +t q . The strain memory
surface is now briefly described.

The strain memory surface is defined in the space of plastic strains

as the surface of a sphere centered at position t +t with radius
3 t +t t +t
q . (In ref LC, is written ). This surface can be
2
written as
2 t +t 2
fm = e P t +t t +t q 2 = 0
3
Figure 3.4-15 shows the strain memory surface.
e3P
t+t
fm=0
t+t
eP
t+t

3 t+tq
2
e1P
e2P
Figure 3.4-15: Strain memory surface in principal

plastic strain space
The differential rules used for the evolution of the strain memory
surface are
dq = (directions of de P ) : (directions of ( t e P t )) de P
and
directions of d = directions of ( t e P t )
in which the symbol <> means that <u>=0 when u<0, and <u>=u
when u>0. The incremental versions of these rules are

q = (directions of e P ) : (directions of ( t + t e P t + t )) e P
and
directions of = directions of ( t + t e P t + t )
Again, the notation is used to denote a configuration between t

and t + t . These concepts are illustrated in Fig. 3.4-16.
n
t t+t
eP e P eP
t+t

t
3 t+tq
2
3 tq
2 parallel to n
Figure 3.4-16: Incremental update of strain memory

surface, for =1
1
is a material constant that must be between 0 and . Typically
2
1
= .
2
These rules ensure that the strain memory surface at time t + t

encloses the strain memory surface for all preceding times. The
evolution of the strain memory surface in 1D uniaxial straining is
shown in Figure 3.4-17.

Shaded area is area

enclosed by strain
q , 1, e1P memory surface
q
e1P
1
Time
Figure 3.4-17: Evolution of strain memory surface in

uniaxial cyclic straining
Isotropic hardening rules: The isotropic hardening rules are

shown in Figure 3.4-18.

y q=q2
q=q1 R
Memory-exponential
Q(q2 ) hardening
Q(q1)
0
y
Q Exponential hardening
0
y
Multilinear hardening
(ePi , yi)
Bilinear hardening
0
y EP
eP
Figure 3.4-18: Dependence of stress radius on accumulated

effective plastic strain
Bilinear hardening:
t +t
y = 0 y + EP t +t e P
where 0 y and EP are material constants, and t +t

e P is the
t +t
accumulated effective plastic strain ( p in ref LC). e P is
t +t 2 P
calculated using e P = t e P + e P , where e P = e : e P
3
.
EET
For bilinear hardening, EP = where ET is the slope of the
E ET
stress-strain curve during plasticity (this formula assumes no
kinematic hardening).

Multilinear hardening:
Pairs of (e P , y ) are given. It is allowed for y to either increase

or decrease as e P increases.
If the multilinear hardening curve is given in terms of the uniaxial

response (e, ) where is the tensile stress corresponding to the
unaxial strain e , then
P
e = e , y = are the corresponding points for the points
E
(e, ) . This formula assumes no kinematic hardening.
Exponential hardening:
y = 0 y + Q(1 exp(b t +t e P ))
t +t
where Q and b are material constants. Q can be positive to

model cyclic hardening, and Q can be negative to model cyclic
softening.
Memory-exponential hardening (exponential hardening with strain

memory surface):
In this model, the yield surface size t +t

y depends on the strain
t +t
memory size q (see above for a description of the strain
memory surface).
The yield surface size is
y = 0 y + t +t R
t +t
where t +t R is the change in the size of the yield surface. t +t

R is
defined using the differential equation
dR = b( tQ t R)de P

t +t
and is calculated using R = t R + R ,
R = b( t +t Q t +t R)e P . t +tQ is the asymptotic change in
yield surface size, and is calculated as
t +t
Q = Q( t +t q ) = QM (QM Q0 ) exp(2 t +t q )
t +t
where q is the size of the strain memory surface.
The material constants for the memory-exponential material model

are 0 y , b , Q0 , QM , b , and the strain memory surface
parameter .
Kinematic hardening rule: The kinematic hardening rule is:
Armstrong-Fredrick nonlinear kinematic hardening:
The back stress is expressed as a sum of partial back stresses
t +t
= t +t ( m )
where t +t ( m ) is partial back stress number m. All of the partial

back stresses are independent of each other.
Each partial back stress evolves according to the differential rule
2 (m) P
d ( m ) = h de ( m ) t ( m ) de P
3
where h ( m ) and ( m ) are material constants (these material

constants are Cl and l for partial back stress number l in ref LC).
Assuming that the directions of plastic strain increments are
constant during a time step, this can be integrated to obtain
t +t
( m ) = A ( m ) ( A ( m ) t ( m ) ) exp(b( m ) )

2 h( m )
where A (m)
= (directions of e P ) , b( m ) = ( m ) e P .
3 (m)
It is allowed to use one partial back stress, with = 0 . Then linear

kinematic hardening is recovered, with E p = h (this formula
assumes no isotropic hardening).
Stress plastic strain curve for uniaxial cycling
In uniaxial cycling, a typical stress plastic strain curve is shown

in Figure 3.4-19. Here only one partial back stress is used with
material constants h and . Since a stable cycle is considered
here, y is taken from the isotropic hardening rule assuming a
very large value of e P . The hardening modulus is given by
EP = h ( y ) on segment A-B. The size of the plastic
strain cycle can be related to the material constants using the
relation
h e P
= tanh + y
2 2
Thus, given several cycles for different cyclic strains, the material
constants can be estimated using the above formula.

A EP 2 y

eP
eP
Figure 3.4-19: Stable uniaxial plastic cycle using one

term in Armstrong-Fredrick nonlinear
kinematic hardening rule
Stress plastic strain curve for initial loading
During initial loading, it can be shown that
=y +
h

(1 exp ( e ))
P
The corresponding stress plastic strain curve is shown in Fig. 3.4-

20. Here only one partial back stress is used with material constants
h and . In this formula, it is assumed that there is no isotropic
hardening, and that the yield stress is y . Hence the hardening
observed during initial loading can be modeled using nonlinear
kinematic hardening, without the use of isotropic hardening.

h
y
eP
Figure 3.4-20: Stress - plastic strain curve during initial loading,

using nonlinear kinematic hardening
Combination of isotropic and kinematic hardening rules: By

combining the Armstrong-Fredrick nonlinear kinematic hardening
rule with the isotropic hardening rules, a wide variety of cyclic
phenomena can be simulated, such as cyclic hardening and
softening, shakedown and ratcheting. See ref LC for details.
Stress-strain integration: When plasticity is detected, the

incremental plastic strains are solved for using an iterative solution
procedure. The maximum number of iterations can be specified,
and the tolerance used in assessing convergence can be specified.
Constitutive tensor: The constitutive tensor (stress-strain matrix)

is constructed to be tangent. This gives good convergence in the
equilibrium iterations. However, in general, the tangent
constitutive tensor is non-symmetric. Therefore it is allowed to use
the non-symmetric equation solver when the plastic-cyclic model is
used. It is also allowed to use a symmetric equation solver; in this
case, the constitutive tensor is symmetrized, and this will reduce
the rate of convergence (increase the number of equilibrium
iterations).
Formulations: When used with the small displacement/small strain

when used with the large displacement/small strain formulation, a
TL is employed, and when used with the large displacement/large
strain formulation, either a ULH or ULJ formulation is employed.

Mixed displacement-pressure formulation: Plane strain,

axisymmetric and 3-D elements that reference these material
models should also employ the mixed displacement-pressure (u/p)
element formulation. This is because the plastic strains are
incompressible. The u/p formulation is the default for these
elements.
Thermal strains: If a thermal load is applied to the structure, the

thermal strains are taken into account but the material
characteristics are considered to be temperature independent.
Modeling of rupture: Rupture conditions can also be included

with this model.
The maximum accumulated effective plastic strain can be specified

for the rupture condition. When rupture is reached at a given
element integration point, the corresponding element is removed
from the model (see Section 11.5).
The option of user-supplied rupture is not available for the plastic-

cyclic material.
Rate-dependent plasticity: The option of rate-dependent plasticity

is not available for the plastic-cyclic model.
3.4.5.2 Specification of input
MATERIAL PLASTIC-CYCLIC: The basic command is
MATERIAL PLASTIC-CYCLIC NAME E NU DENSITY,

ALPHA PLCYCL-ISOTROPIC PLCYCL-KINEMATIC,
PLCYCL-RUPTURE BETA MAXITE RTOL
This command references data defined in commands PLCYCL-

ISOTROPIC, PLCYCL-KINEMATIC, PLCYCL-RUPTURE.
These commands provide the input for the isotropic, kinematic and
rupture parts of the plastic-cyclic model. Each of the isotropic,
kinematic and rupture options given above has a corresponding
PLCYCL-ISOTROPIC, PLCYCL-KINEMATIC and PLCYCL-
RUPTURE command, see below for a detailed description of these
commands.

It is allowed to not specify a value for PLCYCL-KINEMATIC or

PLCYCL-RUPTURE. Then these effects are not included in the
model.
BETA is the integration factor described above. When

BETA=AUTOMATIC, ADINA chooses the value of as
follows: =1 for static or implicit time integration; =0 for
explicit time integration. MAXITE is the maximum number of
iterations used to solve for the incremental plastic strains. RTOL is
a tolerance used to assess convergence of the iterations. RTOL can
be thought of as a reference incremental plastic strain.
PLCYCL-ISOTROPIC: The commands
PLCYCL-ISOTROPIC BILINEAR NAME YIELD EP
PLCYCL-ISOTROPIC MULTILINEAR NAME

aepsi stress-radiusi
PLCYCL-ISOTROPIC EXPONENTIAL NAME YIELD Q B
PLCYCL-ISOTROPIC MEMORY-EXPONENTIAL NAME,

YIELD Q0 QM MU B ETA
are used to specify the constants for the isotropic part of the plastic-
cyclic model.
PLCYCL-KINEMATIC: The command
PLCYCL-KINEMATIC ARMSTRONG-FREDRICK NAME

hi zetai
is used to specify the constants for the kinematic part of the plastic-
cyclic model.
PLCYCL-RUPTURE: The command
PLCYCL-RUPTURE AEPS NAME VALUE
is used to specify the constants for the rupture part of the plastic-

cyclic model. Rupture is based on the accumulated effective

plastic strain.
Input examples:
1) The simplest material that can be defined using this command

is given by a command sequence such as
PLCYCL-ISOTROPIC BILINEAR 1 YIELD=2E8

MATERIAL PLASTIC-CYCLIC 1 E=2.07E11 NU=0.3,
DENSITY=7800 PLCYCL-ISOTROPIC=1
This material is perfectly plastic. This material description

is equivalent to
MATERIAL PLASTIC-BILINEAR 1 E=2.07E11 NU=0.3,

DENSITY=7800 YIELD=2E8
2) The PLASTIC-CYCLIC material can be used to model bilinear

isotropic hardening using a command sequence such as
PLCYCL-ISOTROPIC BILINEAR 1 YIELD=2E8,

EP=2.090909E09
and this material description is equivalent to

DENSITY=7800 YIELD=2E8 ET=2.07E09
3) The plastic-cyclic material can be used to model multilinear

isotropic hardening using a command sequence such as
PLCYCL-ISOTROPIC MULTILINEAR 1
0 2.0E8
1E-3 2.5E8
2E-3 2.7E8

MATERIAL PLASTIC-MULTILINEAR 1 E=2.07E11,

NU=0.3
9.6618E-4 2.0E8
2.2077E-3 2.5E8
3.3043E-3 2.7E8
4) The plastic-cyclic material can be used to model bilinear

kinematic hardening using a command sequence such as
PLCYCL-ISOTROPIC BILINEAR 1 YIELD=2E8

PLCYCL-KINEMATIC ARMSTRONG-FREDRICK 1
2.090909E9
DENSITY=7800 PLCYCL-ISOTROPIC=1,
PLCYCL-KINEMATIC=1

DENSITY=7800 YIELD=2E8 ET=2.07E09,
HARDENING=KINEMATIC
Conversion formulas
Plastic-bilinear to plastic-cyclic:
In general, given a plastic-bilinear material of the form
MATERIAL PLASTIC-BILINEAR ...,

HARDENING=ISOTROPIC,
E=E YIELD=YIELD ET=ET
this material can be converted into an equivalent plastic-cyclic

material as follows:
PLCYCL-ISOTROPIC BILINEAR 1 YIELD=YIELD EP=EP

MATERIAL PLASTIC-CYCLIC... E=E... ,
PLCYCL-ISOTROPIC=1
E ET
using the formula EP = .
E - ET

Plastic-multilinear to plastic-cyclic:
In general, given a plastic-multilinear material of the form
MATERIAL PLASTIC-MULTILINEAR ...,

HARDENING=ISOTROPIC,
E=E
strain1 stress1
strain2 stress2
...
straini stressi
this material can be converted into an equivalent plastic-cyclic

material as follows:
PLCYCL-ISOTROPIC MULTILINEAR 1
aeps1 stress-radius1
aeps2 stress-radius2
...
aepsi stress-radiusi
MATERIAL PLASTIC-CYCLIC ... E=E ...,
PLCYCL-ISOTROPIC=1
using the formulas
aepsi = strain i (stressi / E)

stress-radiusi = stressi
Note that aeps1 will always be zero using these formulas.
3.4.5.3 Output variables
The material model outputs stresses, strains, plastic strains,

accumulated effective plastic strain t +t e P , yield stress t +t y ,
3 t +t t +t
stress radius (defined as s ) and dissipated plastic
2
t +t
work (defined as w p =
0
: de P ).

When there are thermal effects, thermal strains are also output.
When the strain memory surface is used, the strain memory size
t +t
q is also output.
3.5 Thermo-elastic material models
The thermo-isotropic and thermo-orthotropic material models
are used to calculate stress distributions due to imposed
temperatures. For thermal stress analysis, the material moduli can
be constant or temperature-dependent.
The thermo-isotropic model is available for the truss, 2-D solid,

3-D solid, iso-beam, shell and pipe elements.
The thermo-orthotropic model is available for the 2-D solid,

3-D solid and shell elements.
Both models can be used with the small displacement and

large displacement formulations. In all cases the strains are
assumed to be small.
When used with the small displacement formulation, a
materially-nonlinear-only formulation is employed, and when used
with the large displacement formulation, either the TL or UL
formulation is employed.
In the data input for the analysis, the nodal point temperatures
must be defined for all time steps (calculated by the ADINA heat
flow solution option or reading these temperatures from a file
created by ADINA-T, reading these temperatures from a mapping
file or defining them by time functions). If the model includes
axisymmetric shell or shell elements, the nodal point temperature
gradients must also be defined for all time steps. See Section 5.6
for more information on prescribing temperatures and temperature
gradients.
For these models, the elastic moduli, the shear moduli, the
Poisson's ratios and the coefficients of thermal expansion are input
as piecewise linear functions of the temperature, as illustrated in
Fig. 3.5-1. Linear interpolation is used to calculate the material
properties between input points.

3.5: Thermo-elastic material models
For calculation of thermal strains, please refer to Section 3.1.4.
E
l l
t
t
E l l
l l l l
1 t 2 3 4 1 t 2 3 4
Temperature Temperature

l
t
l
l l
1 t 2 3 4
Temperature
Figure 3.5-1: Variation of material properties for thermo-elastic

model
For the evaluation of the temperatures and 0 at the

t
integration point considered, the isoparametric interpolation

functions hi are used; e.g., in two-dimensional analysis we have
q
= hi i
i =1
where i is the temperature at element nodal point i (see Fig.

3.5-2). Note that when higher-order elements are used the
temperatures at the integration point can be significantly different
from the values at the nodal point (for example negative although
all nodal point temperatures are greater than or equal to zero).

s
l 1 Temperature t 1
l
l
l
l
r
l
Integration points
l
Figure 3.5-2: Interpolation of temperature at integration points
For axisymmetric shell and shell elements, the temperature at an

integration point is established using the temperatures and
temperature gradients at the midsurface nodes. Note that the
temperature gradient at a shell midsurface node is defined in the
direction of the mid-surface director vector at that node, see
Section 2.7.3.
For transition shell elements, no temperature gradients are
associated with the top/bottom nodes and only the temperatures at
the top/bottom nodes are used to evaluate the temperatures at the
integration points.
3.6 Thermo-elasto-plasticity and creep material models

This section describes the thermo-plastic, creep, plastic-creep,
multilinear-plastic-creep, creep-variable, plastic-creep-
variable, multilinear-plastic-creep-variable and irradiation
creep material models.
ref. KJB
The thermo-elasto-plasticity and creep models include the
Section 6.6.3 effects of
Thermal strains, t ersTH
Time-independent plastic strains, t ersp
Time-dependent creep strains, t ersC
and the constitutive relation used is

3.6: Thermo-elasto-plasticity and creep material models
t
ij = t Cijrs
E
( t ers t ersP t ersC t ersTH )
where t ij is the stress tensor at time t and t Cijrs
E
is the elasticity
tensor at the temperature corresponding to time t. The tensor t Cijrs
E
can be expressed in terms of Young's modulus tE and Poisson's

ratio tv which are both temperature-dependent.
The irradiation creep model includes the effects of
Thermal strains, ersTH
Irradiation strains, eW
rs
Neutron fluence-dependent creep strains, ersC
and the constitutive relation used is

ij = Cijrs
E
( ers ersC ersW ersTH )
where ij is the stress tensor corresponding to the neutron fluence
at time t and Cijrs
E
is the elasticity tensor corresponding to the
neutron fluence at time t. The tensor Cijrs
E
can be expressed in
terms of Young's modulus E which also corresponds to the
neutron fluence at time t.
These material models can be used with the truss, 2-D solid,
3-D solid, iso-beam, shell and pipe elements, with the exception of
the creep-irradiation model which can only be used with 2-D solid
and 3-D solid elements.
These models can be used with the small displacement/small

strain, large displacement/small strain and large displacement/
large strain formulations. The large displacement/large strain
formulation can only be used with the 2-D solid and 3-D solid
elements.
when used with the large displacement/small strain formulation,

either a TL or a UL formulation is employed and when used with

the large displacement/large strain formulation, the ULH


these material models should also employ the mixed displacement-
pressure (u/p) element formulation. The u/p formulation is the
default for these elements.
Note that the constitutive relations for the thermal, plastic and
creep strains are independent of each other; hence the only
interaction between the strains comes from the fact that all strains
affect the stresses. Figure 3.6-1 summarizes the constitutive
description for a one-dimensional stress situation and a bilinear
stress-strain curve.
Rate-dependent plasticity described in Section 3.4 can also be

applied to thermoplasticity models with isotropic hardening
(bilinear or multilinear).
Since there is no direct coupling in the evaluation of the

different strain components, we can discuss the calculation of each
strain component independently.
ref. M.D. Snyder and K.J. Bathe, "A Solution Procedure for
Thermo-Elastic-Plastic and Creep Problems," J. Nuclear
Eng. and Design, Vol. 64, pp. 49-80, 1981.

Algorithm for Thermo-Elasto-Plasticity and Creep," Int.
J. Numer. Meth. Engng., Vol. 24, No. 8, pp. 1509-1532,
1987.

Rupture of the material based on the following three-invariant

criterion can be specified:
J 3 ( t ) + J1 ( t ) + (1 ) J 2 ( t ) u
in which t is the stress tensor, J1 , J 2 , J 3 are the first, second

and third stress tensor invariants, , are material properties
(input data) and u is the ultimate effective stress (input data).
This rupture criteria can be used in combination with other
rupture criteria described later in this section.
When rupture is reached at a given element integration point,
the corresponding element is removed from the model (see Section
11.5).

Area A
l l
tP
Temperature t
(a) Model problem of truss element under constant load
Stress
t
t (t )
T( )
tE t
yv
tE(t)
Strain
teP (time independent)
teE (time independent)
Creep
strain
teC (time dependent)
t Time
(b) Strains considered in the model
Figure 3.6-1: Thermo-elasto-plasticity and creep constitutive

description in one-dimensional analysis
3.6.1 Evaluation of thermal strains
The thermal strains are calculated from the prescribed nodal

point temperatures with the coefficient of thermal expansion being
temperature dependent. The procedure used is the same as for the
thermo-isotropic material model, see Section 3.5.

The thermal strains of the irradiation creep model are dependent

on temperature and neutron fluence. The value of thermal strain is
determined using the Shifting rule described in Section 3.6.5.
3.6.2 Evaluation of plastic strains
Plasticity effects are included in the thermo-plastic, plastic-

creep, multilinear-plastic-creep, plastic-creep-variable and
multilinear-plastic-creep-variable material models. For the material
models that have the word "multilinear" in their names, the stress-
strain curve is multilinear, otherwise the stress-strain curve is
bilinear.
The plastic strains are calculated using the von Mises plasticity
model (see Section 3.4.1) with temperature-dependent material
parameters (Young's modulus, strain hardening modulus, Poisson's
ratio, yield stress, etc...). Figure 3.6-2 shows the assumptions used
for the dependence of the material parameters on temperature (in
the case of a bilinear stress-strain law). (Note that the plastic
modulus
EP = E ET /(E- ET)
is interpolated as a function of temperature, not the tangent

modulus ET .) Hence, the yield function is, for isotropic hardening
1t t 1t 2
t
fy = s s yv
2 3
and for kinematic hardening
1 t 1
t
fy =
2
( s t ) ( t s t ) t y2v
3
where t s is the deviatoric stress tensor, t yv is the virgin yield
t
stress corresponding to temperature (see Fig. 3.6-2) and t is
the shift of the stress tensor due to kinematic hardening. In the case
of multilinear stress-strain curves, the yield curves are interpolated
during the solution as shown in Fig. 3.6-3.

The expressions for plastic strain increments resulting from the

flow theory are deijP = d t sij for isotropic hardening and
deijP = d ( t sij t ij ) for kinematic hardening, in which d is the
plastic multiplier (positive scalar) which can be determined from
the yield condition tfy = 0. In the case of kinematic hardening, we
express the change of the yield surface position in the form
d ij = t CdeijP
where tC is the modulus
2 t E t ET
t
C= t
3 E t ET
In the case of multilinear yield curves, tET represents the tangent

modulus of the segment on the yield curve corresponding to the
accumulated effective plastic strain t e P .
Care should be exercised in the use of this model in kinematic

hardening conditions. Namely, nonphysical effects can result when
the variation in ET as a function of temperature is large. Hence, it
is recommended to use this model only when the variation in ET as
a function of temperature is small.
Rupture due to plasticity can be modeled using this material

model. When using a bilinear stress-strain curve, a maximum
allowable effective plastic strain eAP can be input. When using a
multilinear stress-strain curve, the maximum allowable effective
plastic strain at a given temperature corresponds to the last point on
the yield curve at that temperature (see Fig. 3.6-3).

1
Stress
ET 1
yv1
2
E1
yv2 ET2 ..
.. .
. E2
yvi i
ETi
Ei
Strain
(a) Uniaxial relation of stress versus total strain
Stress 1
E1 ET1
yv1
E1 - ET 2
1
..
yv2 E2 ET2 .
.. E2 - ET
. 2 i
yvi Ei ETi
Ei - ET
i
Plastic strain
(b) Uniaxial relation of stress versus plastic strain
Figure 3.6-2: Variation of material properties with temperature

for the thermo-elasto-plasticity and creep models;
bilinear stress-strain curves

Stress 3 4
2
1
i
e 1i e i2 ei3 e4i Strain
3 4
1 2
i+1
3
1
e i+1 2
e i+1 e i+1 e 4i+1
Temperature
a) Stress-strain curves input data
Stress 4
3 i
2
Interpolated
yield curve 1
t
4 Plastic
3
1 2 i+1 strain
P2 P3
e i+1 e i+1
Temperature
b) Yield curves
Figure 3.6-3: Interpolation of multilinear yield curves

with temperature

3.6.3 Evaluation of creep strains
Creep, plastic-creep, multilinear-plastic-creep material models
The effective creep strain is calculated using one of the

following creep laws:
Power creep law (creep law 1) :
t
e C = a0 t a1 t a2
in which a0, a1, a2 are material constants.
Exponential creep law (creep law 2) :
t
e C = F (1 e Rt ) + Gt
with
a4
a1 t t t
F = a0 e ; R = a2 ; G = a5 e a6
a3
in which a0 to a6 are material constants.
Eight-parameter creep law (creep law 3) ) :
t
e C = S T e H
with
a7
S = a0 t a1 ; T = t a2 + a3 t a4 + a5 t a6 , H = t

in which a0 to a7 are material constants; note that
t t
In the above equations, t e C , and denote the effective
creep strain, effective stress and temperature at time t. The unit
of the temperature can either be K or C (default). ADINA
automatically applies the conversion to K in the default setting.

LUBBY2 creep law: The LUBBY2 creep law can be employed

to predict long-term behavior of rock salt mass.
The LUBBY2 creep law is implemented for 2-D solid and
3-D solid elements. It can be used in conjunction with strain-
hardening or time-hardening.
The LUBBY2 creep law is specified by the following

equations:
d te C 1 GK ( t ) 1 t
= exp t +
dt K ( t ) K ( t ) M ( t )

or
1 GK ( t ) t 1 t
t
e =
C
exp t + +
GK ( t ) K ( t ) M ( t ) GK ( t )

( ) = e m , GK ( t ) = G*K e k1
t t

where M t *
M
and
K ( t ) = * e k
t

K
2
. Also, as usual, t e C is the effective creep
d te C
strain, is the effective creep strain rate and t is the
dt
effective creep stress.
In these equations, m, k1, k2, M* , G *K and K* are six
material parameters. m, k1, k2 account for the stress effect on
parameters M* , G *K , K* . The six material parameters are
identified in ADINA as a0, a1, a2, a3, a4 and a5 respectively.
Unloading and cyclic loading effects are considered in the
LUBBY2 creep law.
Analytical solutions with the LUBBY2 creep law are very
sensitive to the three parameters m, k1, k2, so these parameters
must be very carefully specified.
For more information about the LUBBY2 creep law, see the

ref. K.H. Lux and S. Heusermann, Creep Tests on Rock

Salt with Changing Load as a Basis for the Verification
of Theoretical Material Laws, Proc. 6th Symp. on Salt,
Toronto, Vol. 1, pp. 417-435, 1983.
Blackburn creep law:
t
e C = C1 {1 exp( r1t )} + C2 {1 exp( r2t )} + Gt
with
A0G A1 A G A3
C1 = , C2 = 2
r1 r2
r1 = A4t R A5 , r2 = A6t R A7 , G = Ft R A8
log10 (t R ) = B0 + B1 log10 t + B2 ( log10 t ) + A9

2
A11
F = A10 exp t
+ 273.16
A12 A14 A15
B0 = t + A13 , B1 = t , B2 = t
+ 273.16 + 273.16 + 273.16
A0 , A1 , A2 ,..., A15 are material constants. The unit of
temperature should be Celsius.
Creep-variable, plastic-creep-variable, multilinear-plastic-

creep-variable material models
The following creep laws can be employed:
Eight-parameter creep law (creep law 3) with variable

coefficients: The same functional form as in creep law 3
described above is employed, but coefficients a0 to a7 are
considered to be functions of temperature and effective stress.
The values of a0 to a7 can also be evaluated within user-
supplied subroutines. In the user-supplied subroutines, these
parameters can be functions of the temperature and the effective
stress.

The user-supplied subroutines are:
UCOEF2 for 2-D solid elements (file ovl30u.f)

UCOEF3 for 3-D solid elements (file ovl40u.f)
UCOEFB for iso-beam elements (file ovl60u.f)
UCOEFS for shell elements (file ovl100u.f)
UCOEFP for pipe elements (file ovl110u.f)
These subroutines are called by ADINA every time the stress-strain

law is updated.
Six-parameter LUBBY2 creep law with variable coefficients: The

same functional form as in the LUBBY2 creep law described above
is employed, but the six parameters are considered to be functions
of the temperature (but not the stress, because the stress effects on
parameters M , GK , K have already been taken into account
using the three parameters m, k1, k2). The six parameters can also
be evaluated using the user-supplied subroutines mentioned above.
Irradiation creep model
The irradiation creep strain is calculated from
a

e C = 1 exp ( a1 ) 2 + M ,
E0
M = a3 exp ( a4 + a5 )
where is the neutron fluence at time t, a1 ,, a5 are material

constants, is the Kelvin temperature at time t, and E0 is the
initial Youngs modulus. The subsequent Youngs modulus can be
dependent on neutron fluence and temperature. In this case, its
value is determined by using the Shifting rule described in Section
3.6.5.

All creep material models
The creep strains are evaluated using the strain hardening

procedure for load and temperature variations, and the O.R.N.L.
rules for cyclic loading conditions.
ref. C.E. Pugh, J.M. Corum, K.C. Liu and W.L. Greenstreet,
"Currently Recommended Constitutive Equations for
Inelastic Design of FFTF Components," Report No. TM-
3602, Oak Ridge National Laboratory, Oak Ridge,
Tennessee, 1972.
The procedure used to evaluate the incremental creep strains is

summarized in the following: Given the total creep strains t eijC and
t +t
the deviatoric stresses sij ,
1) Calculate the effective stress
1
3 2
t +t
= t +t sij t +t sij
2
2) Calculate the pseudo-effective creep strain
1
2 2
t
e = ( t eijC eijorig )( t eijC eijorig )
C
3a) For creep law 1 (and also for creep law 3 with constant
coefficients when the constants correspond to creep law 1),
calculate the effective creep strain and effective creep strain rate
at time t + t using
( eC ) = (teC ) + ( a0 t +t a1 )
1/ a2 1/ a2 1/ a2
t +t
t ,
t +t
t +t e C t e C
eC =
t

3b) For other creep laws (including creep law 3 with variable
coefficients), calculate the pseudo-time t satisfying
e C ( t +t , t +t , t ) =t e C + e C ( t +t , t +t , t ) t
where e C ( t +t
, t +t , t ) is the generalized uniaxial creep law
d e C ( t +t , t +t , t ) . Then
and e C
( t +t
, t +t
, t ) =
dt
calculate the effective creep strain and effective creep strain rate
at time t + t using
t +t
e C = e C ( t +t , t +t , t ) , t +t
e C = e C ( t +t , t +t , t ) .
4) Calculate t +t
using
t +t
3 eC
t +t
=
2 t +t

5) Calculate the incremental creep strains using
eijC = t t +t
t +t
sij
The use of the pseudo-time in step 3b corresponds to a strain

hardening procedure. There is also the option of using time
hardening in step 3b, in which t = t . See ref. KJB, pp 607-608 for
a discussion of the two hardening procedures. The strain hardening
procedure is recommended for general use (and is the default).
The above procedure also assumes that the time integration
parameter is equal to 1 (the default). There is also the option of
using a different value of .
Rupture of the material due to excessive creep strain can be

specified. In this case, a uniaxial creep strain failure envelope must
be defined as a function of temperature and effective stress (Fig.
3.6-4).
The following criterion is then checked at each integration point

during the analysis:

euC,r
t
e C
TF
in which e C
u ,r is the uniaxial creep strain at failure, t e C is the
effective creep strain and TF is the triaxiality factor.
t
m
You can choose to use the triaxiality factor TF = (the
t

default) or to use no triaxiality factor (that is, TF = 1 ).
t
m = t 11 + t 22 + t 33 is the volumetric stress.
Uniaxial creep strain cu,r(31) cu,r(41)

at rupture cu,r(2)1
cu,r(1)1 cu,r
Effective
stress

Temperature

Figure 3.6-4: Multilinear interpolation of uniaxial creep strain
at rupture
You can choose to apply the rupture model only under tensile
conditions ( t m > 0 ) or under both tensile and compressive
conditions. The default is to apply the rupture model only under
tensile conditions.
Rupture can be allowed only in tension, based on the sign of
t
m .

You can apply a time offset to any of the creep material models
(for example, creep law 1 becomes t e C = a0 t a 1 ( t t0 ) 2 , where
a
t0 is the time offset, which can be positive or negative). The time

offset can be modified in a restart run.
The accumulated effective creep strain is calculated as
t
eC =
all solution steps
e C
2 C
where e C = e eC and eC is the tensor of creep strain
3
increments in a solution step. Because of the summation over the
solution steps, we refer to the calculated value of t e C as the
accumulated effective creep strain.
3.6.4 Computational procedures
ref. KJB The stresses at the integration points are evaluated using the
Section 6.6.3 effective-stress-function algorithm.

Algorithm for Thermo-Elasto-Plasticity and Creep," Int.
J. Numer. Meth. Engng., Vol. 24, No. 8, pp. 1509-1532,
1987.
Briefly, the procedure used consists of the following calculations.

The general constitutive equation
t +t
(i ) = t +t C E ( t +t e(i ) t +t e P (i ) t +t eC (i ) t +t eTH ) (3.6-1)
is solved separately for the mean stress and for the deviatoric
stresses. In this equation the index (i) denotes the iteration counter
in the iteration for nodal point equilibrium. For easier writing this
index will be dropped in the discussion to follow. The mean stress
is calculated as

t +t
E
t +t
m =
1 2 t +t

( t +t
em t +t eTH ) (3.6-2)
The deviatoric stresses t +t s depend on the inelastic strains and

they can be expressed as
1
t +t
s= t +t e (1 ) t t s (3.6-3)
t +t
aE + t +
t +t
t +t E t
where aE = , s = deviatoric stress at the start of the
t +t
1+
time step and is the integration parameter used for stress
evaluation ( 0 < 1) The creep and plastic multipliers and
are functions of the effective stress t +t
only, and they
account for creep and plasticity; also
t +t
e = t +t
e t e P t eC
is known since the deviatoric strains t +t e , plastic strains t e P and

creep strains t eC are known from the current displacements and the
stress/strain state at the start of the current time step.
The following scalar function f ( t +t

) is obtained from Eq.
(3.6-3)
f ( t +t ) = a 2 t +t 2 + b c 2 2 d 2 = 0 (3.6-4)
The zero of equation (3.6-4) provides the solution for the effective
stress t +t , where
a = t +t aE + t +

b = 3 (1 )t t +t eij t sij
c = (1 ) t t
3 t +t
d2 = eij t +t
eij
2
with summation on the indices i, j.

Once the solution for t +t has been determined from Eq. (3.6-
4), simultaneously with the scalars and from the creep and
plasticity conditions, the deviatoric stress t +t s is calculated from
Eq. (3.6-3), and the plastic and creep strains at the end of the time
step are obtained as
e = t e P + t +t s
t +t P
e = t eC + (1 ) t s +
t +t C t +t
s t
The above equations correspond to isotropic hardening

conditions and a general 3-D analysis. The solution details for
kinematic hardening conditions and for special problems (for the
plane stress, shell, isoparametric beam elements) are given in the
above cited reference and also in the following reference.
ref. M. Koji and K.J. Bathe, "Thermo-Elastic-Plastic and

Creep Analysis of Shell Structures", Computers &
Structures, Vol. 26, No 1/2, pp. 135-143, 1987.
3.6.5 Shifting rule for the irradiation creep model
Irradiation tests obtaining irradiation strain data and material

property changes of graphite have been carried out generally at a
constant temperature. They cannot be determined directly from the
test data due to variation of temperature.
Therefore, the irradiation strain at temperature and under fast
neutron fluence is calculated as follows

n
W ( , ) = W ( i , i 1 , i i 1 )
i =1
where
W (i , , ) = W ( i , + ) W ( i , )
This is illustrated in Figure 3.6-1, and documented in the

following reference.
ref. T. Iyoku and M. Ishihara, "Development of

Thermal/Irradiation Stress Analytical Code VIENUS for
HTTR Graphite Block", Journal of Nuclear Science and
Technology, 28[10], pp. 921-931, October 1991.
0 1 2 3
(a)
(b) (a) irradiated from zero to

1 1
under temperature 1
(c)
W (b) irradiated from 1 to 2
under temperature 2
(c) irradiated from 2 to 3
under temperature 3
2
Figure 3.6-1: Shifting rule for irradiation creep model
3.7 Concrete material models

The concrete material models in ADINA are the concrete and
data fitted concrete material models.
The concrete model can be employed with the 2-D solid and
3-D solid elements, and composite shell elements.
The concrete model can be used with the small displacement

and large displacement formulations.When used with the small

displacement formulation, a materially-nonlinear-only (MNO)
formulation is employed, and when used with the large
displacement formulation, a total Lagrangian (TL) formulation is
employed.
These material models are supported in explicit and implicit

analyses.
These material models allow three types of temperature effects:
no temperature effects;
isotropic temperature independent effects; and
isotropic temperature dependent effects.
Reinforcement bars can be included in the concrete model using

rebar elements. Rebar elements are truss elements that are
automatically connected to any 2-D or 3-D solid mesh in which
truss elements lie. The AUI does this connection using constraint
equations during data file generation. For more information on
rebar elements, see Section 2.1.7.
Rebar elements of Section 2.1.7 can be incorporated into 2-D or

3-D solid elements for modeling as concrete structures. The
reinforcements in the concrete shell can employ various material
models, such as the elastic-orthotropic, plastic-orthotropic models,
into the shell layers.
3.7.1 Concrete material model
Although the model is entitled "concrete model", the basic

constitutive characteristics are such that the model can also be
useful when representing other materials. The basic material
characteristics are:
Tensile cracking failure at a maximum, relatively small

principal tensile stress
Compression crushing failure at high compression

3.7: Concrete material models
Strain softening from compression crushing failure to an

ultimate strain, at which the material totally fails.
The tensile cracking and compression crushing failures are

governed by tensile failure and compression crushing failure
envelopes.
These material characteristics pertain, for example, to a variety
of rocks.
It is well accepted that concrete is a very complex material. The

model provided in ADINA may not contain all the detailed material
characteristics that you may look for. However, considering the
variability of concrete materials that need to be described in
practice, and recognizing that the model may also be useful in the
modeling of rock materials, the objective is to provide an effective
model with sufficient flexibility to model most of the commonly
used material behaviors.
The notation used in the formulation of the model is given

below.
t
E = equivalent multiaxial tangent Young's modulus at
time t (the left superscript "t" refers to time t)
E0 = uniaxial initial tangent modulus (all uniaxial
quantities are identified with a tilde "~")
Es = uniaxial secant modulus corresponding to uniaxial
c
maximum stress, Es =
ec
Eu = uniaxial secant modulus corresponding to uniaxial
u
ultimate stress, Eu =
eu
t
E pi = uniaxial tangent modulus in the direction of t pi
t
eij = total strains
eij = incremental strains
t
e = uniaxial strain

ec = uniaxial strain corresponding to c ( ec < 0 )

eu = ultimate uniaxial compressive strain ( eu < 0 )
t
ij = total stresses
ij = incremental stresses
= uniaxial stress
t
t = uniaxial cut-off tensile stress ( t > 0 )

tp = post-cracking uniaxial cut-off tensile stress
(0 tp t ) . Note that if tp = 0 , ADINA sets
tp = t .
c = maximum uniaxial compressive stress ( c < 0 )
u = ultimate uniaxial compressive stress ( u < 0 )
t
pi = principal stress in direction i ( t p1 t p 2 t p 3 )
Three basic features are used in the concrete model:
A nonlinear stress-strain relation to allow for the weakening

of the material under increasing compressive stresses
Failure envelopes that define cracking failure in tension and

crushing in compression
A strategy to model the post-cracking and crushing behavior

of the material
Stress-strain relations: The general multiaxial stress-strain

relations are derived from a uniaxial stress-strain relation t
versus t e .
A typical uniaxial stress t to uniaxial strain t e relation is
shown in Fig. 3.7-1. This stress-strain relation shows that there are
three strain phases; namely, t e 0, 0 > t e ec , and ec > t e eu
where ec is the strain corresponding to the minimum (crushing)
stress c that can be reached, and eu is the ultimate compressive

strain.
If t e > 0 , the material is in tension and the stress-strain relation
is linear until tensile failure at the stress t . A constant Young's
modulus E0 is employed, i.e.,
t
= E0 t e (3.7-1)
For t e 0 , we assume the following relation
E0 t e

t
Es ec
=
c e
t
te
2
te
3
1+ A + B + C
ec ec ec
~

~
t
~p

e~u
t
e~c
~
et ~
e
~
u
~
c
Figure 3.7-1: Uniaxial stress-strain relation used in the concrete

model

and hence
te
2
te
3
E0 1 B 2C
ec ec
t
E= 2
(3.7-2)
te te
2
te
3
1 + A + B + C
ec ec ec
where
E0 2 E0
E + ( p 2 p ) E ( 2 p 3 p + 1)
3 3 2
A= u s
( p 2 p + 1) p
2

E
B = 2 0 3 2 A
Es
E
C = 2 0 + A
Es
c eu
and the parameters E0 , c , ec , Es = , u , eu , p = , Eu = u
ec ec eu
are obtained from uniaxial tests.
The stress-strain relation in Eq. (3.7-2) assumes monotonic
loading conditions. For unloading conditions and loading back to
the stress state from which unloading occurred, the initial Young's
modulus E0 is used. For strain states beyond eu in compression, it
is assumed that stresses are linearly released to zero, using the
following modulus:
u c
Eu =
eu ec
Note that confined concrete can therefore be modeled using close

values for u and c .
Under multiaxial stress conditions, the stress-strain relations are
evaluated differently depending on whether the material is loading

or unloading. Poisson's ratio is assumed to be constant under

tensile stress conditions and can vary in the compressive region.
To characterize loading and unloading conditions we define a
t
loading scalar g for each integration point,
t
g = t e (3.7-3)
t
where e is the effective stress at time t. The material is loading at
the integration point except when the unloading conditions are
determined,
t
g < gmax (3.7-4)
where gmax is the maximum value of the loading scalar that has
been reached during the complete solution.
During unloading, the material is assumed to be isotropic and
the initial Young's modulus, E0 , is used to form the incremental
stress-strain matrix, both for stiffness and stress calculations.
In loading conditions, the principal stresses t pi (with
t
p1 t p 2 t p 3 ) are calculated. For each principal stress
direction, a tangent Young's modulus t E pi corresponding to the
given strain state t e pi is evaluated using Eqs. (3.7-1) and (3.7-2).
In Eq. (3.7-2), the strains t e pi are used together with the
parameters c , u , ec and eu defined in Eq. (3.7-8) which
accounts for the multiaxial stress conditions.
The material is considered as orthotropic with the directions of
orthotropy being defined by the principal stress directions. Once
cracking occurs in any direction i, that direction is fixed from that
point onward in calculating t pi .
The stress-strain matrix corresponding to these directions is, for
three-dimensional stress conditions,

1
C=
(1 + )(1 2 )
(1 ) t E p1 t E12 t E13 0 0 0

(1 ) t E p 2 t E23 0 0 0
( ) p3
1 t
E 0 0 0

0.5 (1 2 ) t E12 0 0
symmetric 0.5 (1 2 ) t E13 0

0.5 (1 2 ) t E23
(3.7-5)
t
where v is the constant Poisson's ratio and the Eij with i j are
evaluated using
t
pi t E pi + t pj t E pj
t
Eij = (3.7-6)
t
pi + t pj
Note that the above stress-strain relations for material loading

conditions are only employed in the calculation of the stiffness
matrix at time t. Considering the evaluation of the stress increment
from t to time t + t, we have

= Ce (3.7-7)
is the one defined in Eq. (3.7-5)
where the stress-strain matrix C
but using the Young's moduli E pi . These moduli are evaluated
using three point Gaussian integration of E pi between t e pi and
t +t t +t
e pi , with t e pi and e pi the strain components measured in
the directions of the respective principal stresses t pi and t +t
pi .
Material failure envelopes: The failure envelopes shown in Figs.

3.7-2 to 3.7-5 are employed to establish the uniaxial stress-strain
law accounting for multiaxial stress conditions, and to identify
whether tensile or crushing failure of the material has occurred.

Uniaxial stres-strain law under multiaxial stress conditions:

Having established the principal stresses t pi with
t
p1 t p 2 t p 3 the stresses t p1 and t p 2 are held constant
and the minimum stress that would have to be reached in the third
principal direction to cause crushing of the material is calculated
using the failure envelope, see Fig. 3.7-5(a). Let this stress be c ,
we have
c
1 =
c
and
u = 1 u ; ec = ( C1 12 + C2 1 ) ec ;
(3.7-8)
eu = ( C1 12 + C2 1 ) eu
where C1 and C2 are input parameters. Normally, C1 = 1.4 and C2 =

-0.4. The constants c , u , ec and eu are employed instead of the
unprimed variables in order to establish, using Eq. (3.7-2), the
uniaxial stress-strain law under multiaxial conditions (see Fig.
3.7-5(b)).

~ , 0, 0)
( c
t
p3
~ ,0)
(0, c
t
p2
t
p1
~
~' =

t t t
~ = ~ (1-0.75
p2
t t ~ )
c

c
(1-0.75 t p3 )
=
t t
~c
~ = ~ (1-0.75 t p2 ) (1-0.75 p3 )
t
t t ~ ~
c c
~ = Uniaxial cut-off tensile stress under multiaxial conditions

~t = Uniaxial compressive failure stress under multiaxial conditions
c
Figure 3.7-2: Three-dimensional tensile failure envelope of

concrete model

t
p2
~
c
t
p21
~
c
ADINA
Experimental
Figure 3.7-3: Biaxial concrete compressive failure envelope
t
p3
~ t p2 t
c p3
t p2 t
p3
t t p1
p2
t i
p1
Curve i ~
t = constant
c
Input to ADINA
Curve 1
Experimental
t
p2
~
c
Figure 3.7-4: Triaxial compressive failure envelope

t
p3 t =t p3
~ p2
c t
p2 = p3
t
~
c
t
1 ~ p1
c ~ = constant
c
1 is determined from
failure envelope by
holding t p1 and t p2
constant.
t t
p2 p2
~ = constant
~
c c
~ from given (t , t )
(a) Determination of c p1 p2
~
t
e~u e~u e~c e~c
~ /
~ = t e~
c c 1
u /
~ ~ =
u 1
~e / e~ = C112+ C2 1
c c
~e / e~ = C112+ C2 1
u u
~
t
u
~
t
c
~
t
u
~
t
c
(b) Stress-strain behavior under multiaxial stress conditions
Figure 3.7-5: Definitions used for evaluation of one-dimensional

stress-strain law under multiaxial stress conditions

Tensile failure envelope: The tensile failure envelope used in the

concrete model is shown in Fig. 3.7-2. To identify whether the
material has failed, the principal stresses are used to locate the
current stress state. Note that the tensile strength of the material in
a principal direction does not depend on tensile stresses in the other
principal stress directions, but depends on compressive stresses in
the other directions.
Tensile failure occurs if the tensile stress in a principal stress
direction exceeds the tensile failure stress. In this case it is
assumed that a plane of failure develops perpendicular to the
corresponding principal stress direction. The effect of this material
failure is that the normal and shear stiffnesses and stresses across
the plane of failure are reduced and plane stress conditions are
assumed to exist at the plane of tensile failure, as discussed in more
detail below.
Prior to tensile failure the stress-strain material law is given by
Eqs. (3.7-5) to (3.7-7). Assuming that tp1 is larger than the tensile
failure stress, the new material stress-strain matrix used in the
stiffness matrix calculation is given by

E0n 0 0 0 0 0

1 E p 2 E23
t t

0 0 0

(1 2 ) t
E p 3

0 0 0

C=
E0 s
0 0
2 (1 + )

E0 s
symmetric 0
2 (1 + )

t
E23
2 (1 + )
(3.7-9)

where the t E pi are the uniaxial Young's moduli evaluated in the

principal stress directions using Eqs. (3.7-1) or (3.7-2), and the
t
Eij are evaluated using Eq. (3.7-6).
The constants n and s are the normal and shear stiffness
reduction factors, respectively. Typically, n = 0.0001 and s =
0.5. The factor n is not exactly equal to zero in order to avoid the
possibility of a singular stiffness matrix. The factor s depends on
a number of physical factors and you must use judgement to choose
its value. For the concrete model in ADINA, n and s are both
input parameters.
For the stress calculation, the following stress-strain matrices
are used:
For the tensile stress normal to the tensile failure plane and
the shear stresses in this plane, we use the total strains to
calculate the total stresses with
Ef 0 0

C= G12f 0 (3.7-10a)
sym. G13f
where Ef and G12f , G13f are evaluated as shown in Fig. 3.7-6. In

this figure, is a user input variable that defines the amount of
tension stiffening. Fig. 3.7-6(b) shows that Gijf is evaluated
from the initial shear modulus. Also, Fig. 3.7-6(a) assumes
loading from zero stress directly into the tensile region. If the
tensile stress is reached by unloading from a compressive stress,
the strain normal to the tensile failure plane is measured from
the specific strain value at which the stress is zero (see Fig. 3.7-
7 point 11).
Special care in the analysis must be taken if is chosen to
be greater than 1.0, because strain softening may yield non-
unique solutions.

To obtain a mesh independent solution, the fracture energy

Gf can be provided instead of . In doing so, is evaluated at
each integration point, based on the size of the finite elements
(see Fig. 3.7-6(c)). In this case, ADINA also calculates s
internally, overwriting the user input.
~

~
t
~
tp ~ ~t
etp = ~ p
E0
t
~
E0
1 Ef Strain normal
1 to tensile
failure plane
l
~
et te e~m (= ~etp) ~
e
a) Calculation of Young's modulus Ef, normal to tensile failure plane.

~ is the post-cracking uniaxial cut-off tensile stress.
tp
!f
~
~ E0
G ijf = ! f
1.0 2(1+)
!s Strain normal
to tensile
failure plane
~
et e~m ~
e
b) Calculation of shear modulus in tensile failure plane
Figure 3.7-6: Material moduli for stress calculation

after tensile failure

Element Integration point
Physical state with Model with

multiple cracks unique crack
~
2E G
x = ~ 20 f with Gf energy released per unit area in
s tp h2 direction perpendicular to h direction,
2
where h 2 is the maximum size of the
element dimension
c) Smeared crack approach for the calculation of the parameter x
For the remaining stress components, the increments are

evaluated from the strain increments using

E p 2 E23 0
1

C= 2
E p3 0 (3.7-10b)
1

sym. E23
(1 ) 2 (1 + )
2
where the E pi are the uniaxial Young's moduli used in Eq.

(3.7-7) and E23 is evaluated using Eq. (3.7-6) but with E pi
instead of t E pi .
Compressive failure envelope: The triaxial failure envelope used in

the concrete model is shown in Fig. 3.7-4. Note that the biaxial
failure envelope shown in Fig. 3.7-3 is curve 1 of the triaxial
failure envelope. Also, note that the biaxial and triaxial envelope

curves can be used to represent a large number of specific

envelopes by the input explained below, and this flexibility makes
it possible to model various concrete and rock materials.
The triaxial compression failure envelope is shown in Fig.
ip1
3.7-4. First, the values are input. These values define for
c
which levels of the t
p1 the two-dimensional failure envelopes
functions of stresses t p2 and t p3 are input.
Six failure envelopes must be defined, each with three points
corresponding to the locations t p2 = t p1 , t p2 = t p3 , and
t
p2 = t p3 where is an input parameter ( 0 1) .
To identify compression failure, the largest principal stress,
t
p1 , is employed to establish the biaxial failure envelope function
of t p2 and t p3 , using the interpolation shown in Fig. 3.7-5(a). If
the stress state corresponding to t p2 and t p3 lies on or outside
this biaxial failure envelope, then the material has crushed.
It should be noted that you must choose appropriate values for
the input of the failure surfaces and other parameters. These may
vary significantly for different materials and structures and must be
obtained from experimental data.
Material behavior after failure: The post failure material

behaviors considered in the concrete model in ADINA include the
post tensile cracking, post compression crushing, and strain-
softening behaviors.
Post tensile cracking behavior: Once a tensile plane of failure has

formed, it is checked in each subsequent solution step to see
whether the failure is still active. The failure is considered to be
inactive provided the normal strain across the plane becomes
negative and less than the strain at which the "last" failure
occurred. It is otherwise active. Therefore, a tensile failure plane
can repeatedly be active and inactive. Consider the uniaxial cyclic
loading case shown in Fig. 3.7-7 in which the strain is prescribed.
As the strain reaches the tensile cut-off limit, a tensile failure plane
becomes active. This tensile failure plane remains active while the

strain continues to increase to point 2 and then decreases to point 3

and then increases to point 4 and then decreases to point 5. The
failure plane becomes inactive after 5 and remains so while the
strain decreases to point 8 and increases again to point 11. At point
11, the tensile failure plane again becomes active, and it remains
active until the strain reaches point 13 and beyond.
Stress
1
l
l
2
3l 4
5l
l Strain
Stress
A
12 Strain
11 l l
Point numbers:
2-5,11-12 crack open l
6-10,13-20 crack closed l

l
>20 material crushed
l
l
8
l
20
Figure 3.7-7: Uniaxial cyclic loading stress/strain experiment

with the concrete model
If a tensile failure plane has developed, which may or may not

be active, the material stress-strain relations are always established
as described above, but corresponding to the directions along and
perpendicular to the plane of failure. Hence, instead of using the
principal stresses and corresponding directions as done for the
unfailed material, the stresses in the directions defined by the

material tensile failure plane are used to evaluate the stress-strain

matrix, corresponding to those directions. Once a failure plane has
been initiated, a subsequent failure plane is assumed to form
perpendicular to the direction of the first failure plane whenever a
normal stress along the original failure plane has reached the tensile
failure stress. It follows that at any integration point, the direction
of the third tensile failure plane is fixed once failure has occurred in
two directions.
Post compression crushing, strain-softening behavior: If the

material has crushed in compression, it is assumed that the material
strain-softens in all directions.
Consider first uniaxial stress conditions. As shown in Fig.
3.7-1, when the uniaxial strain is smaller than ec , the material has
crushed and softens with increasing compressive strain, i.e., t E is
negative.
Under multiaxial stress conditions the compression crushing is
identified using the multiaxial failure envelope, and once the
material has crushed, isotropic conditions are assumed. As in
uniaxial conditions, in the subsequent solution steps the Young's
modulus is assumed to be very small but positive in the stiffness
matrix calculations, but the stress increments are computed from
the uniaxial stress-strain law with the constants c , ec and so on,
see Eq. (3.7-8), corresponding to the multiaxial conditions at
crushing. The Young's modulus E corresponding to the current
strain increment ep3 is evaluated using the uniaxial stress-strain
relationship in Fig. 3.7-1

E=
t t
e p3 + e p3 e p3
e p3
where tep3 and ep3 are the strain component and incremental strain
components at time t measured in the direction of the principal
stress tp3. To obtain the stress increment, Eq. (3.7-7) is used where
the matrix C corresponds to isotropic material conditions with
Young's modulus E. Note that when tep3 becomes equal to or less
than eu , the stresses are linearly released to zero.

If unloading of the crushed material in the strain-softening

region occurs, characterized by ep3 0, the initial Young's modulus
E0 is used. At any time during post-crushing calculations, if any
one of the principal stresses checked individually reaches a positive
value, this stress is set to zero.
Temperature effects: In some analyses, thermal strains t eTH need

to be included. These strains are taken into account by replacing
the total incremental strain e in the governing incremental
equations with the strain e - eTH, where eTH is the thermal
incremental strain and is calculated from the temperature
conditions.
The following material properties can be defined as
temperature-dependent:
Uniaxial initial tangent modulus E0

Poisson's ratio
Mean coefficient of thermal expansion
Uniaxial cut-off tensile stress t
Post-cracking uniaxial cut-off tensile stress tp
Uniaxial maximum compressive stress c
Uniaxial compressive strain at c , ec
Uniaxial ultimate compressive stress u
Uniaxial strain at u , eu
Fracture energy Gf
Constant for tensile strain failure
The nodal point temperatures are input to ADINA as discussed

in Section 5.6.
Poisson's ratio in the compressive region: It has been observed in

experiments that the ratio of lateral strain to principal compressive
strain remains constant until approximately 80% of the maximum
compressive stress c .
Usually, we assume that the Poisson ratio is constant throughout
the analysis; however, as an option, the value vs can be used when

the material dilates under compression. The value of vs is given by
p3
t
s = when 2 = a
c
2
2 a
s = f ( f ) 1 1 when 2 > a
a
where v is the initial Poisson's ratio, vf = 0.42 is the maximum

Poisson's ratio at failure and a = 0.7.
3.7.2 Data fitted concrete material model
The data fitted concrete material model is, in essence, an

empirical material model that identifies concrete as a brittle
material, which behaves nonlinearly in compression, and that it can
be solely described by its uniaxial cylinder compressive strength
(i.e., the maximum strength that a cylinder specimen of concrete
can sustain when it is subjected to uniaxial compressive loading).
The model is based on the work of M.D. Kotsovos and other

researchers, and is described in the following references:
ref. M. D. Kotsovos and M. N. Pavlovi, Structural

Concrete: Finite-element analysis for limit-state design,
Thomas Telford, London, 1995.
ref. M. D. Kotsovos and K. V. Spiliopoulos, Modeling of

crack closure for finite-element analysis of structural
concrete, Computers & Structures, 69:383398, 1998.
The experimental data reported by Kotsovos et al. indicate that

one can attribute the following material characteristics to plain
concrete:
can be considered an isotropic material with a nonlinear

behavior in compression followed by a brittle post-peak
behavior, which is characterized by a complete and immediate

loss of load-carrying capacity after the ultimate strength is

attained;
cannot be considered to be a continuum medium beyond its

peak load level since Poisson's ratio values greater than 0.5
were measured near and around the peak load level;
microcracking fracture processes define the compressive

behavior of concrete up to failure (in which the microcracks
extend in the direction of the maximum compressive principal
stress), while macrocracking describes the fracture processes
occurring during failure (in which crack planes form such that
they are orthogonal to the maximum tensile principal stress
existing just prior to cracking);
no appreciable stress-strain hysteresis loop occurs during

unloading and reloading;
failure can be captured in the stress space via a failure

surface and it usually occurs first in all possible triaxial states of
stress except the fully compressive triaxial stress state;
proper characterization of concrete behavior requires the

description of general triaxial stress states, which are
conveniently interpreted by the octahedral stresses;
stress path independence characterizes the experimental data

that define the stress-strain curves as well as the failure surface;
all material constants can be related to the uniaxial

compressive strength of a cylinder specimen; and
the overall mechanical properties are independent of the

loading rate.
In consideration of the previous premise, the data fitted concrete

material model consists of
(i) a nonlinear experiment based stress-strain law in

compression that is combined with a linear behavior when
concrete unloads/reloads and is in tension (see Fig 3.7-8);

(ii) an experiment based stress failure surface function that

dictates when either cracking or crushing occurs in concrete (see
Fig. 3.7-9); and
(iii) a post-failure response for cracking based on a smeared

crack approach that allows cracks to close and reopen (see
Figures 3.7-10 to 3.7-12).
Figure 3.7-8: Stress-strain behavior of the data fitted concrete

material model

Figure 3.7-9: Failure surface of the data fitted concrete material

model

Figure 3.7-10: First crack formation of the data fitted concrete

material model (a) before, and (b) after
Figure 3.7-11: Second crack formation of the data fitted concrete


Figure 3.7-12: Third crack formation of the data fitted concrete

The experimental data reported by Kotsovos et al. is presented

in a compact way in terms of the octahedral normal and shear
stresses ( o , o ) , their corresponding octahedral strains ( o , o ) ,
and the uniaxial cylinder compressive strength (fc) material
constant. Note that the octahedral normal and shear stresses
possess a geometric interpretation in stress space similar to that of
the first stress invariant and second deviatoric stress invariant (I1,
J2), respectively.
Recall that any stress state is a point in the stress space that can
be represented by the principal stresses ( 1 , 2 , 3 ) or,
equivalently, by the stress invariants (I1, J2,, J3). The latter is
composed of
I 1 = ii = 3 m first stress invariant,
1
J2 = sij sij second deviatoric stress invariant, and
2
J 3 = det sij third deviatoric stress invariant,

where ij and sij are the stress and deviatoric stress tensors
expressed in index notation, respectively, and m is the mean
stress. Note that these stress quantities are related via
ij = sij + m ij
where ij is the Kronecker delta. A much more convenient and

geometrically intuitive way to express a stress state is by (I1, J2,, )
or ( o , o , ) , where
1 3 3J
= arccos 3

2 ( J )2
3
3
2
is the angle of similarity or Lode angle defined on a deviatoric

stress plane (i.e., any plane orthogonal to the hydrostatic axis) such
that 0 3 , and the octahedral normal and shear stresses are
defined as
1 2
o = I1 = m and o = J2
3 3
respectively. The combination of octahedral stresses and angle of

similarity ( o , o , ) for expressing any stress state is heavily used
in research that reports experimental data of concrete.
Similar expressions to the ones presented for the octahedral

stresses can be derived for the volumetric strain ( o ) and the
deviatoric strain ( o ) .
The stress-strain relations that describe the nonlinear

deformational behavior of concrete are derived from curves that
best fit the experimental data portrayed in o o , o o and

o o plots. Due to the nature of the data, the relations to be

presented are only valid when the stress units employed are N/mm2
or MPa. It is important to highlight that the experimental data
indicate that when subjecting concrete to hydrostatic pressure ( o )
the deformation consists of mainly volumetric strain ( o ) , while
when subjecting concrete to deviatoric stresses ( o ) , after
applying a certain level of hydrostatic stress, the deformation
consists of both deviatoric and volumetric strains ( o , o ) . The
empirical stress-strain relations capture these phenomena and at
any given time are defined by
o = 3K s ( o TH ) id
and
o = 2Gs o ,
where TH is the thermal strain defined as in Section 3.1.4, Ks and

Gs are are the secant bulk and shear moduli, respectively, and id is
termed internal hydrostatic stress due to an external deviatoric
stress. The secant bulk and shear moduli are defined as
Kc o
for 2
fc
1 + A o
fc
Ks =
Kc o
1
for >2
o fc
1 + 2 Ab + 2 ( b 1) + A
b 1 b
fc
and
Gc
Gs = d 1
,

1+ C o
fc

respectively, and
n
k fc o
id = m .
o f c
1+ l
fc
The following material parameters are used: Ke, Ge, A, b, C, d, k, l,

m and n. These parameters are functions fc and have been derived
for values of fc in the range of 1665 N/mm2. Outside of this
range, all of these parameters remain constant and equal to their
values for either fc = 16 N/mm2 or fc = 65 N/mm2. The expressions
of these material parameters are
K e = 11000 + 3.2 f c2 ,
Ge = 9224 + 136 f c + 3296 1015 f c8.273 ,
0.516 for f c 31.7 N/mm 2 ,

A= 0.516
1 + 0.0027 ( f 31.7 )2.397 for f c > 31.7 N/mm 2 ,
c
b = 2 + 1.81 108 f c4.461 ,
3.573 for f c 31.7 N/mm 2 ,

C= 3.573
1 + 0.0134 ( f 31.7 )1.414 for f c > 31.7 N/mm 2 ,
c
2.12 + 0.0183 f c for f c 31.7 N/mm 2 ,

d =
2.7 for f c > 31.7 N/mm 2 ,
4
k= ,
1 + 1.087 ( f c 15 )
0.23

2.415 for f c 31.7 N/mm 2 ,

m=
3.531 + 0.0352 f c for f c > 31.7 N/mm 2 ,
1 for f c 31.7 N/mm 2 ,

n=
0.3124 + 0.0217 f c for f c > 31.7 N/mm 2 .
As one can clearly notice, these expressions are unit dependent.

Furthermore, Ke and Ge, are the elastic bulk and shear moduli,
N/mm2, while the remaining material parameters do not have units
since they are either coefficients or exponents.
The stress-strain relations that describe concrete when it unloads

from a given compressive stress level, reloads back to such a stress
level, and is in tension are defined by
o = 3K e ( o TH ) and o = 2Ge o .
These expressions are the linear elastic constitutive equations

written in terms of the octahedral stresses and strains, and governed
by the corresponding elastic moduli of concrete.
The criterion that identifies nonlinear compressive loading from

the other possible linear loading states is dictated by comparing the
octahedral shear stress between two time steps such that nonlinear
loading occurs when t +t o > t omax , where t omax is the maximum
octahedral shear stress (during nonlinear loading) up to time t.
In some analyses the temperature effects can be significant and,

therefore, the thermal strains need to be included. These strains
( ) are taken into account by assuming that the material is
TH
either an isotropic temperature independent or dependent material

(see Section 3.1.4). If the former is assumed, then the thermal
strains are calculated following Eq. (3.1-16), while if the latter
applies, then the thermal strains are given by Eq. (3.1-17).
When the material is isotropic temperature dependent, the
following material properties can be defined as temperature
dependent: uniaxial cylinder compressive strength (fc), mean

coefficient of thermal expansion () and fracture energy (see the

description of the tension softening option below). Note that when
temperature effects are included, ADINA will require temperature
data as input (see Section 5.6).
The AUI command that defines the data fitted concrete material
model in ADINA is MATERIAL DF-CONCRETE. In order to
adapt any model to the above constitutive equations, this command
provides the parameter FC, which defines the uniaxial cylinder
compressive strength of the user's concrete, and the parameter
CONFAC, which converts FC from the user's stress unit to N/mm2
such that 16 N/mm2 FC x CONFAC} 65 N/mm2.
When temperature effects are desired, the parameter ALPHA,
which defines the mean coefficient of thermal expansion, must be
specified.
The failure surface that defines the ultimate strength values of

concrete is derived from curves that best fit the experimental data
portrayed in o f c o f c plots. These data is usually presented
on meridian planes (i.e., any plane that contains the hydrostatic axis
and is defined by a constant angle of similarity) as well as on
deviatoric planes (i.e., any plane that is perpendicular to the
hydrostatic axis and is defined by a constant octahedral normal
stress). Again, due to the nature of the data, most of the relations
that follow are only valid when the stress units employed are
N/mm2 or MPa. The failure surface function employed by the data
fitted concrete material model is given by
0.724
0.944 f c o
g =o TP SP f c = 0,
r fc
where r is an elliptic function, TP is a percentage of fc that

represents the amount of tensile hydrostatic stress allowed by g
such that 0.05 TP 0.1, and SP is a percentage of fc that
represents the amount of additional octahedral shear stress allowed
by g such that 0.0 SP 0.05. Note that if g > 0, then the stress
state of concrete is outside the failure surface and it indicates that
failure has occurred in the form of a crack or crush. The elliptic
function is defined as

4 (1 e 2 ) cos 2 + ( 2e 1)
2
r= ,
2 (1 e 2 ) cos + ( 2e 1) 4 (1 e 2 ) cos 2 + 5e 2 4e
where 0 3 and e denotes the eccentricity of the failure

surface on a deviatoric plane. The data fitted concrete material
model defines the eccentricity as
0.133

e = 0.670551 TP o ,
fo
which, in essence, is the ratio of radial distances to the cross section

of the failure surface (on a deviatoric plane) at = 0 and = 3 .
A description of the elliptic function can be found in
ref. Ph. Mentrey and K. J. Willam, Triaxial Failure

Criterion for Concrete and Its Generalization ACI
Structural Journal, 92(3), pp. 311-318, 1995.
Basically, the elliptic function defines the cross section of the

failure surface that best fits the experimental data. Moreover, this
failure surface portrays a somewhat triangular cross section for
tensile and small compressive octahedral normal stresses, which
tends to change to a somewhat circular cross section for larger
compressive octahedral normal stresses (see Fig. 3.7-9).
It is important to highlight that both TP and SP can be user

input and are parameters of the AUI command MATERIAL DF-
CONCRETE. Basically, they provide some flexibility to the
failure surface by stretching it in different directions. The former
parameter stretches it along the tensile side of the hydrostatic
axis, while the latter dilates it on the deviatoric plane.
The smeared crack approach, which governs part of the post-

failure behavior of the data fitted concrete material model follows a
non-orthogonal fixed crack type of modeling, that is, one that
defines the crack planes of an integration point when they first

occur such that they stay valid for the remaining of the simulation
and where the crack planes are not necessarily orthogonal to each
other. The modeling of a crack starts after cracking is detected by
the failure surface function and hinges on the idea that concrete
retains its strength only on the defined crack planes, which
indicates a brittle type of cracking. Consequently, the stress-strain
relations for cracking are understood the best when they are
presented in the local coordinate system (i.e., the coordinate system
of the crack of interest) and always need to be transformed to the
global coordinate system. Following an incremental formulation,
these are
(i ) = C(i ) (i ) ,
where ( i ) , C( i ) and ( i ) denote the stress increment,

constitutive matrix and engineering strain increment in the
coordinate system corresponding to the i-th crack, respectively.
An integration point of a 3-D solid element can take up to three

cracks. The first crack plane is defined to be orthogonal to the
direction of the maximum tensile principal stress existing prior to
failure. For notation purposes, the first axis of the coordinate
system of the first crack is assumed to be aligned with the direction
of such a stress. After failure, this stress is set to zero (see Figure
3.7-10) and the corresponding constitutive matrix is defined as
STIFAC M 0 0 0 0 0

0 M N 0 0 0
0 N M 0 0 0
C(1) =
0 0 0 SHEFAC G 0 0
0 0 0 0 SHEFAC G 0

0 0 0 0 0 G
where STIFAC and SHEFAC are the normal and shear stiffness
reduction factors, respectively, such that 0.0 < STIFAC < 1.0 and
0.0 < SHEFAC < 1.0, and

4 2
M = K t + Gt , N = K t Gt and G = Gt ,
3 3
where Kt and Gt are the tangent bulk and shear moduli prior to
failure, respectively. When the state prior to failure is considered
to be a nonlinear compressive state, these moduli are defined as
Ke o
b 1
for 2
fc
1 + Ab o
Kt =
fc

Ke o
for >2
1 + 2b 1 Ab fc
and
Ge
Gt = d 1
,

1 + Cd o
fc
otherwise they are defined as their elastic moduli counterparts.

Note that the material parameters and terms of the last two relations
have been defined previously.
An integration point presenting a single crack can in a future
time step remain in the one-crack state, which indicates that the
crack stays open, or fail again, that is, either follow the constitutive
relations of the second crack (see next bullet) or crush, or return to
its previous state, which indicates that the crack closes.
A second crack forms when under a one-crack state, failure is

again detected by the failure surface function such that
there is at least one tensile principal stress. The second crack plane
is defined to be orthogonal to the direction of the current maximum
tensile principal stress. Note that the first and second crack planes
need not be perpendicular to each other and this fact conveys that
concrete retains its strength only in the direction defined by the
intersection of these two crack planes. For notation purposes, the
first axis of the coordinate system of the second crack is assumed to
be aligned with the direction of the current maximum tensile

principal stress and the second axis is aligned with the intersection
of the two crack planes. After failure, the stresses along the first
and third axes are set to zero (see Fig. 3.7-11) and the
corresponding constitutive matrix is defined as
STIFAC M 0 0 0 0 0

0 M 0 0 0 0
0 0 STIFAC M 0 0 0
C(2) =
0 0 0 SHEFAC G 0 0
0 0 0 0 SHEFAC G 0

0 0 0 0 0 SHEFAC G
An integration point presenting two cracks can in a future time step

remain in the two-crack state, which indicates that both cracks stay
open, or fail again, that is, either follow the constitutive relations of
the third crack (see next bullet) or crush, or return to the one-crack
state, which indicates that the second crack closes while the first
crack remains open. Note that in the last case, once in the one-crack
state, the integration point can return to its state prior to cracking
within the same time step, which indicates that both cracks close.
A third and last possible crack forms when under a two-crack

state failure is detected one more time by the failure surface
function such that the stress along the intersection of the two
previous crack planes is tensile. This is a fully cracked state that
resembles an almost complete loss of load-carrying capacity. Note
that the coordinate system of the third crack is the same as that of
the second crack, and after failure the stresses along the
axes of such coordinate system are set to zero (see Fig. 3.7-12) and
the corresponding constitutive matrix is defined as

C(3) =
STIFAC M 0 0 0 0 0

0 STIFAC M 0 0 0 0
0 0 STIFAC M 0 0 0

0 0 0 SHEFAC G 0 0
0 0 0 0 SHEFAC G 0

0 0 0 0 0 SHEFAC G
An integration point presenting three cracks can in a future time

step remain in the fully cracked state, which indicates that all three
cracks stay open, or return to the two-crack state, which indicates
that the third crack closes while the first and second crack stay
open. An integration point that is fully cracked cannot fail
again. Moreover, given the right strain conditions (see next bullet)
such an integration point can return within the same time step to the
one-crack state or even to its state prior to cracking.
The criterion that identifies when a crack can be closed is

dictated by comparing strain values along the direction of the stress
that caused the formation of the crack. For the one- and two-crack
cases, the direction is perpendicular to their corresponding crack
planes, while for the fully cracked state it is along the line that
defines the intersection of these two crack planes. In order to close
a crack, the current strain value must be compressive and smaller
than the strain value present when the crack was formed.
An integration point presenting a closed crack can in a future
time step remain with the closed crack or fail again. When failure
is detected by the failure surface function, the corresponding
integration point can either reopen the closed crack or crush. Note
that cracks are assumed to be history dependent and, therefore, a
reopened crack will have the orientation defined when it was
initially opened..
Note that both STIFAC and SHEFAC can be user input and are
parameters of the AUI command MATERIAL DF-CONCRETE.
The values assigned to these factors are constant for the entire
duration of the simulation and must be chosen carefully by the
user.

In essence, these factors dictate how the concrete's stiffness

degrades during cracking. On the one hand they intend to capture
what occurs during cracking, while on the other hand they help
avoid possible numerical difficulties associated with the
degradation of the stiffness matrix (to a singular matrix), especially
when a large number of cracks start forming. The normal stiffness
reduction factor must reflect the brittle nature of cracking, that is,
the loss of strength outside the crack plane. It is recommended that
STIFAC be a very small (positive) value, i.e., STIFAC 1.0.
The shear stiffness reduction factor is related to the notion that
aggregate interlock occurs on the formed crack planes due to the
roughness present on the separating surfaces of a crack. An
adequate value for it must neither be too big that creates an
unrealistically large stiffness nor too small that cause numerical
instability. It is recommended that SHEFAC be within 0.1
SHEFAC 0.5.
The other possible post-failure behavior of the data fitted

concrete material model is crushing, which portrays the brittle
nature of concrete in compression. After failure is detected by the
failure surface function, crushing occurs when all the principal
stresses are compressive. Crushing is modeled as a complete and
permanent loss of load-carrying capacity and, therefore, all the
stress components are set to zero (see Fig. 3.7-13). For numerical
reasons, the corresponding constitutive matrix is defined as the one
for the fully cracked state but in the global coordinate system.
Figure 3.7-13: Crush modeling of the data fitted concrete material

model (a) before, and (b) after
An optional feature of the data fitted concrete material model is

the inclusion of tension softening after cracking. The tension
softening behavior is assumed to follow a linear stress-strain curve
(see Fig. 3.7-14) and, therefore, will replace the default brittle
cracking behavior. Tension softening is possible only for the one-
and two-crack states. In the one-crack state tension softening
applies to X(1) , while in the two-crack state both X(2) and Z(1)
will follow the tension softening behavior. In addition, the
corresponding components of the constitutive matrices C(1) and
C(2) of the one- and two-crack states, respectively, will be
modified to account for tension softening.
Figure 3.7-14: Stress-strain behavior of the data fitted concrete

material model
The tension softening behavior is defined by the fracture energy

value and the characteristic length of the concrete model so as to

ensure a numerical solution that is not dependent on the concrete's
finite element mesh. The former is defined as the area under the
tensile stress-displacement curve and via the parameter of the AUI
command MATERIAL DF-CONCRETE. The latter can be usually
taken as the length of the element along the direction perpendicular
to the first crack plane and is defined by the parameter ELEN
MATERIAL DF-CONCRETE. If the parameter ELEN is not input
by the user, then ADINA will define the characteristic length of the
model as the largest length of the element. However, this is not
always desirable and, therefore, it is recommended that the user
input a value for ELEN. Refer to Fig. 3.7-14 for a graphical
interpretation of the relationship between GF and ELEN.
Note that since tension softening replaces the default brittle
cracking behavior, the value of the parameter STIFAC is controlled
by ADINA.
The post-processing of the simulation results of the data fitted

concrete material model is done via the AUI. For each integration
point of an element, this material model outputs the following
quantities (where their corresponding AUI variable names are
enclosed in parentheses):
(STRESS-(XYZ));
engineering total strains (STRAIN-(XYZ));
thermal strain (THERMAL_STRAIN);
temperature (ELEMENT_TEMPERATURE);
failure surface function value

(FAILURE_SURFACE_FUNCTION);
data fitted concrete material state flag (DF-

CONCRETE_FLAG);
crack state flag (CRACK_FLAG);
normal directions of the crack planes (SIGMA-

P1_ANGLE and FE_SIGMA-P2_ANGLE for 2-D

solid elements, FE_SIGMA-P1_DIRECTION-(XYZ),

FE_SIGMA-P2_DIRECTION-(XYZ) and
FE_SIGMA-P3_DIRECTION-(XYZ)} for 3-D solid
elements);
stress values at the time of crack formation along the

normal directions (SIGMA-P1, FE_SIGMA-P2 and
FE_SIGMA-P3); and
strain values at the time of crack formation along the

normal directions (FE_STRAIN-P1, FE_STRAIN-P2
and FE_STRAIN-P3).
For additional simulation results available for post-processing, refer

to Section 13.1.1.
A useful way of post-processing the results associated with

cracking and crushing is by using the vector plot capabilities of the
AUI. The post-failure-related element vectors applicable to the
data fitted concrete material model, which are displayed with the
AUI command EVECTORPLOT, are:
OPEN_CRACKS;
CLOSED_CRACKS;
CRUSHED;
CRACKS;
CRACK_STRESS; and
CRACK_STRAIN.
Although the above variables have self-explanatory names, it

should be noted that CRACKS displays the open and closed crack
planes as well as the crushed data, CRACK_STRESS and
CRACK_STRAIN display the stress and strain values at the time of
crack formation (see previous bullet), respectively. For additional
information regarding vector plotting, refer to Section 5.5 of the
AUI Command Reference Manual, Volume IV: Display

Processing.
The following is a list of modeling hints and suggestions for

simulations of concrete structures that will use the data fitted
concrete material model.
1. The data fitted concrete material model is intended for

modeling reinforced concrete structures. Although unreinforced or
plain concrete structures can be modeled with this material model
by using the tension softening option, the aim of this model is to
simulate concrete structures that contain proper reinforcement in
the applications of monotonic and/or dynamic loading.
2. Use of the data fitted concrete material model is recommended

when the available material parameters of concrete cannot fully
describe complex stress-strain laws and/or failure surfaces. The
advantage of this particular concrete material model is that it is
fully described by a single material parameter, which is the uniaxial
cylinder compressive strength.
3. As stated above, the inclusion of proper reinforcement is highly

desirable in concrete structures that are simulated using the data
fitted concrete material model. The reason behind this statement is
because this model treats concrete as a brittle material when it
cracks and crushes. Consequently, plain concrete structures that
use the data fitted concrete material model with its default brittle
behavior will, in general, fail with somewhat smaller loads than if
they were to be simulated with a concrete material model that
includes, for example, softening.
4. The values of the parameters TP or SP can be increased to

overcome convergence difficulties. In particular, increase
TP for structures that show large tensile normal octahedral
stresses, e.g., reinforced concrete beams that tend to flex, and
increase SP for structures that show large shear octahedral
stresses, e.g., a structure lacking shear reinforcement.
5. The values of the parameters STIFAC or SHEFAC

can be modified to overcome convergence difficulties. In
particular, keep in mind that STIFAC 1.0 and 0.1 SHEFAC
0.5. The former parameter is intended to model brittle cracking,

while the latter models the aggregate interlock phenomenon due to

cracking.
In the case of plain concrete structures that are simulated with
the data fitted concrete material model, the above recommendation
for the parameter SHEFAC can provide undesirable results. It is
instead recommended that SHEFAC 1.0.
6. The geometry, type of loading, boundary conditions, mesh size

and pattern, and analysis type are other modeling criteria that must
be carefully selected by the user when simulating concrete
structures.
For instance, choosing between force or displacement loading
can affect how far a simulation can be accurately solved. Also, 8-,
20- or 27-node 3-D solid elements; or fine or coarse meshes can
affect the performance of a simulation. The choice of element type
usually depends on the type of deformation the element is expected
to sustain during the simulation. Regarding the mesh size, it is
usually recommended that it be neither too fine that creates
localized failure zones nor too coarse that gives unrealistic results.
The type of analysis used for simulating concrete structures can
affect the quality of the final results. This modeling criterion plays
an important role when a concrete structure starts presenting
convergence difficulties due to cracking and/or crushing, that is,
due to the degradation of the stiffness matrix of the concrete
structure. ADINA has a number of numerical methods that can
help obtain accurate results for nonlinear static analyses (see
Section 7.2). For example, in the case of a static analysis that
presents convergence difficulties, adding line search to improve the
iterative solution procedure, or changing the analysis to a low
speed dynamic analysis, or simply using the automatic-time-
stepping (ATS) method, or alternatively changing the analysis to
use the load-displacement-control (LDC) method can help
overcome such numerical problems. As a general rule of thumb, it
is recommended to use the ATS method with or without line search
to overcome the unavoidable numerical difficulties present in
concrete modeling.
3.7.3 Creep/Shrinkage effects
Creep/Shrinkage effects can be included in the DF-

concrete(data fitted concrete) material model.

Creep/Shrinkage models are adopted from ACI 209R-92 code

and CEB FIP 1990 code, see the following references:
ref. ACI 209R-92, 2008. Modeling and Calculating

Shrinkage and Creep in Hardened Concrete.
ref. CEP-FIP model code 1993. Design of Concrete

Structures.
ref. Bazant, Z.P. Criteria for Rational Prediction of Creep

and Shrinkage of Concrete, Revue Francaise de Genie
Civil 2 (3-4), 1999, 61-89.
The ACI209R-92 creep/shrinkage model is as follows:
Creep strain coefficient:

(t t0 )
(t , t0 ) = (2.35 c )
d + (t t0 )
C = C ,t C , RH C ,VS C , S C , C ,
0
C ,t = 1.25 t0 0.118 for moist curing

0
C ,t = 1.13 t0 0.094 for steam curing

0
t0 is the age of concrete at loading
C , RH = 1.27 0.67 h for h 0.4 , h is relative humidity
2
C ,VS = (1 + 1.13 e{0.0213(VS )} ) , in mm
3
2 {0.54(VS )}
= (1 + 1.13 e ) , in inch
3
VS is the volume-surface ratio

slump factor , C , S = 0.82 + 0.00264 s (mm)

= 0.82 + 0.067 s(inch)
fine aggregate factor , C , = 0.88 + 0.0024 (%)
air content factor , C , = 0.46 + 0.09 (%) 1
with d =10, and =0.6
Shrinkage strain:
(t ts )
sh (t , ts ) =
780 106 sh
f + (t ts )
sh = sh ,t sh, RH sh ,vs sh, s sh , sh,c sh ,

s
sh ,t = 1.202 0.2337 log(ts ) for moist curing

s
= 1 for steam curing
ts is the age of concrete at the beginning of shrinkage
sh , RH = 1.40 1.02 h for 0.4 h 0.8

= 3.00 3.0 h for 0.8 h 1
sh ,vs = 1.2 e{0.00472(VS )} , in mm

= 1.2 e{
0.12(VS )}
, in inch
slump factor , sh , s = 0.89 + 0.00161 s (mm)

= 0.89 + 0.041 s (inch)

fine aggregate factor , sh , = 0.30 + 0.014 (%), for 50

= 0.90 + 0.002 (%), for > 50
cement content factor , sh ,c = 0.75 + 0.00061 c(kg/m3 )

= 0.75 + 0.00036 c(lb/yd 3 )
air content factor , sh , = 0.95 + 0.008 (%) 1
The CEB-FIP 1990 creep/shrinkage model is as follows:
Creep strain coefficient:

0.3
1 (t t0 )
(t , t0 ) = h ( f c )
0.2
0.1 + (t0 ) h + (t t0 )
1 h /100
h = 1 +
0.1 3 L
16.8
( fc ) =
fc
h = 1.5 L 1 + (1.2 h /100)18 + 250

L = 2 Ac / u
t0 is the age of concrete at loading
h is the relative humidity of the environment (%)
Ac is the cross section area and u is the perimeter
f c is the mean compressive strength at 28 days
Shrinkage strain:
t ts
sh = [160 + 10 sc (9 f c /10)] 106 h [ ]0.5
350 ( L /100) + t ts
2

h = 1.55 [1. (h /100)3 ] for 40 h(%) < 99

= +0.25 for h(%) 99
sc = 4 for slowly hardening cements SL
= 5 for normal or rapid hardening cements N and R
= 8 for rapid hardening high strength cements RS
f c is the mean compressive of concrete at 28 days
ts is the concrete age at the beginning of shrinkage
The ADINA creep strain equation is
a a14 a15
a1 + t a2 a5 + t0 a6 a9 + (t t0 ) a10
13
=
cr
a12
E (t ) a3 + t a4 a7 + t0 a8 a11 + (t t0 )
E (t ) = a22 exp[a23 (1 (a24 / t ) a25 ]
The ADINA shrinkage strain equation is
a30
a + (t ts ) a 27
= 26
sh
a
a31
a28 + (t ts )
29
Calculation of the creep strain, using creep multiplier
t +t
t +t 3 cr
= t +t
2
ijcr = t t +t sij
t +t
Shrinkage strain is calculated simply as the thermal strain
The equation of total strain is
= co + th + cr + sh

where
co is the concrete strain

th is the thermal strain
cr is the creep strain
sh is the shrinkage strain
3.8 Rubber and foam material models

The rubber and foam material models in ADINA are the
Bathe and eight-chain material models.
These material models can be employed with the 2-D solid, 3-D
solid and 3D-shell elements.
These material models can be used with the large

displacement/large strain formulation. A TL formulation is
employed.
These material models include the following effects:
isotropic effects (required), see Section 3.8.1

viscoelastic effects using the Holzapfel model (optional), see
Section 3.8.2
viscoelastic effects using the Bergstrm-Boyce model
(optional), see Section 3.8.2
Mullins effects (optional), see Section 3.8.3
orthotropic effects (optional), see Section 3.8.4
thermal strain effects (optional), see Section 3.8.5
These material models allow three types of temperature

dependence:
no temperature dependence
time-temperature superposition for viscoelastic effects, see

Section 3.8.6. This type of material is referred to as a
thermorheologically simple material (TRS material).

full temperature dependence, see Section 3.8.7.
Table 3.8-1 shows a summary of the features.
Table 3.8-1: Summary of features for rubber and foam material models
Viscoelastic
Viscoelastic effects, Thermal
effects, Mullins Orthotropic
Bergstrm- 1 strain
Holzapfel effects effects
model1
Boyce effects
model1,2
No
temperature yes yes yes yes no
dependence
TRS
temperature yes no yes yes yes
dependence
Full
temperature yes yes yes yes yes
dependence
1) Viscoelastic effects and Mullins effects cannot be used together.

2) Orthotropic effects cannot be used with the Bergstrm-Boyce model
When the material temperature dependence is TRS or full, heat

generation from the rubber-like materials is included in TMC
(thermo-mechanical-coupling) analysis.
3.8.1 Isotropic hyperelastic effects
The rubber and foam material models include the following

models for isotropic hyperelastic effects:
Mooney-Rivlin
Ogden
Arruda-Boyce
hyper-foam
Sussman-Bathe

3.8: Rubber and foam material models
Eight-chain
Only the Mooney-Rivlin, Ogden, Arruda-Boyce, Sussman-

Bathe, and eight-chain material models are supported in explicit
dynamic analysis.
The isotropic hyperelastic effects are mathematically described

by specifying the dependence of the strain energy density (per unit
original volume) W on the Green-Lagrange strain tensor ij .
We now give a brief summary of the quantities and concepts

used. For more information, refer to ref KJB, section 6.6.2. Here
and below, we omit the usual left superscripts and subscripts for
ease of writing. Unless otherwise stated, all quantities are
evaluated at time t and referred to reference time 0 .
Useful quantities are the Cauchy-Green deformation tensor Cij ,

given by
Cij = 2 ij + ij (3.8-1)
where ij is the Kronecker delta; the principal invariants of the

Cauchy-Green deformation tensor,
1 2
I1 = Ckk , I2 =
2
( I1 Cij Cij ) , I 3 = det C (3.8-2a,b,c)
the reduced invariants

1 2 1

J1 = I1 I 3 3 , J 2 = I 2 I 3 3 , J 3 = I 3 2 , (3.8-3a,b,c)
the stretches i where the i s are the square roots of the principal
stretches of the Cauchy-Green deformation tensor; and the reduced
stretches
1
i* = i ( 12 3 )

3 (3.8-4)
Note that

J 3 = 12 3 (3.8-5)
is the volume ratio (ratio of the deformed volume to the

undeformed volume).
The strain energy density W is written either in terms of the

invariants or in terms of the stretches. In many cases, the strain
energy density is conveniently written as the sum of the deviatoric
strain energy density WD and the volumetric strain energy density
WV .
With knowledge of how the strain energy density W depends

on the Green-Lagrange strain tensor (through the invariants or
stretches), the 2nd Piola-Kirchhoff stress tensor is evaluated using
1 W W
Sij = + (3.8-6)
2 ij ji
and the incremental material tensor is evaluated using
1 Sij Sij
Cijrs = + (3.8-7)
2 rs sr
3.8.1.1 Mooney-Rivlin model
The Mooney-Rivlin material model is based on the following

expression:
WD = C1 ( I1 3) + C2 ( I 2 3) + C3 ( I1 3) + C4 ( I1 3)( I 2 3) +
2
C5 ( I 2 3) + C6 ( I1 3) + C7 ( I1 3) ( I 2 3) +
2 3 2
(
C8 ( I1 3)( I 2 3) + C9 ( I 2 3) + D1 exp ( D2 ( I1 3) ) 1
2 3
)
(3.8-8)
where C1 to C9 and D1 to D2 are material constants.

This strain energy density expression assumes a totally

incompressible material ( I 3 = 1) and is modified as explained
below for plane strain, axisymmetric or 3-D analysis.
Plane stress analysis In plane stress analysis, the material is

assumed to be totally incompressible. Therefore WV is zero and
W = WD . A displacement-based finite element formulation is
used, in which the incompressibility condition of the material is
imposed by calculating the appropriate thickness of the material.
ref. KJB
Section 6.6.2 Plane strain, axisymmetric and 3-D analysis: In plane strain,
axisymmetric and 3-D analysis, the material is modeled as
compressible (that is, the bulk modulus is not infinite), but you can
set the bulk modulus high so that the material is almost
incompressible.
The Mooney-Rivlin strain energy density equation is modified
by: 1) substituting for the invariants I1 , I 2 the reduced invariants
J1 , J 2 , 2) removing the condition I 3 = 1 , and 3) adding the
volumetric strain energy density
1
WV = ( J 3 1)
2
(3.8-9)
2
where is the bulk modulus. This expression for the volumetric

strain energy density yields the following relationship between the
pressure and the volume ratio:
p = ( J 3 1) (3.8-10)
The Mooney-Rivlin material model is intended for use when the

bulk to shear modulus ratio is large. The large bulk-to-shear
modulus ratio causes numerical difficulties for axisymmetric, plane
strain and 3D elements. The mixed u/p formulation can be used
(and is the default) for these elements. This formulation avoids the
numerical difficulties by use of a separately interpolated pressure,
as described in the following reference:

ref. T. Sussman and K.J. Bathe, "A Finite Element

Formulation for Nonlinear Incompressible Elastic and
Inelastic Analysis," J. Computers and Structures, Vol.
26, No. 1/2, pp. 357-409, 1987.
The displacement-based formulation can also be used, but is not

recommended.
Selection of material constants: The Mooney-Rivlin material

description used here has constants C1 to C9, constants D1 to D2 and
the bulk modulus . Strictly speaking, this material law is termed a
higher-order or generalized Mooney-Rivlin material law.
Choosing only C1 0 yields the neo-Hookean material law, and
choosing only C1 0, C2 0 yields the standard two-term Mooney-
Rivlin material law. In ADINA, constants C3 to C9 can be chosen
to more closely fit experimental data.
Constants D1 and D2 are primarily intended for modeling certain
biological materials and need not be used otherwise.
The small strain shear modulus and small strain Youngs
modulus can be written in terms of these constants as (assuming
=)
G = 2 ( C1 + C2 ) + D1 D2 (3.8-11)
E = 6 ( C1 + C2 ) + D1 D2 (3.8-12)
These moduli must be greater than zero.
The bulk modulus is used to model the compressibility of the

material for plane strain, axisymmetric and 3-D analysis.
ADINA assumes a default for the bulk modulus based on small

strain near-incompressibility, i.e.,
E
= with = 0.499 (3.8-13)
3 (1 2 )
where E is the small strain Young's modulus or, in terms of the

small strain shear modulus G,

2G (1 + )
= = 500G for = 0.499 (3.8-14)
3 (1 2 )
This rule of thumb can be used to estimate the bulk modulus in the
absence of experimental data. However, you can use lower values
of the bulk modulus to model compressible materials.
In explicit analysis, the program assumes the same bulk

modulus based on small strain near-incompressibility. However,
this can significantly reduce the stable time step. In such cases, is
better to use a bulk modulus that results in = 0.49.
When explicit analysis with automatic time step calculation is

used for a Mooney-Rivlin material, the critical time step is
governed by the dilatational wave speed. This is most frequently an
acceptable assumption since the material is almost incompressible.
As the material deforms, the bulk to shear modulus ratio may

change, because the instantaneous shear modulus is dependent on
the amount of deformation. A value of the bulk modulus that
corresponds to near incompressibility for small strains may not be
large enough to correspond to near incompressibility for large
strains.
3.8.1.2 Ogden material model
The Ogden material model is based on the following expression:
9

WD = n 1 n + 2 n + 3 n 3 (3.8-15)
n =1 n
where n and n are the Ogden material constants.

Plane stress analysis: The material is assumed to be totally

incompressible. Therefore WV is zero and W = WD . A

displacement-based finite element formulation is used, exactly as
for the Mooney-Rivlin material model described above.
Plane strain, axisymmetric and 3-D analysis: The material is

ref. KJB modeled as compressible (that is, the bulk modulus is not infinite),
Section 6.6.2 but you can set the bulk modulus high so that the material is
almost incompressible.
The Ogden strain energy density equation is modified by 1)
substituting the reduced stretches for the corresponding stretches,
2) removing the condition 12 3 = 1 , and 3) adding the
1 1
WV = ( 12 3 1) = ( J 3 1)
2 2
(3.8-16)
2 2
where is the bulk modulus. The relationship between the

pressure and the volumetric ratio is the same as for the Mooney-
Rivlin material description.
Either the u/p formulation (default) or the displacement-based

formulation can be used. For comments about the u/p formulation,
see the corresponding comments in the Mooney-Rivlin material
description.
Selection of material constants: The Ogden material description

used here has 19 constants: n , n , n = 1,...,9 and the bulk
modulus. Choosing only n , n 0, n = 1, 2,3 the standard three-
term Ogden material description is recovered.
The small strain shear modulus and small strain Youngs
modulus can be written as (assuming = )
1 9
G= n n
2 n =1
(3.8-17)
3 9
E= n n
2 n =1
(3.8-18)

These moduli must be greater than zero.
When explicit analysis with automatic time step calculation is

used for an Ogden material, the critical time step is governed by the
dilatational wave speed. This is most frequently an acceptable
assumption since the material is almost incompressible.
The bulk modulus is used to model the compressibility of the

material for plane strain, axisymmetric and 3-D analysis. For
comments about the bulk modulus, see the corresponding
comments about the bulk modulus in the Mooney-Rivlin material
description.
3.8.1.3 Arruda-Boyce material model
The Arruda-Boyce model is based on the following expression:
5
C
WD = 2ii 2 ( I1i 3i ) (3.8-19)
i =1 L
1 1 11 19
in which C1 = , C2 = , C3 = , C4 = ,
2 20 1050 7000
519
C5 = and in which there are two material constants:
673750
and L .
The Arruda-Boyce material model is described in the following

reference:
ref. E.M. Arruda and M. C. Boyce, A three-dimensional

constitutive model for the large stretch behavior of
rubber elastic materials, J. Mech. Phys. Solids, Vol. 41
(2), pp 389-412 (1993).
The Arruda-Boyce material model is an approximation to the

eight-chain material model described in Section 3.8.1.6.



ref. KJB
Section 6.6.2 Plane strain, axisymmetric and 3-D analysis: The material is
modeled as compressible (that is, the bulk modulus is not infinite),
but you can set the bulk modulus high so that the material is
The Arruda-Boyce strain energy density equation is modified
by: 1) substituting for the strain invariant I1 the reduced strain
invariant J1 , 2) removing the condition I 3 = 1 , and 3) adding the
volumetric energy term
WV = ( J 3 ln J 3 ( J 3 1) ) (3.8-20)
where is the small-strain bulk modulus. The relationship between

the pressure and the volume ratio is
p = ln J 3 (3.8-21)

description.
When the u/p formulation is used, there should be at least one

solution unknown. This is because the constraint equation used in
the u/p formulation is nonlinear in the unknown pressures.
Therefore equilibrium iterations are required for convergence, even
when all of the displacements in the model are prescribed.
Selection of material constants:
Material constant is only an approximation to the initial shear

modulus. The relationship between material constant and the

initial shear modulus 0 is
3 99 513 42039
0 = 1 + + + +
5 2
L 175L 875L 67375L8
4 6
3.8.1.4 Hyper-foam material model
The hyper-foam model is based on the following expression:
N
n n 1
W = n n
1 + 2 + 3 3 + (
J 3 n n 1 ) (3.8-22)
n =1 n n
in which there are the material constants n , n , n , n = 1,..., N .

The maximum value of N is 9.
A material model similar to the hyper-foam material model is

described in the following reference:
ref. B. Storkers, On material representation and

constitutive branching in finite compressible elasticity,
J. Mech. Phys. Solids, Vol. 34(2), pp 125-145 (1986).
In this reference, n is the same for all values of n.
When it is required to compute the deviatoric and volumetric

parts of the strain energy density, we use
N
n
WD = 1 + 2 + 3 3 J 3 / 3
n n n n
(3.8-23)
n =1 n
N
n 1
WV = ( n / 3
3 J 3 1 + ) (
J 3 n n 1 ) (3.8-24)
n =1 n n
Notice that W = WD + WV . This decomposition of the strain

energy density has the advantage that the stresses obtained from the
deviatoric and volumetric parts separately are zero when there are
no deformations:
1 W WD
SijD = D+ = 0,
ij = 0 2 ij ji
ij = 0
1 W W
SijV = V + V =0
ij = 0 2 ij ji
ij = 0
Notice that WD contains the volumetric part of the motion through

/3
the term 3 J 3 n . Therefore WD is not entirely deviatoric.
Plane stress analysis: The material is not assumed to be totally

incompressible. A displacement-based finite element formulation
is used.
Plane strain, axisymmetric and 3-D analysis: The material is not

assumed to be totally incompressible.
Because both WD and WV contain the volumetric part of the
motion, the mixed u/p formulation cannot be used with the hyper-
foam material. A displacement-based formulation is used.
Selection of material constants: The hyper-foam material

description used here has 27 constants: n , n , n , n = 1,...,9 .
The small strain shear modulus and small strain bulk modulus
can be written as
1 9
G= n n
2 n =1
(3.8-25)
9
1
= n + n n (3.8-26)
n =1 3
These moduli must be greater than zero, hence we note that n

should be greater than 1/3.

When all of the n are equal to each other = , then the
Poissons ratio is related to using

= (3.8-27)
1 2
The default values of the material constants are 1 = 1.85 ,

1 = 4.5 , 1 = 9.2 , 2 = 9.2 , 2 = 4.5 , 2 = 9.2 . It is
likely that you will need to change these values in your analysis.
The hyper-foam material model is generally used for highly

compressible elastomers.
If the ratio of the bulk modulus to shear modulus is high
(greater than about 10), the material is almost incompressible and
we recommend that one of the other hyperelastic materials be used.
3.8.1.5 Sussman-Bathe material model
The Sussman-Bathe model is based on the following equation:
WD = w(e1 ) + w(e2 ) + w(e3 ) (3.8-28)
where w(e) is a function of the principal logarithmic strain

(Hencky strain) and e1, e2, and e3 are the principal logarithmic
strains.
The primary goal of the model is to fit given uniaxial

tension/compression data very well. This goal is accomplished by
using a spline to fit the derivative of w(e), as described in detail
below.
Of course, when uniaxial tension/compression data is known, a

curve fitting approach can, in theory, be used to determine the
constants for the other hyperelastic models, e.g. the Ogden material
model. But this curve fitting in practice does not provide good fits
to the data under many circumstances.

incompressible material (I3 = 1) and is modified as explained below

for plane strain, axisymmetric or 3-D analysis.
The Sussman-Bathe model is given in the following reference
ref. T. Sussman and K.J. Bathe, A model of incompressible

isotropic hyperelastic material behavior using spline
interpolations of tension-compression test data,
Commun. Numer. Meth. Engng, Vol. 25, Issue 1, pp. 53-
63, January 2009.
The following gives a quick summary of the Sussman-Bathe

model. In this summary, we assume that the material is totally
incompressible. Differences due to slight compressibility are
small.
Theoretical background:
1) The Cauchy stress i corresponding to the principal strain ei is
WD
i = + p = w(ei ) + p (3.8-29)
ei
where w(ei ) dw / de .
2) In uniaxial tension/compression (Figure 3.8-1), e1 = e,

e2 = e3 = 12 e so
= w(e) w( 12 e) (3.8-30)
(3.8-30) can be inverted to obtain
(( )

w(e) = 2)
1 k
e (3.8-31)
k =0
The series converges when (e) 0 as e 0 .

Original
Deformed
L2
F1
L3 0
L1
A1 L2 L3 0 A1
L1
L1
L1 1
e e1 ln e2 e3 # e1
0
L1 2
F1
"1 "2 "3 0
A1
Figure 3.8-1: Uniaxial tension/compression test
3) The asymptotic conditions for w are w(e) as e ;

w(e) as e . These asymptotic conditions correspond to
the asymptotic conditions of infinite stresses for infinite strains.
4) For a stable material, it is necessary (but not sufficient) that

w(e) > 0 for all e. Not all materials for which (e) > 0 have
w(e) > 0 . For example, the material with
(e) = ET e, e > 0
= EC e, e < 0
where ET and EC are constants greater than zero, has w(e) > 0
only if
1
2 ET < EC < 2 ET
5) Given only simple tension data for (e) , there are multiple
w(e) that exactly correspond to (e) , for positive e only. Two
such w(e) are

w(e) = 0, e < 0
= (e), e > 0
and
w(e) = (2e), e < 0

= 0, e > 0
Hence the material is not uniquely described given only uniaxial

tension (or uniaxial compression) data. Both uniaxial tension and
uniaxial compression data must be provided to uniquely describe
the material.
6) The small-strain Youngs modulus E is found by differentiating

the uniaxial stress-strain curve, and evaluating at e = 0, and, since
the material is almost incompressible, the small-strain shear
1
modulus G is G = E . The results are
3
3 1
E= w(0) , G = w(0) (3.8-32a, b)
2 2
8) The Ogden material model can be considered a special case of

(3.8-28), since the Ogden material model can be written in terms of
w(e) :
w(e) = n ( exp( n e) 1) (3.8-33)

n
Spline representation of w(e) :
In the Sussman-Bathe model, we choose w(e) to fit given uniaxial

tension/compression data very well, as follows.
The uniaxial tension-compression data is in the form of user-

specified data points ( ei , i ) . From these data points, we build a
non-uniform cubic spline representation of the uniaxial

tension/compression stress-strain data = (e) , as shown in

Figure 3.8-2. For the non-uniform cubic spline representation of
( e) ,
1) A spline segment is placed between two successive user-input

data points. The user-input data points need not be equally spaced.
2) The range of the cubic spline is between the first and last user-
input data points.
3) Outside the range of the cubic spline, the slope of (e) is greater
than zero. This ensures that the asymptotic conditions of
() = , () = are met.
"(e)
Segment of Slope > 0

cubic spline
emin,"
emax," e
Measured data point

Slope > 0
Figure 3.8-2: Uniaxial tension/compression stress-strain spline
Using the non-uniform cubic spline representation of (e) and

(3.8-31), we build a uniform cubic spline for w(e) as shown in
Figure 3.8-3. For the uniform cubic spline representation of w(e) ,
1) The same number of spline segments is used in tension and in

compression.
2) The range of the cubic spline is the same in tension and in

compression. This range includes the range of the user-input data

points.
3) Outside the range of the cubic spline, the slope of w(e) is

greater than zero, whenever possible.
w(e)
Segment of Slope > 0

cubic spline
emin,w
emax,w e
Slope > 0
Figure 3.8-3: w(e) spline
In order to measure the accuracy of the spline representation of

w(e) , we define the relative interpolation error
( e ) ( e)
r = max (3.8-34)
e (e )
in which (e) is the stress evaluated from the spline representation

of w(e) (using 3.8-30), and (e) is the stress evaluated from the
spline representation of (e) .
The number of spline segments is automatically chosen to make the

interpolation error r smaller than a user-specified value. Typically
only a few spline segments need be used for w(e) in order to
reduce the interpolation error to a value smaller than experimental
error.

Each cubic spline segment in w(e) can be written
w(e) = Ai +1 z + Bi +1 z 3 + Ai (1 z ) + Bi (1 z )3 (3.8-35)
for the segment ei e ei +1 where z = (e ei ) /(ei +1 ei ) . With

this definition, w(e) can be written
z2 z4 1 (1 z ) 2 1 (1 z ) 4
w(e) = Ci + (ei +1 ei ) Ai +1 + Bi +1 + Ai + Bi
2 4 2 4
(3.8-36)
The AUI determines the constants Ai, Bi, Ci from uniaxial stress-
strain data, as described above. The constants Ai, Bi, Ci are listed
in the ADINA output file when the model definition is listed.

incompressible. Therefore WV is zero and W = WD. A
Plane strain, axisymmetric and 3-D analysis: The material is

Equation (3.8-28) is modified by 1) substituting the deviatoric
principal strains for the corresponding principal strains, 2)
removing the condition e1 + e2 + e3 = 0, and 3) adding the
WV = ( J 3 ln J 3 ( J 3 1) ) (3.8-37)
where is the bulk modulus. The relationship between the

pressure and the volume ratio is
p = ln J 3 = ( e1 + e2 + e3 ) (3.8-38)
which is a generalization of the small-strain pressure-strain

relationship. Either the u/p formulation (default) or the

displacement-based formulation can be used. For comments about
the u/p formulation, see the corresponding comments in the
Mooney-Rivlin material description.
Data input considerations:
1) Data input is in the form of a set of stress-strain data points, with

positive stresses/strains corresponding to uniaxial tension and
negative stresses/strains corresponding to uniaxial compression.
Compression and tension data are entered together in the same data
set.
2) The data set should contain both tension and compression data
(compression data is possibly converted from equibiaxial tension
data, see below). If the data set contains only tension data, the
program will assume that the true stress / true strain curve in
compression is a straight line, which is most likely not a good
assumption.
3) The stresses and strains in the set of stress-strain data points can
be either
a) True stresses and logarithmic strains

b) Engineering stresses and engineering strains
c) Engineering stresses and stretches
In the latter cases, the program internally converts the stresses and
strains to true stresses and logarithmic strains.
4) Data points from equibiaxial tension experiments can be

converted into equivalent uniaxial compression data. The
conversion formulas are:
eu = 2eb , u = b2 , 0 eu = (1 + 0 eb ) 1
2
(3.8-39)
u = b , 0 u = 0 b b3
where eu is the is the equivalent uniaxial logarithmic strain (< 0), eb

is the equibiaxial logarithmic strain (> 0), u is the equivalent

uniaxial stretch, b is the equibiaxial stretch, 0 eu is the equivalent

uniaxial engineering strain, 0 eb is the equibiaxial engineering
strain, u is the equivalent uniaxial true (Cauchy) stress, b is the
equibiaxial true (Cauchy) stress, 0 u is the equivalent uniaxial
engineering stress, 0 b is the equibiaxial engineering stress. All of
these conversion formulas assume that the material is
incompressible.
5) The Sussman-Bathe model fits the data so closely that roughness

and waviness in the data causes roughness and waviness in the
w(e) splines. The program does not smooth the data in order to
eliminate roughness and waviness. If the original data set contains
roughness and waviness that should not be present in the analysis,
the data set should be smoothed before entering the data into the
program.
6) If the data set corresponds to a stable material, then the

Sussman-Bathe model is stable, otherwise the Sussman-Bathe
model may not be stable.
7) The strain range of the data set should contain the range of
strains anticipated during the analysis.
8) Do not confuse uniaxial compression with hydrostatic

compression. These two test cases are very different.
3.8.1.6 Eight-chain material model
The eight-chain model is based on the following expression:

WD = R b + ln ( g + ln ) (3.8-40)
sinh sinh
in which
3L
R= (3.8-41)



b= , = L1 (b) (3.8-42a, b)
L
1
g= = L1 ( g ) (3.8-43a, b)
L ,
1 1
= 12 + 22 + 32 = I1 (3.8-44)
3 3
The function b = L( ) is the so-called Langevin function, defined

as
b = L( ) = coth (1/ ) (3.8-45)
The inverse Langevin function = L1 (b) cannot be written in

closed-form.
There are two material constants: and L . is the initial

shear modulus and L is the locking stretch. The locking stretch
will be discussed in more detail below.
The eight-chain material model is described in the following

references:
ref. E.M. Arruda and M. C. Boyce, A three-dimensional

constitutive model for the large stretch behavior of
rubber elastic materials, J. Mech. Phys. Solids, Vol. 41
(2), pp 389-412 (1993).
ref. M. C. Boyce and E.M. Arruda, Constitutive models of

rubber elasticity: a review, Rubb. Chem. & Tech., Vol.
73 (3), pp 504-523 (2000).
The Arruda-Boyce material model in Section 3.8.1.3 is an

approximation to the eight-chain material model.
Theoretical background
The following gives a brief introduction to the theory.

The material response is governed by the deformations of material

chains that connect the corners of a unit cube of material to its
center (Fig 3.8-4(a)). This unit cube is selected so that the
deformation state is principal, and therefore the deformed side
lengths of the unit cube are the principal stretches.
As there are eight chains that connect the corners of the unit cube
to its center, the material model is called the "eight-chain model".
However, all of these eight chains undergo the same deformation.
Therefore attention is focused on the single chain connecting two
opposite corners of the unit cube. The initial length of this chain is
just 0 r = 3 (Fig 3.8-4(b)) and the current length of the chain is
t
r = 12 + 22 + 32 (Fig 3.8-4(c)). Thus the stretch of this chain is
t
r 1
chain = 0
= 12 + 22 + 32 (3.8-46)
r 3
Due to the assumed incompressibility of the material, the stretch

chain is always greater than or equal to 1.

(a) Eight chains 1

(b) Single representative chain,
undeformed
l2
l3
l1
(c) Single representative chain,

deformed
Figure 3.8-4: Unit cube and chains used in eight-chain model
From the paper of Boyce and Arruda (2000), the following

expression is given for the strain energy function of the eight-chain
model.

W = Nk n chain + n ln (3.8-47)
sinh
Here N is the number of chains in the chain structure, k is

Boltzmann's constant, and is the absolute temperature. is
defined implicitly by

chain
= L( ) = coth (1/ ) (3.8-48)
n
For ease of writing, we define L = n , = chain , b = / L

and R = Nk n . The result is

W = R b + ln (3.8-49)
sinh
Since this quantity is not zero for zero deformation, we subtract the
value of W at zero deformation to obtain the strain energy density
expression (3.8-40).
Equation (3.8-41), which gives the value of R in terms of the shear

modulus and locking stretch, is obtained by considering the case of
simple tension, as described below.
Differentiating equation (3.8-40) with respect to b gives
dWD
= R (3.8-50)
db
Thus is seen to be a stress-like quantity conjugate to the strain-

like quantity b . A plot of the function = L1 (b) is given in Fig.
3.8-5. Notice that as the stretch approaches the locking stretch,
becomes infinite. Thus the stretch cannot exceed the locking
stretch. Also notice that for stretch much smaller than the
locking stretch (very small b ), the function = L1 (b) is
approximated by 3b .

10.
9. Eight-chain model,
exact relationship used
8.
Arruda-Boyce model,
7. series approximation used
Asymptote
6. neo-Hookean model corresponding to
b 5. locking stretch
4.
3.
2.
1.
0.
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1
b
Figure 3.8-5: Relationships between b and b
Also it is evident that the locking stretch L must be greater than 1.
Simple tension
It is instructive to consider the cases of simple tension, pure shear

and equibiaxial tension. For simple tension, the engineering stress
is calculated using
dWD dWD db d R 1
0 =
d
= =
db d d L 3
( 2 )

(3.8-51)
Here depends on through the relationships = L1 (b) ,

1 1
b = / L , = 12 + 22 + 32 = 2 + 2 1 .
3 3
The derivative of the engineering stress with respect to the stretch,

evaluated at zero deformation, is
d 0 d R 1 R R
d
=
d
( 2 ) = = (3.8-52)
=1 =1 L 3 L =1
L
Since this is equal to the initial Young's modulus E = 3 , R must

be defined by equation (3.8-41). Notice that, in the limit of large

locking stretch,
3L 3L
R= = L2 (3.8-53)
3g
and for stretch much smaller than the locking stretch,
( ) 3
2
L
R 1 1 =
L
(3.8-54)
L 3 L 3
so the neo-Hookean case is recovered for stretch much smaller

than the locking stretch.
The maximum tensile and compressive stretches in simple tension

1
are obtained by setting = L = 2 + 2 1 . An
3
approximate solution in tension is obtained by neglecting 2 1 and
an approximate solution in compression is obtained by neglecting
2 2
2 . The result is L < < 3L . Thus we see that the
3
locking stretch L does not correspond directly to an
experimentally observed tensile or compressive stretch. However,
if by experiment the maximum tensile and compressive stretches
are known, the corresponding locking stretch can be obtained with
the above expressions.
Pure shear
For pure shear, the engineering stress is calculated using
dWD R 1
0 =
d
=
L 3
( 3 )

(3.8-55)

1 1
b = / L , = 12 + 22 + 32 = 2 + 2 + 1 .
3 3
The maximum stretch in pure shear is obtained by setting

1
= L = 2 + 2 + 1 , with an approximate maximum
3
stretch of about 3L (very similar to the maximum tensile
stretch in uniaxial tension.)
Equibiaxial tension
For equibiaxial tension, the engineering stress is calculated using
1 dWD R 1
0 =
2 d
=
L 3
( 5 )

(3.8-56)

1 1
b = / L , = 12 + 22 + 32 = 2 2 + 4 .
3 3
The maximum tensile stretch in equibiaxial tension is obtained by

1
setting = L = 2 2 + 4 , with an approximate maximum
3
3
tensile stretch of about L (significantly different than the
2
maximum tensile stretch in uniaxial tension).
Comparison with Arruda-Boyce model
Although both the eight-chain model and its series approximation

are presented in the references by Arruda and Boyce, we use the
name Arruda-Boyce to refer only to the series approximation. This
is primarily because the series approximation is already widely
used under the name Arruda-Boyce.
In terms of terminology given in this section, the Arruda-Boyce

series approximation takes the form
1 1 3 11 5
6 b + 6 b + 27 b
2 20 1050
(3.8-57)
19 7 519 9
+ 108 b + 405 b
7000 673750
in which the original constants of the Arruda-Boyce approximation

are shown. This curve is also plotted in Fig. 3.8-5. The Arruda-
Boyce series approximation gives results very close to the eight-
chain model only when the stretch is not close to the locking
stretch. Note that the Arruda-Boyce series approximation does not
give a limiting value for the stretch, in other words, the stretch
can exceed the locking stretch L when the Arruda-Boyce series
approximation is used.
Also, in the Arruda-Boyce model, material constant is only an

approximation to the initial shear modulus, whereas in the eight-
chain model, material constant is the initial shear modulus.
For these reasons, the eight-chain model is preferred to the Arruda-

Boyce model.
Comparison with neo-Hookean model
The eight-chain model can also be interpreted as an extension of

the neo-Hookean model (Mooney-Rivlin model with just C1
nonzero). In terms of terminology of this section,
3b (3.8-58)
This curve is also plotted in Fig. 3.8-5. Material constant C1 is

equal to 1/2 of the initial shear modulus.
Finite element inplementation
The strain energy density expression (3.8-40) assumes a totally



ref. KJB
Section 6.6.2 Plane strain, axisymmetric and 3-D analysis: The material is
The eight-chain strain energy density equation is modified by:
1) substituting for the strain invariant I1 the reduced strain
invariant J1 , 2) removing the condition I 3 = 1 , and 3) adding the
volumetric energy term
WV = ( J 3 ln J 3 ( J 3 1) ) (3.8-59)
where is the small-strain bulk modulus. The relationship between

the pressure and the volume ratio is
p = ln J 3 (3.8-60)

description.
When the u/p formulation is used, there should be at least one

solution unknown. This is because the constraint equation used in
the u/p formulation is nonlinear in the unknown pressures.
Therefore equilibrium iterations are required for convergence, even
when all of the displacements in the model are prescribed.
Selection of material constants: Both of the material constants

and L have a clear physical interpretation. is the initial
shear modulus and L is the locking stretch. The locking stretch

can be estimated by examination of the experimental stress-strain

curves for simple tension/compression, pure shear or equibiaxial
tension, as explained above.
3.8.1.7 Curve fitting
You can have the AUI determine the material constants for the
Mooney-Rivlin, Ogden, Arruda-Boyce , hyper-foam, or eight-chain
materials by a curve-fit to experimental stress-strain data. Note that
the Sussman-Bathe model does not require curve-fitting.
The AUI allows for data from three experimental cases as

shown in Figure 3.8-6: simple tension, pure shear and equibiaxial
tension. Input from a single experiment from one of these cases or
a combination of data from any two, or all three of these
experimental cases is allowed.
A single set of constants representing a best fit in the least
squares sense is calculated. You must select the chosen order of
the approximation (giving the number of constants used in the
model). Note: the total number of data points must be greater than
the number of material constants to be determined.
For each experimental case, you can enter either the stretch or
engineering strain in the strain column and you enter the
engineering stress in the stress column. Refer to Figure 3.8-6 to see
the definitions of stretch, engineering strain and engineering stress
for the experimental cases.

% = (L+)/L e=/L
Y
%1 = %; %2 = %3 = %-1/2
L $
L
0 p1 = F/A
X
Deformed area A/%
Z Original area A
a) Simple tension test (incompressible material)
Y % = (L+)/L e=/L
L $
0 p1
L = F/A
X
Deformed area A/%
Z Original area A
%1 = %; %2 = 1; %3 = 1/%
b) Pure shear test (incompressible material)

Figure 3.8-6: Experimental test cases for rubber and foam materials

% = (L+)/L e=/L
Y %1 = %; %2 = %; %3 = %-2
F Original area A
Deformed area A/%
0 p1 = F/A
Z L+
L
c) Equibiaxial tension test (incompressible material)

Mooney-Rivlin model: The strain energy density for total

incompressibility is given in Eq. 3.8-8 with the material constants
D1 and D2 set to zero.
Given M non-zero material constants, Ci, then M is equal to 2, 5
and 9 for first, second and third order approximations, respectively.
The principal stress in the direction of principal stretch 1 is
given by
W
p1 = = 0 p1 ( Ci , ) (i = 1,..., M ) (3.8-61)

0
Notice that this is the engineering principal stress, that is, the force
divided by the original area of the specimen. Thus, we have an
expression for the principal stress that is linear in terms of the
material constants Ci, and dependent on the principal stretch
determined from the allowed experimental cases. Hence, given a
set of data points ( k , k ) from one or more of the test cases, we
can establish the least squares fit by a curve of the form given by
Eq. (3.8-61).
Consider the sum of discrete error terms at the given data points
of one of the experimental test cases (simple tension, pure shear or
equibiaxial tension)

N
S = ( k p1 ( k ) )
2
(3.8-62)
k =1
where
p1 is the approximating function dependent on the
constants Ci (i = 1,...,M) given by Eq. (3.8-61). We can expand
this sum to include data from all three test cases, i.e.:
N 2
j
( )
3
S = kj 1j ( kj ) (3.8-63)
j =1 k =1
where the superscript j denotes the experimental test case (simple

tension j = 1, pure shear j = 2, equibiaxial tension j = 3). Note: the
form of 1j differs for j = 1, 2, 3 due to the different principal
stretch terms 2 and 3 .
The least squares fit is determined by minimizing S in Eq. (3.8-
63) with respect to the constants Ci,, which gives M algebraic
equations for the constants Ci (i = 1,...,M).
Ogden model: The strain energy density function is, for total
incompressibility, given by Eq. (3.8-15). Thus, for an M-term
Ogden model, there are 2 M constants, i.e., the pairs ( i , i ) for i
= 1,...,M.
Assuming the principal stretches for the three test cases shown
in Figure 3.8-4, the following form for the principal stress 0 p1
can be derived:
0 p1 = i ( 1 c 1 )
i i
(3.8-64)
i =1
with c = -0.5 for simple tension, -1.0 for pure shear, -2.0 for
equibiaxial tension.
Equation (3.8-64) is linear in terms of the coefficients i , but
nonlinear in terms of the exponents i . A nonlinear least squares
algorithm could be employed to iterate to a solution for the

constants ( i , i ) . However, in our curve-fitting implementation

in the AUI, you specify the desired exponent terms i (with the
default i = i), and the least squares fit of the resulting function
yields the corresponding coefficients i . You can select the
number of terms M desired (1 M 9).
Arruda-Boyce model: The strain energy density function is, for

total incompressibility, given by Eq. (3.8-19). There are only two
material constants, and L .
in Figure 3.8-6, the following forms for the principal stress 0 p1
can be derived:
Uniaxial tension:
5
iCi
p1 = 2 ( 2 ) I , I1 = 2 + 2 1
i 1
(3.8-65)

0 2i 2 1
i =1 L
Pure shear:
5
iCi
p1 = 2 ( 3 ) I , I1 = 2 + 2 + 1 (3.8-66)
i 1

0 2i 2 1
i =1 L
Equibiaxial tension:
5
iCi
p1 = 2 ( 5 ) I , I1 = 2 2 + 4
i 1
(3.8-67)

0 2i 2 1
i =1 L
The AUI uses a nonlinear curve-fitting algorithm to determine

and L .
Hyper-foam model: The strain energy density function is given by

Eq. (3.8-22). However, we seek a curve-fit for which all of the n
terms are equal. Thus, for an M-term hyper-foam model, there are
2 M + 1 material constants, i.e., the constants ( n , n ) for n =
1,...,M, and .
First is determined using information about the lateral stretch
T . There are two options for entering the lateral stretch

information:
1) Specify the Poissons ratio. is then calculated using (3.8-

27).
2) Enter the lateral stretches as part of the stress-strain data

input. Then is determined with the use of the following
formulas:
Uniaxial tension:

T = 1+ 2
(3.8-68)
Pure shear:

T = 1+
(3.8-69)
Equibiaxial tension:
2

T = 1+
(3.8-70)
Since each of these formulas can be also written in the form

ln T = f ( ) ln , a linear curve-fit is performed to obtain
.
Next, the constants ( n , n ) are determined. Assuming the

principal stretches for the three test cases shown in Figure 3.8-6,
the following form for the principal stress 0 p1 can be derived:
( )
M
1 n
0 p1 =
n
J3 (3.8-71)

n
n =1
where J 3 = T2 for uniaxial tension, J 3 = T for pure shear and

J 3 = 2 T for equibiaxial tension. (3.8-71) is then used as the
basis of a nonlinear curve-fitting algorithm to obtain the constants
( n , n ) .

Eight-chain model: The strain energy density function is, for total
incompressibility, given by Eq. (3.8-40). There are only two
material constants, and L .
in Figure 3.8-6, the principal stress 0 p1 is given by equations
(3.8-51), (3.8-55), (3.8-56).
The AUI uses a nonlinear curve-fitting algorithm to determine

and L .
Notes for curve-fitting for the Mooney-Rivlin and Ogden

material models: The system of equations used to solve for the
least squares fitted constants is likely to be ill-conditioned. Direct
use of Gaussian elimination may yield constants with large
magnitudes, which are balanced to cancel out when the fitted
function is evaluated. Furthermore, and perhaps more
importantly, there is no guarantee that the fitted function has
physically feasible behavior outside the range of the data supplied.
The reason for the ill-conditioning of the least squares system
matrix is that the data may be fitted equally well by two or more of
the "basis" functions or combinations thereof, thus rendering the
matrix nearly singular.
The AUI provides an alternative matrix solution algorithm to
the Gauss-Jordan elimination technique, namely that of Singular
Value Decomposition (SVD) described in
ref. Forsythe, G.E., Malcom, M.A. and Moler, C.B.,

Computer Methods for Mathematical Computations,
Chapter 9, Prentice Hall, Inc. Englewood Cliffs, NJ,
1977.
The essence of the SVD algorithm is that combinations of the basis

functions which correspond to near zero eigenvalues of the matrix
system are eliminated.
Successive elimination of the near singular terms always
increases the error (in the least squares sense). However, the
increase in error is often small compared to the overall
improvement in the fitted solution the constants obtained are

usually smaller in magnitude and physically reasonable.

The SVD algorithm is controlled by successively eliminating
combinations corresponding to the smallest eigenvalues until a
fitted function is obtained which gives a monotonic increase in
stress with strain.
Depending on data, the number of terms eliminated from a
higher order approximation may be large, giving a poor fit, even
though the monotonicity criterion has been achieved.
As an example, consider the curve fitting results illustrated in
Fig. 3.8-7. Fig. 3.8-7(a) shows the fitted curves for a 9-term ( i =
i, i = 1, . . ., 9) Ogden model approximating a set of data points for
a simple tension test. The sum of error squares S is given in the
key of the graphs.

2.0
STRESS
-2.0
a) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
STRETCH
Input data Ogden (9-term; Gauss) S=1.56E-03

Ogden (9-term; SVD(1)) S=1.61E-03
2.0
STRESS
b)
-2.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

STRETCH

2.0
STRESS
c)
-2.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

STRETCH
Figure 3.8-7: Curve fitting results for Ogden material model
While the Gaussian elimination curve has the minimum S, its

solution is asymptotically unacceptable. Two terms were removed
to give a monotonic increasing solution (SVD(2)) which still gives
a good approximation to the input data. Fig. 3.8-7(b) shows the
fitted curves for a 6-term Ogden model for the same input data. A
monotonic increasing solution was obtained after removal of one
term (SVD(1)), while the Gauss elimination solution was not
asymptotically feasible. For the 3-term Ogden fit to the data, the
Gauss elimination solution was accepted since it provided a

monotonic increasing curve, as shown in Fig. 3.8-7(c).
Notes for curve-fitting for the Arruda-Boyce and hyper-foam

material models: The Marquardt-Levenberg algorithm from the
following reference is used in the curve fitting:
ref. W.H. Press, B.P. Flannery, S.A. Teukolsky and W.T.

Vetterling, Numerical Recipes the Art of Scientific
Computing (Fortran Version), Cambridge University
Press, pp 521-528 (1990).
For the curve-fitting procedure for the hyper-foam material, the

starting values for the material constants are the user-provided
values for the material constants. It is suggested that the curve
fitting first be done using one term only. If the curve-fit is not
accurate enough, then a two-term curve fit can be done with
starting values for the first term obtained from the curve-fit for the
first term. This procedure can be repeated for each additional term
until an accurate curve fit is obtained.
General comments: It can be seen that the curve fitting technique

provided gives a valuable tool, but one which must be used with
caution. Always check the resulting fitted function. If a physically
reasonable fit is not achieved, we recommend the following
strategy
a) Use a lower order approximation.

b) Use a different material model
c) Provide more data points, covering a greater range of
anticipated strains, including values for compression ( < 1).
Specification of input: All of the information required for the

curve fitting is contained in a data set of type curve-fitting. The
curve-fitting data contains the numbers of the data sets for each of
the three test cases, the order of the curve fit and other information
required for the curve fit.
3.8.2 Viscoelastic effects
Viscoelastic effects can be included in the Mooney-Rivlin,

Ogden, Arruda-Boyce, hyper-foam, Sussman-Bathe, and eight-

chain material models.
Two viscoelastic models are available:
Holzapfel viscoelasticity (Section 3.8.2.1)

Bergstrm-Boyce viscoelasticity (Section 3.8.2.2)
3.8.2.1 Holzapfel viscoelasticity
In the following, we give a brief discussion of the Holzapfel model

for finite strain viscoelasticity, see the following references for
more information:
ref. G. A. Holzapfel, On large strain viscoelasticity:

continuum formulation and finite element applications to
elastomeric structures, Int. J. Num. Meth. Engng., Vol.
39, pp 3903-3926, 1996.
ref. G. A. Holzapfel, Nonlinear solid mechanics. A

continuum approach for engineering. John Wiley &
Sons, Chichester, pp 278-295, 2000.
ref. G. A. Holzapfel, Biomechanics of soft tissue, in
Lemaitre (ed.), Handbook of Materials Behavior
Models: Nonlinear Models and Properties, Academic
Press, 2001, pp 1057-1071.
Equivalent 1D model: The equivalent 1D model is shown in Fig

3.8-8. It is the same as a generalized Maxwell model with many
chains. A generic chain is denoted with superscript , as shown
in the figure.

Strain e,
&
stress
E
E1 !1
..
E . !
Strain Strain
g '
Figure 3.8-8: Equivalent 1D model for viscoelastic effects
The spring E is equivalent to the elastic stiffness of the

model. Each chain contains a spring with stiffness E and dashpot
with viscosity . (Note that the superscripts and do not
denote exponentiation.) The strain in each chain is the sum of the
strain in the spring g and the strain in the dashpot . The
observed stress is
= + q (3.8-72)

where = E e is the elastic stress and q = E g = is

the stress in chain . Using the definition = and the
E
assumption E = E , the following expression is obtained:
1
q + q = (3.8-73)

Assuming that 0 q = 0 , (3.8-73) can be written in convolution

form as
t t
t
q = exp t dt (3.8-74)
0

from which the total stress is
t t t
t

= E 1 + exp
e dt (3.8-75)
0
Evidently the relaxation modulus is

t
E (t ) = E 1 + exp which is a Prony series

expression.
The dissipation in dashpot is
1 q
D = q = q ( e g ) = q e (3.8-76)
E
and the total dissipation is D =

D . In the above, the
viscoelastic material constants for each chain are and .
Potential-based 1D model: The 1D model can be written in terms

of a potential as follows:
= (e) + ( g ) (3.8-77)

1 2
where (e) = E e is the strain energy of the elastic chain
2
1
and ( g ) = E ( g ) is the strain energy in the spring of
2
2
chain . In terms of , ( g ) = ( g ) . The 1D
model is recovered by defining

= , q = (3.8-78a,b)
e fixed e fixed

Notice that (3.8-72) and (3.8-78a) imply

q = = .
e fixed
g
Finite strain model: The finite strain model is derived from the
potential-based 1D model as follows. The elastic potential is
defined as
( ) = W ( ) (3.8-79)
( )
where W ij is the strain energy density from the elastic part of
the material model. The potential of each chain is defined as
( ij , ij ) = W ( Gij ) , usage=combined
= WD ( Gij ) , usage=deviatoric (3.8-80a,b,c)
= WV ( Gij ) , usage=volumetric
in which the usage flag (which is a user-input flag) determines

whether the entire elastic strain energy density, deviatoric strain
energy density or volumetric strain energy density is taken for
chain . Here Gij is analogous to the strain in the 1D spring g ,
and we assume Gij = ij ij . Note that with this definition of
Gij , we have

Sij = , Qij = = =
ij ij ij Gij
ij fixed ij fixed ij fixed
(3.8-81a,b)
where Sij are the 2nd Piola-Kirchhoff stresses and Qij is analogous
to the stress q .
Following exactly the same arguments as in the 1D case, we
obtain

1
Qij + Qij = Sij (3.8-82)

Assuming that 0Qij = 0 , (3.8-82) can be written in convolution

form as
t t
t
Qij = exp t Sij dt

(3.8-83)
0
and (3.8-83) can be numerically approximated using
t
1 exp
t t +t S t S .
t +t
Qij = exp tQij +
t
( ij ij )

(3.8-84)
(3.8-84) is exact when Sij does not change during the time step,
and is more accurate than the formula given by Holzapfel:
t t
t +t
Qij = exp tQij + exp ( t +t Sij t Sij )
2
(3.8-85)
t
especially when is large.

Dissipation calculations: If the dissipation is required, it is

calculated using
D = Qij ij = Qij ij Gij( ) (3.8-86)
where
2

Grs = Qij (3.8-87)
Gij Grs

is used to compute the unknown Grs from the known Qij .

If usage=combined,
2 2W
=
= Cijrs . (3.8-88)
Gij Grs ij rs
where the tensor Cijrs is evaluated at the strain state Gij . The
dissipation calculation requires the solution of a set of simultaneous
linear equations of at most order 6 (in the 3D case) at each
integration point.
If usage=deviatoric,
2 WD
2

= = ( CD )ijrs (3.8-89)
Gij Grs ij rs
where the tensor ( CD )ijrs is evaluated at the strain state Gij . Here,
the dissipation calculation requires a singular value decomposition
of ( CD )ijrs , since ( CD )ijrs has a zero eigenvalue. A similar
situation applies when usage=volumetric, except that the
corresponding material tensor has only one nonzero eigenvalue.
The procedure given in (3.8-86) to (3.8-89) is only approximate,
since the fundamental assumption Gij = ij ij strictly speaking
only holds for small strain analysis.
Restrictions and recommendations: The allowed values of the

usage flag depend upon the material model and finite element type,
as shown in Table 3.8-2.
In view of the restrictions, we recommend that usage=combined

be used in conjunction with the hyper-foam material model, and
that usage=deviatoric be used in conjunction with the Mooney-
Rivlin, Ogden, Arruda-Boyce, Sussman-Bathe and eight-chain
material models.
Time-temperature superposition: The preceding derivation

assumes that the viscoelastic response is not temperature-

dependent. One method of including the effects of temperature is
the method of time-temperature superposition.
In time-temperature superposition, the actual time t is replaced
by the reduced time . The relationship between the actual time
and reduced time is given by
d 1
= (3.8-90)
dt aT ( t )
where t is the temperature and aT ( t ) is the shift function.
Table 3.8-2: Allowed values of the usage flag
Mooney-Rivlin, hyper-foam
Ogden, Arruda-Boyce,
Sussman-Bathe, eight-
chain
Plane Plane strain, Plane Plane strain,
stress1 axisymmetric, stress3 axisymmetric,
3D2 3D
usage=combined yes no yes yes
usage=deviatoric yes yes no yes
usage=volumetric no no no yes
1) Usage cannot be equal to volumetric. This is because the

material is assumed to be fully incompressible, hence the
volumetric strain energy density is zero.
2) When the u/p formulation is used, the usage flag cannot be

combined or volumetric. This is because the modification to
the volumetric stresses caused when the usage flag is combined
or volumetric is not taken into account in the u/p formulation.
3) The only allowable value of the usage flag is combined. This

is because the out-of-plane stress component S xx must be zero, and
in the Holzapfel finite strain viscoelastic model, the only way that
this can happen is if S xx is zero.

Evidently
t
1
=
t
t
dt (3.8-91)
0
aT ( )
The shift function used here is either the WLF shift function,
C1 ( t ref )
log10 aT ( t ) = (3.8-92)
C2 + t ref
or the Arrhenius shift function
1 1 t
log10 aT ( t ) = C1 t , ref

ref
(3.8-93)
1 1 t
= C2 t
, < ref
ref
where ref is the reference temperature and C1 , C2 are material

d
constants. Notice that as t increases, aT ( t ) decreases and
dt
increases. When using the Arrhenius shift function, the
temperature unit must be absolute (either Kelvin or Rankine).
For the viscoelastic model used here, the differential equation of

the 1D model (3.8-73) becomes
dq 1 d
+ q = (3.8-94)
d d
and using (3.8-90), (3.8-94) can be written as
1
q + q = (3.8-95)
aT ( )

It is seen that the effect of temperature is to modify the time

constants. As the temperature increases, the modified time
constants become smaller, that is, the material relaxes more
quickly.
The convolution equation of the finite strain model becomes
d t Sij
t

Qij = exp

d (3.8-96)
0 d
and (3.8-96) is numerically approximated by

1 exp
t +t S t S
t +t
Qij = exp t Qij +

( ij ij )

(3.8-97)
The only additional consideration is to calculate , and this is

done using
t +t
1
= t
aT ( t )
dt (3.8-98)
This integration is performed numerically assuming that ln aT ( )

varies linearly over the time step.
Heat generation: A user-specified fraction of the energy

dissipated by the viscoelastic model can be considered as heat
generation. This heat generation can cause heating in a TMC
Specification of input: To add viscoelastic effects to the rubber-

like material model, you need to define the viscoelastic data using a
data set of type rubber-viscoelastic, then specify the number of the
rubber-viscoelastic data set in the rubber-like material model.

The rubber-viscoelastic data set includes:
A flag indicating whether the WLF or Arrhenius shift

function is used.
The shift function material constants C1 , C2
A table with one row for each chain. Each row contains ,
, the heat generation factor (fraction of dissipation
considered as heat generation, default value is 0.0), and the
usage flag (default value is deviatoric). There is no restriction
on the number of chains permitted. The usage flag can be
different for each chain.
In order to include time-temperature superposition, the material

must have a TRS temperature dependence.
It is seen that the dissipation calculation can be quite expensive.

Furthermore the dissipation is not required for the stress solution.
Therefore it is the default in ADINA to not perform the dissipation
calculation. The dissipation is only calculated for the chain
when the heat generation factor is non-zero.
3.8.2.2 Bergstrm-Boyce viscoelasticity
This subsection describes ADINAs implementation, with some

modifications, of the model described by:
ref. J.S. Bergstrm and M.C. Boyce, Constitutive modeling

of the large strain timedependent behavior of
elastomers, J. Mech. Phys. Solids, Vol. 46, pp 931954,
1998.
The viscoelastic model of Bergstrm and Boyce (1998) captures

well the time-dependent effects of macro-molecular reptation, or
the gradual, relative movement of entangled polymer chains.
This viscoelastic model is especially well suited for modeling the

finite viscoelastic response of filled rubbers, nitrile rubbers,
silicone rubbers, and other elastomers. It is available to users via

the command RUBBER-VISCOELASTIC BERGSTROM-BOYCE.
3.8.2.2.1 Rheological representation
This viscoelastic model is comprised of 2 parallel chains and is

implemented within ADINA's Parallel-Network Framework (see
Section 3.17). The first chain A is purely hyperelastic, and the
second chain B includes both viscous (time-dependent) and elastic
effects.
The parallel network assumption requires X = X A = X B , and the

multiplicative Krner-Lee decomposition for viscoelastic network
B is
X B = XeB XvB , (3.8-99)
where X eB and X vB are the elastic and inelastic distortions,

respectively.
3.8.2.2.2 Elastic response
Bergstrm and Boyce (1998) employ the eight-chain hyperelastic

function for both the purely hyperelastic chain A and the elastic
portion of chain B. ADINA, however, permits users to employ
Bergstrm-Boyce viscoelastic effects with any of ADINA's
hyperelastic models.
Users may scale the effects of the hyperelastic response in network

B using the factor RATIOBA. For example, if RATIOBA = 1.6 and
if both elastic networks are experiencing the same motions, then
the hyperelastic contribution of network B will be 60% greater than
that of network A.
When used with the Arruda-Boyce or eight-chain hyperelastic

models, RATIOBA is equivalent to the ratio CR( B ) CR( A) as
employed by Bergstrm and Boyce.

3.8.2.2.3 Viscous response
Bergstrm and Boyce (1998) postulate time-dependent viscous

behavior in network B of the form
B = K1 ( Bv 1)
K2
( B )
K3
, (3.8-100)
tr CvB T
where Bv = is the inelastic chain stretch, CvB = X vB X vB
3
is the inelastic right Cauchy-Green tensor, and
1
B = B : B , (3.8-101)
2 dev dev
1
( )
where Bdev = B tr B I is the deviatoric Cauchy stress
3
resulting from the elastic distortions.
However, rather than use equation (3.8-101), we use
B = K1 ( Bv 1 + K 4 ) (MB )
K2 K3
, (3.8-102)
where
1
MB = Bdev : Bdev , (3.8-103)
2
1
( )
and Bdev = B tr B I is the deviatoric Mandel stress.
3
This is consistent with the assumption of Mandel stress-driven
inelastic behavior in ADINA's parallel-network framework,
described in Section 3.17.2.
Further, unlike Bergstrm and Boyce (1998) who assume

l = l eB + DvB where l is the total spatial velocity gradient, and l eB

and DvB are, respectively, the elastic velocity gradient and inelastic
deformation rate for network B, we take the plastic spin to vanish
WBv = 0, and thus the relation (cf., equation 3.17-6)
1
l = l eB + XeB DvB XeB , (3.8-104)
is assumed. For more information about these kinematic

assumptions, see Section 3.17.1.
Heat generation: A user-specified fraction of the energy dissipated

by the viscous effects can be considered as heat generation. This
heat generation can cause heating in a TMC (thermo-mechanical-
coupling) analysis.
3.8.2.2.4 Constants and their interpretation
The Bergstrm-Boyce viscoelastic model is particularly powerful

given that its flow rule in equation (3.8-102) requires only 3
physically interpretable material constants and one numerically
required constant. Each parameters corresponding designation in
ADINA is given parenthetically in monospaced font and its
dimensions in brackets, when applicable.
1. The constant K1 (C1B) is a positive viscous deformation rate

1 M K3
constant of dimensions 2 (e.g.,
T L T
s 1 ( MPa )
K3
). As K1 0 , the viscous response vanishes,
and the model recovers the material's equilibrium response
(i.e., the special case of infinitely slow motions or load
application).
2. Constant K 2 1, 0] (C2B) is the dimensionless chain
stretch exponential. When K 2 = 0 , the viscous response is
insensitive to inelastic chain stretch.
3. Constant K 3 (MMB) is a positive, dimensionless stress
exponential which scales the effects of the deviatoric Mandel

stress on inelastic flow. As K 3 0 , the viscous response

becomes insensitive to the Mandel stress.
4. Parameter K 4 (EOFFSET) is a small positive number (say,
1.0E-6) that prevents numerical difficulties as the inelastic
chain stretch Bv 1 .
In addition to those constants associated with equation (3.8-

102), the Bergstrm-Boyce viscoelastic model requires the
following parameters.
1. RATIOBA: The ratio of hyperelastic contribution of network B

to that of network A, as described in section 3.8.2.2.2.
2. GENFAC: The heat generation factor (fraction of dissipation
energy convered into heat generation). If GENFAC = 0, the
dissipation is not calculated and there is no heat generation.
3. MAXITE: The Bergstrm-Boyce model requires an iterative
material stress update to obtain the Mandel stress and inelastic
distortion. The parameter MAXITE specifies the maximum
number of iterations that ADINA should use for this stress
update convergence.
4. RTOL: The tolerance required for convergence of the stress
update. It is not recommended to loosen this tolerance from the
default of 1.0E-12.
3.8.3 Mullins effect
When rubber is loaded to a given strain state, unloaded, then

reloaded to the same strain state, the stress required for the
reloading is less than the stress required for the initial loading.
This phenomenon is referred to as the Mullins effect.
The Mullins effect can be included in the rubber-like materials.

The material model used is the one described in the following
reference:
ref. R.W. Ogden and D. G. Roxburgh, A pseudo-elastic

model for the Mullins effect in filled rubber, Proc. R.
Soc. Lond. A (1999) 455, 2861-2877.
We briefly summarize the main concepts below.

Fig 3.8-9 shows the Mullins effect in simple tension. On initial

loading to force Fc , the specimen follows the force-deflection
curve a-b-c. When the load is removed, the specimen follows the
unloading curve c-d-a. On reloading to force Fc , the specimen
follows the reloading curve a-d-c, and on further loading to force
Ff , the specimen follows the loading curve c-e-f. When the load
is removed, the specimen follows the unloading curve f-g-a, and,
on reloading to force Fg , the specimen follows the reloading curve
a-g-f.
Force
h
f
e
c
b
d g
a
Deflection
Figure 3.8-9: Mullins effect loading-unloading-reloading curves
Note that any permanent set associated with the Mullins effect
is not included in the Ogden-Roxburgh model used here.
The Ogden-Roxburgh model, as implemented in ADINA, uses
the following strain energy density expression:
W = WD ( ij ) + ( ), all except hyper-foam

(3.8-105a,b)
=W ( ij ) + ( ), hyper-foam
where W ( ij ) is the total elastic strain energy density, WD ( ij ) is

the deviatoric elastic strain energy density, is an additional
solution variable describing the amount of unloading and ( ) is
the damage function. W is referred to as the pseudo-energy
function. In our implementation, the deviatoric strain energy
density is used for the (almost) incompressible materials and the

total strain energy density is used for compressible materials. For

ease of writing, we discuss only the case of compressible materials;
for incompressible materials, replace W by WD in the equations
below.
is computed as
1 1
= 1 erf (Wm W ) (3.8-106)
r m
where erf( x) is the error function
x
2
erf( x ) = exp ( u ) du
2
(3.8-107)
0
Wm is the maximum value of W encountered during the

deformation history and m and r are material constants.
( ) is defined by
d ( )
= W (3.8-108)
d
and is computed by numerical integration of ( ) = W . For a

given value of Wm , there is a minimum value of computed as
1 W
m = 1 erf m (3.8-109)
r m
The value of ( ) at = m is written ( m ) . (Note: the

subscript m in the term Wm means maximum, but the subscript m
in the term m means minimum.) The time rate of change of
( m ) can be written as
1 W
( m ) = (1 m ) Wm = erf m Wm (3.8-110)
r m

Physically, ( m ) is interpreted as the dissipation.

During loading, Wm = W , Wm > 0 and = 1 . Therefore
( ) = 0 and ( m ) > 0 during loading.
During unloading or reloading, Wm > W , Wm = 0 and
m < 1 . Therefore ( ) 0 and ( m ) = 0 during unloading
or reloading.
Material constants m and r do not have any direct physical
significance. However Fig 3.8-10 shows the dependence of an
unloading-reloading curve in simple tension on these parameters.
It is seen that, for an unloading-reloading loop in which Wm >> m ,
the initial slope of the reloading curve is reduced by the factor
1
1 . r must therefore be greater than 1.
r
Force
Slope kb
Fb
2 1
Slope k b $ Fb2
Slope ka 4 rm
Deflection
( 1 *W + )
Slope , 1 # erf . m / - k a
2 r 0 m 13
Wm = strain energy density at b
Figure 3.8-10: Dependence of reloading curve on Mullins effect

material constants
It can also be shown that the dissipation of a loading-unloading

cycle, as shown in Fig 3.8-11, can be written as

m Wm W 1 W 2
E
= 0 de = erf m 1 exp m
r m m
A m
(3.8-111)
where
C
Wm = 0 de (3.8-112)
A
0 is the engineering stress, e is the engineering strain. Therefore,

given and Wm from two loading-unloading cycles of different
amplitude, m and r can be computed.
=Force/original area
0
B
Dissipation = shaded area
D
A
E e=displacement/
original length
Figure 3.8-11: Dissipation in Mullins effect in a loading-reloading

cycle
Heat generation: A user-specified fraction of the energy

dissipated by the Mullins effect model can be considered as heat
generation. This heat generation can cause heating in a TMC
Specification of input: To add Mullins effects to the rubber-like

material model, you need to define the Mullins data using a data set
of type rubber-Mullins, then specify the number of the rubber-
Mullins data set in the rubber-like material model.
The rubber-Mullins data set includes:

The material constants r and m.
The heat generation factor (fraction of dissipation considered

as heat generation). The default value is 0.
3.8.4 Orthotropic effect
A 2D network of orthotropic fibers can be included in the rubber-

like material descriptions, see Figure 3.8-12. The directions of the
fibers are described by the normal vectors n a , nb , with
components ( na )i and ( nb )i respectively. These normal vectors
are defined in the undeformed configuration, and can change
directions due to material deformations.
nb na
Figure 3.8-12: Network of orthotropic fibers
The orthotropic fibers are included by appending to the strain

energy density the following additional term
WO =
k1
2k 2
{
exp k2 ( J 4 1) + exp k2 ( J 6 1) 2 (3.8-113)

2

2
}
where J 4 = I 4 I 31/ 3 , J 6 = I 6 I 31/ 3 , I 4 = Cij ( na )i ( na ) j ,
I 6 = Cij ( nb )i ( nb ) j , Cij is the Cauchy-Green deformation tensor,

and k1 , k2 are material constants. k1 has the dimension of stress
and k2 is dimensionless. In the expression in brackets in (3.8-113),
the first term represents the response of the n a fibers and the

second term represents the response of the nb fibers.

This material description is described in the following
reference:
G. A. Holzapfel, T. C. Gasser, R. W. Ogden, A New

Constitutive Framework for Arterial Wall Mechanics and a
Comparative Study of Material Models, Journal of Elasticity,
61: 1-48, 2000.
We note:
The normal directions n a and nb are defined as shown in
Figure 3.8-13. In 2D analysis, the figures show how n a and nb
are defined in terms of the input material axis angle and the
offset angles a , b . n a and nb lie in the plane of the element.
In axisymmetric analysis, it is also allowed to set nb to the hoop
(x) direction. In 3D analysis, the figure shows how n a and nb are
defined in terms of the material directions a and b . n a and nb lie
in the plane defined by a and b .
n a and nb need not be perpendicular to each other.
1 nb
2 b

a
b
4
a
3
na
a) 2D element, hoop option not used

z
1
y
2

a
b
4
a
3
na
b) 2D element, hoop option used, nb = x
Figure 3.8-13: Orientation of orthotropic axes
2 1
3 4
7 8 c
nb
b
a
b
a
na
c) 3D element, na, nb lie in the a-b plane

I 4 is interpreted as the square of the stretch of the fiber initially

orientated in direction n a ; I 6 is interpreted as the square of the
stretch of the fiber initially orientated in direction nb ; see Ref KJB,

Example 6.7.
J 4 is interpreted as the deviatoric part of I 4 ; J 6 is interpreted
as the deviatoric part of I 6 . It is assumed that the material is nearly
incompressible.
When J 4 < 1 , the material is in compression in direction n a ;
when J 6 < 1 , the material is in compression in direction nb . It is
possible to remove the orthotropic effect when the material is in
compression in either of the fiber directions.
Specification of input: To add orthotropic effects to the rubber-

like material model, you need to define the orthotropic data using a
data set of type rubber-orthotropic, then include the number of the
rubber-orthotropic data set in the rubber-like material model
description.
The rubber-orthotropic data set includes:
The material offset angles a , b (default values are 0).
The material constants k1 , k2 .
A flag indicating whether the orthotropic effect is removed

when the material is in compression in either of the fiber
directions (the default is to remove the orthotropic effect
when the material is in compression).
A flag indicating whether direction nb is the x (hoop)

direction (the default is that nb is not the x (hoop) direction).
3.8.5 Thermal strain effect
When the material is temperature-dependent, you can include a

coefficient of thermal expansion. The coefficient of thermal
expansion is ( ) , where is the temperature. Notice that the
coefficient of thermal expansion is temperature dependent.
The thermal strain is calculated as

eth = ( ) ( REF ) ( 0 ) ( 0 REF ) (3.8-114)
where 0 is the initial temperature and REF is the material

reference temperature. This is the same formula as is used for the
other thermo-elastic materials, see Section 3.5.
The thermal strain is assumed to be isotropic.
When the thermal strain is non-zero, the deformation gradient
X is assumed to be decomposed into a thermal deformation
gradient Xth and a mechanical deformation gradient X m , using
X = X m Xth (3.8-115)
The thermal deformation gradient is

Xth = (1 + eth )I (3.8-116)
therefore the mechanical deformation gradient is
X m = (1 + eth ) 1 X (3.8-117)
the mechanical Cauchy-Green deformation tensor is
Cm = (1 + eth ) 2 C (3.8-118)
and the mechanical Green-Lagrange strain tensor is
1
m = (1 + eth ) 2
2
(1 (1 + eth ) 2 ) I (3.8-119)
For small thermal strains, (3.8-119) reduces to m eth I , so

that the strains are nearly the sum of the mechanical and thermal
strains, as in small strain analysis. However, we do not assume that
the thermal strains are small.
The strain energy densities are computed using the mechanical
deformations. This is done by computing all invariants and
stretches using the mechanical deformations, e.g. the mechanical
Cauchy-Green deformation tensor.
The 2nd Piola-Kirchhoff stresses are obtained by differentiating
the strain energy density with respect to the total strains. Since the
strain energy density is a function of the mechanical strains, we

obtain
1 W W
Sij = +
2 ( )ij () ji
1 W ( m ) ab W ( m )ba
= + (3.8-120)

2 ( m ) ab ( )ij ( m )ba ( ) ji
2 1 W W
(
= (1 + eth ) ) +
2 ( m )ij ( m ) ji
With this definition, the 2nd Piola-Kirchhoff stresses are conjugate

to the Green-Lagrange strains.
3.8.6 TRS material
The rubber-like materials can include the TRS

(thermorheologically simple) material assumption. In this
assumption, the material properties are independent of temperature,
but the viscoelastic time constants of the Holzapfel viscoelastic
model are functions of temperature as described in the viscoelastic
material effect (time-temperature superposition).
A TRS material requires a rubber-table of type TRS. The
number of the rubber-table data set is specified as part of the
rubber-like material description.
The rubber-table of type TRS data set is a table with a different
temperature for each row of the table. Each row of the table
contains the coefficient of thermal expansion (enter 0 if there are
no thermal effects).
We do not include the absolute temperature in the strain energy
density function, so we are not capturing the Gough-Joule
thermoelastic effects, described in
ref. Gurtin, Fried, Anand, The Mechanics and

Thermodynamics of Continua, Cambridge University
Press, 2010, Section 58.
3.8.7 Temperature-dependent material

The rubber-like materials can be fully temperature-dependent.
Specification of input: The input of the material properties for a

fully temperature-dependent material is done using a rubber-table.
The rubber-table has different forms depending upon whether the
Bathe or eight-chain relationships are used, hence there is a rubber-
table of type Mooney-Rivlin, a rubber-table of type Ogden, etc.
For brevity, we discuss only the case of a Mooney-Rivlin material.
A Mooney-Rivlin material that is fully temperature-dependent
requires a rubber-table of type Mooney-Rivlin. The number of the
rubber-table data set is specified as part of the Mooney-Rivlin
material description.
The rubber-table of type Mooney-Rivlin data set is a table with
a different temperature for each row of the table. Each row of the
table contains
The coefficient of thermal expansion.
The material properties (C1 to C9, constants D1 to D2 and the
bulk modulus ).
The curve-fitting number (0 if curve-fitting is not used).
The rubber-viscoelastic data set number, if the model
includes viscoelastic effects.
The rubber-Mullins data set number, if the model includes
Mullins effects.
The rubber-orthotropic data set number, if the model
includes orthotropic effects.
You can specify a different curve-fitting number for each

temperature.
You can specify a different rubber-viscoelastic data set number,
rubber-Mullins data set number or rubber-orthotropic data set
number for each temperature. In this way, the material constants
for the viscoelastic, Mullins or orthotropic effects can be
temperature-dependent. However, if one row of the rubber-table
has viscoelastic data, all rows must have viscoelastic data.
Similarly, if one row of the rubber-table has Mullins or orthotropic
data, all rows must have this data.

Solution process: The strain energy density is computed by

interpolation from the strain energy densities of the bounding
temperatures as follows: Let be the current temperature, 1 be
the temperature in the rubber-table just below the current
temperature and 2 be the temperature in the rubber-table just
above the current temperature. Then the program computes the
strain energy densities W1 and W2 , and the derivatives of the strain
energy densities, using the material properties for 1 and 2
respectively. Finally the program computes the strain energy
density and its derivatives using linear interpolation, for example
2 1
W= W1 + W2 (3.8-121)
2 1 2 1
3.8.8 Rubber stability indicators
In this section we give the theory underlying the stability indicators

for the rubber-like material models. The stability indicators can be
plotted in the AUI, using the command
MATERIALSHOW STRAIN CURVETYPE=STABILITY
3.8.8.1 Introduction
We consider the stability of a unit cube of incompressible

hyperelastic material. In this investigation, we only consider
homogeneous deformations, and only stability considering
perturbations that are also homogeneous.
The cube is subjected to applied loads R1 , R2 , R3 (Fig 3.8-14).

These loads are deformation-independent. Since the cube is a unit
cube, these loads can also be interpreted as engineering stresses.
The corresponding displacements of the cube are u1 , u2 , u3 . The

stretches and true strains corresponding to these displacements are
i = 1 + ui and ei = ln i , again using that the cube is a unit cube.

R3, u3
R2, u2
R1, u1
Figure 3.8-14 Unit cube under applied loads
The strain energy of the cube is W = W (e1 , e2 , e3 ) , with different

strain energy functions for different material models (e.g. Mooney-
Rivlin, Ogden, Arruda-Boyce, Sussman-Bathe, eight-chain).
Since the material is incompressible, we must enforce the condition

e1 + e2 + e3 = 0 .
Also note, in the derivation below, the summation convention is

not used.
Equilibrium
Equilibrium, along with incompressibility, is determined using a

Lagrange multiplier approach:
W = W + m(e1 + e2 + e3 ) (3.8-122)
where m is the Lagrange multiplier. Taking variations of (3.8-122)

gives
W
W = + m ei + (e1 + e2 + e3 ) m (3.8-123)
i ei
The virtual work done by the loads is
R = Ri ui = Ri i ei (3.8-124)
i i

and equating the external and internal virtual work gives
W
+ m = Ri i , no sum on i , (3.8-125a)
ei
e1 + e2 + e3 = 0 (3.8-125b)
3.8.8.2 Stability
The stability criterion is obtained taking increments of the loads

and determining the corresponding increments of the
displacements. The resulting matrix equation is
2W 2W 2W
2 R11 1
e1 e1e2 e1e3 e
2W 2W 1 1R1
R2 2 1 e2 2 R2
e2 e3 = (3.8-126)
2
e2
e3
3 R3

2W
R33 1 m 0
e3
2

symm 0
Equation (3.8-126) can of course be solved for ei and m given

Ri , but this does not give insight into the stability. The
eigenvalues and eigenvectors of the coefficient matrix could be
solved for, but the eigenvectors need not satisfy incompressibility.
To continue, we note that the stability matrix equation (3.8-126)

has the form
1 e1 r1
KEE 1 e2 r2
= (3.8-127)
1 e3 r3

1 1 1 0 m 0
where KEE is a symmetric 3x3 matrix. We now choose a new

basis for the strain increments:

e1 y1
e = A y (3.8-128)
2 2
e3 y3
2 1
0
3 3
1 2 1 1
where A = . A is orthogonal with
2 3 2 3

1 2 1 1
2 3 2 3
determinant=1. We see that
2 1 2 1 2

3 2 3 2 3 e

1y 1
e 1
y = AT e = 1

1
2 2 0 e2 (3.8-129)
2 2
y3 e3 e
1 1 1 3

3 3 3
Note that y1 and y2 are insensitive to the volume change and y3

is proportional to the volume change e1 + e2 + e3 .
In the new basis, the stability matrix equation has the form
0 y1 s1
KYY 0
y2 = s2 (3.8-130)
3 y3 s3

0 0 3 0 m 0
s1 r1
where KYY = A KEE A and s2 = AT
T r .
2
s3 r3

Equation (3.8-130) is solved noting that the 4th row equation is

satisfied by the choice y3 = 0 ; thus the matrix stability equation
reduces to
KYY11 KYY12 0 y1 s1

KYY12 KYY22 0 y2 = s2 (3.8-131)
KYY KYY23 3 m s3
13
and (3.8-131) is solved by
KYY11 KYY12 y1 s1
KYY = (3.8-132a)
12 KYY22 y2 s2
1
m = ( s3 KYY13 y1 KYY23y2 ) (3.8-132b)
3
Thus we conclude that stability is governed by the eigenvalues of

KYY11 KYY12
KYY , and the stability criterion is the smallest
12 KYY22
eigenvalue of this matrix.
3.8.8.3 Stability in uniaxial tension
W
In uniaxial tension, R2 = R3 = 0 , so, from equilibrium, m =
e2
W W 1
and R11 = . Also, at equilibrium, e2 = e3 = e1 .
e1 e2 2
However, there are still three admissible strain increments e1 ,
e2 , e3 , so the 4x4 matrix stability equations given above ((3.8-
126) to (3.8-132)) are applicable.
3.8.8.4 Stability in pure shear
In pure shear, e2 = 0 , hence e2 = 0 and there are only two

admissible strain increments. Here R3 = 0 , so, from equilibrium,

W W W
m= and R11 = . Also, at equilibrium, e3 = e1 .
e3 e1 e3
The matrix stability equation becomes
2W 2W
2 R11 1
e1 e1e3 e r
2W 1 1
1 e3 = r3 (3.8-133)
e32 m 0
symm
0

This is solved using the change of basis
e1 y1
e = A y (3.8-134)
3 2
1 1
2 2
where now A = , so that the matrix stability equation
1 1
2 2

(3.8-133) becomes
0 y1 s1
KYY
2 y2 = s2 (3.8-135)

0 2 0 m 0
and (3.8-135) is solved using
1
KYY11y1 = s1 , y2 = 0 , m = ( s2 KYY12 y1 ) (3.8-136a,b,c)
2
Stability is governed by coefficient KYY11 . (Note, this is a different

KYY11 than is used for the cases of 3 admissible strain increments.)

3.8.8.5 Stability in equibiaxial tension
In equibiaxial tension, R1 = R2 and R3 = 0 , so, from equilibrium,

W W W
m= and R11 = R2 2 = . Also, at equilibrium
e3 e1 e3
e2 = e1 and e3 = 2e1 . Since there are still three admissible strain
increments, e1 , e2 , e3 , the 4x4 matrix stability equations given
above ((3.8-126) to (3.8-132)) are applicable.
3.8.8.6 Comments
The results from the stability indicators can be checked by running

a corresponding ADINA finite element model, using dead loads.
When the stability indicator is positive, the stiffness matrix has all
positive pivots; when the stability indicator is negative, the
stiffness matrix has at least one negative pivot.
3.9 Geotechnical material models

3.9.1 Curve description material model
The curve description model can be employed with the 2-D

solid (plane strain and axisymmetric) and 3-D solid elements.
The curve description model can be used with the small

displacement and large displacement formulations. In all cases,
small strains are assumed.
materially-nonlinear-only formulation is employed, and when used
with the large displacement formulation, a TL formulation is
employed.
The curve description model is a simple incremental stress-

strain law used to represent the response of geological materials.
The model describes the instantaneous bulk and shear moduli as
piecewise linear functions of the current volumetric strain, as
shown in Fig. 3.9-1. An explicit yield condition is not used and
whether the material is loading or unloading is determined by the
history of the volumetric strain only.

3.9: Geotechnical material models
To present the governing constitutive relations, let

t
eij = total strains (the left superscript "t" always refers to
time t)
eij = incremental strains
t
eii
t
em =
i 3
= mean strain (negative in compression)
eii
em = 3
i
= incremental mean strain
t
gij = t eij t em ij = deviatoric strains
gij = incremental deviatoric strains
t
ij = total stresses (negative in compression)
ij = incremental stresses
t
ii
t
m =
i 3
= mean stress
ii
m =
i 3
= incremental mean stress
pmin = minimum mean stress ever reached

t
sij = t ij t m ij = deviatoric stresses
sij = incremental deviatoric stresses
t
G, t K = shear and bulk moduli
The incremental stress-strain relations using the curve description

model are then
sij = 2 t G gij
m = 3 t K em

K LD
Loading bulk modulus

l
l
t
KLD l
l l l
e1v e2v te
v e3v e4v e5v e6v ev
KUN
l Note: Compressive volume strain
Unloading bulk modulus
is assumed positive
l
t
KUN l
l
l l
e1v e2v te v e3v e4v e5v e6v ev
G LD t
K t
Note: tG UN = t UN GLD
Loading shear modulus
K LD
l l
t l
GLD
l
l
l
e1v e2v te v e3v e4v e5v e6v ev
Figure 3.9-1: Input of the curve description model

t
The instantaneous bulk and shear moduli, K and tG, are
t
functions of the loading condition, and the volumetric strain ev is
defined as
t
e = egrav + ( 3 t em )
where egrav is the volumetric strain (three times the mean strain and
t
taken positive in compression) due to the gravity pressure and em
is the mean strain at time t. Defining emin as the minimum mean
strain ever reached during the solution, we have that the material is
loading if t em emin and the material is unloading if t em > emin
i.e.,
tK when t em emin
t
K = t LD
KUN when t em > emin
and
tG when t em emin
t
G = t LD
GUN when t em > emin
Note that the loading conditions for both the bulk and the shear
t
moduli are determined by the history of em only. The values of
t
KLD, tKUN, and tGLD are obtained using the curves in Fig. 3.9-1,
t KUN
and the modulus t GUN = t GLD t .
K LD
The incremental solution at time t is obtained using the

equations
t +t
sij = 2 t G ( t +t
gij t g ij ) + t sij
and
t +t
ij = 3 t K ( t +t
eij t eij ) + t ij
t t
To obtain G and K, we check whether the loading or unloading
conditions are active by comparing the current volumetric strains
and previous volumetric strains. To start the procedure at time 0,

loading conditions are assumed. It should be noted that the stresses

at time t + t are calculated using the material moduli pertaining to
time t.
An important additional analysis option is that the material can
weaken under loading conditions if tensile stresses exceed
preassigned values. Since the curve description model has been
developed primarily for the analysis of geological materials, the
material weakening is assumed to occur once the principal tensile
stresses due to the applied loading exceed the in-situ gravity
pressure p (taken positive). The gravity pressure p is calculated as
N
p = hi pi where hi are the element interpolation functions and
i =1
pi is the pressure at the element nodes. The nodal pressure pi is
calculated as pi = Z i where is the material density and Zi the
nodal Z coordinate, which coincides with the vertical direction.
The material weakening can be included using a tension cut-off
model or a cracking model.
In both modes of behavior (tension cut-off and cracking) the

principal stresses due to the external loading are calculated and
compared with the in-situ gravity pressure. Once the principal
tensile stress is equal to the in-situ gravity pressure, the material
is treated as being orthotropic, with the modulus corresponding
to the direction of the principal tensile stress being multiplied by
a stiffness reduction factor (an input parameter). Another
factor, also an input parameter, is applied to reduce the shear
stiffness.
In the tension cut-off mode, the normal and shear stiffnesses

corresponding to the direction of the maximum principal tensile
stress are reduced, but the stresses are fully retained. The model
therefore simulates elastic-plastic flow of the material.
On the other hand, in the cracking mode the stiffnesses are
reduced, and in addition the normal and shear stresses due to
external loads and corresponding to the direction of stiffness
reduction are released. The reduction of stiffness together with
the stress release models a crack, i.e., a tensile failure plane, that
forms at right angles to the direction of the maximum principal
tensile stress.

In subsequent load steps, tension cut-off or cracking only in

the direction already determined and/or other(s) perpendicular
to it may be active, i.e., no change in the tension cut-off or
cracking direction is considered. The tension cut-off or
cracking mode is considered to be inactive provided the strain
normal to the crack (in the direction of tension cut-off) becomes
both negative and less than the strain at which the failure
occurred initially; otherwise it remains active.
Fig. 3.9-2 illustrates the tension cut-off and cracking options.
Note that the Z direction should be the vertical direction and

that the ground level is assumed to be at Z = 0 in the calculation of
the in-situ gravity pressure.
a
x2
a
ab
Maximum a =
x1 in-situ gravity pressure
ea
(a) Option of tension cut-off
a
x2
a
ab a failure stress =
in-situ gravity pressure
x1
ea
(b) Option of cracking
ADINA checks for tension cut-off or cracking at each integration point.
Figure 3.9-2: Tension cut off and cracking options

The in-situ gravity pressure is evaluated using the input for

material density which corresponds to the weight per unit volume.
This is different than the mass density which is used in the element
mass matrix calculation.
3.9.2 Drucker-Prager material model
The Drucker-Prager model is based on
The Drucker-Prager yield condition (see p. 604, ref. KJB)
A non-associative flow rule using the Drucker-Prager and

cap yield functions
A perfectly-plastic Drucker-Prager yield behavior
Tension cut-off
Cap hardening
The Drucker-Prager model can be used with the 2-D solid and
3-D solid elements.
The Drucker-Prager model can be used with the small

displacement/small strain and large displacement/small strain
formulations.
and when used with the large displacement/small strain
formulation, a TL formulation is employed.
Fig. 3.9-3 summarizes some important features of the Drucker-

Prager model (with tension cut-off and cap hardening).
The Drucker-Prager yield function is given by:
t
f DP = t J1 + t
J2D k
where and k are material parameters and are functions of the

friction angle and cohesion coefficient c , and t J1 and t J 2 D are

the first stress invariant and the second deviatoric stress invariant,
respectively.
The corresponding potential function is given by
t
g DP = t J1 + t
J2D u
where is a function of the dilatation angle , u is not required

as input parameter.
The parameters , k and must be properly defined. They
are dependent on the type of application to be modeled. In
particular, they can be determined by matching the Drucker-Prager
criterion to the Mohr-Coulomb criterion (see Section 3.9.4 and
Figs. 3.9-6 through 3.9-9 for details).
A compressive meridian match gives:
2sin 6c cos 2sin

= k= =
3 ( 3 sin ) 3 ( 3 sin ) 3 ( 3 sin )
A tensile meridian match gives:
2sin 6c cos 2sin

= k= =
3 ( 3 + sin ) 3 ( 3 + sin ) 3 ( 3 + sin )
A plane strain match gives:
tan 3c
= k=
9 + 12 tan 2
9 + 12 tan 2
The cap yield function depends on the shape of the cap. For a
plane cap:
t
f C = t J1 + t J 1a
where t
J 1a is a function of the volumetric plastic strain t eVP :

1 e
t P
t
J 1 = ln 1 V + 0 J 1a
a
D W
0
where J 1a is the cap initial position and W and D are material
constants.
#

a) Drucker-Prager yield surface in principal stress space
Plane cap
J2D
Yield surface
Tension
cut-off H
1
Elliptical cap
k Elastic region
O A C -J1
T
a
- J1 (plane cap)
a
- J 1 (elliptical cap)
b) Drucker-Prager model in J1 versus J2D space
Figure 3.9-3: Drucker-Prager model
For an elliptical cap:

f C = ( t J1 + t L ) + R 2 ( t J 2 D t B 2 )
t 2
t
where B is the vertical semi-axis of the ellipse (AH), R is the cap
t
ratio (AC/AH) and L is equal to OA. The dependence of 0 J 1a on
the volumetric plastic strain is the same as for the plane cap. In the
case of tension cut-off, T is the maximum value that t J 1a can take.
We note that:
Unlike for the von Mises model, the Drucker-Prager yield

function cannot be used with hardening (the material is always
assumed to be elastic-perfectly-plastic, except for cap
hardening).
If (f) approaches zero (minimum value of is taken to be

10 ), the initial position of the cap 0 J 1a is moved far to the
-5
right in Figure 3.9-3 and is not reached in the analysis, and the
position of the tension cut-off T is moved far to the left and is
also not reached in the analysis, then the Drucker-Prager yield
condition approaches the von Mises elastic-perfectly-plastic
yield condition.
In the case of Drucker-Prager yielding, the dilatency

(volume expansion of the material in shear) is governed by the
magnitude of (y) (for the von Mises yield condition = 0
and there is no dilatency).
Cap yielding leads to an increase of compressive plastic

volumetric strain. If the plane cap is used, there is no change of
deviatoric plastic strains, while if the elliptical cap is used, the
deviatoric components of the plastic strain change during the
cap yielding. Fig. 3.9-4 shows the relation between the plane
cap position t J 1a and the volumetric plastic strain
e = t e11P + t e22
t P
V
P
+ t e33P
that is, the constraint that exists between these two quantities.
We note that the plane cap movement corresponds to hardening

and that the volumetric plastic strain is constrained to be smaller

than the input parameter W in absolute value.
t a
J1
Material parameters
W and D are negative.
- _1
D
0 a
J1
Possible volumetric te P
v
plastic strain (negative)
W
a
Figure 3.9-4: Relation between plane cap position tJ 1 and volumetric
plastic strain tePv
In the case of vertex yielding (the stress state is represented

by point H in Fig. 3.9-3), the plastic deformation corresponds to
the Drucker-Prager and cap yielding. The vertex yielding leads
to changes in volumetric and deviatoric plastic strains with the
cap hardening behavior.
Stress states beyond the tension cut-off value T (input to

ADINA as a positive value) are not possible. When this
limit is reached or exceeded, the program sets all the shear
stress components equal to zero and the normal stress
T
components all equal to . The elastic constitutive law is
3
used for the stiffness matrix.

3.9.3 Cam-clay material model
The Cam-clay material model is a pressure-dependent plasticity

model. It is based on
An associated flow rule using an elliptical yield function
The critical state line, which controls the failure of the

material
The consolidation behavior of clayey materials
The Cam-clay model can be used with the 2-D solid and 3-D
solid elements.
The Cam-clay model can be used with the small

formulations.
materially-nonlinear-only formulation is employed and when used
with the large displacement/small strain formulation, a TL
The Cam-clay model is able to simulate the following

mechanical behaviors for clayey materials, which are confirmed by
lab tests and in-situ tests:
Strain hardening and softening under normal consolidation

states or overconsolidation states
Nonlinear dependence of the elastic volumetric strain on the

hydrostatic pressure
An ultimate condition of perfect plasticity at which plastic

shearing can continue indefinitely without changes in volume or
effective stress
The Cam-clay yield function is given by
t
q2 t t
t
F= 2
+ p ( p t p0 ) = 0
M

where the mean stress t p and the distortional stress t q at time t

are related to the first stress invariant t I1 and the second deviatoric
t
I1
stress invariant J 2 by p =
t t
and t q = 3 t J 2 . M, a material
3
constant, is the slope of the critical state line and t p0 , called the
pre-consolidation pressure, is the diameter of the ellipsoid at time t
along the axis p. All of the variables are shown in Fig. 3.9-5(a).
The hardening rule is written as
q
Critical state line
p=tpcs
Elliptical yield
surface
(tp,tq)
M
1
p=t p0 p
a) Cam-clay yield surface
v
N Isotropic consolidation line
% Critical state line
' 1 Overconsolidation line
%
1
5
1
(tp,tv)
p=1 p=tpcs p=t p0 ln p
b) Compression behavior
Figure 3.9-5: Cam-clay model

t
p0
t
v = N ln t p0 + ln t
p
where t v is the specific volume at time t and t v P is the plastic

specific volume rate at time t. , and are material constants,
in which is the slope of the isotropic consolidation line, is the
slope of the overconsolidation line and is the specific volume at
the critical state when t p is 1.0. is related to N, the specific
volume at the isotropic consolidation state when t p is 1.0, by
= + ( ) ln p0
The effective bulk modulus at time t can be expressed as
t
t vt p
K=

The corresponding shear modulus at time t is obtained using

3 (1 2 ) t
t
G= K , in which is the Poissons ratio (constant
2 (1 + )
throughout the analysis).
In an analysis using the Cam-clay model, ADINA requires

either the initial stresses or an initial stiffness to be defined.
Initial stresses: The initial stresses are either directly input or

elastically calculated before or at the first load step for any
specified element group. They can be tiny isotropic
compressive stresses (such as 5% of the gravity-induced
stresses), or can be the full gravity-induced stresses.
When the analysis starts with initial stresses, the initial size
of the yield surface can either be directly input or computed
from the initial stresses. If the initial stresses are induced by the
full gravity load, the initial size of the yield surface (pre-
consolidation pressure) would be estimated by employing the
input parameters OCR and KNULL, unless the initial size of the

yield surface is specified in the Cam-clay model definition.

If the initial stresses are not directly input, ADINA calculates
the initial stresses as the stresses computed at load step 1. Then,
(a) The results for load step 1 are calculated using the input
elastic material properties E and .
(b) The results for the remaining load steps depend upon the
solution for load step1.
When the analysis starts with initial stresses, ADINA takes
the true initial size of the yield surface as the greater of the
specified initial size and the computed initial size.
Initial stiffness: The analysis can be performed with an initial

stiffness, which is calculated using the initial Youngs modulus
and Poissons ratio. In this case, the initial size of the yield
surface must be directly specified.
Parameter KNULL is defined as the coefficient of earth pressure

at the normal consolidation state. Usually it is different than the
coefficient of current earth pressure. KNULL is approximated by
K 0 = 1 sin where is the internal friction angle of the soil.
The material constant M is computed based on . If triaxial

6sin
compression tests are performed, M is given by M = . If
3 sin
triaxial extension tests are performed, M is given
6sin
by M = .
3 + sin
For more information, see references:
ref. D.M. Wood, Soil behavior and critical state soil

mechanics, Cambridge University Press, pp. 84-225,
1990.
ref. A.M. Britto and M.J. Gunn, Critical state soil mechanics
via finite elements, Ellis Horwood Limited, pp. 161-184,
1987.

3.9.4 Mohr-Coulomb material model
The Mohr-Coulomb model is based on
a linearly elastic and perfectly plastic yield behavior;
a yield function defined by the Mohr-Coulomb criterion;
a potential function defined by the Drucker-Prager criterion;
a piecewise tension cut-off analysis; and
a non-associative flow rule.
The Mohr-Coulomb model can be used with the 2-D solid and
3-D solid elements.
The Mohr-Coulomb model can be used with the small

formulations.
When used with the small displacement/small strain formulation, a
materially-nonlinear-only (MNO) formulation is employed, and
total Lagrangian (TL) formulation is employed.
The Mohr-Coulomb yield function is given by
t
f MC =sin t I1
+
1
2
(
3 (1 sin ) sin t + 3 ( 3 + sin ) cos t ) t
J 2 3c cos
where

1 3 3 t J3
= arc cos
t
,
( 2)
3 2 t J 3/ 2
is the angle of similarity or Lode angle defined in the deviatoric

stress plane (see Fig. 3.9-8), and 0 t . In addition, note that
3

is the friction angle (a material parameter) such that

0 < ;
2
c is the cohesion (a material parameter) such that c 0 ;

t
I1 is the first stress invariant at time t;
t
J 2 is the second deviatoric stress invariant at time t; and
t
J 3 is the third deviatoric stress invariant at time t.
In the stress space, the Mohr-Coulomb yield function is an irregular

hexagonal pyramid with its apex on the hydrostatic axis (see
Fig. 3.9-6).
The Mohr-Coulomb potential function is defined by the

Drucker-Prager criterion such that its surface is inscribed in the
Mohr-Coulomb surface defined by the dilatation angle (a
material parameter), where 0 (see Figs. 3.9-7 and 3.9-10).
Consequently, the Mohr-Coulomb potential function is given by
g MC = MC I1 +
t
t t
J 2 kMC
where
2sin
MC = .
3 (1 sin ) sin MC + 3 ( 3 + sin ) cos MC
The angle MC is determined by inscribing a Drucker-Prager

surface to the Mohr-Coulomb surface defined by the dilatation
angle (see Fig. 3.9-10 assuming that = ) and kMC is not
needed by the model. It is important to mention that this is the
default Mohr-Coulomb potential function. Moreover, this function
provides a plastic return to the yield surface that is unique and
radial in the deviatoric stress plane. However, in the meridional

plane the plastic return is not unique at the apex of the cone or in
the tension cut-off region.
The tension cut-off (T) defines a limiting surface in the stress

space via
I1 T = 0,
where T 0 (see Fig. 3.9-7). The tension cut-off (a material

parameter) can be taken to be three times the tensile strength of the
material. This limiting surface combined with the Mohr-Coulomb
yield surface can better approximate the behavior of a material
when tensile stresses occur and completely define ADINA's Mohr-
Coulomb yield surface. Note that when a tension cut-off analysis is
applied by the model, the hydrostatic stress components are shifted
to the hydrostatic pressure corresponding to the tension cut-off
value.
Drucker-Prager
criterion
# 1'
Mohr-Coulomb
criterion
# 3'
# 2'
Figure 3.9-6: The yield surfaces of the Mohr-Coulomb model

and the Drucker-Prager model in stress space

Mohr-Coulomb yield surface
J2 Mohr-Coulomb potential surface
Tension
cut-off
Mohr-Coulomb potential surface
(with temperature effects)
# I1
Figure 3.9-7: The Mohr-Coulomb model in the

I1 vs J2 plane (or meridional plane)
Drucker-Prager yield
surface at a tensile Drucker-Prager yield
# 1' surface at a compressive
meridian match ( = 0)
meridian match ( = 4/3)
Mohr-Coulomb
yield surface
# 3'
# '2
Figure 3.9-8: Shapes of the Mohr-Coulomb and the

Drucker-Prager yield criteria in the 4 plane

Drucker-Prager
B p
ij
yield surface
Mohr-Coulomb
yield surface
ijp ijp
A C
Figure 3.9-9: Corner treatment of the

Mohr-Coulomb model
Mohr-Coulomb yield surface Mohr-Coulomb potential surface

1
Mohr-Coulomb potential surface
(with temperature effects)
MC
2 3
Figure 3.9-10: Shapes of the Mohr-Coulomb yield and

potential surfaces in the 4 plane

In the limiting cases of vertex yielding, that is, the stress

states represented by A or C (tensile meridian, = 0) and B

(compressive meridian, = ) in Fig. 3.9-9, the
3
analysis boils down to a Drucker-Prager-type analysis.
A couple of important and practical suggestions apply to the

Mohr-Coulomb material model.
Since the model follows a non-associative flow rule, the

constitutive matrices at the integration points are, in
general, non-symmetric. Therefore, it is recommended
to use the non-symmetric sparse solver instead of the
default sparse solver when a user's model exhibits slow
convergence or it cannot converge (see the AUI
command MASTER parameter SOLVER=NONSYM-SP).
If tensile conditions are present in a user's model, then it is

recommended to use the potential function implemented for
the Mohr-Coulomb material model with temperature effects
(see below). In order to successfully use this potential
function, some type of temperature input to ADINA is
required in the user's model; however, these can be set to
zero so as to neglect the temperature effects.
For more information, see the following references:
ref. W. F. Chen and E. Mizuno, Nonlinear Analysis in Soil

Mechanics, Theory and Implementation, Developments
in Geotechnical Engineering 53, Elsevier, Amsterdam,
1990.
ref. M. A. Crisfield, Non-linear Finite Element Analysis of

Solids and Structures, Volume 2: Advanced Topics, John
Wiley & Sons, Chichester, 1997.

3.9.4.1 Mohr-Coulomb material model with temperature effects
The Mohr-Coulomb material model with temperature effects

has basically the same characteristics of the default Mohr-Coulomb
material model except for the points discussed below.
t
In some geotechnical analyses, the thermal strains eTH need to
be included. These strains are taken into account by assuming that
the Mohr-Coulomb material is an isotropic temperature
independent material (see Section 3.1.4) affecting the material
behavior only hydrostatically. The thermal strains are calculated
from the temperature conditions and the mean coefficient of
thermal expansion, where 0 , as given in Eq. (3.1-16).
ADINA will require temperature data as input. Please refer to

Section 5.6 for a discussion on the several ways to input these data
to ADINA.
The Mohr-Coulomb potential function used describes a

completely smooth surface in the stress space, as shown in Figs.
3.9-7 and 3.9-10, and it is given by
( ) 3c
2
( 2aMC sin )
2
t
g MC = + t r M MC t
J2 + 2sin t I1 ,
tan
where
3c 3c
tan if T tan with 0 < < 1,
aMC =
3c T if T < 3c ,
tan tan
M MC = 3 ( 3 sin )
and
4 (1 e 2 ) cos 2 t + ( 2e 1)
2
t
r= ,
2 (1 e ) cos + ( 2e 1) 4 (1 e ) cos + 5e 4e
2 t 2 2t 2

is the elliptic function described in
ref. Ph. Mentrey and K. J. Willam, Triaxial Failure

Criterion for Concrete and Its Generalization, ACI
Structural Journal, 92(3), pp. 311-318, 1995.

and is defined for 0 t . In order for such an elliptic
3
function to be convex and smooth, it is required that 0.5 < e 1 ,
where e is known as the eccentricity (in the deviatoric plane) and it
will be assumed to be
3 sin
e= ,
3 + sin
as suggested in the above reference. Basically, this ensures that the

surface described by the above potential function matches the
Mohr-Coulomb yield surface at the tensile meridian ( = 0 ) and

compressive meridian ( = ). This potential function contains
3
parameters and terms that are mentioned in the description of the
default Mohr-Coulomb material model. Note that the accepted
value ranges of some of these quantities have changed due to the
nature of the new function. This potential function contains the
following parameters and terms:
c is the cohesion such that c > 0 ;

is the friction angle such that 0 < < ;
2
is the dilatation angle such that 0 < ;
aMC is the distance along the hydrostatic axis from the apex
of the Mohr-Coulomb yield surface to the surface described
by the potential function;

is an eccentricity parameter in the meridional plane (at the

apex of the Mohr-Coulomb yield surface) that helps define
aMC such that 0 < < 1 and it is recommended to be
0.1 ;
T is the tension cut-off such that T 0 ; and

t
is the angle of similarity defined earlier and is a function
of t J 2 and t J 3 .
The Mohr-Coulomb potential function used provides a smooth

surface in the stress space and, therefore, in both the deviatoric and
meridional stress planes. In the former plane, it has an elliptic
shape, which is basically dictated by the elliptic function (see Fig.
3.9.10), while in the latter plane its shape is that of a hyperbola (see
Fig. 3.9.7). The flow rule is non-associative and presents a unique
plastic return in the stress space, which is guaranteed by the
smoothness of the potential surface. Note that all the above
characteristics ensure the need for no special cases at the corners of
the Mohr-Coulomb yield surface, its apex or in the tension cut-off
region.
Due to the non-associative flow rule applied in the model, it is

recommended to use the non-symmetric sparse solver instead of the
default sparse solver when a user's model exhibits slow
convergence or it cannot converge (see the AUI command MASTER
parameter SOLVER=NONSYM-SP).
3.10 Fabric material model with wrinkling

The elastic-orthotropic material model (discussed in Section
3.2) can be used with the 2-D solid element with the 3-D plane
stress (membrane) option and the large displacement/small strain
formulation to model the behavior of fabric structures.
Fabric materials are typically very thin and flexible. They can
support only in-plane tensile stresses, and will wrinkle under
compressive stresses.

The analysis of fabric structures with ADINA is performed in

two steps: first, initial strains must be specified, Young's moduli
must be set to a fraction (usually 10-4) of their actual values, and
the wrinkling of the material must not be taken into account. Then,
in a second step (restart analysis), the same orthotropic material
model is used except that the Young's moduli are now set to their
actual values, and wrinkling is taken into account.
For each integration point, wrinkling occurs if the smallest of

the two in-plane principal stresses (t
p1 , t p2 ) becomes negative.
If wrinkling occurs, then this smallest principal stress t p2 is set
to zero. If the other in-plane principal stress t p1 is greater than
zero, the program calculates a new "wrinkling stress" t p1 using
t
p1 = A t p1
where t p1 is the strain in the direction of t p1 and A is calculated

from the orthotropic material law, using the condition t p2 = 0 . If
t
p1 is smaller than zero, ADINA sets t p1 to zero; in this case,
wrinkling occurs in two directions.
The element constitutive matrix corresponding to the principal
directions is also modified for the wrinkling behavior and becomes
A 0 0
0 A 0

0 0 A
where is a reduction factor equal to 10-5, and = 1 if t p1 0 or

= if t p1 < 0 .
3.11 Viscoelastic material model

The viscoelastic model can be used with the truss, 2-D solid,
3-D solid and shell elements.

3.11: Viscoelastic material model
The viscoelastic model can be used with the small

large displacement/large strain formulations (2-D solid and 3-D
solid elements only).
TL formulation is employed and when used with the large
displacement/large strain formulation, the ULH formulation is
employed.
The mechanical behavior for an isotropic and linear viscoelastic

material may be expressed in tensor notation as
dG ( )
t
sij (t ) = 2G (0)eij (t ) + 2 eij (t ) d (3.11-1)
0 + d
dK ( )
t
kk (t ) = 3K (0) kk (t ) + 3 kk (t ) d (3.11-2)
0+
d
1
where t is the time, sij = ij ij kk is the deviatoric stress, ij
3
1
is the Kronecker delta, ij is the stress, eij = ij ij kk is the
3
deviatoric strain, ij is the strain, G(t) is the shear modulus and
K(t) is the bulk modulus.
In the presence of a temperature variation ( t ) , the stresses for
an isotropic and thermorheologically linear viscoelastic material
may be written as

dG ( )
sij (t ) = 2G (0)eij (t ) + 2 eij ( ) d (3.11-3)
0 + d

dK ( ) dK ( )
kk (t ) = 3K (0) ( kk (t ) (t ) ) + 3 [ kk ( )]
th
d 3 kkth (t ) d
d d
kk
0+ 0+

(3.11-4)
where
t
= ( ) d , = ( ) d (3.11-5)
0+ 0+
and the thermal strain is given by
kkth ( t ) = 3 ( (t ) ) ( t ) TALPHA 3 ( 0 ) [ 0 TALPHA ] (3.11-6)
( (t ) ) is the temperature-dependent coefficient of thermal

expansion and (t ) is the shift function, which obeys
d
(TALPHA ) = 1, ( ) > 0, >0 (3.11-7)
d
Note that TALPHA is the reference temperature used for thermal

strain calculation.
In equations (3.11-3) and (3.11-4) it is assumed that the
mechanical and thermal responses are uncoupled. Furthermore if
the temperature is constant, equations (3.11-3) and (3.11-4) reduce
to equations (3.11-1) and (3.11-2).
We assume the following thermo-material properties:

G
G (t ) = G + Gi e it (3.11-8)
i =1
K
K (t ) = K + K i e it (3.11-9)
i =1
(t ) 0 (3.11-10)
= ( (t ) ) (3.11-11)
where G and K are the long-time shear modulus and bulk

modulus respectively, i and i are the decay constants for the

3.11: Viscoelastic material model
shear modulus and bulk modulus respectively and G and K are

the number of time-dependent terms for the shear modulus and
bulk modulus respectively. Equations (3.11-8) and (3.11-9) are
referred to in the literature as Prony or Dirichlet series. G and K
are limited to a maximum value of 15.
The shift function used is either the Williams-Landell-Ferry
(WLF) equation, written as follows
C1 ( ref )
log10 ( ) = (3.11-12)
C2 + ( ref )
or the Arrhenius shift function
1 1
log10 ( ) = C1 , ref

ref
(3.11-13)
1 1
= C2
, < ref
ref
in which C1 and C2 are material constants, and ref is defined as

the reference temperature of the WLF equation (TREF in the the
MATERIAL VISCOELASTIC command).
You can either enter the Prony series constants in the AUI, or
you can enter the measured data for the shear and bulk modulus
and let the AUI construct the Prony series. In the latter case, the
AUI constructs the Prony series using a least-squares technique, see
the reference by Frutiger and Woo given below.
The nodal point temperatures are input to ADINA as discussed

in Section 5.6.
For more information, see the following references:
ref. W.N. Findley, J.S. Lai and K. Onaran, Creep and

relaxation of nonlinear viscoelastic materials, Dover
Publications, 1976.
ref. R.L. Frutiger and T.C. Woo, A thermoviscoelastic

analysis for circular plates of thermorheologically simple

material, Journal of Thermal Stresses, 2:45-60, 1979.
3.12 Porous media formulation

The porous media formulation is applicable to porous structures
subject to static or dynamic loading. It deals with the interaction
between the porous solids and pore fluids, which flow through the
porous solid skeleton.
The porous media formulation only applies to 2D and 3D solid

elements. These elements have pore pressure variables at their
corner node points.
The pore pressures are considered to be independent variables,

along with the displacements. The pore pressures are output by
ADINA along with the displacements, velocities, stresses and other
solution results.
The porous media model is implemented with a fully coupled

formulation, as described in detail below.
The ADINA porous media formulation is based on the

following assumptions:
The pore fluids are incompressible or slightly compressible.
The porous solid skeleton is fully saturated with pore fluid.
The pore fluid flows through the porous solid according to

Darcys law.
With the porous media formulation, undrained analysis (static

pore fluid) and consolidation or swelling analysis (pore fluid
flowing through the porous solid skeleton) can be performed.
In the analyses of porous structures, all the nonlinear

formulations in ADINA (small displacement/small strain, large
displacement/small strain and large displacement/large strain) can
be used. A materially-nonlinear-only formulation is employed for
the small displacement/small strain formulation; the TL
formulation for the large displacement/small strain formulation and

3.12: Porous media formulation
the ULH formulation for the large displacement/large strain

formulation.
The mixed (u/p) formulation can be employed in the analyses of
porous structures. However the incompatible modes feature cannot
be employed in the analyses of porous structures.
The types of problems for which the porous media model can be
employed are:
Transient static analysis (consolidation): The pore pressure

distribution, the displacement and stress distribution, and their
changes with time are of great interest.
Transient dynamic analysis: In addition to the pore pressure,

displacement and stress results, the potential failure of the
porous structure will draw much more attention. Liquefaction,
which is an extremely important failure mode for infrastructures
in an earthquake disaster, can be studied with the porous media
model.
Undrained analysis: No specific formulation is given to relate

the pore pressure to total stresses under undrained conditions.
The undrained analysis can be performed by specifying
undrained boundary conditions and small time steps.
The orthotropic, thermo-isotropic, thermo-orthotropic, thermo-

plastic, Drucker-Prager, Mohr-Coulomb, Cam-clay, creep, and
plastic-creep material models can be employed in porous media
analysis.
The general equations governing the porous media model are as

follows:
The macroscopic stresses must satisfy the equilibrium

condition:
ij , j + f i = 0 (3.12-1)
where the total (Cauchy) stress is

ij = C e kle + Pf ij ,
ijkl
i, j , k , l = 1, 2,3 (3.12-2)
Pf is the pore pressure, C ijkle is the fourth order elasticity tensor,

kle is the elastic strain tensor and ij is the Kronecker delta.
The continuity condition for slightly compressible fluid

flows must be satisfied:
ii n Pf
( k Pf ) = + (3.12-3)
t f t
where k is the permeability matrix of the porous material, ii is

the volumetric strain of the porous skeleton, n is the porosity of
the material and f is the bulk modulus of the fluid. If the
fluid is assumed incompressible, the last term in (3.12-3) will be
zero.
After a finite element discretization, equations (3.12-1) and

(3.12-3) yield the following equations.
Transient statics:
0 0 t +t U t +t K uu t +t
K u pf t +t U
t +t K T t +t t +t + t +t
u pf K (1)
p f pf Pf 0 t +t K (2) Pf
p f pf
t +t R u
= t +t (3.12-4)
R pf
Transient dynamics:

t +t M 0 t +t U C 0 t +t U
+ t +t T t +t
K (1) t +t +
0 0 t +t Pf K u pf pf p f Pf
t +t K uu t +t
K u pf t +t U t +t R u
t +t = t +t
0 t +t K (2) Pf R pf
pf pf
(3.12-5)
For a finite element, these matrices are defined as follows

(omitting the left superscript for brevity):
K uu = BTu DB u dV
V
K u pf = BTu I H pf dV
V
1
p p =
K (1) nH
T
H pf dV
f
f f pf
V
K (2)
pf p f = B k B p dV T
pf f
V
(H ) f dS
T
R u = HTu f dV +
S
u f
V Sf
( ) q dS
S
T
Rp = Hp q
f f
Sq
In the above expressions, D is the constitutive matrix (possibly

elasto-plastic), f and q are load vectors (force, traction and flux),
B u and B p are shape function matrices for the displacements U
f
and pore pressures Pf , H u and H p are shape function matrices

f
for the displacements U and pore pressures Pf , and I is the

identity tensor.
Numerically integrating equations (3.12-4) and (3.12-5) with

respect to time gives the final equations used for the porous media
model:

Transient statics: the Euler backward method is used:
t +t K uu t +t
K u pf t +t U
t +t T t +t t +t t +t
K u pf pf pf t
K (1) K (2)
pf pf Pf
t +t
Ru
= t +t t +t t +t
t R pf + K u pf U + K pf pf Pf
T t (1) t
(3.12-6)
Transient dynamics: the Euler backward method is used for

the continuity equation and the Newmark method is used for the
equilibrium equation:
t +t K uu + a0 t +t M + a1 t +t C t +t
K u pf t +t U
t +t t +t t +t t +t
K Tu pf pf pf t
K (1) K (2) Pf
pf p f
t +t
R ud
= t +t t +t t +t
t R pf + K u pf U + K pf pf Pf
T t (1) t
(3.12-7)
in which
t +t
R ud =t +t R u + t +t
M ( a0 t U + a2 t U + a3 t U ) +
t +t
C ( a1 t U + a4 t U + a5 t U )
(3.12-8)
and where a0, a1, ... , a5 are the integration constants for the
Newmark method (see Ref. KJB, Section 9.2.4).
Please note the following:
Usually, the global stiffness matrix is no longer positive

definite in porous media analysis. This is because the diagonal
components of t +t K (1)
p f p f t
t +t
K (2)
pf pf in equations (3.12-6)
and (3.12-7) are usually negative. Therefore, you should

request that ADINA continue execution after ADINA detects a

non-positive definite stiffness matrix (see Section 7.1.1).
It is important to have a reasonable scheme for choosing

time steps. Theoretically, the solution process is
unconditionally stable for any time step size because of the use
of the Euler backward method. However, this does not
necessarily imply that any time step size is permissible. A small
time step size and large mesh size will result in the zigzag pore
pressure distribution at the initial stage of pore fluid dissipation
(Britto & Gunn, 1987). It is logical to use progressively larger
time steps in the porous media analysis.
The pore pressures are only computed by ADINA at the

corner nodes of solid elements. During postprocessing, the AUI
interpolates the pore pressures from the corner nodes to the
other nodes (e.g. mid-side nodes) in higher-order elements.
Once the porous solid material properties are specified, the

corresponding pore fluid properties are automatically computed.
Certainly, they can be redefined if necessary.
Porous media cannot be used in explicit dynamic analysis.
For more information, see the following reference:
ref. Britto, A.M. and Gunn, M.J., Critical state soil

mechanics via finite elements, Ellis Horwood Limited,
1987.
3.13 Gasket material model

Gaskets are relatively thin components placed between two
bodies/surfaces to create a sealing effect and prevent fluid leakage
(see Figure 3.13-1). While most gaskets are flat, any arbitrary
gasket geometry can be modeled in ADINA.

Gasket thickness direction

(orthotropic axis A)
Gasket in-plane
directions
Only one element

through the thickness
Figure 3.13-1: Schematic of gasket
The sealing effect is created when the compressive load, applied

in the direction of the gasket thickness, exceeds the initial yield
stress of the gasket. The sealing effect is maintained as long as the
compressive stress does not drop beyond a specified threshold
value. The gasket ruptures if the compressive stress exceeds the
gaskets ultimate stress. Unlike rupture, if a gasket leaks it still
maintains its load-deflection characteristics.
The gasket model can be used with 2-D solid and 3-D solid
elements. It can also be used with small displacement/small
strain, large displacement/small strain formulations.
The gasket behaves as a nonlinear elastic-plastic material when

compressed in the thickness or gasket direction. Its load-
deformation characteristics are represented by pressure-closure
curves. Tensile stiffness can be assumed to be constant or zero.
Hardening is assumed to be isotropic.
Since the gasket material has different properties in different

directions, it is considered by ADINA to be an orthotropic material.
The gasket thickness direction is defined as the material axis a
direction. The AUI attempts to determine this direction
automatically, see below.
The closure strain is always measured as the change in gasket

thickness divided by the original gasket thickness.

3.13: Gasket material model
The gaskets uni-directional plasticity model speeds up

computations, and allows more flexibility in defining the shape of
the loading and unloading curves.
Two gasket types are implemented in ADINA, the full or

regular gasket, and the simple gasket.
In the full gasket, elastic material properties are used for the in-
plane deformations. Therefore the full gasket element has stiffness
for all translational degrees of freedom.
In the simple gasket, the element has no stiffness corresponding

to the in-plane deformations. Therefore the nodal degrees of
freedom for the in-plane directions must be eliminated, otherwise
the model will have zero pivots in the stiffness matrix. The AUI
attempts to automatically eliminate the nodal degrees of freedom
for the in-plane directions, see below.
User input
The user input for the gasket material is now described in terms of
the command-line interface.
MATERIAL GASKET NAME TREF DENSITY,

YIELD-CURVE G-G E-G ALPHA-G,
LEAKAGE-PRESSURE,
E-INPLANE NU-INPLANE
ALPHA-INPLANE,NPOINTS
lcurvei
E-G is the Young's modulus for tensile loading in the thickness

direction ( Etensile ).It is allowed to set E-G to a very small number,
such as 1E-5, to model a gasket with very small tensile stiffness.
G-G, ALPHA-G are the transverse shear modulus and the

coefficient of thermal expansion in the thickness direction.
LEAKAGE-PRESSURE specifies the leakage pressure. The

LEAKAGE-PRESSURE does not affect the material response, but
the LEAKAGE-PRESSURE affects the possible gasket states, see
below.

E-INPLANE, NU-INPLANE and ALPHA-INPLANE are the in-

plane values of the Young's modulus, Poisson's ratio and
coefficient of thermal expansion. These properties are ignored for a
simple gasket.
YIELD-CURVE, NPOINTS and the lcurvei are used to enter the

pressure - closure strain relationship. YIELD-CURVE and the
lcurvei are curves that are defined using the LCURVE command.
Pressure - closure strain input
We now describe the pressure - closure strain input in more detail.

Figure 3.13-2 shows a typical pressure - closure strain input. The
main loading curve is specified by parameter YIELD-CURVE. The
main loading curve can contain any number of points, and
successive points are connected by straight lines.
Last yield
point (rupture
point)
Main loading curve

Initial yield 2nd yield
point point
Gasket pressure
Unloading curves
Leakage
pressure
Closure strain
Etensile
Figure 3.13-2: Pressure-closure strain input for a gasket material

The point number of the first yield point on the yield curve is given
by parameter NPOINTS. NPOINTS must be greater than or equal
to 3.
For each point on the main loading curve higher than the first yield
point, there is an associated unloading curve, as shown. Notice that
each unloading curve has the same number of points, and this
number of points is equal to NPOINTS. There must be at least
three unloading curves, and therefore the total number of points on
the yield curve must be at least NPOINTS + 2.
There are two options for specification of each unloading curve:
1) The unloading curve can be specified using one of the lcurvei,

data input lines. The lcurve must contain NPOINTS points, the first
point must have zero pressure, and the last point must correspond
to the (closure strain, pressure) of the main loading curve yield
point.
2) The AUI can automatically construct the unloading curve by

interpolation of the user-input unloading curves.
It is necessary to specify at least one of the unloading curves using

the lcurvei, data input lines.
User input for nonlinear elastic gasket
We now describe how to input material data for a nonlinear elastic

gasket.
Figure 3.13-3 shows an example for input of pressure - closure

strain data for a nonlinear elastic gasket. In this example, we
assume that points 2 to 4 correspond to experimental data. Points 5
to 7 are on a line extended from segment 3-4.

Original nonlinear elastic loading curve

Extended nonlinear elastic loading curve
7
Unloading curve 1 5
Unloading curve 2
6 4
NPOINTS=5 in this example
5
Gasket pressure
5 4
3
3
4
2
2
3
1 2 1 1
Closure strain
Etensile
Figure 3.13-3: Example input for nonlinear elastic gasket
NPOINTS for this example is 5.
Points NPOINTS to NPOINTS+2 should correspond to very high

closure strain, so that the program never reaches the closure strain
corresponding to point NPOINTS.
The gasket material will be nonlinear elastic as long as the closure

strain is less than the closure strain at point NPOINTS, and the
relationship between closure strain and gasket pressure is given by
the segments connecting points 1 to NPOINTS.
Two unloading curves must be entered, corresponding to points

NPOINTS+1 and NPOINTS+2 on the loading curve, as shown.
Each unloading curve has NPOINTS equally spaced points, with
unloading point 1 for pressure =0 and unloading point NPOINTS
for the point coincident with the loading curve.
The tensile behavior of the gasket is governed by constant E-G

( Etensile ).

Pressure - closure strain response
Figure 3.13-4 shows a typical pressure - closure strain response.

The gasket is loaded to point B, then unloaded to point C. When
the closure strain becomes negative, the response is governed by
Etensile . Then the gasket is loaded to point D. Up to this time, the
gasket behaves as a nonlinear elastic material. Then the gasket is
loaded to point E and unloaded to point F. The unloading to point F
is done on an unloading curve based at point E; this unloading
curve is interpolated from the input unloading curves. The gasket is
loaded to point G and unloaded to point H. The gasket follows an
unloading curve based at point G and interpolated from the input
unloading data. During the loading to point I, the gasket follows
the unloading curve to point G, then follows the loading curve to
point I. The gasket is then unloaded to points J and K. When the
pressure becomes negative, the response is again governed by
Etensile . During reloading to point L, the gasket follows the same
unloading curve as used for points J and K. The gasket is then
loaded to point M, at which time the gasket ruptures. The pressure
drops to zero and remains zero thereafter (point N).
Interpolated unloading
curve based on highest M
closure strain I
Initial yield
Gasket pressure
point
G
E L
D
H
J
B
Leakage F
pressure A
N
C K Closure strain
Figure 3.13-4: Typical pressure-closure strain history

Each gasket can have one of the following five states:
Open: The gasket pressure is less than the leakage pressure.
Closed: The gasket pressure is higher than the leakage

pressure but has not yet caused plasticity.
Sealed: There has been plastic gasket deformation and the

current pressure is above the gasket leakage
pressure.
Leaked: After plastic deformation, the gasket pressure has

dropped below gasket leakage pressure.
Crushed: Gasket closure strain has exceeded the rupture value.
Modeling issues
The gasket must be modeled as a single layer of 2-D or 3-D

elements. Linear elements are recommended (4-node quad elements
in 2-D, 6-node prism and 8-node brick elements in 3-D). Higher
order 2-D quad elements and 3-D prism and brick elements can
also be used. In this case, the AUI automatically generates
constraints for the midside nodes in the gasket direction to keep a
linear thickness displacement variation.
If the user does not explicitly set the material axes directions in
gasket elements, the AUI attempts to automatically determine the
gasket material axes directions. This automatic determination is
made based on the element layout. Fig 3.13-5 shows element
layouts in which the AUI can and cannot automatically determine
the gasket material directions. If the AUI cannot automatically
determine the gasket material directions, and if the material axes
for the gasket are not explicitly defined by the user, then the AUI
cancels dat file generation.

(a) Each element is surrounded by neighboring elements on adjacent sides.

Gasket direction can be automatically determined.
(b) At least one element is surrounded by neighboring elements on opposite sides only.
Gasket direction cannot be automatically determined.
Figure 3.13-5: Automatic determination of gasket material directions
The AUI eliminates the in-plane deformations of simple gasket

elements using nodal skew systems. The procedure is as follows.
Skew systems are constructed with the skew a direction aligned
with the gasket material direction, and the skew b and c directions
aligned with the in-plane directions. The skew systems are assigned
to the nodes of the simple gasket elements. Then the AUI deletes
the b and c degrees of freedom at these nodes.
Note that this procedure is applied to all nodes attached to
simple gaskets, even nodes that are also attached to other elements.
It is allowed to deactivate the AUI's computation of the material

axes directions and skew systems. This is done using command
MISC-OPTIONS GASKET-AUTOMATIC=NO
(the default is GASKET-AUTOMATIC=YES). When GASKET-

AUTOMATIC=NO is used, the user must explicitly specify the

material axes directions. And, if simple gaskets are used, the user
must explicitly eliminate the degrees of freedom corresponding to
in-plane deformations.
The top and bottom surfaces of a gasket can be separate from

those of the mating surfaces. In this case, they should be connected
via contact. The gasket can also share a common surface with the
intended mating surface. In this case, contact is not needed,
however, the gasket cannot separate from its target. A gasket can
also be attached to its mating surface via tied contact, mesh
glueing, constraint equations, or rigid links.
Output variables: The following gasket output variables are

available for both gasket types: GASKET_PRESSURE,
GASKET_CLOSURE_STRAIN, GASKET_THERMAL_STRAIN,
GASKET_STRESS-*, GASKET_STRAIN-*,
GASKET_THERMAL_STRAIN-*, GASKET_YIELD_STRESS,
GASKET_PLASTIC_CLOSURE_STRAIN, GASKET_STATUS,
GASKET_DEFORMATION_MODE
See Section 13.1.1 for the definitions of these variables. For the 2D
gasket type, the components represented by the * are AA, BC and
CC, and for 3D gasket type, the components represented by the *
are BB, CC, AB, AC, BC.
3.14 Shape Memory Alloy (SMA) material model

The Shape Memory Alloy (SMA) material model is intended
to model the superelastic effect (SE) and the shape memory
effect (SME) of shape-memory alloys.
These materials can undergo solid-to-solid phase

transformations induced by stress or temperature. The high
temperature phase is called austenite (A) with a body-
centered cubic structure and the low-temperature phase is
called martensite (M) with a monoclinic crystal structure in
several variants.
Fig.3.14-1 shows a schematic SMA stress-temperature

3.14: Shape Memory Alloy (SMA) material model
diagram. The martensite phase in two generalized variants

and the austenite phase are shown, as well as stress- and
temperature-dependent transformation conditions. The
martensite phase is favored at low temperatures or high
stresses. Upon heating from low temperature the material
begins transforming from martensite to austenite at
temperature As. The transformation is 100% complete at
temperature Af. If the material is then cooled again, the
austenite starts transforming back to martensite at
temperature Ms. This transformation is 100% complete at
temperature Mf. These four temperatures are also stress
dependent with high stresses favoring the martensite phase.
This stress dependence is assumed to be linear with slope CM
and CA for the martensite and austenite temperatures,
respectively. A typical variation of volume fraction of
martensite in the SMA material with temperature is shown in
Fig. 3.14-2.
eff
Detwinned
martensite CM CM CA CA
Austenite
Twinned
martensite
Mf Ms As Af
Temperature,
Figure 3.14-1: SMA stress-temperature phase diagram
A typical uniaxial isothermal stress-strain curve is shown

in Fig. 3.14-3.
The superelastic effect is evident when the material is

deformed at temperature >Af and is displayed in Fig.3.14-

3(a). The stress cycle application induces transformations
from AM and then from MA to exhibit the hysteresis
loop. The shape memory effect is evident when the material
is deformed at temperature < As and is displayed in
Fig.3.14-3(b). A residual transformation strain remains after
unloading; however heating the material to temperature above
Af leads to thermally induced MA transformation and the
recovery of transformation strain.
Transformation at eff = 0
Transformation at eff > 0
Mf Mf As As
Volume fraction of martensite, m
1.0
0.8
0.6
0.4
0.2
0.0
Ms Ms Af Af
Temperature,
Figure 3.14-2 : Volume fraction of martensite vs. temperature


> Af < As
(a) (b)
Figure 3.14-3: Schematic of stress-strain curves for shape-memory alloys:

(a) superelasticity; and (b) shape memory effect
Both shape memory effects due to transformation from

martensite to austenite and due to re-orientation of the
martensite are captured by modeling the twinned and
detwinned martensites as different phases.
The SMA material model is based on the following

equations:
The total strain,
= e + t +
where
e= elastic strain
= thermal strain
t = transformation strain; to be evaluated
The one-dimensional macro-scale model,
= s + t ; 0 1
+ = 1

t = tmax s
= ((1 ) E A + EM )( t max s )
t = twinned martensite volume fraction

s = detwinned martensite volume fraction
A = austenite volume fraction
tmax = maximum recoverable residual strain; a material
property usually obtained from a simple tension test
when the material is fully detwinned martensite (s = 1)
The flow rule of three-dimensional constitutive

model,
ijt = s max
t
nijt
3 sij
nijt = ; for the martensitic transformation
2
3 ij
t
n =
t
; for the reverse transformation
2 t
ij

where
sij = deviatoric stresses
3
= sij sij is the effective von Mises stress
2
3 t t
t = ij ij is the effective transformation strain
2
This results in the following equation for deviatoric stress

calculation:

t +t
E ( )
t +t
sij =
1+ t +t
( )
( t +t
ij ijt )
where
t +t
ij = t +t ij t ijt
Four phase transformation conditions,
1. Starting condition for the martensitic transformation
f M S = 3J 2 CM ( M S )
2. Ending condition for the martensitic transformation
f M f = 3 J 2 CM ( M f )
3. Starting condition for the reverse transformation
f AS = 3J 2 C A ( AS )
4. Ending condition for the reverse transformation
f Af = 3J 2 C A ( Af )
The phase transformation rate using linear kinetic rule

(Auricchio and E. Sacco, 1999),
R
= f f
ff

t +t
3 J2 t J2
f = c( t +t t )
2
t +t
where, for the austenite to martensite transformation,
f f = - f M f , c = CM and R = 1 -
and for the reverse martensite to austenite transformation,
f f = f A f , c = C A and R =
Evolution of single-variant detwinned martensite:

Martensite re-orientation is based on the following
condition
f R = 3 J 2 CR R
where
R = material yield property at = 0
CR = slope of yield function temperature variation
Austenite to martensite transformation leads to
s =
Martensite to austenite transformation leads to proportional

transformation of the twinned and detwinned phases:
s
s =

t
t =


Computational steps for the stress-integration of the SMA

model are as follows (Koji and Bathe, 2005):
1. Calculate the trial deviatoric stresses, assuming no

additional phase transformation or re-orientation,
E ( t ) t +t
t +t TR
s = ( ij )
1 + ( t )
ij
2. Check for martensitic re-orientation,
f R > 0 and t s < t and t < 1
Check for austenite to martensite transformation,
f M f f M s < 0 and t < 1 and f > 0
Check for martensite to austenite transformation,
f Af f As < 0 and t > 0 and f < 0
3. In case of martensitic re-orientation solve the following

governing equation:
3 t +t t +t
g 2 ( s ) = sij sij CR t +t R = 0
2
(3.14-1)
E ( ) t +t "
t +t
t +t
sij = ( ij s max
t t +t t
nij )
1 + ( )
t +t
The martensite reorientation calculation step is optional; it is

activated when R > 0 is input.

4. In case of austenite to martensite transformation, solve the

following governing equation:
t +t
R 3 t +t
J 2 ( ) t J 2
g ( ) = c( t +t t ) = 0
t +t
ff 2
t +t

(3.14-2)
where
E ( ) t +t "
t +t
t +t
sij = ( ij s ( ) max
t t +t t
nij )
1 + ( )
t +t
R = 1 , f f = f M f and c = CM
5. In case of martensite to austenite transformation, solve the

governing equation (3.14-2) with
R = , f f = f Af and c = CA
6. Update history-dependent variables for this time step/iteration

step.
7. Calculate the consistent tangent constitutive matrix.

Structures, Springer, 2005
ref. F. Auricchio and E. Sacco, A Temperature-Dependent

Beam for Shape-Memory Alloys: Constitutive
Modelling, Finite-Element Implementation and
Numerical Simulation, Computer Methods in Applied
Mechanics and Engineering, Vol. 174, pp. 171-190
(1999)

3.15: The Anand material model for soldering
3.15 The Anand material model for soldering

Isotropic elasticity
Isotropic, isochoric viscoplastic flow
The material model is described in the following references:
G. Weber and L. Anand, Finite deformation constitutive

equations and a time integration procedure for isotropic,
hyperelastic-viscoplastic solids, Comp. Meth. Appl. Mech.
Engng., 79 (1990) 173-202.
S. B. Brown, K. H. Kim and L. Anand, An internal variable

constitutive model for hot working of metals,
Int. J. Plasticity, 5 (1989) 95-130.
The material model can be used with the 2-D plane strain, 2-D
axisymmetric and 3-D solid elements.
The material model can be used with the small

large displacement/large strain formulations. When used with the
small displacement/small strain formulation, a materially-
nonlinear-only formulation is employed, when used with the large
displacement/small strain formulation, the TL formulation is used,
and when the large displacement/large strain formulation is used,
the ULH formulation is used.

Plane strain, axisymmetric and 3-D elements that reference this

material model should employ the mixed displacement-pressure
(u/p) formulation. The u/p formulation is the default for these
elements.

The material model does not possess a yield criterion, and no

loading/unloading criteria. Instead, viscoplastic flow is assumed to
take place at all levels of (deviatoric) stress, although at low levels,
the rate of viscoplastic flow may be immeasurably small.
The resistance to viscoplastic flow offered by the material is

modeled using a single, scalar parameter s called the isotropic
deformation resistance. This parameter has the dimension of stress.
The material behavior is characterized by an evolution equation

for the effective viscoplastic strain rate, and an evolution equation
for the isotropic deformation resistance (Brown, et al.). These
equations are
1/ m
Q
VP
= A1 exp sinh (3.15-1)
R s
A2
s s
s = h0 1 * sgn 1 VP (3.15-2)
s s*
n
VP Q
with s* = s exp (3.15-3)
A1 R
where A1 is the pre-exponential factor, Q is the activation energy,

R is the gas constant, is the absolute temperature, is the
multiplier of stress, m is the strain rate sensitivity of stress, h0 is
the hardening (or softening) constant, A2 is the strain rate
sensitivity of hardening (or softening), s is the coefficient for the
saturation value of the isotropic deformation resistance, n is the
strain rate sensitivity of the deformation resistance, and is the
equivalent tensile stress.
The initial value of the isotropic deformation resistance s is s0 .
The material behavior is temperature dependent, but it can be

3.15: The Anand material model for soldering
made temperature independent by setting Q = 0 .
Thermal strains are calculated from the prescribed nodal point

temperatures using a constant coefficient of thermal expansion.
The Anand model is used in the modeling of solder materials.
The uniaxial stress-strain response at room temperature is

shown in Fig. 3.15-1 for a solder material with the following
material constants:
E = 20 103 MPa
= 0.31
A1 = 2.3 107 s 1
Q / R = 11262 K
= 11
m = 0.303
s = 80.79MPa
n = 0.0212
h0 = 4121.31MPa
A2 = 1.38
s0 = 42.32MPa

Stress (MPa)
120
Deformation resistance, s
100
80
60 Uniaxial tensile stress,
40
20
0
0 0.2 0.4 0.6 0.8 1.0
Strain
Fig. 3.15-1 The evolution of the isotropic deformation resistance in uniaxial

tension
3.16 Piezoelectric material model

The piezoelectric material model is available for 2-D and 3-D
solid elements and can be applied to analyze the coupling behavior
between the electric field and the corresponding mechanical
deformation in structural analysis.
The piezoelectric analysis discussed here is based on the direct

coupling formulation, which is different than the iterative coupling
formulation. Modeling with the iterative coupling formulation can
set up a user-coded material (see Sections 3.17 and 12.4 for general
information).
In the direct coupling formulation, the electric potential is

interpreted as a voltage degree of freedom (DOF), and each node of
a solid element is assigned with a voltage DOF. In most cases, the
number of the voltage DOFs in a solid element equals to the
number of nodes in the element.

3.16: Piezoelectric material model
For the 2D-solid element, the piezoelectric material model can

be applied in the cases of plane strain, plane stress, axisymmetric,
generalized plane strain and 3D plane stress. In the case of
generalized plane strain, there is no voltage DOF assigned to the
auxiliary node.
The constitutive equation of the piezoelectric material model is

considered to be linear. The TL formulation is available for
piezoelectric analysis, which is thus applicable for large
displacement/rotation problems.
The piezoelectric material model can by used in static, dynamic-

implicit, frequency, and mode superposition analysis. Currently,
only the sparse solver is available for piezoelectric analysis. Due to
the non-positive definite stiffness matrix of the piezoelectric
material model, the setting of the control parameter "Continue Even
When Non-Positive Definite Stiffness Matrix Encountered" in the
AUI ControlSolution Process dialog box is ignored and the
program always continues its solution.
The piezoelectric material model cannot be used with:
explicit dynamic analysis
substructures
cyclic symmetry
mixed formulations
3.16.1 Constitutive equation
The constitutive equation of the piezoelectric material model is
C e
= T
D e E
where

: elastic stress vector [force/area]
D: electric displacement field [electric charge/area]
C: elastic stiffness matrix [force/area]
e: piezoelectric coupling matrix [force/(voltlength)]
: dielectric matrix [electric charge/(voltlength)]
: elastic strain vector
E: electric field [volt/length]
For 2-D solid elements,
1 C11 C12 C13 C14 e1a e1b 1

C22 C23 C24 e2 a e2b 2
2
3 C33 C34 e3a e3b 3
=
4 C44 e4 a e4b 4
Da symm 1a 1b Ea

Db 2b Eb
where the subscripts 1, 2, 3 and 4 respectively represent the local

axes a, b, ab and c. There are totally 21 independent constants in
the full constitutive matrix.

1 C11 C12 C13 C14 C15 C16 e1a e1b e1c 1

C22 C23 C24 C25 C26 e2 a e2b e2 c 2
2
3 C33 C34 C35 C36 e3a e3b e3c 3

4 C44 C45 C46 e4 a e4b e4 c 4

5 = C55 C56 e5 a e5b e5c 5
C66 e6 a e6b

e6 c 6
6
Da symm aa ab ac Ea

Db bb bc Eb
Dc cc Ec
where the subscripts 1 - 6 respectively represent the local axes a, b,

c, ab, ac and bc. There are totally 45 independent constants in the
full constitutive matrix.
3.16.2 Orthotropic properties
The piezoelectric material model is generally orthotropic.

Therefore the elastic stiffness C can also be defined in terms of the
elastic modulus and Poisson's ratios (OPTION1 in command
'MATERIAL PIEZOELECTRIC') as follows:
1 ac
E ab 0
Eb Ec
a
1 bc
0
Eb Ec
C1 =
1
0
Gab
1
symm 0
Ec

1 ab ac
E 0 0 0
Eb Ec
a
1 bc
0 0 0
Eb Ec

1
0 0 0
Ec
C1 =
1
symm 0 0
Gab
1
0
Gac
1

Gbc
More information about the definition of the elastic-orthotropic

material can be found in Section 3.2.2.
3.16.3 Strain-form coupling matrix
The stress-form constitutive equation of the piezoelectric

material model is often written in the strain-form
= C eE = C-1 + dE
where d is called the strain-form piezoelectric coupling constants,

and e = Cd.
Thus, the piezoelectric coupling constants can be defined either

in stress-form as e, or in strain-form as d (OPTION2 in command
'MATERIAL PIEZOELECTRIC'):
Matrix d has a similar form as e.

d1a d1b
d d 2b
d = 2a
d3a d 3b

d 4 a d 4b
and for 3-D solid elements,
d1a d1b d1c

d d 2b d 2 c
2a
d d 3b d 3c
d = 3a
d4a d 4b d4c
d5a d 5b d5c

d 6 a d 6b d 6 c
3.16.4 Polarization direction of the piezoelectric material
The coupling matrix e (or d) of a piezoelectric material model is

generally not a full matrix due to the orthotropic property.
Corresponding to different polarization directions, the nonzero
entries are distributed in different places in the matrix e (or d). For
example, in the 3-D case, if the polarization is along c, b and a
direction respectively, the coupling matrix has the following form
for the respective directions:
0 0 e1c 0 e1b 0
0 0
e2 c 0 e2b 0

0 0 e3c 0 e3b 0
e= , e = and
0 0 0 e4 a 0 0
e5 a 0 0 0 0 0

0 e6b 0 0 0 e6 c

e1a 0 0
e 0 0
2a
e 0 0
e = 3a ;
0 e4b 0
0 0 e5c

0 0 0
and in the 2-D case, if the polarization is along a and b direction

respectively, the coupling matrix has the following form for the
respective directions:
0 e1b e1a 0
0 e2b e 0
e= and e =
2a
e3a 0 0 e3b

0 e4b e4 a 0
It is sometimes necessary to change the polarization from one

direction to another. This is achieved by
Directly defining the coupling matrix e (or d) in its

corresponding form as shown above. This is an easy operation
only if the two polarizations are both along the coordinate axes.
For a more general case when the polarization is transformed

between two arbitrary directions, the orthotropic axis system
must be used, as discussed in next section.
3.16.5 Orthotropic axis system for the piezoelectric material model
For the 3-D solid element, if the local axes system is determined
by three unit vectors a, b and c as shown in Figure 3.16-1, then the
direction cosine matrix from local the global axes is defined as

A11 A12 A13 ax ay az

A = A21 A22 A23 = bx by bz
A31 A32 A33 cx cy cz
Z
c
b
a Y
X
Figure 3.16-1: Material axes defined by unit vectors a, b and c
Then the stress vector in the global system is calculated as
= A
with
A112 A122 A132 2 A11 A12 2 A11 A13 2 A12 A13

2 2 2
A21 A 22 A 23 2 A21 A22 2 A21 A23 2 A22 A23
A2 A 2
A 2
2 A31 A32 2 A31 A33 2 A32 A33
A = 31 32 33

A11 A21 A12 A22 A13 A23 A11 A22 + A12 A21 A11 A23 + A13 A21 A12 A23 + A13 A22
A A A12 A32 A13 A33 A11 A32 + A12 A31 A11 A33 + A13 A31 A12 A33 + A13 A32
11 31
A21 A31 A22 A32 A23 A33 A21 A32 + A22 A31 A21 A33 + A23 A31 A22 A33 + A23 A32
The strain vector in global system is calculated as
= A
with
AT = A1
The electric field and electric displacement field vectors in global

system are transformed as
= AE
E
and
= AD
D
The elastic stiffness matrix in global system is transformed from C
= A CAT
C
The piezoelectric coupling matrix in the global system is also

transformed as
e = A eAT
and
d = A dAT
For the 2-D solid element, the direction cosines can be simplified
as
cos sin 0
A = sin cos 0
0 0 1
and the stress transformation matrix can be simplified as
c 2 s 2 2cs 0
2 2
s c 2cs 0
A =
cs cs c s 2
2
0

0 0 0 1
where, c = cos and s = sin .
Modeling hint:

In ADINA, the piezoelectric material properties are always defined

in material local axes. You may need to rotate the material axes to
the directions that you want them to be in the global axes by
applying the orthotropic axis system. Figures 3.16-2(a) and (b)
respectively show the local axis correspondence for 2D and 3D
cases when the option 'Show Material Axes' is turned on in the
AUI.
b c
b
a
a
(a) (b)
Figure 3.16-2: Material axes displayed in ADINA AUI.

(a) For 2D solid element (b) For 3D solid element
3.16.6 Electrical loading and boundary condition
In piezoelectric analysis, the electric loading includes prescribed

voltage, concentrated electric charge and distributed electric
charge, which can be applied to 2-D and 3-D solid elements with
the piezoelectric option. The voltage DOF is also one of the fixity
options in the boundary condition definition.
In 2-D analysis, the distributed charge can be applied on lines,

edges, and element edge sets; in 3-D analysis, it can be applied on
surfaces, faces, and element face sets. Both charge and voltage
loads are scalar values without direction.
It is recommended that at least one voltage DOF of each

separate piezoelectric body be fixed or prescribed with a voltage
value to suppress the numerical singularity.

3.17 Parallel-Network Framework

We now consider a material undergoing finite viscoelastic or
elastoplastic motions modeled as N rheologically-parallel networks,
each network potentially possessing both an elastic and inelastic
response. For the purposes of this section, we will present the
relevant continuum-mechanical quantities for each individual
network.
The various aspects of this framework are described in the

ref. M.E. Gurtin et al., The Mechanics and Thermodynamics

of Continua. Cambridge University Press, 2010.
ref. E. Krner, Allgemeine Kontinuumstheorie der

Versetzungen und Eigenspannungen, Arch. Rat. Mech.
Anal. Vol. 4, pp 273334, 1960.
ref. E.H. Lee, Elastic-plastic deformations at finite strains,

ASME Journal of Applied Mechanics, Vol. 35, pp 16,
1969.
ref. F.J. Montns and K.J. Bathe, Computational Issues in

Large Strain Elasto-Plasticity: An Algorithm for Mixed
Hardening and Plastic Spin, Int. J, Numer. Meth,
Engng., Vol. 63, pp 159196, 2003.
3.17.1 Network-specific kinematics
Rheologically, the parallel network assumption requires that all

networks undergo the same overall motions, giving rise to the
kinematic constraint
X = Xi , (3.17-1)
with i = 1, , N. Furthermore, in this framework we adopt the

ref. KJB approach of Krner (1960) and Lee (1969) and multiplicatively
Section 6.6.4
decompose each network's deformation gradient tensor

3.17: Parallel-Network Framework
X i = Xie Xiv , (3.17-2)
where Xie and Xiv are tensors associated with the elastic and
inelastic (viscous or plastic) parts, respectively. The inelastic part
describes those motions that remain after the body is elastically
unloaded; thus Xiv can be thought of as the motions from the
reference configuration to an intermediate, unstressed structural
configuration. The elastic part Xie defines motions from the
intermediate configuration to the current configuration.
Furthermore, det Xiv = 1 ; that is, the inelastic motions are
isochoric (volume-preserving). Thus, J = J i = det Xie .
This section generalizes for each of the N networks, thus we

will suppress the subscript i for the remainder of the discussion.
The elastic right Cauchy-Green tensor and Green-Lagrange strain

are
1 e
Ce = X e X e ;
T
Ee =
2
(C I ) , (3.17-3)
respectively. Inelastic strain and the total strain are thus equal
when, at any instant, the body is unloaded (and unstressed).
The spatial velocity field is v = u , and u is the displacement field.

The spatial velocity gradient l = v can be written in terms of the
elastic and inelastic distortions. We take the time derivative
X = Xe Xv + Xe Xv, (3.17-4)
1 1
and since X -1 = X v Xe , we can write
l = ( Xe Xv + Xe Xv ) Xv Xe ( 1 1
),
1 1 1 (3.17-5)
= Xe Xe + Xe Xv Xv Xe .
le Lv

Finally, and with no loss of generality,
1
v = l = l e + Xe Lv Xe , (3.17-6)
where l e is the spatial velocity gradient due to elastic motions and
Lv = Dv + W v, (3.17-7)
is the structural velocity gradient for inelastic motions, Dv is the

inelastic deformation rate (the symmetric part of Lv ), and W v is
the skew-symmetric inelastic spin tensor. Note that Lv is deviatoric
(its trace is zero), since det X v = 1 .
ref. KJB A key assumption within the framework for isotropic (visco)
Section 6.6.4 plasticity is that of plastic irrotationality that the inelastic spin
tensor W v vanishes. The irrotationality assumption allows us to
write
Lv = Dv , (3.17-8)
1
and recalling that Lv = X v X v ,
X v = D v X v. (3.17-9)
Finally, the so-called flow direction is
Dv
N = v ,
v
(3.17-10)
D
and the inelastic deformation rate is related to the network-specific

flow rule via
Dv = N v , (3.17-11)

or, equivalently Dv = , where constitutively describes the

networks time-dependent inelastic behavior.
3.17.2 Power expenditure, virtual velocities, and force balance
This subsection draws heavily from the following reference:
ref. M.E. Gurtin et al., The Mechanics and Thermodynamics

of Continua. Cambridge University Press, 2010, sections
22, 84, and 91101.
We can postulate a power balance of the form
t ( n ) v da + b v dv = ( T : l e + J 1 Tv : Lv ) dv ,(3.17-12)
e
B B B
for a body B. The left and right sides of equation (3.17-12)

represent external and internal power expenditures, respectively.
t ( n ) is a surface traction, and n is an outward unit normal to
the boundary B of body B.
b is a body force, possibly due to inertial effects.
e e
T is the work conjugate of l .
v v
T is the work conjugate of L .
da and dv are differential area and volume elements,
respectively, in the current configuration.
Note that all kinematic quantities are related by equation (3.17-6).
At this point, it is useful to introduce the concept of virtual

velocities. Let us now consider v , l e , and Lv at an arbitrary, but
fixed, instant. We will denote the virtual velocity v and virtual
velocity gradients l e and Lv . These virtual quantities may be
specified independently, though they must always satisfy the
relationship (cf., equation 3.17-6)

1
v = l e + Xe Lv Xe . (3.17-13)
Substituting these virtual quantities into equation (3.17-12) yields

the so-called balance of virtual power
t ( n ) v da + b v dv = ( T )
: l e + J 1 Tv : Lv dv. (3.17-14)
e
B B B
Careful selection of admissible virtual velocities can yield powerful

results. We will concentrate on two special cases:
Zero virtual inelastic flow

Zero virtual total velocity, but with internal relaxation
Let us first specify that the virtual inelastic velocity gradient

vanishes Lv = 0 . In this case, all virtual motions are due to the
elastic response, equation (3.17-13) requires v = l e , and the right
(T : v ) dv . The
e
hand side of equation (3.17-14) simplifies to
B
divergence theorem, substitution into (3.17-14) and some
simplifications reveal that
( t ( n ) T n ) v da + ( T + b ) v dv = 0,
e e
(3.17-15)
B B
and that t ( n ) = Ten and Te + b = 0 . Recalling Cauchys

postulate and the balance of linear momentum, it is clear that
Te , (3.17-16)
which is the Cauchy stress.
We may now write the elastic stress-power as : l e . Since = ,

-1 -T
then X e X e is also symmetric, by construction. Recalling that
1
l e = Xe Xe , one can define

-1 -T
S e J Xe Xe (3.17-17)
as the so-called elastic second Piola-Kirchhoff stress.
1
We will now select v = 0 so that l e = Xe Lv Xe . This
corresponds to the special case of zero total velocities, but with
virtual inelastic flow and internal stress relaxation. Equation (3.17-
14) simplifies to
(T : l e + J 1 Tv : Lv dv = 0, )
e
(3.17-18)
B
and we can write J 1 Tv : Lv = Te : X e Lv X e ( 1

) . Recalling that
Lv is deviatoric and that Te admits
T -T
Tv = J Xe dev() Xe , (3.17-19)
1
where dev ( A ) = A tr ( A ) I is the deviator operator. Using the
3
definition from equation (3.17-17), we can further simplify
equation (3.17-19)
Tv = J dev Xe Xe ( T -T
),
((
= dev Xe Xe S e ,
T
) ) (3.17-20)
= dev ( Ce S e ) .
The Mandel stress is
M Ce S e , (3.17-21)
and Tv dev ( M ) . The deviatoric Mandel stress has thus been

shown to be the work-conjugate of the inelastic velocity gradient,
and therefore flow. Hence, the deviatoric Mandel stress is assumed

to drive inelastic flow and so (cf., equations 3.17-10 and 3.17-11)
dev ( M )
= ( dev ( M ) ) ; N v = . (3.17-22)
dev ( M )
It is important to note that the deviatoric Cauchy and Mandel

stresses are identical for cases where the elastic and inelastic
distortions are diagonal (as for uniaxial tension or compression and
biaxial extension of isotropic materials) and when the motions are
isochoric (strict incompressibility).
3.17.3 Energy dissipation
Viscoelastic/plastic motions are dissipative processes and thus

involve hysteresis and associated energy dissipations. Within the
context of a purely mechanical framework (i.e., neglecting
temperature and entropy), the local free-energy inequality resulting
from the first and second laws of thermodynamics can be written as
W ( Ce ) e
S 2
e
: E + dev ( M ) : Dv 0, (3.17-23)
Ce

where W is the strain energy function constitutively describing the

T
hyperelastic response, and Ee = X e X e is the time rate of change
of the elastic Green-Lagrange strain tensor. Assuming a purely
W ( Ce )
hyperelastic response S = 2 e
, we arrive at the so-called
Ce
reduced dissipation inequality
Dint = dev ( M ) : Dv 0, (3.17-24)
where Dint is the internal energy dissipation.
3.17.4 Stress update procedure
t +t
Given the full deformation gradient tensor 0 X and the inelastic

deformation rate t +t Dv , integrating equation (3.17-9) over the

time interval [t , t + t ] using a backward-Euler exponential
mapping yields the updated inelastic distortion
t +t
0 Xv = exp ( t +t Dv t ) 0t Xv , (3.17-25)
from which we compute the elastic distortion
Xv ( t +0t X ) .
t +t t +t 1
0 Xe = 0 (3.17-26)
Recalling the definition of the Mandel stress and that

Ce = ( t +0t Xe )
t +t T t +t
0 0 Xe , the updated Mandel stress and flow
direction are
t +t t +t
W ( t +0t Ce )
M=2 0 C e
,
t +0t Ce
(3.17-27)
t +t
dev ( t +t M )
Nv = ,
dev ( t +t M )
respectively. The correct Mandel stress satisfies
dev ( t +t M )
(
dev ( t +t M ) ) dev ( t +t
M)
t +t Dv = 0. (3.17-28)
3.17.5 Three-network model
This subsection describes ADINAs implementation, with some

modifications, of the model described by:
ref. J.S. Bergstrm and J.E. Bischoff, An Advanced

Thermomechanical Constitutive Model for UHMWPE,
Int. J. Struct. Changes Sol., Vol. 2, pp 3139, 2010.

The Three-network model is especially well-suited for capturing

the highly nonlinear and time-, temperature-, and load history-
dependent behaviors of synthetic polymers including Polytetra-
fluoroethylene (Teflon) and Polyethylene. In particular, it has the
ability to model viscoelastic effects, irreversible, load history-
dependent changes to the materials effective stiffness, and entropic
stiffening at high strains.
This model is implemented within ADINA's Parallel-Network

framework and made available to the user via the command
MATERIAL THREE-NETWORK.
3.17.5.1 Rheological representation
The model's three networks A, B, and C are rheologically parallel

and thus experience the same total motions
X = X A = X B = XC . (3.17-29)
That is, all three chains are constrained to move together. Network
A captures viscoelastic effects, and network B captures both visco-
elastic effects and irreversible material softening similar to
permament yielding. Finally, network C is purely hyperelastic and
captures the effects of entropic stiffening at high strains.
Note that although the effects of material softening depend on

the loading history and are irreversible, this model is viscoelastic
and not plastic. That is, inelastic flow may occur under any loading
conditions, and there is no history-dependent back-stress. Thus,
there is no permanent set.
3.17.5.2 Thermoelastic response
Networks A, B, and C all possess Eight-chain type hyperelastic

responses. See section 3.8.1.6 for details. All three networks share
the same locking stretch L . The reduced network-specific elastic
chain stretch is i =
1
3
( )
tr det ( Cie ) Cie , where i = A, B, C.
1/ 3
When THERMAL-EFFECTS=YES, thermal effects are

incorporated directly through the shear modulus
0
A ( A , ) = A 1 + ,

0
B ( B , ) = B 1 + , (3.17-30)

0
C ( A , ) = C 1 + ,

where is the temperature, 0 is a reference temperature, and

describes the temperature response. When temperature effects are
not included A = A , B = B , and C = C .
The user provides initial shear moduli (at the reference

temperature) A and C directly. However, B evolves according
to the differential equation
B = ( B B final
) A , (3.17-31)
with the initial condition B = Binit . The user supplies material

parameters , Binit , and Bfinal Binit . This equation allows the
model to capture irreversible changes to the material's effective
stiffness. It also reveals that the elastic responses of networks A and
B are not completely independent.
Given t +t A , equation (3.17-31) is integrated over the time

interval [t , t + t ] using a backward-Euler method to obtain the
updated shear modulus
t +t
B = B
final
( )
+ t B Bfinal exp ( t +t
A t ) , (3.17-32)
and B monotonically approaches Bfinal as a function of network

A's inelastic flow rate history.

Network C is purely hyperelastic with reduced right and left

Cauchy-Green tensors C = J 2 / 3 X T X and b = J 2 / 3 X X T ,
respectively. Its deviatoric (volume-preserving) Cauchy stress
contribution is
1 C ( C , ) L ( C / L )
1
Cdev = dev ( b )
1 + q J C
( )
L 1 L 1
(3.17-33)
q C ( C , ) 2J
+ J1b 2 I b 2 ,
1+ q J 3
where q 0 scales the effect of the first and second terms, L 1 is

the inverse of the Langevin function L ( x ) = coth ( x ) 1/ ( x ) ,
and
1
J1 = tr ( C ) ; J 2 = tr ( C ) tr ( C2 ) ,
2
(3.17-34)

2
are the first and second reduced invariants, respectively. The J 2

dependence in the second term of equation (3.17-33) phenomeno-
logically captures the effects of entropic stiffening at high strains.
All three networks share the same bulk modulus , which enters
the model through the volumetric strain energy contribution
WV ( J ) = ( J ln J ( J 1) ) . (3.17-35)
The pressure is then p = ln J . The u/p formulation is always

used with the Three-Network model.
3.17.5.3 Viscous response
The temperature-dependent viscous behavior of networks A and B

is of the form

mi n
Mi
i = 0 , (3.17-36)
i + a R ( pi ) 0
1
for i = A,B. M i = M i : M idev is the norm of the deviatoric
2 dev
Mandel stress M idev = dev ( M i ) , pi is the pressure, i , a , and
mi are material constants governing the rate of viscous flow,
R ( x ) = max ( x, 0 ) is the ramp function, n is the temperature
response exponential, and 0 = 1 s 1 for dimensional consistency.
Note that although Bergstrm and Bischoff (2010) use the Cauchy
stress to drive the viscous response, we use the Mandel stress for
the reasons outlined in section 3.17.2.
Further, we take the plastic spin to vanish Wiv = 0, and thus the
relation (cf., equation 3.17-6)
1
l = l ie + Xie Div Xie , (3.17-37)
must hold at all times. For more information about these kinematic
assumptions, see section 3.17.1.
Heat generation: A user-specified fraction of the energy dissipated

by the viscous effects can be considered as heat generation. This
heat generation can cause heating in a TMC (thermo-mechanical-
coupling) analysis.
3.17.5.4 Constants and their interpretations
Three-Network model involves several material parameters,

described below. Each parameters corresponding designation in
ADINA is given in parentheses and its dimensions in brackets,
when applicable.

L (LAMBDL): The locking stretch for all networks. L must be

greater than or equal to 1.0.
M
(KAPPA) : The bulk modulus for all networks.
L T
2
a (PD): The pressure dependence. When a = 0 , the viscous

response is insensitive to the hydrostatic pressure.
M
A (MUA) : The initial shear modulus of network A at
L T
2
the reference temperature 0 .
M
A (FRA) : Resistance to viscous flow in network A.
L T
2
As A , the viscous response in network A vanishes.
mA (MA): Stress exponential for network A. When mA = 0 , the

viscous response in network A is insensitive to the Mandel
stress and hydrostatic stress.
M
B (MUBI) : Network Bs initial shear modulus at
L T
init 2
the reference temperature and its virgin state.

M
B (MUBF) : Network Bs final shear modulus. This
L T
final 2
must be less than or equal to Binit .
(BETAB): This dimensionless constant controls how quickly

the shear modulus for network B approaches B . Settingfinal
= 0 recovers the special case of B = B and completely

init
decouples networks A and B.

M
B (FRB) : Resistance to viscous flow in network B.
L T
2
mB (MB): Stress exponential for network B.
M
C (MUC) : Initial shear modulus for network C at the
L T
2
reference temperature.
q (QI2C): This positive, dimensionless constant governs the

relative effects of the two terms of network Cs deviatoric
stress. As q 0 , the J 2 -dependence (the effects of entropic
stiffening) vanishes.
(TEMPHAT): Stiffness response to the change in

temperature. When = 0 , the shear modulus is insensitive to
temperature.
0 (TREF): Reference temperature.

n (TEMPN): Temperature exponential. When n = 0 , the
viscous response is insensitive to temperature.
In addition to the 16 constants described above, the Three-

Network model requires the following parameters.
TEMPUNIT: If the problem is to include temperature effects, the

user may specify the temperature unit for TREF, TEMPHAT as
Celsius, Fahrenheit, Kelvin, or Rankine.
GENFAC: The heat generation factor (fraction of dissipation

energy convered into heat generation). If GENFAC = 0, the
dissipation is not calculated and there is no heat generation.
MAXITE: As discussed in section 3.17.4, Parallel Network

framework requires an iterative material stress update to obtain
the correct Mandel stresses and inelastic distortions. The
parameter MAXITE specifies the maximum number of iterations

that ADINA should use for this stress update convergence.
RTOL: The tolerance required for convergence of the stress

update. It is not recommended to loosen this tolerance from the
default of 1.0E-12.
3.17.5.5 Additional notes
The network-specific pressure pi is taken to be 1/3rd the total

hydrostatic pressure.
The total hydrostatic pressure cannot be accurately computed

from the strains in the u/p formulation. Consequently, the stress
update will not converge quadratically and the constitutive tensor
will not be tangent.
The Three-network model is highly sensitive to the stress

exponentials mA and mB . Very large values for mA and/or mB
may well slow or prevent material stress update convergence. In
that case, the user may increase the number of material stress
update iterations MAXITE and/or decrease the time step or use
ATS. If at all possible, it is recommended that the user not loosen
the stress update tolerance RTOL.
3.18 User-coded material model

The user-coded material model is provided in ADINA to allow
you to construct any type of material model for use with 1-D truss,
2-D or 3-D solid elements.
This section first explains the general concepts involved in

defining and using user-coded materials. The input and output
parameters are then explained. This is followed by the details of
the stress integration, then specific issues related to 1-D, 2-D and 3-
D cases. Finally, the user-coded material models for which the
source code is provided with the ADINA installation CD are
described.

3.18: User-coded material model
The constitutive relation must be arranged in the form of an

algorithm and inserted into subroutine UCMAT1 for 1-D truss
elements, subroutine UCMAT2 for 2-D solid elements or
subroutine UCMAT3 for 3-D solid elements.
The material model can be used with the following

formulations:
Materially-nonlinear-only, for infinitesimally small

displacements and strains (MNO)
total Lagrangian, in which the displacements and rotations

can be very large, but the strains may have to be small or can be
large depending on the material characteristics (TL)
updated Lagrangian Hencky, in which case the strains can be

large (ULH) (ULH formulation is not available for 1-D truss
elements.)
Either the displacement based elements or the mixed-

interpolated displacement/pressure elements can be used with the
user-coded material model. The mixed-interpolated elements are
preferred for (nearly) incompressible materials. The mixed-
interpolated element formulation is not applicable to 1-D truss
elements.
The user-supplied material subroutine UCMAT* is called for all

integration points and is used to perform the following five types of
operations during the various phases of the analysis. The type of
operation to be performed is controlled by the integer variable
KEY.
KEY = 0: Set size of working arrays (if requested). Set size of

integer and real working arrays (LGTH2 and LGTH 1,
respectively). Also select portion of these arrays to write the results
to the (porthole) file (LGTH 4 and LGTH 3 for the integer and real
arrays, respectively). This call with KEY = 0 is only performed if
requested from the user-coded material command, by setting
AUTOLEN = YES.

KEY = 1: Initialization call. Called once at the beginning of the

analysis.
This call is for the initialization of the integration point variables,

and is performed only once per integration point during the input
phase. The arrays STRAIN, STRESS, ARRAY and IARRAY are
automatically initialized by ADINA to zero. The user coding must
set ARRAY and IARRAY to their proper initial values, if different
from zero. Error checks on the material data and working array
sizes should be performed here.
KEY = 2: Stress calculation call. Called for each iteration within a

load (time) step. It is called for each subincrement if
subincremenation is used.
This call is for the calculation of the element stresses at time t + t.

It is made during the solution of the governing equilibrium
equations, and in the stress calculation phase after convergence is
reached. In addition to the stress calculation, the integer and real
history arrays IARRAY and ARRAY must also be updated when
KEY = 2 (if they are used).
The stress integration can be performed in a single step using
the total increment of strains and displacements or in subincrements
of time. In this case, each subincrement spans a time of t/n where
n is the number of subincrements. This subincrementation may be
useful for incremental rate type formulations such as
viscoplasticity.
The actual stress integration adopted by the user within the
increment or subincrement can be implicit or explicit. Details of the
stress integration procedure are provided in Section 3.18.3.
If temperature-dependent material properties are present, the
variables ALFA, ALFAA, CTD(*), CTDD(*) containing the
properties at times t and t + t , are available for the user-supplied
subroutine.
Note that if the ULH formulation is employed or if mixed-
interpolated displacement/pressure elements are used, then the
STRESS array will be internally modified by ADINA immediately
after this stage to reflect the correct stresses at the element level.
KEY=3: Stress-strain matrix calculation calls. Called when the

tangent stiffness matrix is to be calculated.
This call is for the calculation of the constitutive matrix D which is

to be employed in the evaluation of the tangent stiffness matrix.
The variables available for the constitutive matrix calculation are
the values defined or calculated in the last subdivision of the stress
calculation when KEY=2, as described above. Note that the
algorithmic tangent constitutive matrix should be used if possible,
to achieve quadratic convergence.
In the 1-D truss element case, D is the tangent modulus.
KEY=4: Stress printout call. Called for stress printout to the .out
file.
This call is for the printing of the element response in the stress
calculation phase. No stress calculations or updates should be
performed when KEY=4.
Note that if the TL formulation is employed, ADINA
automatically transforms, just for printout, the stresses used in the
constitutive relations (see Table 3.18-1) to the Cauchy stresses in
the global coordinate system before calling the user-supplied
subroutine with KEY = 4.
In all cases of 1-D elements, the cross-sectional area of the
element is assumed to remain unchanged, and the strain is equal to
the longitudinal displacement divided by the original length.
If the ULH formulation is employed, total stretches PRST(3) are
calculated and passed to the user-coded material subroutine.
In 2-D, PRST(1) stores the maximum in-plane stretch
(maximum deviation from 1.0), PRST(2) stores the smaller in-
plane stretch and PRST(3) is the stretch in the X-direction. Also,
the angles of the first and the second principal stretches with
respect to the Y-axis, stored in the array ANGLE(2), are passed to
UCMAT2.
In 3-D, the PRST(3) vector stores the principal stresses and
PHIST(3.3) stores the direction cosines.
Table 3.18-1: Stress and strain measures used in the user-coded

material model. Note that the stresses are transformed internally by
ADINA into true (Cauchy) stresses for printout (KEY=4)

Kinematic formulation Strain measure Stress measure
MNO (materially- nonlinear- Infinitesimal strains Cauchy stresses

only): small displacements,
small strains
TL (total Lagrangian): large Green-Lagrange strains 2nd Piola-Kirchhoff
displacements, small or large stresses
strains (depending upon the
material model)
ULH (updated Lagrangian Logarithmic Hencky Cauchy stresses
Hencky): large displacements, (elastic) strain and
large strains plastic deformation
gradient
The contents of the real and integer working array are written to
the porthole file for post-processing at every integration point (the
first LGTH3 entries for the real array, and the first LGTH4 entries
for the integer array) . They can be accessed via the
USER_VARIABLE_I and INT_USER_VARIABLE_I variables in
ADINA-PLOT.
3.18.2 Input and output parameters
The user-supplied subroutines get numerous variables and arrays

from ADINA to provide maximum flexibility to the user. The
variables and arrays can be categorized as material properties,
history dependent, algorithm control and output variables. Further
documentation on these variables can be seen in the listings of the
sample user-coded material subroutines provided in the ADINA
installation. See, for example, the file ovl30u_vp1.f.
History dependent variables and arrays. Stored at integration

points and characterize the history of the material behavior:
STRAIN(*): Strain components at time t + t. For the MNO

and TL formulations, this vector stores the total strains
(including elastic, inelastic, thermal). For the ULH formulation,

STRAIN has the (right) elastic strains (total minus inelastic).

The size of this vector is 4 for 2-D and 6 for 3-D. For the 1-D
elements, STRAIN is the axial strain at time t + t.
EPS(*): Strain components at time t. For MNO and TL

formulations, this vector stores the total strains. For the ULH
formulation, it stores the (right) elastic strains at time t. For the
1-D elements, EPS is the axial strain at time t.
THSTR1(*): Thermal strain components at time t. For the 1-D

elements, THSTR1 is the axial thermal strain at time t.
THSTR2(*): Thermal strain components at time t + t. For the

1-D elements, THSTR2 is the axial thermal strain at time t + t.
ARRAY(LGTH1): Working array of material history

parameters (floating point variables).
LGTH1: Length of array ARRAY.
LGTH3: Number of terms in array ARRAY to save as results in

porthole file.
IARRAY(LGTH2): Working array of material history

parameters (integer variables).
LGTH2: Length of array IARRAY.
LGTH4: Number of terms in array IARRAY to save as results

in porthole file.
PHIST(3,3): For the TL formulation, this array stores the

components of the deformation gradient. For the ULH
formulation, PHIST(3,3) stores the direction cosines
corresponding to the principal (right) stretches.
PHIST1(3): For the ULH formulation, 3-D elements, and when

KEY=4, this vector stores the direction cosines of the first
principal (right) stretch (stretch which has the maximum
deviation from 1.0).

RN(*,*): For the ULH formulation, this array stores the

components of the rotation tensor (from the polar
decomposition theorem).
PRST(3): For MNO, TL: Original coordinates of integration

point.
For ULH: Principal right stretches.
Note that the PHIST, PHIST1, RN, and PRST arrays are not
applicable to 1-D elements.
Note also that the ARRAY and IARRAY arrays are generated
per integration point. Therefore, the size of these arrays should
not be overestimated. Temporary local arrays should be used
instead for temporary calculations if needed.
Material properties variables and arrays. These properties are

provided via the user-coded material model in ADINA.
Three types of inputs are available for the user: a list of material
constants, a list of solution control constants, and a table of all
properties that are temperature dependent. The material and
solution control constants serve the same purpose, and either of
them can be used to provide the real-numbered material constants
to the user-coded material. If, for example, the material used a
Youngs Modulus E which is constant it can be passed to the user-
subroutine preferably in the material constant array. However, if E
varies with temperature, it should be passed in the temperature
dependent table. This information is passed to subroutine
UCMAT* as follows:
CTI(NCTI): Array of material constants.
NCTI: Number of material constants (maximum 99).
SCP(NSCP): Array of solution control constants.
NSCP: Number of solution control constants (maximum 99).
CTD(NCTD): Temperature-dependent material properties

constants for the given temperature at time t ( for

subincrementation). ADINA calculates CTD(NCTD) from the

table of material parameters vs. temperature (if any) provided
by the user-coded material model.
CTDD(NCTD): Temperature-dependent material properties

constants for the given temperature at time t + t (or t + for
forward subincrementation). ADINA calculates CTDD(NCTD)
from the table of material properties vs. temperature (if any)
provided by the user-coded material model.
NCTD: Number of temperature-dependent material properties

constants (maximum 98).
ALFA: Coefficient of thermal expansion for the given

temperature at time t. ADINA calculates ALFA from the table
of vs. temperature (if any) provided by the user-coded
material model.
ALFAA: Coefficient of thermal expansion for the given

temperature at time t + t for forward stress integration or t -
t for backward stress integration. ADINA calculates ALFAA
from the table of vs. temperature (if any) provided by the
user-coded material model.
Algorithm control variables. The algorithm control variables are

passed to subroutine UCMAT* as follows:
TMP1: Temperature at time t.
TMP2: Temperature at time t + t (forward stress integration)

or t - t (backward stress integration).
TIME: Time at t + t.
DT: Increment of time t.
INTEG: Flag for stress integration scheme. 0 = forward, 1 =

backward.

ITE: Iteration number.
INTER: Number of subincrements for incremental stress

integration.
Element information. These are the variables providing extra

information about the element to subroutine UCMAT*.
NG: Element group number.
NEL: Element number.
IPT: Integration point number.
ETIMV: Birth time of the current element.
ETIMV2: Death time of the current element.
NODNUM(NNODE): Global node numbers of element NEL.
XYZ(*, NNODE): Initial nodal coordinates of element NEL.
NNODE: Number of nodes in element.
Output variables and arrays. These are variables that must be

calculated by the user-supplied material at different phases of the
analysis.
STRESS(*): The stress vector. Input to subroutine with

stresses at time t and should output with stresses
at time t + t. Calculated when KEY = 2.
The size of this vector is 4 for 2-D and 6 for
3-D elements. For the 1-D elements, STRESS
is the axial stress.
DP(*, *): Stress-strain matrix. Calculated when KEY = 3.

The size of this array is 4 x 4 for 2-D and 6 x 6
for 3-D. This array is not applicable for 1-D
elements.

DPSP(*): For the ULH formulation, this vector stores the

increment of inelastic strains (plastic and/or
creep and/or viscoplastic, etc.), calculated by the
user-supplied code when KEY=2. This array is
not applicable for 1-D elements.
3.18.3 Stress integration procedure
If the stress calculation at t + t is solely based on the state at t +

t,
t +t
= f ( t +t
e, t +tT , t +t ,...)
then it is the most straightforward to implement, and includes
elastic and hyperelastic materials. However, for incremental or rate
type functions, or such as
= f (e)
the user must decide between single and multi-step calculation and
also between implicit or explicit stress calculation. An example of
an explicit stress calculation is
t +t
= t + t C E ( e eth e IN ) (3.18-1)
An example of an implicit stress calculation is
t +t
= t +t
C ( t +t e t +t eth t +t e IN ) (3.18-2)
where CE is the elastic constitutive matrix and eth and eIN are the
increments of thermal and inelastic strains in the subincrement. For
the 1-D elements, CE is the elastic modulus and eth and eIN are
the increments of axial thermal strain and axial inelastic strains i
the subincrement.
The calculation of Eq. (3.18-2) can be done iteratively or a
predictor-corrector algorithm can be used (such as the elastic

predictor followed by a radial vector corrector).
Backward stress integration
The stress calculation is as follows:
First, the subroutine UCMAT* is called for the evaluation of the

trial elastic stress *E corresponding to the total and thermal strains
at time t + t, t + te and t + teth, respectively, and to the inelastic
strains at time t, teIN. For example
t +t
*E = t +t C E ( t +t e t +t eth t e IN ) (3.18-3)
where t + tCE is an elastic constitutive matrix corresponding to

t + t
temperature . Calculation of the trial elastic stress must
be provided by your coding.
The stress correction is then calculated. For example
t +t
= C E e IN
where t +t C E is the elastic constitutive matrix corresponding to

temperature t + t.
Note that in the case of the updated Lagrangian Hencky

formulation only the backward stress integration can be employed.
Also, the increments of inelastic strains evaluated by your coding
must be passed to ADINA for the calculation of the inelastic
deformation gradient in ADINA (array DPSP).
In the example of Eq. (3.18-3), arrays STRAIN and THSTR2

contain the total strains and thermal strains as described in the
listing of subroutine UCMAT2 (see file ovl30u_vp1.f). Your
coding must store into the working arrays ARRAY and/or
IARRAY the history of deformation variables, e.g., the inelastic
strains eIN, etc. and use them in the calculation of stresses.
In the calls to UCMAT* following calculation of the trial elastic
stress, you must provide the coding for the stress increments due to
the inelastic deformation.

Note that when using the ULH formulation, the array STRAIN
contains the trial elastic strains for KTR = 0; the array is further
updated in ADINA by subtracting increments of inelastic strains
DPSP for each subdivision. Stresses and strains passed to
subroutine UCMAT* correspond to deformation without rigid body
rotations. The stresses STRESS which are true (Cauchy) stresses
are passed to subroutine UCMAT* only for printout (KEY = 4).
The transformation that takes into account the material rotation is
performed in ADINA prior to calling subroutine UCMAT*.
An example of coding for the backward stress integration is
given in the listing of the viscoplastic material example.
Note that when using the TL formulation, the stresses in array

STRESS correspond to the second Piola-Kirchhoff stress tensor.
The Cauchy stresses, obtained by the transformation of the second
Piola-Kirchhoff stress tensor in ADINA, are passed to the
subroutine UCMAT* only for printout (KEY=4).
If you require the stress invariants, your coding can call subroutine
XJ1232 or XJ123 described in Section 3.18.4.
Note that if the large displacement/large strain options are used,

stresses energetically conjugated with the corresponding strains
(see Table 3.18-1) are employed in these calculations, and you
cannot control this.
3.18.3.1 Subincrementation
Subincrementation can be used for forward stress integration. In

this case, instead of going from time t to time t + t in one step,
subincrements are used. Subincrements can be forward in time
from time t to t + t in k steps as shown in Fig. 3.18-1 when =
t/k.. The time forward subincrementation can be used with both
implicit or explicit stress integration.

Strain
Time " Time "$"
or t + (k-1)" or t + k "
e(k)
ij
Time t Time t+t te " "+"e t+teij

ij eij ij
Time
" "$"
t t+t
Subincrements Direction of
subincrementation
Figure 3.18-1: Schematic of forward time incrementation
Values provided and passed back through

the calling arguments of subroutine UCMAT*
Variable or At entry, the arguments At exit, the arguments

Array name correspond to time: correspond to time:
EPS(*) t t
STRAIN(*) t+t t t
THSTR1(*) " "
THSTR2(*) " " " "
TMP1 " " No change
TMP2 " " " " in value
ALFA " "
ALFAA " " " "
CTD(*) " "
CTDD(*) " " " "
ARRAY(*) " " "
IARRAY(*) " " "
STRESS(4) " " "
Figure 3.18-2: Variable definition for forward time incrementation
The variables passed to the user-coded material subroutines are

defined as shown in Fig. 3.18-2.
If an explicit stress integration is employed, we get

t + k t t + ( k 1) t
= + k (3.18-4)
The stress increment k in this example corresponds to strain

and other quantities at time t + (k-1)t.
In the case of an implicit time integration, we can have
k = t + ( k 1) t C E ( k e k eth k e IN ) (3.18-5)
where CE is the elastic constitutive matrix and keth and keIN are
the increments of thermal and inelastic strains at time t + k.
For the 1-D element case, equations (3.18-4) and (3.18-5)
become scalar forms.
The evaluation of the strain subincrements is performed by

ADINA prior to calling subroutine UCMAT*. Your coding must
update the stresses (in array STRESS) and other history dependent
variables for each subincrement, and print the stresses once the
final stresses for the load (time) step have been evaluated.
The total strain increment is divided equally among the

substeps.
If temperature-dependent material properties are present, the

variables ALFA, ALFAA, CTD(*), CTDD(*) containing the
properties at times and + , are available for the user-supplied
subroutine.
3.18.4 Coding guidelines
The ADINA user-coded materials use FORTRAN 90/95

conventions but maintain a few restrictions from FORTRAN 77
such as the 72-character line width.
The user is responsible for updating variables only at
appropriate function calls (KEY = 2). It is advisable also to
perform error checks during KEY = 1. These include checking the
integrity of the material data input in the MATERIAL USER-
SUPPLIED command, as well as the lengths of the working arrays
assigned by the user. Errors can be reported to the .out file which
is accessible via parameter IOUT which is passed to the user-coded
material subroutine.

Several internal ADINA subroutines are available to facilitate

the user implementation. These include:
SUBROUTINE XJ123 (STRESS, XI1, XI2, XI3, XJ1, XJ2, XJ3)
Variable descriptions:
Description: Calculates the invariants of the 3-D stress

tensor
Input: Stress
Output: XI1 XI3. The 3 invariants of the stress tensor.
XJ1 XJ3. The 3 invariants of the deviatoric stress tensor.
They are calculated as follows:
XI1 = ii , XI2 = 1/ 2 ij ij , XI3 = 1/ 3 ij jk ki

XJ1 = 0, XJ2 = 1/ 2 sij sij , XJ3 = 1/ 3 sij s jk ski
where ij is the stress tensor passed in STRESS and sij is the

deviatoric stress tensor corresponding to ij . The summation
convention is used in the above expressions.
IMPLICIT DOUBLE_PRECISION (A-H, O-Z)

DIMENSION STRESS (6)
SUBROUTINE XJ232 (STRESS, XI1, XI2, XI3, XJ1, XJ2, XJ3)
Description: Calculates the invariants of the 2-D stress

tensor
Input: STRESS
Output: XI1 XI3. The 3 invariants of the stress tensor.
XJ1 XJ3. The 3 invariants of the deviatoric stress tensor.
They are calculated as in subroutine XJ123.


DIMENSION STRESS (4)
SUBROUTINE BISECT (X, XMN, XPL, DX, XMIN, F, FMN,

FPL, ACCEL, IMINUS, IACCEL,
INDB)
Description: Computes a new trial value of X and its

increment DX to find a solution for a
nonlinear equation F(X) = 0 using the
bisection method.
Input: F, ACCEL, IMINUS, IACCEL, INDB, described as
follows:
F: current value of the function F

ACCEL: acceleration factor to increase DX
IMINUS: indicator for the slope of the function in the
interval (XMN, XPL). Value of 1 for a
positive slope, 2, for a negative slope.
IACCEL: indicator for application of acceleration
factor during bisection inside the interval
(XMN, XPL). Value of 0 when acceleration
is not applied to the slope TT, and 1 when
applied. Currently not used.
INDB: indicator, value of 1 when both XMN and
XPL still not found, and value of 2 when
XPL and XMN have been obtained.
Input & Output: X, FMN, FPL, XMN, XPL, described as

follows:
X: current value of the argument X

FMN: negative value of F closest to 0
FPL: positive value of F closest to 0
XMN: X corresponding to FMN
XPL: X corresponding to FPL

Output: DX, described as follows:
DX: Increment of X (positive)
3.18.5 1-D truss elements
The user-supplied material algorithm for 1-D elements must be

placed in subroutine UCMAT1.
A sample version of UCMAT1 may be found in file

ovl20u_vpl.f. The coding within this sample subroutine is fully
explained in Section 3.18.8.1.
The stress in this 1-D element is the axial stress.
3.18.6 2-D solid elements
The user-supplied material algorithm for 2-D elements must be

placed in subroutine UCMAT2.

ovl30u_vpl.f. The coding within this sample subroutine is fully
explained in Section 3.18.8.1.
The convention used for stresses and strains is shown in Fig.

3.18-3.
"zz
"yz
STRESS(1)="yy
STRESS(2)="zz
"yy
STRESS(3)="yz
Z "xx
STRESS(4)="xx
Y
X
Figure 3.18-3: Stress convention for 2-D elements

3.18.7 3-D solid elements
The user-supplied algorithm for the user-coded material model

for 3-D solid elements must be placed in subroutine UCMAT3.

ovl40u_vp1.f. The coding within this sample version is fully
discussed in the discussion of the viscoplastic model (see below).
The convention used for stresses and strains is shown in Fig.

3.18-4.
"zz STRESS(1)="xx
"yz
STRESS(2)="yy
"xz
STRESS(3)="zz
"yy
STRESS(4)="xy
Z "xx
"xy STRESS(5)="xz
Y STRESS(6)="yz
X
Figure 3.18-4: Stress/strain convention for 3-D elements
When the ULH formulation is employed, the principal stretches

and the direction cosines of the maximum principal stretch (with
maximum deviation from 1.0) are passed to the subroutine
UCMAT3 for the printout phase (KEY=4).
If you require the stress invariants, your coding can call

subroutine XJ123. This subroutine is described in Section 3.18.4.
It returns the three invariants XI1, XI2 and XI3 of the stress tensor,
and the three invariants XJ1, XJ2 and XJ3 of the deviatoric stress
tensor.
3.18.8 User-coded material models supplied by ADINA R&D, Inc.
User-coded material model subroutines are provided by ADINA

(during installation) and can be modified by users for the following

materials:
- Viscoplastic
- Elasto-plastic
- Thermoplasticity and creep
- Concrete with creep
- Viscoelastic
- Ramberg-Osgood with mixed hardening
- Nonlinear elastic
- Mooney-Rivlin and Ogden
Details on some of these material models are provided in the

sections below.
3.18.8.1 Viscoplastic material model
A viscoplastic material model for axisymmetric or plane strain

conditions is given as an example of the use of the UCMAT2
subroutine. A viscoplastic model for fully 3-D conditions is given
as an example of the UCMAT3 subroutine, and a 1-D truss element
example is also provided in the UCMAT1 subroutine of
ovl20u.vol.f. The coding in subroutine UCMAT2 is based on the
equations given hereafter.
The UCMAT2 and UCMAT3 subroutine coding can be found

in files ovl30u_vp1.f and ovl40u_vp1.f respectively. The
UCMAT1 subroutine can be found in the file ovl20u.vol.f.
The effective stress and accumulated effective plastic strain are

stored as the first and second variables in the working array and
provided as output in the porthole file. This is followed by the
components of the viscoplastic strain tensor (4 in 2-D and 6 in 3-
D). In the 1-D element case, the effective stress, accumulated
effective plastic strain and viscoplastic strain are stored in the
working array.

The rate of viscoplastic strain is given by
t
3 sij
t VP
e = < > t (3.18-6)
2
ij
where
t
sij = component of deviatoric stress
3t t
t
= sij sij = effective stress
2
= fluidity parameter
< > = if > 0

= 0 otherwise
In the above,

t
= 1 (3.18-7)
y
where y is the yield stress (above which the material is considered

perfectly plastic).
The index i has a range of 1 to 4 for 2-D and 1 to 6 for 3-D, and
follows the conventions defined in Figures 3.18-3 and 3.18-4.
The rest of this section assumes a 2-D material but all the
expressions can be generalized to the 3-D case.
The material is subjected to mechanical and thermal loading and

the material constants are considered to be temperature
independent.
Stress integration is performed using forward or backward

integration.
Forward time incrementation: In this case, the explicit Euler

forward stress integration is used, where the increment of stresses

in the subincrement (Eq. (3.18-2)) are
i( k ) = CijE ( e(jk ) ethj ( k ) eVP

j
(k )
) (3.18-8)
where
e (k )
=
( t +t
ej tej )
(3.18-9)
j
NSUBD
and NSUBD is the number of subincrements,
ethj ( k ) = ( + ) for j = 1, 2, 4
3 sj
e VP ( k )
= ( )( ) 1 (3.18-10)
j
2
y
Here is the mean coefficient of thermal expansion, and are

defined in Fig. 3.18-1, ej are the strains.
The same convention is used for increments of viscoplastic strains

and deviatoric stresses. The coefficients CijE represent the elastic
constitutive matrix

1 (1 ) 0

(1 )

CE =
E (1 ) 1 0
(1 )
(1 + )(1 2 )
SYMM
(1 2 ) 0
2 (1 )

1
(3.18-11)
where E and are the Young's modulus and Poisson's ratio.

In the 1-D case, e quations (3.18-6), (3.18-8), (3.18-9), and (3.18-

10) can be expressed as
3 t
t VP
e = < > (3.18-6a)
2 t
( k ) = C E ( e( k ) eth ( k ) eVP ( k ) ) (3.18-8a)
e (k )
=
( t +t
e te ) (3.18-9a)
NSUBD
eth ( k ) = ( + )
3
eVP ( k ) = ( )( ) 1 (3.18-9a)
2 y
where CE is the elastic modulus.
Backward time incrementation: In this case, the implicit Euler

backward stress integration is used. The trial elastic state is
determined according to Eq. (3.18-5),
t +t
iE* = CijE ( t +t e j t +t ethj t eVP
j ) (3.18-12)
The increments of stresses are obtained from
i( k ) = CijE eVP
j
(k )
(3.18-13)
and the eVP

j
(k )
are given by Eq. (3.18-10).
If during subincrementation the effective stress ( k ) is
(k ) < y (3.18-14)
the subdivision loop is stopped by setting the parameter ISUBM to

-1. Also the stresses are corrected as follows:
t +t
= x ( k ) + ( k ) (3.18-15)
y (k ) t +t
where x = and the are the final stresses at time
( k )
t+t. Similarly, the increments of the viscoplastic strains are
corrected using
t +t VP
e = x eVP ( k ) + eVP ( k ) (3.18-16)
t +t VP
where e are the final viscoplastic strains at time t+t.
In the 1-D element case, equations (3.18-12), (3.18-13), (3.18-15)

and (3.18-16) are
t +t
*E = C E ( t +t e t +t eth t eVP ) (3.18-12a)
( k ) = C E eVP ( k ) (3.18-13a)
t +t
= x ( k ) + ( k ) (3.18-15a)
t +t VP
e = x eVP ( k ) + eVP ( k ) (3.18-16a)
Constitutive matrix: For the calculation of the constitutive matrix,

the constitutive relation (3.18-8) can be written as
m = cm ( em eth ) (3.18-17)
1
si =
aE
( ei eiVP ) (3.18-18)
where

1
em = ( e1 + e2 + e4 )
3
(3.18-19)
1
e3 = e3
2
are increments of the mean and deviatoric strains, and cm and aE

are elastic constants
E
cm =
1 2
(3.18-20)
1+
aE =
E
The derivatives of the mean stress and of the deviatoric stress

follow from Eqs. (3.18-17) and (3.18-18):
m 1
= cm j = 1, 2, 4
e j 3
(3.18-21)
si 1 e ( eiVP )
= i
e j aE e j e j

ei
The derivatives can be written in matrix form as
e j
2 1 1
ei 1
= 1 2 1 i, j = 1, 2, 4 (3.18-22a)
e j 3
1 1 2
e3 1
= (3.18-22b)
e3 2
ei
=0 for i = 3 or j = 3 and i j (3.18-22c)
e j

In the case of viscoplastic flow in the time step, an approximate
expression for
( e ) VP
i
can be obtained from Eq. (3.18-10) as
e j
( eiVP ) si
= cVP (3.18-23)
e j e j
where
3 t +t
cVP = 1 (3.18-24)
2 t +t y

It follows from Eqs. (3.18-23) and (3.18-21) that the derivatives

si
are
e j
si e
= c i (3.18-25)
e j e j
where
1
c = (3.18-26)
aE + cVP
VP
The constitutive matrix C can be obtained using the relation
i = si + m i = 1, 2, 4
(3.18-27)
3 = s3
from which it follows that
si m
CijVP = + i = 1, 2, 4 (3.18-28a)
e j e j

s3
C3VPj = (3.18-28b)
e j
By substituting Eqs. (3.18-21) and (3.18-25) into (3.18-28) and

VP
using Eq. (3.18-22), the constitutive matrix C is obtained as
1 1 1
3 ( cm + 2c ) 3 ( cm c ) 0
3
( cm c)

1 1

( cm + 2c) 0 ( cm c)
3 3
CVP =
SYMM 1
c 0
2
1

( cm + 2c )
3
In the 1-D element case, the viscoplastic modulus is
1
C VP =
aE + cVP
3.18.8.2 Plasticity model
An isothermal von Mises plasticity model with isotropic

hardening is given as an example of the use of the UCMAT1,
UCMAT2 and UCMAT3 subroutines.
The implementation provided includes the MNO, TL and ULH

formulations for 2-D and 3-D cases, and MNO and TL for the 1-D
case.
The UCMAT1, UCMAT2 and UCMAT3 subroutine coding can

be found in files ovl20u_pl1.f, ovl30u_pl1.f and ovl40u_pl1.f
respectively.
The implementation is based on the equations given in Section

6.6.3 of Ref. KJB. The effective-stress-function algorithm is used
for stress integration.

3.18.8.3 Thermo-elasticity and creep material model
A model for thermoelasticity and creep is given as an example

of the use of the UCMAT1, UCMAT2 and UCMAT3 subroutines.
The model includes coding for the MNO, TL and ULH
formulations for 2-D and 3-D cases, and MNO and TL for the 1-D
case.

be found in files ovl20u_cn1.f, ovl30u_cn1.f and ovl40u_cn1.f
respectively.
The implementation is based on the equations given in Section

6.6.3 of ref. KJB.
The creep strain increments are defined using the Prandtl-Reuss

equations:
d ijC = sij
where ijC are the creep strain components, sij are the deviatoric
stress components and is the creep multiplier. Note that the
effect of the hydrostatic pressure on the creep deformations is
considered to be negligible.
The creep law is the eight-parameter creep law, i.e.:
C = f1 ( ) f 2 ( ) f3 ( t )
where C is the effective creep strain, is the effective stress,
is the temperature and t is the time, and
f1 ( ) = a0 a1 ; f 2 ( ) = exp ( a7 / ( + 273.16 ) )
f3 ( t ) = t a2 + a3t a4 + a5t a6
The same creep law also is used in the standard ADINA creep
material model (see Section 3.6).
For multiaxial creep deformation, the calculation of the stresses
is based on the use of an effective stress versus effective (creep)

strain curve. The curve is assumed to be the same as the uniaxial

stress-strain curve of the material.
In case of cyclic loading, the origin of the creep strain tensor
depends on the stress tensor reversal and is determined using the
ORNL rule (subroutines XCRCY1, XCRCY2 and XCRCY3).
The effective-stress-function algorithm is used for stress
integration (subroutines XEFSF1, XEFSF2 and XEFSF3). Either
the time hardening procedure or the strain hardening procedure can
be used in the calculation of the creep multiplier (subroutines
XGAMA1, XGAMA2 and XGAMA3). A root-finding algorithm
without acceleration (subroutine UBSECT) is used for computation
of both the pseudo time (strain hardening) and the effective stress.
t +t t +t i
The tangent constitutive matrix C EC
= t +t , where
j
ij
t +t
i are the stress components and t + dt
j are the total strain
components, is calculated using an approximation based on the
effective-stress-function algorithm (subroutines XELMA2 and
XELMA3). The corresponding 1-D tangent modulus
t +t t +t
C EC = is calculated in subroutine XELMA1.
t +t
3.18.8.4 Concrete material model, including creep effects
A model for concrete is given as an example of the use of the

UCMAT2 and UCMAT3 subroutines. The model includes coding
for the MNO, TL and ULH formulations for 2-D and 3-D cases,
and MNO and TL for 1-D cases.

be found in files ovl20u_cn2.f, ovl30u_cn2.f and ovl40u_cn2.f
respectively.
This user-coded material model has been developed for the

analysis of concrete structures, in which creep deformations depend
on the loading, on material ageing, on the temperature, and
possibly on other material or load related parameters.
The creep strain increments are defined by the Prandtl-Reuss

equations

d ijC = sij
where ijC are the creep strain components, sij are the deviatoric
stress components and is the creep multiplier. Consequently,
note that the effect of the hydrostatic pressure on the creep
deformations is considered to be negligible.
A generalized creep law of the following form is assumed:
= f1 ( ) f 2 ( ) f3 ( t ) f 4 ( x ) + f5 ( ) f 6 ( ) f 7 ( t ) f8 ( x )
where is the effective creep strain, is the effective stress,

is the temperature, t is the time, and x is a user-defined variable.
Appropriate coding for the creep member functions f1 to f8 is
provided (subroutines XFNC1, XFNC2 and XFNC3). In addition,
coding for the time derivatives of f3 and f7 is also provided
(subroutines XFNC1, XDFNC2 and XDFNC3).
Two different time variables are used (see Figure 3.18-5): t =

time since concrete was cast; t0 = time since ADINA analysis (and
presumably loading) started. The quantity (t - t0) represents the age
of the concrete at the start of the finite element analysis. The
quantity (t - t0) is stored as the first entry in the SCP array.
Therefore, t = t0 + SCP(1).
t0
t - t0
Time
Time when Time when Current time

concrete was finite element
cast analysis started
Figure 3.18-5: Time variables used for creep in concrete

The second entry in the SCP array (SCP(2)) is stores the flag for
choosing between the strain hardening (value of 1) or the time
hardening (value of 0) procedure in the effective-stress-function
algorithm.
The elastic modulus is assumed to be time dependent and given

by (see Figure 3.18-6):
(
E ( t ) = a1 exp a2 1 ( a3 / t )

a4
)
Elastic
modulus Asymptotic limit
Time
Figure 3.18-6: Time dependent elastic modulus for creep in

concrete
The creep law member functions are:
f1 ( ) =
f 2 ( ) = a5 + a7 a8
a6
with =exp ( a9 + a10 )
a25
1 a11 + t a12
a23
a15 + t0a16
a24
a19 + ( t t0 )a20
f 3 ( t , t0 ) = a18

E ( t ) a13 + t a14 a17 + t0 a21 + ( t t0 ) 22
a
f4 ( x ) = 1
and

f 5 ( ) =
f 6 ( ) = a26 ( + a27 )
a28
1
f7 ( t ) =
E (t )
f8 ( x ) = 1
Note that the term of f3 which depends only on (t - t0) could also be
programmed as the member function f4 (t - t0).
The creep constants are taken from array CTI as follows: a1 =

CTI(1), a2 = CTI(2), ... . One possible set of values is a1 = 2.0 H
1010, a2 = 0.5, a3 = 28, a4 = 0.5, a5 = 1.0, a12 = 0.6, a13 = 10, a14 =
0.6, a23 = 1.0.
In the case of multiaxial creep deformation, the calculation of

the stresses is based on the use of an effective stress versus
effective (creep) strain curve. The curve is assumed to be the same
as the uniaxial stress-strain curve of the material.
In case of cyclic loading, the origin of the creep strain tensor

depends on the stress tensor reversal and is determined using the
ORNL rule (subroutines XCRCY1, XCRCY2 and XCRCY3).
The effective-stress-function algorithm is used for stress
integration (subroutines XEFSF1, XEFSF2 and XEFSF3). The
time hardening procedure or the strain hardening procedure can be
used in the calculation of the creep multiplier, in case of variable
stress and/or variable temperature conditions (subroutines
XGAMA2 and XGAMA3). A root-finding algorithm without
acceleration (subroutine UBSECT) is used for computation of both
the pseudo time (strain hardening) and the effective stress.
t +t t +t i
The tangent constitutive matrix CEC
= t +t , where
j
ij
t +t
i are the stress components and t +t
j are the total strain
components, is calculated using an approximation based on the
effective-stress-function algorithm (subroutines XELMA1,
XELMA2 and XELMA3). The tangent modulus in the 1-D element

t +t t +t
case is C EC =
t +t
3.18.8.5 Viscoelastic material model
A model for linear viscoelasticity is given as an example of the

UCMAT2 and UCMAT3 subroutines. The model includes coding
for the MNO, TL and ULH formulations for 2-D and 3-D cases,
and for the MNO and TL formulations for the 1-D case.

be found in files ovl20u_vel.f ovl30u_vel.f and ovl40u_vel.f
respectively.
The implementation is fully described in Section 3.11.
Subincrementation in the viscoelastic material model is only

possible with forward time incrementation.
The material constants must be specified in the CT1 array of

floating point material constants.
Constant, 1: G
Constant, 2: G
Constant, 3: G1
Constant, 4: 1
...
Constant, 2 (G + 1) + 1 : K
Constant, 2 (G + 1) + 2 : K
Constant, 2 (G + 1) + 3 : K1
Constant, 2 (G + 1) + 4 : 1
...
Constant, 2 (G + K + 2 ) + 1 : T0
Constant, 2 (G + K + 2 ) + 2 : 0

Constant, 2 (G + K + 2 ) + 3 : C1
Constant, 2 (G + K + 2 ) + 4 : C2
and the total number of constants is 2 (G + K + 2 ) + 4 . In

addition, the user must set the length of the floating point working
array to 70 or larger in 2-D solid elements and 90 or larger for 3-D
solid elements.
In the 1-D element case, only one series of modulus and decay
constants is needed for the property description, i.e.,
Constant: E
Constant: E
Constant: E1
Constant: 1
...
Constant, 2 ( E + 1) + 3 : T0
Constant, 2 ( E + 1) + 4 : 0
Constant, 2 ( E + 1) + 5 : C1
Constant, 2 ( E + 1) + 6 : C2
The size of the work array is 10 + 4 E for the 1-D elements.
3.18.8.6 Ramberg-Osgood material model with mixed hardening
The Ramberg-Osgood material model with mixed hardening is

provided as an example of the UCMAT2 and UCMAT3
subroutines. The model includes coding for the axisymmetric,
plane strain, plane stress and 3-D elements and for the MNO, TL
and ULH formulations. For 1-D elements, only the MNO and TL
formulations are applicable.
The UCMAT2 and UCMAT3 subroutine coding can be found

in files ovl30u_pl2.f and ovl40u_pl2.f respectively.

Additional files ovl20u_pl3.f, ovl30u_pl3.f and ovl40u_pl3.f

are shown to use a more effective bisection algorithm, calling the
ADINA subroutine BISECT.
The Ramberg-Osgood stress-strain description of the yield

curve is in the form
= 0 y + C y ( e p )
t n
where Cy and n are material constants and 0 y is the initial yield

stress.
The mixed hardening behavior is described by
de p = de pi + de pk = Mde p + (1 M ) de p
p p
where de i and de k are, respectively, the isotropic and kinematic
contributions, and the yield function of the mixed hardening model
is
1 t 1
t
fy =
2
( s t ) ( t s t ) t y2 = 0
3
where the back stress t is evolved by
d = C p (1 M ) de p
Cp is Pragers hardening parameter and is calculated as
2 E p ME p
M
Cp =
3 1 M
where E pM is the derivative of the yield curve at the plastic strain

Me p , E p is the plastic modulus corresponding to e p , and M is the

factor used in general mixed hardening (0 < M < 1) which can be a
variable, expressed as
M = M + ( M 0 M ) exp ( e p )
The effective-stress-function algorithm is used to calculate the

plastic multiplier, see subroutines XEFSA and XEFSB in 2-D
(subroutine UCMAT2) and XEFSF3 in 3-D (subroutine
UCMAT3). The corresponding subroutine is XEFS1 for the 1-D
case.
3.18.8.7 Nonlinear elastic material model
A model for nonlinear elasticity is given as an example of the

UCMAT1, UCMAT2 and UCMAT3 subroutines which can be
found in files ovl20u_ne.f ovl30u_ne.f and ovl40u_ne.f
respectively. For the theory and algorithm of this material model,
please refer to Section 3.3.
The material constants must be specified in the Constant

Properties table of the User-Coded Material dialog box of the AUI
as follows:
Constant 1:
Constant 2: 1
Constant 3: 1
Constant 4: 2
Constant 5: 2
Constant (2*n): n
Constant (2*n+1): n

( i, i ) ( n , n )

( 1, 1 )
( 2 , 2 )
Figure 3.18-7: Input order of data point on stress-strain curve
Note that discontinuities are not allowed in user-supplied stress-

strain curve. The table look-up is performed using linear
interpolation within the table and linear extrapolation outside the
table using the two starting or ending points.
In Solution Control table, the first row is a flag for using tangent
or secant stress-strain matrix when the stress-strain curve enters
into a softening region. The secant stress-strain matrix is employed
for softening when this flag set to 1. Otherwise, the tangent stress-
strain matrix will always be used.
The backward stress integration scheme should be selected with

just one strain subdivision.
The length of the real working array must be set to a minimum

of 3 if Set in user coding is not selected. The length of the integer
working array may be set to zero, since no integer history variables
are required.
The nonsymmetric sparse solver will be used when the non-

symmetric stiffness matrix flag and the nonsymmetric sparse solver
are selected.


4. Contact conditions
4. Contact conditions
Contact conditions can be specified in ADINA to model contact
involving solid elements (2-D and 3-D solids) and/or structural
elements (truss, beam, iso-beam or axisymmetric shell, plate, shell
and pipe elements).
Very general contact conditions are assumed:
The points of contact are assumed not known a priori.

Friction can be modeled according to various friction laws
(only standard Coulomb friction for explicit dynamic
analysis).
Both sticking and sliding can be modeled.
Repeated contact and separation between multiple bodies is
permitted in any sequence.
Self-contact and double-sided contact are permitted.
Tied contact can be modeled.
Node-to-segment or node-to-node contact can be assumed.
A small displacement contact feature is available.
Some of the contact algorithms used in ADINA are described in the

ref. KJB ref. Bathe, K.J. and Chaudhary, A., "A Solution Method for
Section 6.7 Planar and Axisymmetric Contact Problems," Int. J.
Num. Meth. in Eng., Vol. 21, pp. 65-88, 1985.
ref. Eterovic, A. and Bathe, K.J., "On the Treatment of

Inequality Constraints Arising From Contact Conditions
in Finite Element Analysis," J. Computers & Structures,
Vol. 40, No. 2, pp. 203-209, July 1991.
ref. Pantuso, D., Bathe, K.J. and Bouzinov, P.A."A Finite

Element Procedure for the Analysis of Thermo-
mechanical Solids in Contact," J. Computers &
Structures, Vol. 75, No. 6, pp. 551-573, May 2000.

Chapter 4: Contact conditions
Contact in ADINA is modeled using contact groups, contact

surfaces, contact segments and contact pairs, as explained in much
greater detail below.
Note that the presence of contact groups renders the analysis

nonlinear even when no nonlinear element groups are defined.
Most of the settings and tolerances needed for contact are

provided in the contact group definition command (CGROUP2 or
CGROUP3). Some contact settings however apply to all contact
groups (such as contact convergence tolerances, suppression of
contact oscillations). These settings are provided in the MASTER
or CONTACT-CONTROL commands.
4.1 Overview
Contact groups (and their contact surfaces) in ADINA can be
either 2-D or 3-D. The contact surfaces should be defined as
regions that are initially in contact or that are anticipated to come
into contact during the solution.
2-D contact surfaces are either axisymmetric or planar and

must lie in the global YZ plane, with all X coordinates equal to
zero. Typical two-dimensional contact surfaces are shown in
Fig. 4.1-1.
A 2-D contact surface is a closed surface if all the contact
segments on the surface taken together form a closed path (see
Fig. 4.1-1, contact surface 3). If the segments do not form a
closed path, then the contact surface is open (see Fig. 4.1-1,
contact surfaces 1 and 2).

4.2: Contact algorithms for implicit analysis
Contact surface 3
Body 2 l
l l
l
D l l
l Body 3 l
Contact surface 2 l
l l
l l l
l
l l l l l
l
l l l
A
l
l l
l l
l l l
l
C l
Bl Contact surface 1
Body 1
Figure 4.1-1: Typical 2-D contact surfaces
A 3-D contact surface is made up of a group of 3-D contact

segments (faces) either on solid elements, shell elements, plate
elements, or attached to rigid nodes. See Fig. 4.1-2 for an
illustration.
Contact surface 1
(surface of cylinder)
Body 1
Contact surface 2
(top surface of
body 2)
Body 2
Figure 4.1-2: Typical contact surfaces and contact pair
A contact pair consists of the two contact surfaces that may

come into contact during the solution. One of the contact surfaces
in the pair is selected to be the contactor surface and the other
contact surface to be the target surface. In the case of self-contact,
the same surface is selected to be both contactor and target in the

contact pair. (Self-contact is when a contact surface is expected to

come into contact with itself during the solution.)
Within a contact pair, the nodes of the contactor surface are

prevented from penetrating the segments of the target surface, and
not vice versa.
Fig. 4.1-3 shows the effect of contactor and target selection on

the different contact configurations.
No penetration Target
surface
Contactor
surface
Contactor
No penetration surface
Target
surface
Figure 4.1-3: Contactor and target selection
At least one of the two contact surfaces in a contact pair must

not be rigid. If one surface is rigid, this surface should, in most
cases, be the target surface.
In explicit dynamic analysis, both contactor and target surfaces

can be rigid if the penalty algorithm is used. Otherwise, the same
restriction mentioned above applies.
Rigid surfaces have no underlying elements and therefore no

flexibility apart from rigid body motions. All their nodal degrees
of freedom must be either fixed, have enforced displacement, or be
rigidly linked to a master node.
Symmetric contact pairs can be defined, where in one contact

pair surface A can be the contactor and surface B the target, and in
another contact pair surface B is the contactor and surface A is the
target.
A non-zero contact surface compliance should always be used
with symmetric contact pairs. Without compliance, the system of
contact equations can become overconstrained, that is, the same
contact equation can be present more than once in the system of
contact equations. When this happens, the global system of
equations becomes singular and the solution may diverge.
Basic concepts
ref. KJB The normal contact conditions can ideally be expressed as

Section 6.7.2
g 0; 0; g = 0 (4.1-1)
where g is a gap, and is the normal contact force. Different

algorithms may vary in the way they impose this condition.
For friction, a nondimensional friction variable can be defined

as
FT
= (4.1-2)

where FT is the tangential force and is the normal contact force.
The standard Coulomb friction condition can be expressed as
1
and < 1 implies u = 0 (4.1-3)
while = 1 implies sign ( u ) = sign ( )
where u is the sliding velocity.
In static analysis, the sliding velocity is calculated by dividing

the incremental sliding displacement by the time increment. Hence,
time is not a dummy variable in static frictional contact problems.
When (Coulomb) friction is used, the friction coefficient can be

constant or calculated from one of several predefined friction laws.
The possible states of the contactor nodes and/or segments are
No contact: the gap between the contactor node and target

segment is open.
Sliding: the gap between the contactor node and the target
segment is closed; a compression force is acting onto the
contactor node and the node kinematically slides along the
target segments (either due to frictionless contact or due to a
frictional restrictive force less than the limit Coulomb force.
Sticking: as long as the tangential force on the contactor node

that initiates sliding is less than the frictional capacity (equal to
the normal force times the Coulomb friction coefficient), the
contactor node sticks to the target segment.
Old and new contact surface representations
Two types of contact surface representation are supported in

ADINA, an old and a new contact surface representation (set via
the CSTYPE parameter in the CONTACT-CONTROL command).
The new contact surface representation is the default. The main
differences between the two representations are:
- In the old representation, contact segments are linear (2 nodes

for 2D contact; 3- or 4 nodes for 3D contact). In the new
representation, contact segments can be linear or quadratic (up to 3
nodes for 2D contact, up to 9 nodes for 3D contact).
- In the new representation, contact is based on the actual faces of

the contact segments, which results in more accurate contact
constraints.
- The new representation uses a more accurate contact traction

calculation algorithm.
- The new representation generates more accurate contact

constraints for 3-D contact segments resulting from 10, 11 node tet
elements and 20, 21 node brick elements. These elements generate

zero (10, 11 node tets) or negative (20, 21 node bricks) contact

forces at their corner nodes when subjected to a uniform contact
pressure.
- More detailed output is provided in the new representation, such

as gap, contact status, slip velocity. Tractions are also reported as
nodal quantities in the new surface representation.
The new contact surfaces cannot be used with the following

features:
- Segment method algorithm

- Rigid target algorithms
- 16-node shell elements
- 4-node beam elements
- Cyclic symmetry
The direction of a contact surface is marked by a normal vector

n pointing towards the interior of the contact surface as shown in
Fig. 4.1-4 and Fig. 4.1-5.
Continuum
n
l
l k+2
k-1 nr
l l
Contactor
k k+1 surface
Z
Contactor segment
n =normal vector, acting into the continuum
Y nr=tangent vector
Figure 4.1-4: Normal vector of a contact segment in 2-D analysis

Figure 4.1-5: Normal vector of a contact segment in 3-D analysis
Single-sided contact
For single-sided contact (see Fig. 4.1-6), one side of the contact
surface is assumed to be internal and the other side to be external.
Any contactor node within the internal side of a target surface is
assumed to be penetrating and will be moved back to the surface.
This single-sided option is ideal for contact surfaces on the faces of
solid elements since in that case it is clear that one side is internal
to the solid while the other is external. In this case, the external
side can usually be predicted from the geometry. This option is
also useful for shells when it is known that contact will definitely
occur from one direction. In this case, however, the program
cannot intuitively predict the internal side of the contact surface.
Single-sided contact is selected using CGROUP ...
PENETRATION-ALGORITHM=ONE (the default).

Internal side
(no contactor nodes allowed)
Target
surface
External side
Figure 4.1-6: Single-sided contact surface
Double-sided contact
In double-sided contact (see Fig. 4.1-7), there are no internal or
external sides. The contactor surface nodes in this case are
prevented from crossing from one side of the target contact surface
to the other during solution. This option is more common for shell-
based contact surfaces. If a contactor node is on one side of the
target surface at time t, it will remain on the same side at time t +
t.
Double-sided contact is selected using CGROUP ...
PENETRATION-ALGORITHM=TWO. Double-sided contact is
only available for 3D contact.
Contactor node cannot

penetrate lower side
Target
surface
Contactor node cannot

penetrate upper side
Figure 4.1-7: Double-sided contact
Tied contact
When the tied contact feature is selected for a contact group,
ADINA performs an initial contact check at the start of the
analysis. All contactor nodes that are found to be in contact or

overlapping are permanently attached to their respective target

segments. Contactor nodes that are not in contact are also set to be
tied if the contact gap is less than a user-specified tied contact
tolerance. This tolerance is useful when the contact gap is due to
non-matching finite element discretizations of the contact surfaces.
The tied contact feature is conceptually similar to using rigid
links or constraints to attach the node to the target surface. The
main difference is that the coefficients for the rigid links are
automatically determined by the program and they are only applied
for the nodes that are initially in contact. The basic idea is
illustrated in Fig. 4.1-8.
Permanent rigid Contactor surface

connections
Gap between
contact surfaces
(exaggerated)
Target surface
Figure 4.1-8: Tied contact option
Tied contact is not "real" contact because there can be tension

between tied contact surfaces. Also no sliding can occur between
tied contact surfaces.
The tied contact option can be used to glue two surfaces
together if they have incompatible meshes. However, the surface
glueing feature described in Section 5.16 produces more accurate
glue equations.
If the contact surfaces initially overlap, they are not pushed
back to eliminate the overlap. Similarly, if there is an initial gap, it
is not eliminated.
The tied contact constraint equations are computed based on the
initial nodal positions only. The constraints generated in tied
contact are not updated during the analysis. Hence, the constraints
will be inaccurate if the bodies experience large rotations.
Tied contact is selected using CGROUP ... TIED=SMALL

Small displacement contact

If the small displacement contact feature is used, the contact
constraints are generated once in the beginning of the analysis and
are kept constant, as shown in Figure 4.1-9. A target location is
identified for each contactor node if possible, and its gap and
normal direction are determined. The local coordinates of the target
point and the normal direction are then kept constant for the
remainder of the analysis. This is in contrast to the standard large
displacement contact, where the contact constraints are updated
every iteration, and the contactor nodes can undergo any amount of
sliding.
This feature is useful when there is very little relative
deformation around the contact region. For such problems, it is
much more computationally efficient to perform only one detailed
contact search at the beginning of the analysis, rather than
repeating the search every iteration. Also, in some cases,
convergence can also be slow or unachievable with the general
algorithm, for example as nodes oscillate between one target
segment and another equally valid neighboring target segment.
Small displacement contact is selected using CGROUP ...
DISPLACEMENT=SMALL.
Target surface
x*1
N x*1
x1 x1
Contactor surface
Original geometry: Determine target Small displacement contact acceptable

point x*1 and normal vector N
x*1
x1 x*1
x1
Small displacement contact not Small displacement contact not

acceptable due to excessive displacement acceptable due to excessive rotation
Figure 4.1-9. Small displacement contact feature

Node-node contact
When node-node contact is used, each node of the contactor
surface is paired with a node on the target contact surface (a node-
node contact pair). A node on the contactor surface can come into
contact with the infinite plane passing through the paired node on
the target surface; the infinite plane is perpendicular to the normal
direction indicating the direction within the target body (Fig. 4.1-
10). This normal direction, and hence the infinite plane, is constant
during the response. The normal directions can be based on the
geometry of the target segments or on the vector connecting each
contactor and target nodes.
Node-node contact is selected using CGROUP ...
NODETONODE=YES.
Contactor node
in node-node
contact pair
Node-node contact pair
Infinite plane passing

Target node through target node
in node-node
contact pair Continuum Target node normal
Figure 4.1-10: Node-to-node contact
4.2 Contact algorithms for implicit analysis

ADINA offers three contact solution algorithms (set via the
CONTACT-CONTROL CONTACT-ALGORITHM flag):
Constraint-function method,
Segment (Lagrange multiplier) method, or

Rigid target method
Each contact group must belong to one of these three contact

algorithms. However, different contact groups can use different

algorithms.
All 3 contact algorithms can be used with or without friction.
4.2.1 Constraint-function method
In this algorithm, constraint functions are used to enforce the

no-penetration and the frictional contact conditions.
The inequality constraints are replaced by the following normal
constraint function:
2
g + g
w ( g, ) = + N
2 2
where N is a small user-defined parameter. The function is shown

in Fig. 4.2-1. It involves no inequalities, and is smooth and
differentiable. The parameter N is set via the EPSN variable in the
CGROUP commands. The default value of 1.0 x 10-12 is suitable
for most applications and should rarely be modified.
It is possible to set EPSN=0.0. In this case ADINA
automatically determines EPSN. However, this determination may
not result in correct results for some problems. Hence EPSN=0.0
should not be used in general.
g
w(g,%)
Figure 4.2-1 Constraint function for normal contact
The constraint function method is also used to approximate the

rigid non-differentiable stick-slip transition of Eq. (4.1-3). This

results in a smooth transition from stick to slip and vice versa, and
it also results in a differentiable friction law that is less likely to
cause convergence difficulties.
Two friction regularization algorithms are available in ADINA.
Both constraint functions take the form v ( u , ) = 0 .
The newer default algorithm involves a more accurate
linearization of the frictional constraints and, in general, converges
much faster than its predecessor. The v function is defined
implicitly as a multilinear function as shown in Fig. 4.2-2. Here T
is a small parameter (EPST parameter in the CGROUP commands)
which has the physical meaning of the "sticking velocity", that is,
the maximum velocity corresponding to sticking conditions.
"
.
-10 -2 -1 1 2 _
10 u
T
-1
Figure 4.2-2: Frictional contact constraint function for new friction

algorithm
In the old version 8.3 friction algorithm, the v function is

defined implicitly via
2 uv
+ v arctan =0
T
Here T is a small parameter (EPST parameter in the CGROUP

commands) which provides some elastic slip to the Coulomb
friction law as shown in Fig. 4.2-3. Guidelines for selecting T are
provided in Section 4.8.3.

"
_
2
4
1
.
-10 -2 -1 1 2 _
10 u
T
-1
Figure 4.2-3: Frictional contact constraint function for old friction

algorithm
The friction algorithm can be selected using the CONTACT-

CONTROL FRICTION-ALGORITHM flag.
4.2.2 Segment (Lagrange multiplier) method
In this method, Lagrange multipliers are used to enforce the

contact conditions of Eq. (4.1-1). The kinematic conditions are
enforced at the contactor nodes, and the frictional conditions are
enforced over the contact segments.
This method involves distinct sticking and sliding states. It also

calculates this state for each contactor node based on the contact
forces on the target segment.
4.2.3 Rigid target method
This is a simplified contact algorithm intended primarily for

metal forming applications. The algorithm is fully described in
Section 4.9.
4.2.4 Selection of contact algorithm
Our experience is that in most contact problems the constraint

function method is more effective than the segment method. The
constraint function method is the default.

For problems involving rigid targets, either the constraint

function or the rigid target algorithm can be employed.
Note that the target surface can be rigid in all three contact
algorithms. The presence of a rigid target does not mean that the
rigid target algorithm must be used.
4.3 Contact algorithms for explicit analysis

ADINA offers three contact solution algorithms for explicit
analysis:
Kinematic constraint method,

Penalty method, or
Rigid target method
Each contact group must use one of these three contact

algorithms. However, different contact groups can use different
algorithms.
All of these contact algorithms can be used with or without

friction.
4.3.1 Kinematic constraint method
This is the default explicit contact algorithm.
A predictor step is first done without applying contact

constraints or forces. Then displacements are evaluated and
penetration is detected and corrected. The exact correction of
displacements requires the solution of a non-diagonal system of
equations. Instead, a good approximation is done. In this case, for
each penetrating contactor node, a penetration force
N
FCN = M C aCN * = M C N
t 2
is calculated. This is the force required to remove the penetration at

the contactor node. However, not all the penetration will be
removed by moving the contactor. The target will get some motion
depending on its mass relative to the contactor and how many

4.4: Contact group properties
contactor nodes are touching it. So, the FCN force above is
projected to the target segment nodes:
FTNi = N i FCN
where Ni is the shape function relating the contactor displacement

to that of each target node. Similarly, the mass of the contactor
node is projected to the target in the same way:
M Ti = N i M C
and this mass is added to that of the target node itself. Then the
acceleration of the target node is determined as
aTN ( M T + M Ti ) = FTNi
This correction is then used to update the target displacements. The

contactor acceleration is
aCN = aCN * aTN N i
For friction, a similar approach is used. A correction force is

calculated
vT
FCT * = M C
t
where vT is the tangential sliding velocity. However, this force

cannot exceed the limit force based on the normal force and the
coefficient of friction
FCT = min ( FCN , FCT * )
The rest of the procedure is very similar to the case of normal

contact. The form of the equations is different if there is damping,
and is also different if the previous and current time steps are not
the same.
A modification is also required for rigid targets, which are

common in contact. The form of the equations in this case depends

on whether the rigid target has natural or essential boundary
conditions.
The kinematic constraint algorithm should not be used when the

target surface dofs are fixed.
4.3.2 Penalty method
In this algorithm, contact conditions are imposed by penalizing

the inter-penetration between contacting surfaces. When a
penetration is detected, a normal force of
F = A P = A N ( K N N + K D N )
is applied to the contactor node, where KN is the normal stiffness,

KD is a normal rate stiffness, N is the penetration, N is the
penetration rate, N is the normal vector pointing towards the
contactor, A is the contact area and P is the normal contact traction.
An opposing force is distributed to the target nodes.
Similarly, in the presence of friction, the relative sliding

velocity between the two bodies is penalized as follows:
x
FT = min A KT xT , FN T
xT
where xT is the relative tangential incremental displacement.
The normal and tangential penalty stiffnesses KN and KT can be

selected by the user, or determined automatically by the program.
The penalty rate stiffness KD can be explicitly selected by the user,
or determined by the program as a ratio of critical damping for the
contact node.
When penalty stiffnesses are automatically determined they are

chosen based on the masses of the contactor nodes and the time
step. They are selected such that they have a minimal effect on the
existing time step.

Note that unduly small penalty stiffnesses will lead to excessive

penetrations, and unduly large penalty stiffnesses will lead to
excessive oscillations or unstable explicit time integration.
In analysis with friction, it is recommended that the slip velocity

and the contact status be compared. If the slip velocity is large and
the contact status is 'sticking', then the tangential penalty stiffness
should be increased.
4.3.3 Rigid target method
This algorithm is fully described in Section 4.9.
4.3.4 Selection of contact algorithm for explicit analysis
The kinematic constraint method is the default.
The penalty method is the simplest and fastest of the explicit

contact algorithms. It can also handle rigid contactor and target
surfaces. It also allows a contactor node to be in contact with
multiple targets simultaneously.
The main disadvantage of the penalty method is that contact

conditions are not exactly satisfied and it usually shows oscillations
in contact forces. These oscillations can sometimes be removed by
using the penalty rate stiffness factor K D . It is also sensitive to the
choice of the penalty stiffness. If that stiffness is too large it leads
to instability and oscillations, and if it is too small it leads to
excessive penetrations.
The default penalty stiffness selected by the program is, in most

cases, a suitable compromise.
4.4 Contact group properties

This section describes the main options available to contact groups.

Contact surface offsets

Penetration of a contact surface occurs when the plane or line
defined by the contact segment nodes is penetrated. However, an
offset distance can be specified which causes the actual contact
surface to be offset from the plane defined by the contact surface
nodes. In the case of double-sided contact, the offset creates two
separate surfaces above and below the reference surface. Fig. 4.4-1
shows the possibilities for single and double-sided contact. Note
that the offset distance should be small compared to the contact
surface length.
Offsets for a whole contact group are specified in the CGROUP

commands, using parameters OFFSET and OFFSET-TYPE.
Offsets for a specific contact surface can also be defined using the
CS-OFFSET command. If one of the contact surfaces has a
defined offset, it will overwrite the contact group offset.
Note that if the contact surface is on a shell, then half the true shell
thickness can automatically be used as the offset. When this option
is used, the shell thickness at a contact node is computed as the
average of all of the shell thicknesses at that node.
The use of contact surface offsets in double-sided contact is not

recommended.
Offset
t Actual contact
surface
(a)
Internal side Defined contact
surface
Offset
Actual contact
t surfaces
t
(b) Offset
Defined contact surface
Figure 4.4-1: Contact surface offsets for: (a) single-sided contact, and
(b) double-sided contact

Continuous normals (not available in explicit analysis)

The normal direction to a contact segment will in general not be
continuous between segments as illustrated in Fig. 4.4-2. This
sometimes causes convergence difficulties due to the non-unique
normals at nodes and segment edges. The continuous normals
feature first calculates nodal normals as averages of all the normals
from the attached segments, and then interpolates these nodal
normals across the segment. This leads to a uniformly varying
normal direction.
The continuous normals feature is specified in the CGROUP
commands, using parameter CONTINUOUS-NORMAL.
Continuous normals is the default setting.
In modeling target surfaces with sharp corners, either use
discontinuous normal vectors, or use small segments near the
corners, to ensure that the normal vectors for segments near the
corners are computed correctly. See Section 4.8.2 for modeling
tips related to this feature.
Nodal normals
(a) Discontinuous normals (b) Continuous normals
Figure 4.4-2: Contact surface normals
Contact surface depth

By default, the contact region extends for an infinite distance below
the contact surface (for single-sided contact). However, a contact
surface depth can be defined (by setting the DEPTH parameter in
the CGROUP commands), below which the contact surface is no
longer active. The default DEPTH=0.0 results in an infinite
contact depth extension. Fig. 4.4-3 shows some of the possibilities.

Contact surface is used in self-contact.

A B A B
C t C
G
Target depth G
D D
F F
E E
Continuum created
by segment A-B Nodes F and G have not
penetrated segment A-B.
Nodes F and G have
penetrated segment A-B.
a) Target depth option not used b) Target depth option used
Figure 4.4-3: Contact surface depth
Initial penetration
The treatment of initial penetrations in ADINA is governed by the
INITIAL-PENETRATION parameter of the CGROUP commands.
By default, if there is initial overlap (penetration) between a
contact node and a target segment in the first solution step, ADINA
attempts to eliminate the overlap. ADINA can eliminate the overlap
at the first step or over a user-specified time using the TIME-
PENETRATION parameter of the CGROUP commands. This
feature is useful if the initial penetrations are too large to be
eliminated in a single step.
The program can also calculate initial penetrations at the start of
solution and ignore them in future steps. In this case the program
does not detect penetration for a contactor node if the amount of
penetration is less than or equal to the recorded amount. Fig. 4.4-4
shows some of the possibilities. See Section 4.8.2 for modeling tips
related to this feature.
This feature can be interpreted as a redefinition of the contact
gap, as follows:
Eliminate penetration option (INITIAL-PENETRATION

=ALLOWED): The contact gap is equal to the geometric
gap. This is the default.

Ignore penetration option (INITIAL-PENETRATION=

DISCARDED):
Initial geometric gap is positive or zero: The contact

gap is equal to the geometric gap.
Initial geometric gap is negative: The contact gap is

equal to the geometric gap minus the initial geometric
gap.
Initial penetrations can also be set to gap override (see below).
Contactor node Eliminate penetration

(INITIAL-PENETRATION=ALLOWED)
Target surface
Continuum
Solution start, contactor node First solution time,

initially penetrates target surface overlap is eliminated
Eliminate penetration
over time period Ignore penetration
(INITIAL-PENETRATION=DISCARDED)
(INITIAL-PENETRATION=ALLOWED,
TIME-PENETRATION>0)
Overlap
distance
First solution time,

Overlap is gradually eliminated overlap is recorded
Figure 4.4-4: Initial penetration options

Gap override
In the gap override feature, the gaps and penetrations calculated
from the finite element mesh are replaced by a fixed user-specified
value. A positive value represents an initial gap, zero means that
the contact is touching the target, and a negative value represents
an initial penetration (which can be removed either immediately or
over a user-specified time as explained in the above Section on
initial penetration).
To activate this feature, set INITIAL-PENETRATION
=GAP-OVERRIDE in the CGROUP commands, and also set GAP-
VALUE.
This feature can also be interpreted as a redefinition of the

contact gap, as follows:
Gap-override option: The contact gap is equal to the

geometric gap minus the initial geometric gap plus the
gap override value. This formula is applied regardless of
whether the initial geometric gap is positive, zero or
negative.
This feature is useful for problems involving curved meshes in

close proximity, such as the shrink fit example shown in Fig. 4.4-5.
The gaps and penetrations measured from the discretized finite
element mesh are sometimes inaccurate for such problems (unless
matching meshes are used). In some problems, such as that shown
in the figure, a constant geometry based overlap should be applied
to all nodes, which corresponds to a gap override value of .
Note that mesh refinement and quadratic elements reduce the
error in the measured overlaps but frequently a very high mesh
density would have to be used if gap override is not used.
Note also that the error in mesh based gaps and penetrations for
curved surfaces can be more significant when low precision
numbers are used for the node coordinates. Gap override is also
useful for such cases.

Two rings with a geometric overlap 6 (shrink fit)
Directly measured overlaps

are incorrect at most nodes
Proper initial overlap at
all nodes should be 6
Figure 4.4-5. Significance of gap override for curved non-matched

geometries
Contact surface extension

The target surface can be enlarged beyond its geometric bounds, so
that contactor nodes that slip outside the target can still be
considered in contact (via the CS-EXTENSION parameter of the
CGROUP commands). This feature is useful where the edge of the
contactor and target surfaces coincide, as shown in Fig. 4.4-6. Each
target segment is enlarged by an amount equal to the contact
surface extension factor multiplied by the length of the segment.

Target
surface e = L x contact surface extension
L
e
Extended target
surface
Corner contactor node may slip

outside bounds of target due to
numerical round-off, or lateral
displacements
Figure 4.4-6: Contact requiring contact surface extension
Contact surface compliance*

Contact surface compliance is set via the CFACTOR1
parameter of the CGROUP commands and is only available with
the constraint function algorithm. The contact surface compliance
feature can be used to simulate soft or compliant surfaces. When
the contact surface compliance is zero, the contact surface is
assumed to be rigid meaning that no interpenetration is allowed
between the contactor node and the target segment. However, when
the contact compliance is nonzero, the contact surface is no longer
rigid, and contactor nodes may penetrate (overlap) the target
segment. The amount of allowed interpenetration between the
contacting surfaces in this case is
penetration = p normal contact pressure (4.4-1)
where
normal contact pressure = normal contact force / contact area
The constraint function in the presence of a compliance factor is

modified as shown in Fig. 4.4-7. A is the contact area of the
contactor node. If the contactor contact surface is defined as a
* not available in explicit analysis

contact body, then the contact area A of each contactor node is set
to 1.
P /A
Figure 4.4-7: Constraint function for compliant contact
The benefits of contact compliance are it:
Improves the convergence rate. By adding compliance to the

contact surface, more contactor nodes come into contact with
the target surface. It is better for convergence to have many
nodes in contact with small forces at each contactor node, than
to have few nodes in contact with large forces at each contactor
node.
Improves the contact tractions. Without contact compliance, the

discretization error of the mesh often cause some contactor
nodes on the contact area to not be in contact which results in
spotty contact (or patchy) contact tractions. An example of
spotty contact is shown in Fig. 4.4-8.
Reduces numerical oscillations that sometimes result from

dynamic time integration. Numerical oscillations, such as
chatter or ringing, can be triggered by contact reversal. A node
that is out of contact at time t t , in contact at time t and out
of contact at time t + t is said to have had contact reversal.
Contact surface compliance can be used to prevent these
oscillations. For example, consider Fig. 4.4-9, where
verification problem B.120 is solved using the Bathe method
rather than the central difference method. The results show that

no numerical oscillations are triggered when contact compliance

is used.
Figure 4.4-8: Spotty contact on a plate
spurious oscillations
Figure 4.4-9: Reaction force from a thin rod impacting a rigid

surface solved using the Bathe method (verification problem
B.120). Contact reversal causes numerical oscillations unless
contact compliance is used.

However, the drawbacks of contact compliance are:
Nodal contact overlap. The contactor nodes penetrate the target

segment when contact compliance is used. The compliance
factor must be selected such that the contact pressures do not
cause excessive nodal overlap.
Additional (spurious) energy is stored in the contact pair.

Elastic energy is stored in the compliant contact surface. This
energy is released when the contact pair opens which can have
an effect in certain dynamic problems.
Can reduce the resultant contact force and stresses. If a

prescribed displacement is applied, the contact surface
compliance feature will reduce the stress in the body for a given
displacement. The stresses are not reduced if the body is
subjected to an applied load.
By default, the compliance factor is set to zero ( p = 0 ). However,

in general, some amount of contact compliance should always be
used. It is not default because the appropriate compliance factor is
model dependent. Since contact compliance is recommended, the
following AUI Warning is printed when it is not used:
*** Warning: Contact surface compliance not used for contact group ? .
Contact compliance can improve the solution accuracy and convergence
rate.
The value of contact compliance factor is usually set by trial and

error. A good starting value is 105 . The objective is to find the
smallest compliance factor that results in convergence with good
tractions (i.e., not spotty contact), but does not cause excessive
penetration (nodal contact overlap).
After contact compliance is used, the amount of penetration should

always be checked to make sure the maximum overlap is
acceptable. The post-processing NODAL_CONTACT_OVERLAP
variable can be used for this purpose.
When deciding on what penetration is acceptably small we must

consider the mesh used. For example, if the target surface is

curved, there will be a geometric error associated with using a
coarse contractor surface. There is no advantage if the maximum
penetration is less than the geometric error. So, if the target surface
is coarse, a large maximum penetration can be used.
Consistent contact stiffness*

The consistent contact stiffness feature is set via the
CONSISTENT-STIFF parameter of the CGROUP commands.
Changes in the direction of the contact normal provide an
additional contribution to the stiffness matrix that is proportional to
the value of the contact force and the change in the normal
direction. Therefore, higher convergence rates (closer to quadratic)
can sometimes be obtained by selecting the consistent contact
stiffness option, which accounts for these additional stiffness
contributions. This results, however, in an increase in the size of
the stiffness matrix, which is detrimental for large problems. This
option is more beneficial when continuous contact normals are not
selected, because the derivation assumes that the contact normals
are discontinuous.
The consistent contact stiffness feature is not used when the
target surface is rigid.
Consistent contact stiffness is not used in dynamic analysis.
The consistent contact stiffness feature is activated by default
when all of the following conditions are satisfied: 1) the skyline
equation solver is not used; 2) continuous normals are not used; 3)
old contact surfaces are used (in other words, new contact surfaces
are not used).
Contact birth/death
The contact birth feature activates a contact group at a specific
time, while the contact death feature disables a contact group at a
specific time. They are set via the TBIRTH and TDEATH
parameters of the CGROUP commands. A 0.0 birth time means
that the contact group starts active at the beginning of the analysis,
and a death time less than or equal to the birth time means that the
contact group does not die.
Friction delay*

When the friction delay feature is activated, frictional conditions

are applied to a contactor node one time step after contact is
established. This feature can be useful in many problems, since it
delays the non-linearity associated with friction until contact is
established.
Note that the relative sliding velocity cannot be uniquely
determined when a node was not in contact at time t, and is in
contact at time t+t. That velocity depends on the exact time at
which contact started, which is somewhere between times t and
t+t (see Figure 4.4-8). Delaying friction is equivalent to assuming
that contact was established close to time t+t, and hence the
sliding velocity is zero and so is the frictional force.
The friction delay feature is activated using the FRIC-DELAY
parameter of the CGROUP commands. Friction delay is off by
default.
Contact slip*
Contact slip allows a tangential or slip velocity to be added to the
geometric relative velocity between the contact surfaces. With this
feature, you can model out-of-plane slip in 2-D axisymmetric
analysis. This feature is considered to be a load added to the
model. See Section 5.11 for further details.
Time = t
t+2t t+t
x1t
t+2t t+t
FT FT .t
x1 R= Relative sliding velocity
t+2t t+t t+t
FN FN Friction delay FT = 0
. .
x1t+2Rt unique x1t+Rt non-unique
Figure 4.4-8. Friction delay feature

Contact surface output

You can request that ADINA save contact tractions and/or forces.
Different traction outputs are obtained for the old and new (default)
contact surface representations.
Tractions, old contact surface representation:
Traction output is available only on contactor surfaces, and only if

the surface is segment-based (not nodal based). Each contactor
contact segment outputs, at its center, the following output
information to the porthole file. This information is accessible in
the AUI using the given variable names.
2-D contact: SEGMENT_LENGTH, NORMAL-Y, NORMAL-Z,

NORMAL_TRACTION, TANGENTIAL_TRACTION
3-D contact: SEGMENT_AREA, NORMAL-X, NORMAL-Y,

NORMAL-Z, NORMAL_TRACTION-X,
NORMAL_TRACTION-Y, NORMAL_TRACTION-Z,
TANGENTIAL_TRACTION-X, TANGENTIAL_TRACTION-Y,
TANGENTIAL_TRACTION-Z
In addition, the following information is output for solitary nodes:

(solitary nodes are nodes having concentrated contact forces not
resulting from segment tractions, e.g., when an isolated node is in
contact):
SOLITARY_CONTACT_FORCE-X (3-D contact only),

SOLITARY_CONTACT_FORCE-Y,
SOLITARY_CONTACT_FORCE-Z,
SOLITARY_CONTACT_FORCE-A (3-D contact only),
SOLITARY_CONTACT_FORCE-B,
SOLITARY_CONTACT_FORCE-C.
It should be noted that if symmetric contact pairs are used, then the
tractions output are one-half of the expected values. This is
because the tractions are computed solely from the contactor
forces, and when symmetric contact pairs are used, the contactor
forces are one-half of the total forces.
Tractions, new contact surface representation:

Only nodal contact quantities are output for the new contact surface
representation. Output is restricted to the contactor nodes, and only
if the surface is segment-based (not nodal based). The output data
is available to the AUI using the variable names given below.
NODAL_NORMAL_TRACTION (magnitude of normal

traction)
NODAL_NORMAL_TRACTION-{XYZ}
NODAL_TANGENTIAL_TRACTION (magnitude of
tangential traction)
NODAL_TANGENTIAL_TRACTION-{XYZ}
The traction output also generates the following additional output:
NODAL_CONTACT_GAP
NODAL_CONTACT_STATUS
NODAL_SLIP_VELOCITY (magnitude of slip
velocity)
NODAL_SLIP_VELOCITY-{XYZ}
All of these variables are output at the nodes. As a convenience,

band plots still allow the variable names
NORMAL_TRACTION
NORMAL_TRACTION-{XYZ}
TANGENTIAL_TRACTION
TANGENTIAL_TRACTION-{XYZ}
Note: The DISTRIBUTED_CONTACT_TRACTION variable in the

REACTIONPLOT command can be used to plot tractions or nodal
tractions on contactor surfaces.
Comments regarding the traction calculation for the new contact

surfaces.
The traction calculation for the new contact surfaces is revised in

version 9.0. The version 9.0 traction calculation is selected using
CONTACT-CONTROL ... TRACT-CALC=CURRENT (the
default). It is also possible to use the traction calculation of
versions 8.9 and lower, and the 8.9 traction calculation is selected
using CONTACT-CONTROL ... TRACT-CALC=V89.

When the version 9.0 traction calculation is used, then
If symmetric contact pairs are used, then the tractions output

are the expected values. The program computes tractions from the
contactor forces, then the program doubles the calculated tractions
to account for symmetric contact.
The geometry of the contact affects the calculations as

shown in Fig. 4.4-9. Consider the calculation of the traction at node
B. The topology of Figs. 4.4-9(a) and 4.4-9(b) are identical, but in
4.4-9(a), segment B-C contributes to the traction calculation at
node B, and in 4.4-9(b), segment B-C does not contribute to the
traction calculation at node B. The criterion TRACT-ANGLE can
be set by CONTACT-CONTROL ... TRACT-ANGLE; the default
is 45 degrees.
Contacting parts separated for clarity.

C
C
q
q
A Contactor B A Contactor B
Target Target
a) q < TRACT-ANG, segment B-C b) q > TRACT-ANG, segment B-C does

contributes to the traction calculation not contribute to the traction calculation
at node B at node B
Figure 4.4-9: Explanation of TRACT-ANGLE parameter
The value of the tractions at the corner-nodes of higher-order

segments is computed by interpolating the tractions from the
adjacent mid-side and mid-face nodes. It is possible to change this
interpolation using CONTACT-CONTROL ... TRACT-
QUADRATIC. Allowed values of TRACT-QUADRATIC are:
EDGEAVERAGE, averages values from two adjacent nodes;
AVERAGE, averages values from all adjacent nodes; MAXIMUM,

uses maximum value of any adjacent node. EDGEAVERAGE is

the default.
When the version 8.9 traction calculation is used, then
If symmetric contact pairs are used, then the tractions output

are one-half of the expected values. This is because the tractions
are computed solely from the contactor forces, and when
symmetric contact pairs are used, the contactor forces are one-half
of the total forces.
Forces:
Force output is available for all contact nodes under all contact
conditions. The AUI variables are: CONTACT_FORCE-X (3-D
contact only), CONTACT_FORCE-Y, CONTACT_FORCE-Z,
CONTACT_FORCE-A (3-D contact only),CONTACT_FORCE-B,
CONTACT_FORCE-C. These forces are sometimes called
"consistent contact forces".
4.5 Friction
ADINA has a general Coulomb type friction model, where the
coefficient of friction can be a constant or calculated based on
several user-defined friction laws. Explicit analysis however, only
supports standard Coulomb friction.
4.5.1 Basic friction models
By default, a constant coefficient friction is used.
Two other basic models in which the coefficient of friction depends

on the contact tractions are
( )
Model 1: Tt = A1 1 exp ( A2Tn ) , where A1, A2 are specified
constants. This implies that the instantaneous coefficient of friction
1 exp ( A2Tn )
is = . Model 1 is graphically shown in Fig.
Tn / A1

4.5-1.
Model 2: Tt = A2 + ( A1 A2 ) exp ( A3Tn ) Tn , where A1, A2, A3

are specified constants. This implies that the instantaneous
coefficient of friction is = A2 + ( A1 A2 ) exp ( A3Tn ) . Model
2 is graphically shown in Fig. 4.5-2.
Tt
A1
Slope A1 A2 = for small tractions

Tn
a) Maximum tangential traction vs normal traction
A1 A2
Tn
b) Coefficient of friction vs normal traction
Figure 4.5-1: Frictional contact using model 1

4.5: Friction
Tt
Slope A1
Slope A2
Tn
a) Maximum tangential traction vs normal traction
A1
A2
A2 can be less than, equal to,
or greater than A1.
Tn
b) Coefficient of friction vs normal traction
Figure 4.5-2: Frictional contact using model 2
4.5.2 User-supplied friction models (not available in explicit analysis)
User supplied friction models are provided in subroutine FUSER.

They offer you more flexibility in implementing your own friction
models. They can be used with 2D and 3D contact surfaces, and
can be used with both segment and constraint-function methods.
They can also be used with the default node-to-segment contact and
with node-node contact. They cannot be used with the rigid target
contact algorithm.

Some algorithmic considerations
The subroutine FUSER in file ovlusr.f accepts two user-

specified arrays of constants, an integer array and a real array. In
the current implementation, the first integer constant I1 is used to
select a friction law. The real constants starting with A1 are used as
parameters for the specified friction law.
The input parameters passed to subroutine FUSER are: contact

traction Tn, consistent contact force Fn, normal direction, direction
of sliding v, other tangential direction, original and current
coordinates of contactor nodes, normal constraint-function
parameter N, friction constraint-function parameter T, magnitude
of relative sliding velocity u , frictional force, user-selected
parameters (real and integer), number of user-selected parameters,
contact pair number, time, load step number, iteration number,
initial temperature of contactor, current temperature of contactor ,
flag to show if temperatures are present, algorithm used (constraint
function or segment method), contact type (node-to-segment, node-
to-node) and contactor node number.
The output parameters from subroutine FUSER are the

coefficient of friction SCF, an error flag IERR, and an error string
ERMES.
Not all input parameters are available with all contact

algorithms, for example contact tractions are not available with
node-node contact. More details are provided in the documentation
of the FUSER subroutine.
Several friction laws are already implemented in subroutine

FUSER.
Existing user-supplied friction models
Constant coefficient of friction (I1 = 1)
= A1

4.5: Friction
Different static and dynamic friction coefficients (I1 = 4)
A1 if u A3
=
A2 if u > A3
Friction coefficient varying with sliding velocity (I1 = 5)
u
A1 + ( A3 A1 ) if u < A2
= A2
A3 if u > A2

Anisotropic friction model (I1 = 6)
( A v ) 2 + ( A v ) 2 + ( A v ) 2 if u > A5
= 1 (1) 2 (2) 3 (3)
A4 if u A5
where v(1), v(2) and v(3) are the x, y and z components of the sliding
direction.
Friction coefficient varying with consistent contact force (I1 = 7)
= A1 + A2 Fn , 0 1
Time varying friction model (I1 = 8)
t
A1 + ( A3 A1 ) if t < A2
= A2
A3 if t > A2

Coordinate-dependent friction model (I1 = 9)
A1 + A3 x (2) + A4 x (3) in 2D
= , 0 A2
A1 + A3 x (1) + A4 x (2) + A5 x (3) in 3D

Temperature-dependent friction model (I1=10)
A2 if < A1

A A1
= A2 3 + A4 if < A3
A3 A1 A3 A1
A4 if A3
Friction model 1a (I1 = 2)
1 exp( A2Tn )
=
Tn A1
Friction model 1b (I1 = 12)
1 exp( A2 Fn )
=
Fn A1
Friction model 2a (I1 =3)
= A2 + ( A2 A1 ) exp( A3Tn )
Friction model 2b (I1 = 13)
= A2 + ( A2 A1 ) exp( A3 Fn )
4.5.3 Frictional heat generation
The heat generation resulting from frictional contact can be

accounted for in a thermomechanically coupled analysis. The user
can select the fraction of this heat that is lost, as well as the
fractions going into the contactor and target surfaces. Details are
provided in Chapter 11 of the ADINA-T Theory and Modeling
Guide.

4.6: Contact definition features
4.6 Contact definition features

4.6.1 Analytical rigid targets
There is the possibility of describing target contact surfaces as

smooth analytically described surfaces for some simple types of
geometry. These analytically described surfaces are termed
analytical rigid targets (ARTs).
The smooth surface representation with ARTs allows for more
accurate and more easily converging solutions, especially when
friction is present. Most of the applications of ARTs are in metal
forming problems.
The possible geometries are infinite planes, spheres, circles,
infinite cylinders or strips (the intersections of infinite cylinders
with the YZ plane for 2-D problems).
Each ART is connected to one reference node, which defines
the motion of the ART. ARTs can move translationally and rotate
with the reference node as a rigid body, but cannot deform.
Usually the motion of the reference node is controlled with
prescribed displacements. In general the reference node can be any
node in the model, free to move or with boundary conditions or
constraints.
There can be several ARTs in the model, but only one ART in
each contact group. The usual node-to-node contact can be used
with an ART in a contact group. Contact can either be frictionless
or with friction, and all types of static and dynamic analyses in
ADINA are possible when ARTs are present.
There are the following limitations. The ARTs are not plotted
by the AUI and are therefore invisible. Contact tractions cannot be
computed with ART contact.
Analytical rigid targets are defined using the ANALYTICAL-
RIGID-TARGET command.
4.6.2 Contact body
Unlike the contact surface, the contact body consists of a group

of nodes that are not connected through segments (see Fig. 4.6-1).
The contact body can only be the contactor in a contact pair, and
the target has to be a contact surface (i.e. the target cannot be a
contact body). This is useful for modeling metal cutting
applications where the internal nodes of a body can also come in

contact once the outer ones are removed. This option can also be
used to model 3D contact problems involving beams (since the
beams cannot define a 3D contact surface).
Contact bodies are defined using the CONTACTBODY
command.
Contact bodies cannot be used in tied contact.
Segment on
target surface
Contact body
(a collection of nodes)
No segments on
contactor surface
Figure 4.6-1: Contact body
4.6.3 Automatic generation of contact surfaces between Parasolid bodies
Contact surfaces and contact pairs can be automatically created

between two Parasolid bodies, using the CONTACT3-SEARCH
command. All surfaces within a user-specified distance of each
other are accounted for.
4.7 Contact analysis features

4.7.1 Implicit dynamic contact/impact
Oscillations in velocities and accelerations can sometimes be

present in implicit dynamic contact analysis especially for high
speed impact problems. These oscillations can be reduced by
- applying post-impact corrections,
- setting the Newmark parameter = 0.5,

4.7: Contact analysis features
- adding compliance to the contact surfaces,
- using the Bathe time integration method.
In post-impact corrections, the velocities and accelerations of

the contactor and target can be forced to be compatible during
contact (only in the normal contact direction). This is achieved by
modifying the velocities and accelerations of the contact nodes
once convergence is reached such that they satisfy conservation of
linear and angular momentum.
The post-impact correction option requires additional memory
and computations.
The post-impact correction feature should not be used together
with compliant contact surfaces, since the velocities and
accelerations of the contactor and target surfaces are no longer
expected to be identical.
If post-impact correction is activated, all target nodes, except
those with all degrees of freedom fixed or enforced displacements,
must have a positive non-zero mass. The contactor nodes can have
zero mass.
Setting the Newmark = 0.5 instead of the default = 0.25

(trapezoidal rule see Section 7.3) results in an accurate solution
of rigid body impact problems, and frequently has a positive effect
on reducing numerical oscillations in flexible body contact. It is,
however, recommended that the Bathe method be used instead,
whenever possible.
Adding compliance to the contact surface can also significantly

reduce the numerical oscillations that result from dynamic time
integration. In this case, the compliance factor must be selected
based on Eq. (4.4-1) such that the contact pressures do not cause
excessive penetration. Allowing penetration to the order of 1% of
the element size usually eliminates numerical oscillations.
The Bathe time integration method provides some numerical

damping to the high frequency content of the solution, which
includes the contact oscillations.

4.7.2 Explicit dynamic contact /impact
Oscillations in velocities and accelerations can sometimes be

present in explicit dynamic contact analysis especially for high
speed impact problems. These oscillations are more common with
the penalty contact algorithm. In that case, they can be reduced by
- reducing the normal penalty stiffness,
- adding penalty contact damping.
See Section 4.3.2 for details on the explicit penalty contact

algorithm.
In addition, other sources of damping such as Rayleigh damping

can reduce contact oscillations by damping the high frequency
modes that generate them.
Oscillations in results can also occur when using the kinematic

constraint algorithm. These oscillations can be due to a mismatch
in the masses of the two contacting surfaces. See Section 4.8.3 for
more details.
4.7.3 Contact detection
As explained earlier in this chapter, the contact conditions

prevent the contactor nodes from penetrating the target segments.
During each equilibrium iteration, the most current geometry of the
contactor and target surfaces is used to determine and eliminate the
overlap at the contactor nodes.
For single-sided contact, the calculation of overlap at a

contactor node k consists of a contact search, followed by a
penetration calculation. The contact search starts by identifying all
possible target surfaces where node k can come into contact. For
each of these target surfaces:

- Find the closest target node n to node k.
- Find all the target segments attached to node n.
- Determine if node k is in contact with any of these

segments.
- If node k is in contact, update the information.
- If no appropriate target segment is detected, the contact

search is expanded beyond the target segments attached
to node n.
For 3-D single-sided contact in explicit analysis only, a fast

search algorithm is available (Command CONTACT-CONTROL
3D-SEARCH=FAST). Based on the bucket sort and a node-to-
facet search, it is more accurate when the target facets are of very
different sizes. This algorithm is significantly faster when the
contact pair has a very large number of contactors and facets.
For double-sided contact, the contact search algorithm uses time

tracking and checks whether the contactor node penetrated a target
segment between times t and t + t.
4.7.4 Suppression of contact oscillations

(not available in explicit analysis)
In some problems contactor nodes may oscillate during

equilibrium iterations between several (usually two) neighboring
target segments. Frequently, both solutions are acceptable. A
special procedure can be used to prevent such oscillations. This is
done by selecting a non-zero NSUPPRESS parameter in the
CONTACT-CONTROL command. In this case, the program
records the pairing target segment for each contactor node in the
previous NSUPPRESS iterations. Once this array is full, and the
contactor node is still in contact, and the pairing target segment is
one of those recorded in previous iterations, the suppression feature
is activated. The contactor node from this iteration onwards is
associated with only that target segment. It may remain in contact

with the segment, or in contact with an infinite plane passing

through the segment, or it can separate from contact completely.
The node is released from its restrictions once iteration ceases,
either because convergence is reached, or due to non-convergence.
The oscillation suppression feature is not activated by default.
If this oscillation suppression feature is used, it is recommended

that NSUPPRESS be set greater or equal to 5 and at least 5 less
than the maximum number of iterations.
Note that there is memory overhead associated with this feature,

where an integer array of size NSUPPRESS is defined for all
contactor nodes.
4.7.5 Restart with contact
Changes in contact parameters are allowed between restarts,

with some exceptions. Some restrictions exist, such as no restart
from friction to frictionless and vice versa.
The contact algorithm itself for a certain contact group can also
change in a restart. For this purpose, the contact algorithms can be
divided into two categories. The first catergory includes the
constraint function (implicit), Lagrange multiplier segment
(implicit), kinematic constraint (explicit), and penalty (explicit).
Restarts are possible between different algorithms in this catergory.
The second category includes the implicit and explicit Rigid Target
algorithms. Restarts are possible between these algorithms.
However, restarts are not allowed between the two categories.
4.7.6 Contact damping
The contact damping feature allows the user to add normal and
tangential grounded viscous dampers to all contactor and target
nodes in the model. Using the same value for both normal and
tangential direction results in isotropic viscous damping. The
damping force on each node is

FDamp = CN u N + CT uT
Contact damping is activated using the DAMPING, DAMP-

NORMAL and DAMP-TANGENTIAL parameters of the
CONTACT-CONTROL command. Contact damping is not
activated by default.
This damping can be useful in static problems for stabilizing the

model especially when there are insufficient boundary conditions
to remove rigid body modes. It can also be useful in dynamic
analysis to dampen out high frequency numerical oscillations. The
damping can be set to act only at the initial time step, or to be
constant throughout the analysis.
Using the initial damping option, the damping will be active at

the beginning of the first time step, and be reduced gradually
(between iterations) until it fully dies out by the end of the first
time step. Thus the final solution at the first time step should be
free of any damping. Note that if contact is not established and
nothing else stabilizes the model, the program will not converge
and will give an appropriate warning message.
Constant damping remains active throughout the analysis. In

this case, the program outputs the sum of all damping forces in the
output file, and the user must check that these forces are
significantly smaller than the sum of the reaction forces (also
written to the output file).
See Section 4.8.6 for modeling hints on using contact damping to

handle improperly supported structures and how to choose the
damping constants.
4.8 Modeling considerations

4.8.1 Contactor and target selection
For some contact problems, the contactor and target surfaces in

a contact pair can be interchanged without much effect on the

solution. However, for many cases, one of the two alternatives is

better.
If it is more important for the nodes of one surface not to

penetrate the other, then that surface should be the contactor. This
factor is usually important when one surface has a much coarser
mesh than the other as shown in Fig. 4.8-1. The coarse surface
should preferably be the target in this case. A related condition
occurs around corners or edges as shown in Fig. 4.8-2. The upper
surface should preferably be the contactor in this case.
If one of the surfaces has mostly dependent degrees of freedom,

it should be the target. This dependency can be due to boundary
conditions, constraints or rigid elements. The surface can also be
rigid if its nodes are not attached to any elements. In that case too
it has to be the target (except in the explicit penalty algorithm
where this is permitted).
If one surface is significantly stiffer than the other, it should

preferably be the target, unless one of the two conditions above
also exist.
Body 1 (contactor)
Body 2 (target)
Contactor nodes
Target segments
Figure 4.8-1: Effect of incorrect contactor-target selection due to

mesh density

4.8: Modeling considerations
Contactor
surface
Target
surface
Figure 4.8-2: Target selection for surfaces of different sizes
4.8.2 General modeling hints
ADINA automatically determines the direction of the contact

surfaces on the faces of solid elements. For contact surfaces
defined on structural elements (beams, trusses and shells) the user
has to ensure that the correct direction is defined (except in double-
sided contact).
Rigid target surfaces can be modeled using nodes with no

degrees of freedom or nodes with enforced displacements for all
active degrees of freedom. As a result, a fine discretization of a
complex rigid surface geometry is possible with only a small
increase in the solution cost.
If the contact surface is rigid it can be a slave to a master node

that will control the motion of the rigid surface. Rigid links can be
conveniently used between the master node and all the nodes on the
rigid surface.
It is acceptable for the nodes on the contactor and target

surfaces to be coincident (have identical coordinates). In this case,
it is important to ensure that the two surfaces do not share the same
nodes.
In general it is recommended that the lengths of segments on the

contactor and target surfaces be approximately equal. This is
particularly important if multiple contact surface pairs are

considered in the analysis or if the contact surface geometries are

complex.
If required, a contactor surface can be modeled as almost rigid

by choosing a reasonably high Young's modulus for the finite
elements modeling the contactor surface. However, the stiffness of
the surface elements should not be excessively high and make the
model ill-conditioned.
If the degrees of freedom of a node on a contactor surface are

used in constraint equations or attached to a rigid element (see
Section 5.15.2), the contactor node degrees of freedom should
preferably be independent.
If a contactor node has all of its translational degrees of freedom

dependent, the node is dropped from the contact surface.
If the contact surfaces are smooth (i.e., the coefficient of friction

is small), the frictionless model is recommended as it is less costly
to use. It is also recommended that prior to any contact analysis
involving friction, the frictionless solution is first obtained,
whenever possible.
It is not recommended that contact pairs with friction coexist

with contact pairs without friction in the same contact group.
A contactor node should preferably not belong to more than one

contact surface, otherwise the contactor node may be over-
constrained. If it is necessary for a contactor node to belong to
more than one contact surface, then contact compliance should be
used.
For problems in which the contactor and target surfaces are

initially relatively close to each other and no significant sliding
between these surfaces is expected throughout the analysis, the
small displacement contact feature may be used. The analysis will
be faster in this case, since the relatively time consuming contact
search is only performed once, and convergence difficulties
associated with a contactor node oscillating between one target to
another are eliminated. It is the users responsibility however, to
make sure that the problem is suitable for small displacement
contact.

The friction delay feature can sometimes lead to better

convergence since friction will only act once a converged contact
solution is established. This feature is also very useful for many
problems involving initial penetrations. In this case, the first time
step during which these initial penetrations are removed will be
frictionless.
Restarting from frictionless contact to contact with friction and

vice versa is not possible. However, it can be done if the
frictionless analysis is replaced by a frictional analysis with a very
small friction coefficient.
Ignoring initial penetrations is a useful option when these

penetrations are just a product of the finite element discretization,
meaning that they do not exist in the physical model. Fig. 4.8-3
illustrates one such case involving contact between concentric
cylinders. In this situation, if initial penetrations are eliminated, the
contact algorithm will try to push the penetrating contactor nodes
to the target surface segments in the first step, creating initial
prestressing. These initial penetrations and any prestressing that
they might cause are unrealistic. Ignoring them is useful in this
case. Note however, that if either cylinder is significantly rotated
the initial penetrations calculated at each contactor node (in the
initial configuration) will no longer be valid. In this case, the best
alternative would be to use a much finer mesh.
When higher order elements are used in contact, specifically the

10-node tetrahedral and the 20-node brick elements, tensile
consistent nodal contact forces can develop even when the
underlying contact tractions are compressive. The program can
accept such tensile forces as if they are compressive. This is done
via the TENSION-CONSISTENT parameter in the CONTACT-
CONTROL command. Accepting these tensile forces gives more
uniform results for problems involving the above mentioned
elements. However, it may slow down or even prevent convergence
in other problems. It is off by default.

Outer cylinder
Inner cylinder
Contactor surface,
marked with
Target surface,
marked with
Geometry before
discretization
Overlap to ignore
Figure 4.8-3: Analysis of contact between concentric cylinders,

initial penetration is ignored
4.8.3 Modeling hints for implicit analysis
It is recommended that the ATS method be used in contact

analysis (see Section 7.2.1). It can also be effective to use the low-
speed dynamics option of the ATS method (AUTOMATIC TIME-
STEPPING RESPS=YES).
Line search can sometimes be beneficial for contact problems.
Frictional contact problems using the constraint function

algorithm can be sensitive to the choice of frictional regularization
constant (EPST parameter of the CGROUP commands). For most
problems, this parameter should be one or two orders of magnitude
smaller than the expected sliding velocity. Using an excessively
large value leads to a smoother friction law, which generally
converges faster but results in smaller frictional forces or more
sliding. Using an excessively small value enforces the Coulomb
law more accurately but is more likely to experience convergence
difficulties.

Friction is not regularized or smoothed in the Lagrange

multiplier segment algorithm. This results in accurate enforcement
of stick and slip, but is more likely to experience convergence
difficulties.
Geometric and material nonlinearities can highly depend on the

sequence of load application. Thus, for problems involving such
features, the load steps should be small. The time step t should
also be small in dynamic analysis and when time dependent
material constitutive relations (e.g., creep) are used.
If a contact surface with corners or edges is modeled with

continuous contact normals, the normal vectors may be inaccurate
as shown in Fig. 4.8-4(a). In this case, switch to discontinuous
normals, use different contact surfaces for each smooth part (Fig.
4.8-4(b)) or use a fine mesh close to the corners or edges (Fig. 4.8-
4(c)).

Contact surface
with sharp corners
Contact surface 1
Arrows correspond to normal a) Single contact surface

vectors pointing to exterior side
Contact
Contact surface 3
surface 1 Contact surface 1
(with fine
corner mesh)
Contact surface 2
b) Three separate contact surfaces c) Single contact surfaces with

fine mesh at corners
Figure 4.8-4: Defining contact surfaces (with continuous normal

vectors) in the presence of corners
4.8.4 Modeling hints for explicit analysis
The penalty algorithm is preferred when both surfaces are rigid

or have many fixed or prescribed nodes.
Large oscillations in the contact forces may occur when using

the penalty method even though the model is stable. These can be
reduced by adding a penalty damping term and/or reducing the
penalty stiffness.
When using the penalty contact algorithm it is important to

check that the contact stiffnesses are properly selected. Unduly

small penalty stiffnesses will lead to excessive penetrations, while

unduly large penalty stiffnesses will lead to excessive oscillations
or unstable time integration.
Large mismatches between the masses of contacting surfaces

should be avoided when using the kinematic constraint method.
This mismatch is common when contact involves a rigid surface
with a small mass and an applied force, as shown in Fig. 4.8-5. The
best solution in such cases is to minimize the mismatch by
increasing the mass of the rigid surface.
The inaccuracy in this case results from the way the contact is
enforced. The kinematic constraint method first predicts
displacements without contact then applies a contact correction.
The contact conditions are satisfied more accurately when the
penetrations in the predicted configuration are small which is
usually the case due to the small time step size of explicit analysis.
However, some cases such as that mentioned above lead to large
projected penetrations which results in incorrect contact conditions
and tensile contact forces.
Large mismatches between the masses of contacting surfaces

can also lead to problems when using the penalty method. In this
case, the normal penalty stiffness required to avoid instability
(without reducing the time step) can be unduly small leading to
excessive penetrations. The best solution in such cases is to
minimize the mismatch by increasing the mass of the rigid surface,
or increase the penalty stiffness by setting it manually or by
reducing the time step.

F Small mass assigned to rigid surface
One node
in contact
Original configuration Correct solution
Tensile forces
Wrong contact
region
Projected configuration
(before contact correction) Final configuration
Figure 4.8-5: Performance of kinematic contact algorithm when

contact surfaces have a large mass mismatch.
4.8.5 Convergence considerations (implicit analysis only)
When ADINA fails to converge during the incremental analysis,

the intermediate printout given by ADINA in the output listing can
provide some useful information (see Fig. 4.8-6).
Three non-contact related norms are given: first for the energy
convergence criterion, the displacement and rotation convergence
criterion (boxes d and e), and the force and moment convergence
criterion (boxes b and c). Each box has 3 lines of output with the
top one giving the norm of the quantity, the second one giving the
node corresponding to the maximum value, and the third line
giving the maximum value. See Chapter 7 for definitions and more
details on these norms.
Box f of Fig. 4.8-6 shows the contact related norms. Parameter

CFORCE indicates the norm of the change in the contact forces
(between two iterations), and parameter CFNORM gives the norm
of the contact force vector.

The following additional convergence criterion is used since

contact is present:
CFORCE
RCTOL
max(CFNORM,RCONSM)
where RCTOL is the contact force convergence tolerance and
RCONSM is the reference contact force to prevent possible
division by zero in the contact convergence criterion above.
When the maximum number of iterations is reached without

convergence, and all norms are decreasing, the maximum number
of iterations should be increased.
OUT-OF- NORM OF ...
BALANCE OUT-OF-BALANCE NORM OF INCREMENTAL ...
ENERGY FORCE MOMENT DISP. ROTN. CFORCE ...
NODE-DOF NODE-DOF NODE-DOF NODE-DOF CFNORM ...
MAX VALUE MAX VALUE MAX VALUE MAX VALUE
box b box c box d box e box f

ITE= 0 1.14E+00 1.41E+02 9.99E-17 5.35E-02 5.12E-02 1.27E-15 ...
36-Z 35-X 31-Z 31-X 0.00E+00 ...
-1.00E+02 4.71E-17 -5.68E-03 -3.30E-02
ITE= 1 -1.29E-03 2.56E+01 1.92E-04 1.56E-02 2.45E-01 2.65E+03 ...

121-Z 31-X 64-Z 34-X 1.27E-15 ...
-9.85E+00 -1.06E-04 5.07E-03 1.26E-01
ITE= 2 3.32E-04 2.51E+01 1.88E-04 1.80E-02 1.77E-01 1.95E+03 ...

117-Z 31-X 64-Z 32-X 5.08E+01 ...
-9.66E+00 -1.04E-04 4.97E-03 -9.02E-02
ITE= 3 7.69E-02 4.46E+02 8.18E-04 1.04E-03 1.17E-02 1.95E+03 ...

64-Z 34-X 120-Z 33-X 2.00E+03 ...
3.21E+02 5.15E-04 -1.33E-04 -7.92E-03
... CONVERGENCE RATIOS CONVERGENCE RATIOS OUT-OF-BALANCE LOAD

... FOR OUT-OF-BALANCE FOR INCREMENTAL VECTOR CALCULATION
... ENERGY FORCE DISP. CFORCE BETA RATIO
... MOMENT ROTN. (ITERNS)
COMPARE WITH COMPARE WITH

ETOL RTOL DTOL RCTOL
1.00E-03 1.00E-02 (NOT USED) 5.00E-02
... 1.00E+00 1.41E+01 0.00E+00 1.27E-05

... 9.99E-17 0.00E+00
... -9.69E-03 2.56E+00 0.00E+00 2.65E+05 1.00E+00 -5.54E-02

... 1.92E-04 0.00E+00 ( 1)
... 2.49E-03 2.51E+00 0.00E+00 3.85E+01 1.92E-02 5.08E-03

... 1.88E-04 0.00E+00 ( 9)
... 5.76E-01 4.46E+01 0.00E+00 9.77E-01 1.00E+00 3.94E+03

... 8.18E-04 0.00E+00 ( 2)
Figure 4.8-6: ADINA output listing of convergence criteria during

equilibrium iterations
When the norms are rapidly changing before convergence fails,

it is commonly caused by applying the load too quickly or using a
large time step.
When CFNORM is stable but CFORCE changes rapidly during

equilibrium iterations, the contact can be oscillating between 2 or
more close solutions. In this case, try to change the time stepping,
or turn on the suppression of contact oscillations feature. When
CFNORM varies rapidly, usually the other three norms also vary.
4.8.6 Handling improperly supported bodies
Many static problems depend on contact to provide the boundary

conditions necessary for a stable problem (one in which there are
no rigid body modes). Some examples are shown in Fig. 4.8-7.
F
Applied Blankholder
Contact pair 1 load pressure
Symmetry
Fixed die
Contact Contact
pair 2 pair 3
Figure 4.8-7. Examples of improperly supported bodies
In such cases, the stiffness matrix is singular if the contact

constraints are inactive. Even if the constraints are active the
stiffness matrix is still not positive definite. The problem is more
serious if natural boundary conditions are applied (forces,
moments, or pressure). Weak springs can be added by the user to
make the model stable. However, the selection of appropriate
locations and stiffnesses for such springs may not be feasible.
Some models may be better suited for a dynamic or the low-speed
dynamics feature. However, in many cases, this too is not a feasible
option. Therefore, several other modeling techniques are available
in ADINA to handle such problems. These are stiffness
stabilization, contact damping and limiting incremental

displacements. These techniques can be used separately or

combined in the same model.
Stiffness stabilization (see Section 11.15 for details).

This feature provides a stabilizing effect by scaling all diagonal
stiffness terms without affecting the right-hand-side load vector.
The outcome of each iteration will be affected, but the final
converged solution will not be (within the bounds of the
convergence tolerances). Since the stabilization constant is non-
dimensional, it should always be a small number. Typical values
are between 10-12 and 10-9.
Contact damping (see Section 4.7.6 for details).

Contact damping adds grounded viscous dampers to all contactor
and target nodes. Setting the damping to be only at the initial time
step is sufficient for some problems such as those in Fig. 4.8-7.
When the first time step converges contact must be established and
damping will have been removed. This way, the converged solution
will be free of any contact damping. Other problems however,
require the damping to be constantly present. In this case, ADINA
outputs the damping forces at every time step. These forces should
be compared with the reactions in order to ensure that damping is
not excessive.
The damping constants have units of force per unit velocity. Hence,
their proper value is problem dependent. If initial contact damping
is used to stabilize a problem involving two contact bodies at least
one of which is unsupported, and with a gap between them, then a
good estimate of the damping constants CN and CT is one in which
the gap is nearly closed in the first iteration. Starting with the
dynamic equations of motion (see Equation 7.3-1) and canceling
out the inertial term (static analysis), and the stiffness term (since
one or both bodies are initially unsupported), we obtain
C U=R
where C is the total damping matrix, which in this case is diagonal,

and R is the applied load vector. We can assume the normal and
tangential damping constants to be equal, the total damping
contribution to be the damping constant times the number of
contact nodes on the unsupported contact surface N (the top

circular body in the example in Fig. 4.8-7), and the velocity to be

approximately equal to the minimum initial gap between the two
bodies, g, divided by the time step size t. This leads to the
following value of the damping constants
R t
CN = CT =
Ng
where R is the sum of the applied loads at the first time step.
Note that this is only an estimate, but is frequently an acceptable
one.
Limiting maximum incremental displacement (see Section 7.2.1

for details).
Limiting the maximum incremental displacement per iteration is
useful when a load is applied to a body that is not initially in
contact. The model at that stage is unstable and even when stiffness
stabilization or viscous damping is used, the initial displacement
can be excessive leading the program away from the converged
solution, and thus making the return to the proper solution difficult.
Setting the limiting displacement to about the element length size
in this case would scale down the potentially huge displacement in
the first iteration so that the results remain close to the converged
solution.
Note that this feature does not stabilize the stiffness matrix, so
in many cases it may be necessary to use it together with stiffness
stabilization or viscous contact damping or both.
4.9 Rigid target contact algorithm
4.9.1 Introduction
The rigid target contact algorithm is intended for use in
applications in which the target surfaces are considered to be rigid.
Fig. 4.9-1 shows a typical application in metal forming.

4.9: Rigid target contact algorithm
Prescribed punch
displacement
Blank holder force
Punch
Blank holder
Blank
Die
Drawbead
a) Physical problem
Contact Drawbead
surface Contact
4 (target) surface 2 Drawbead
(target) projection
Contact
Contact surface 3
surface 1 (target)
(contactor),
offsets used
to model blank
thickness
b) Modeling with contact surfaces and drawbead
Figure 4.9-1: Sample metal forming analysis using the rigid target
contact algorithm
A target surface can either be stationary, can translate as a rigid

body or can rotate as a rigid body.
Contact can be frictionless or can include Coulomb friction.
Drawbeads can be included.
2D or 3D contact conditions can be considered.
For 2D contact conditions, the rigid target contact algorithm is the

one used in ADINA version 8.3 and earlier.
For 3D contact conditions, the rigid target contact algorithm is

completely revised in ADINA version 8.4. However the rigid
target contact algorithm of ADINA version 8.3 and earlier is
retained in ADINA for backwards compatibility. The revised rigid
target contact algorithm of ADINA version 8.4 is the default.
Throughout this section, the rigid target contact algorithm of

version 8.3 and earlier is referred to as the V83 rigid target
contact algorithm. This section does not describe the V83 rigid
target contact algorithm; for information on the V83 rigid target
contact algorithm, see the ADINA Theory and Modeling Guide for
version 8.3 (report ARD 05-7).
Models that were set up using the V83 rigid target contact
algorithm may need to be revised when using the current rigid
target contact algorithm, see the conversion hints in Section 4.9.6.
We suggest that new models not be set up using the V83 rigid
target contact algorithm.
It is also possible to solve many problems involving rigid targets

using the constraint function and segment contact algorithms
described earlier in this chapter. However, the rigid target contact
algorithm described here is frequently more effective, because the
rigid target contact algorithm uses the assumption of rigid targets to
simplify the contact searching.

4.9.2 Basic concepts
4.9.2.1 Contactor surfaces
Similar to the other contact algorithms in ADINA, the contact

surfaces are organized into contact groups. Each contact surface
consists of 3- or 4-node contact segments. A contact pair consists
of a contactor surface and a target surface. In the rigid target
algorithm, it is allowed for a contactor surface to be in contact with
more than one target surface simultaneously.
Contactor surface: The contactor surface definition includes the

possibility of offsets.
When there are no offsets specified, the contactor surface is

described entirely by the contactor nodes. (Fig. 4.9-2(a)).
When there are offsets specified, the offsets can either be described
using spheres centered around the contactor nodes (Fig. 4.9.2(b)),
or using the contactor normals (Fig. 4.9.2(c)). In either case, the
offset magnitude is either constant or taken from the current
thickness of attached shell elements, as described below.
Contact Contact
segment 1 segment 2
Segment normals
a) Contactor segments without offsets. Segment normals are not used.
Figure 4.9-2: Contacting segments
When the offsets are described using the contactor normals, offset
vectors are constructed using the averaged contactor normals and
the offset magnitude. The upper and lower contactor points are
constructed from the contactor nodes and the offset vectors.

Radius of sphere
Upper surface = contactor offset
Lower surface
b) Contactor segments with spherical offsets. Segment normals are not used.
Upper contactor point
Offset
vector
Contactor
node normal
Lower contactor point
Contactor node normal = average of all segment normals

Offset vector = contactor node normal offset magnitude
c) Contactor segments with offsets, normals used to describe offsets.
When the target surface is concave, it is possible for the contact

situation to be similar to the one shown in Fig. 4.9-3. In this case,
when the offsets are described using spheres, the center contactor
node cannot be in contact with both target segments at once, hence
the center contactor node will oscillate between them. The center
contactor node cannot be in contact with target edge 1 since edge 1
is farther away than either of the target segments. Equilibrium

iterations in static and implicit dynamic analysis will not converge,

because of the oscillation. However contact is correctly modeled
when the offsets are described using normals, because the center
contactor node can be in contact with target edge 1.
Contactor and target surfaces viewed from the side

for ease of visualization.
This contactor node cannot be in contact In contact with target segment 2

with both target segments at the same time.
In contact with target segment 1 Spherical offset
Contactor surface
Targ t2
et seg gmen
men et se
t1 Targ
a) Spherical offsets, contact incorrectly modeled
Figure 4.9-3: Concave target surface, contactor surface with offets

In contact with target segment 2

In contact with target segment 1
In contact with target edge 1
Normals offset
Targ t2
et seg gmen
men et se
t1 Targ
Target edge 1
b) Normals offsets, contact correctly modeled
4.9.2.2 Target surfaces
Each target surface is either stationary, or can rigidly move

(translate, rotate or a combination of translations and rotations).
4.9.2.3 Determination of contact between contactor and target
No contactor offsets: It is allowed for a contactor node to be in

contact with a target segment, target edge or target node. The
program searches for the target segment, edge or node for which
the absolute value of the distance d between the contactor node and
the target segment, edge or node is minimized, where the distance
is measured in the direction opposite to the target normal (Fig. 4.9-
4). A positive distance corresponds to a geometric gap; a negative
distance corresponds to geometric overlap.
Notice that, for interaction between a contactor node and target

edge, or between a contactor node and target node, the target
normal direction is taken from the line segment connecting the
contactor node and the target, as shown in Fig. 4.9-4.

Contactor node
Target segment
Closest point on
target segment
Target segment normal
n
a) Interaction between contactor node and target segment
Contactor node
Target edge
n Closest point on
target edge
b) Interaction between contactor node and target edge
Figure 4.9-4: Interaction between contactor node and target surface
Fig. 4.9-5 shows two target segments with a common target edge.
The shaded volumes indicate which of the target entities any
contactor node is closest to. Notice that the shaded volume in
which the contactor node is closest to the target edge depends upon
the angles between the target segments attached to the edge.
Once the target segment, edge or node is determined, then the

contact gap is computed using
g = d GAPBIAS
where GAPBIAS can be chosen to, for example, not model contact
even if there is geometric overlap. (The default for GAPBIAS is
0.)

Contactor node in this shaded volume

Contactor node in this shaded volume is closest to edge 1.
is closest to segment 1.
Contactor node in this shaded volume

is closest to segment 2.
Edge 1
Segment 1
Segment 2
Figure 4.9-5: Interaction of contactor node with target segments

and edges
If the corresponding gap is negative, and less than DEPTH, then

contact occurs, in other words DEPTH g 0 is the contact
condition. DEPTH can be chosen to limit the depth of the target
surface, exactly as in the other contact algorithms.
Contactor offsets described using spheres: In this case, the

distance d is determined exactly as if there are no offsets. Then the
contact gap is computed using
g = d OFFSET GAPBIAS
where OFFSET is the offset magnitude. The process is illustrated

for interaction between a contactor node and target segment,
assuming that GAPBIAS = 0 (Fig. 4.9-6). The same idea is used
for interaction between a contactor node and target edge or node.

Contactor node
OFFSET
d
d - OFFSET
Target segment
Closest point on
target segment
Figure 4.9-6: Interaction between contactor node and target

segment, spherical offsets
Contactor offsets described using normals: In this case, contact is

detected using the upper and lower contactor points instead of the
contactor nodes.
Oscillation checking: The search for the nearest target segment,

edge or node is performed every equilibrium iteration in static and
implicit dynamic analysis. During the equilibrium iterations, it is
possible for the contactor node to move in such a way as to be
alternately in contact with two neighboring segments. This is
especially true if the target surface is concave. When the contactor
node oscillates between two neighboring segments, the solution
cannot converge unless oscillation checking is turned on. When
oscillation checking is turned on, then, when oscillation is detected
between two neighboring segments, the contactor node is placed
into contact with the shared target edge. In many cases, this
procedure allows the iterations to converge, if in fact the contactor
node should have been in contact with the shared target edge.
Oscillation checking only forces the contact between the contactor

node and shared target edge for the current equilibrium iteration.
For the successive equilibrium iterations, the contactor node is
always in contact with the nearest target segment, edge or node. So
oscillation checking cannot force contact to the wrong target
segment, edge or node in a converged solution.
Contact normal force: The normal force corresponding to contact

is computed as Fn = kn g where the normal force acts in the

direction opposite to the target normal direction (Fig. 4.9-7). kn is
the contact normal stiffness, entered as a parameter (see Section
4.9.3 for hints about choosing kn ). kn can be considered to be a
penalty parameter.
Slope -kn
Fn
-DEPTH g
Free overlap curve

Tensile contact curve
Figure 4.9-7: Normal contact force vs gap
Tensile contact and free overlap: During equilibrium iterations in

static and implicit dynamic analysis, the convergence can be
affected by the use of two related features: tensile contact and free
overlap. These features can be turned on or off in the model input,
see Section 4.9.3. The default is to use tensile contact and not to
use free overlap.
Tensile contact: The basic ideas for tensile contact are illustrated
in Fig. 4.9-8 to 4.9-10.
Fig. 4.9-8 shows the iteration history when tensile contact is not
used. For iteration ite-2, the contactor node and target segment
overlap. Hence contact is assumed between the contactor node and
target segment. For iteration ite-1, because of the relative motion
of the contactor node and target segment, the contactor node and
target segment do not overlap. For this iteration, no contact is
assumed between the contactor node and target segment. For
iteration ite, there is a large overlap because the contactor spring
unloads, since there are no forces acting on the contactor spring,
and the target does not provide any stiffness to the contactor node.

This large overlap causes large contact forces, which can cause
trouble in convergence in the successive iterations.
Spring is compressed
Contactor node Spring is uncompressed
Target segment
a) Iteration ite-2, b) Iteration ite-1, c) Iteration ite,

contact force Fite-2 no contact force contact force Fite
is compressive is large and compressive
Figure 4.9-8: Iterations when tensile contact is not used
Fig. 4.9-9 shows the iteration history when tensile contact is used.
Now, in iteration ite-1, tensile contact is assumed between the
contactor node and target segment. In tensile contact; the target
surface still provides stiffness to the contactor node. Hence the
overlap in iteration ite is small.
Fig. 4.9-10 shows the iteration history when tensile contact is used,
and the gap is large. In iteration ite-1, tensile contact is assumed,
and in iteration ite, no contact is assumed.
Contactor node
Target segment

contact force Fite-2 contact force Fite-1 contact force Fite
is compressive is tensile is compressive
Figure 4.9-9: Iterations when tensile contact is used
It is seen that tensile contact speeds up the convergence when the

converged solution is in contact, and slows down the convergence
when the converged solution is not in contact.
It is not permitted for an iteration in which tensile contact is present

to converge, unless the tensile forces are all less than the value of a
user-input parameter (see Section 4.9.3). Hence the use of the
tensile contact feature does not affect the converged solution.
Contactor node
Target segment

contact force Fite-2 contact force Fite-1 no contact force
is compressive is tensile
Figure 4.9-10: Iterations when tensile contact is used, converged

solution not in contact
Free overlap: The basic ideas for free overlap are illustrated in Fig.
4.9-11 to 4.9-13.
Fig. 4.9-11 shows the iteration history when free overlap is not
used. For iteration ite-3, there is a gap between the contactor node
and target segment. Hence no contact is assumed between the
contactor node and target segment. For iteration ite-2, because of
the relative motion of the contactor node and target segment, the
contactor node and target segment overlap. For this iteration,
contact is assumed between the contactor node and target segment.
For iteration ite-1, the contactor spring is stretched because there is
a large stiffness between the contactor and target due to the contact.
There is a small gap between the contactor and target, hence no
contact is assumed. For iteration ite, the contactor spring is
unloaded, and there is a gap between the contactor and target.
Fig. 4.9-12 shows the iteration history when free overlap is used.
Now, in iteration ite-2, free overlap is assumed between the
contactor node and target segment. Hence there is no contact
assumed for this iteration. In iteration ite-1, the contactor spring is
unaffected by the motion of the target.

a) Iteration ite-3, b) Iteration ite-2, c) Iteration ite-1, d) Iteration ite,

no contact contact no contact, no contact
large spring force
Figure 4.9-11: Iterations when free overlap is not used
a) Iteration ite-3, b) Iteration ite-2, c) Iteration ite-1,

no contact free overlap no contact
Figure 4.9-12: Iterations when free overlap is used
Fig. 4.9-13 shows the iteration history when free overlap is used,
and the overlap remains between target and contactor. In iteration
ite-2, free overlap is assumed, and in iteration ite-1, contact is
assumed.
It is seen that free overlap speeds up the convergence when the

converged solution is not in contact, and slows down the
convergence when the converged solution is in contact.
It is not permitted for an iteration in which free overlap is present

to converge. Hence the use of the free overlap feature does not
affect the converged solution.

a) Iteration ite-3, b) Iteration ite-2, c) Iteration ite-1, d) Iteration ite,

no contact free overlap contact, contact,
large overlap small overlap
Figure 4.9-13: Iterations when free overlap is used
4.9.2.4 Frictional contact
The friction force is calculated using the relative sliding velocity

between the target and contactor. The relative sliding velocity u f
is calculated from the velocities of the contactor and target using
u f = ( u c ut ) ( ( u c ut ) n ) n
where u c is the velocity of the contactor node, u t is the velocity

of the target and n is the target normal.
In static analysis, the contactor and target velocities are calculated

using the nodal incremental displacements divided by the time step.
In dynamic analysis, the contactor and target velocities are taken
from the nodal velocities.
The friction force magnitude is computed using
uf
Ff = Fn , u f < u f min
u f min
= F, u f u f min
where Fn is the normal contact force and is the Coulomb

friction constant (Fig. 4.9-14). u f min is the minimum sliding

velocity, entered as a parameter (see Section 4.9.3 for hints about
choosing u f min ). The direction of the friction force is always
opposite to u f .
When u f < u f min , the friction is sticking, otherwise the friction is

sliding.
Ff
Fn
Sliding
. . .
#uf uf u
#Fn
Sliding Sticking
Figure 4.9-14: Friction force vs velocity
Oscillation checking with friction: During equilibrium iterations in

static or implicit dynamic analysis, it is possible for the contactor
node to undergo sliding reversals. Namely, the contactor node
slides in one direction for an equilibrium iteration, then reverses
sliding direction for the next equilibrium iteration. When sliding
reversals occur, the solution cannot converge unless oscillation
checking is turned on. When oscillation checking is turned on,
then, when sliding reversals are detected, the contactor node is
placed into sticking contact, even if the sticking force is larger than
the sliding force, and convergence is prevented for the current
equilibrium iteration.
Oscillation checking only forces sticking friction for the current

equilibrium iteration. For successive equilibrium iterations, the
frictional state (sliding or sticking) is determined as usual from the
sliding and sticking forces. So oscillation checking cannot

converge to a solution in which the frictional state is wrong.
4.9.2.5 Drawbeads
The rigid target contact algorithm includes a simplified drawbead

model. Fig. 4.9-15 shows a diagram of a drawbead, as seen from
the side, along with the corresponding drawbead model definition
and drawbead tractions.
Fig. 4.9-16 shows the drawbead model in three dimensions. In this

drawbead model, a drawbead is modeled as a series of connected
drawbead segments, in which each drawbead segment connects the
successive nodes used to model the drawbead. The drawbead
segments are projected in the global z direction, and the projected
drawbead segments onto the contactor surface, target surface 1 and
target surface 2 are used in the drawbead calculation. Note that as
the contactor surface moves, the projected drawbead segments will
interact with different contactor segments.
Restrictions: The two target surfaces should not rotate or translate

relative to each other in such a way as to cause relative sliding
between the two target surfaces. Also the two target surfaces
should not rotate or translate so as to cause the projection of the
drawbead segments onto the target segments to change. Hence, the
two target surfaces can only translate in the z direction.
Consistent with these assumptions, the drawbead velocity is taken

from the absolute velocity of the contactor surface underneath the
projected drawbead segments.
The elements attached to the contactor surface are assumed to be 3-

or 4-node shell elements.
Activation of drawbead: If the distance between the target surfaces,

minus the thickness of the contactor, is less that the drawbead
height, then the drawbead is active (Fig. 4.9-17). The thickness of
the contactor is taken as twice the offset of the contactor. The
drawbead height is a user-input parameter. If the drawbead is not
active, it does not apply restraining and uplifting tractions to the
model.

Blank holder
Blank
Blank motion
Die
a) Drawbead diagram
Drawbead node
Projection of drawbead node
Target 1
Contactor
Contactor motion
Target 2
b) Drawbead model definition
Restraining traction, acting on target 1 Uplifting traction, acting on target 1

Restraining traction, acting on contactor
Restraining traction, acting on target 2 Uplifting traction, acting on target 2
c) Drawbead tractions
Figure 4.9-15: Drawbead concepts

Drawbead node
Drawbead segment
Projection of
drawbead onto
target surface 1
Projection of
drawbead onto
contactor surface
Projection of
drawbead onto
target surface 2
Figure 4.9-16: Specification of a drawbead

Distance between
Contactor target surfaces
thickness
Figure 4.9-17: Activation of a drawbead
Drawbead restraining tractions: The magnitude of the restraining

traction acting onto the contactor surface is computed from the
velocity of the contactor surface underneath the projected drawbead
segments. The formula is
u DB
TR = TRMAX , u DB < u DB min
u DB min
= TRMAX , uDB uDB min
as shown in Fig. 4.9-18. u DB min is the minimum drawbead sliding

velocity and TRMAX is the restraining traction corresponding to
sliding. Both of these values are user-input. u DB is the velocity of
the contactor surface underneath the projected drawbead segments,
measured normal to the drawbead segments, but in the plane of the
contactor. TR is the restraining traction (force per unit length), and

always acts in the direction opposite to u DB (Fig. 4.9-19).
TR
TRMAX
Sliding
.
#uDBmin
. .
uDBmin uDB
#TRMAX
Sliding Sticking
Figure 4.9-18: Drawbead restraining traction vs velocity
The restraining traction is applied to the contactor, and tractions

equal and opposite to half of the restraining traction are applied to
each of the targets (Fig. 4.9-20).
Drawbead restraining tractions are like friction forces in that the

restraining tractions always oppose the contactor velocity, and that
for contactor velocities higher than u DB min , the restraining tractions
no longer increase. For contactor velocities lower than u DB min , the
drawbead acts as a damper, with damping constant equal to
TRMAX
. (See Section 4.9.3 for hints about choosing u DB min .)
uDB min
Drawbead uplifting tractions: The drawbead also applies uplifting

tractions TU (force per unit length) to each of the target surfaces:
TR
TU = TUMAX
TRMAX
TUMAX is a user-input drawbead property. TUMAX = 0 is allowed.

The uplifting traction always creates forces that tend to separate the
target surfaces (Fig. 4.9-20).

Restraining and uplifting

tractions acting onto
target surface 1
Restraining traction
acting onto
contactor surface
Restraining and uplifting

tractions acting onto
target surface 2
Figure 4.9-20: Restraining and uplifting tractions
4.9.3 Modeling considerations
Algorithm used: The current rigid target contact algorithm is

selected by default.
To select the rigid target contact algorithm of version 8.3, use the
command
CONTACT-CONTROL RT-ALGORITHM=V83
Modeling of target surfaces: If the target surface translates or

rotates, all of the nodes on the target surface must be connected to a
master node, either using constraint equations or using rigid
links. For example, in Fig. 4.9-21, all of the nodes on the lower
target surface are connected to a master node using rigid links.

It is not allowed for the nodes on a target surface to have

independent degrees of freedom. All degrees of freedom for the
nodes on a target surface must be fixed or constrained.
Fixed target surface
Fixed nodes
Moving target surface

Slave nodes
Rigid links
Master node, independent degrees of freedom
Figure 4.9-21: Modeling of fixed and moving target surfaces
Modeling of contactor surfaces: The amount and description of

offset is determined by the CGROUP parameters
OFFSET-TYPE={NO, CONSTANT, TRUE}
OFFSET-DETECT={AUTOMATIC, SPHERES, NORMALS}
and

OFFSET= (constant offset, used only if OFFSET-

TYPE=CONSTANT)
OFFSET-DETECT=AUTOMATIC means that ADINA determines

the offset description (SPHERES or NORMALS). The criterion
used by ADINA is that an offset description of SPHERES is used
for each target surface that is convex or flat, and an offset
description of NORMALS is used for each target surface that is
concave.
When normals are used for the offset description, small steps
should be used in static and implicit dynamic analysis. This is
because the offset vectors are assumed to remain constant during
the equilibrium iterations. In particular, at convergence, the offset
vectors corresponding to the previous converged solution are used.
Determination of contact, target surface buckets: The target

segment, edge or node that minimizes the distance between the
contactor node and the target surface is determined. To speed up
the search, the target surface volume is partitioned into subvolumes
(buckets). ADINA automatically determines the number of
buckets, but you might want to override the programs choice.
Parameters
CONTACTPAIR NX, NY, NZ
allow you to specify the number of buckets in each of the global

coordinate directions (=0, ADINA automatically determines the
number of buckets). Increasing the number of buckets speeds up
the search, but increases memory consumption.
Determination of contact, modeling issues: It is possible for the

closest target segment, edge or node to not be the expected one.
An example is shown in Fig. 4.9-22. In this example, the rim of the
wheel is modeled with target segments. Because the distance
between a contactor node and a target segment is measured in the
direction of the target segment normal, a contactor node interacts
with the lower target surface, and the contact algorithm detects a
large overlap between this contactor node and the lower target
surface.

Another example is shown in Fig. 4.9-23. In Fig. 4.9-23(a), there

is a gap between the contactor node and the closest target segment,
as expected. In Fig. 4.9-23(b), the punch has moved upward
relative to the contactor node. Now there is a large overlap
between the contactor node and the closest target segment. This
segment is the only segment with a normal that points in the
direction of the contactor node.
In both Fig. 4.9-22 and Fig. 4.9-23, the large overlap is unintended.
In static and implicit dynamics, the equilibrium iterations would
most likely not converge. In explicit analysis, the large forces
between contactor and target would overdistort the elements
attached to the contactor node.
One way to avoid the large overlaps is to use the DEPTH feature so
that contact is not detected between the contactor node and the
incorrect target segment. Another way to avoid this issue is to
create additional target segments as shown. Then the contactor
node is closest to one of the additional target segments.
Choice of kn : The default value of the normal contact stiffness kn

is 1E11. However, kn can be chosen for optimal convergence.
Note that increasing kn causes the maximum overlap between the
contactor and the target to become smaller. Also, increasing kn
can lead to convergence difficulties.
We recommend that you use the smallest value of kn such that the
maximum overlap is still acceptably small. For example, if the
target surface is curved, there will be a geometric error associated
with using a coarse contractor surface (Fig. 4.9-24). There is no
advantage if the maximum overlap is less than the geometric error.
So, if the mesh is coarse, a large maximum overlap can be used.

Contactor node is closest Contactor node is closest

to lower target. Contactor to upper target
node is not closest to upper
target, since there is no Upper target
upper target segment with
normal that points in the
direction of the contactor
node.
Lower target
a) Orthographic view of wheel b) Side view of wheel
Contactor node
cannot be in contact
with lower target Contactor node is closest
Upper target to target edge
Upper target
Lower target Lower target

DEPTH
c) DEPTH feature used d) Wheel modeled with additional target surfaces
Figure 4.9-22: Meshing of a wheel
Another consideration for the choice of kn is the following.

Because the rigid target algorithm is node-based, and because the
contact stiffness is the same for each node in contact, the stresses
computed in higher-order elements on the contactor surface will be
inaccurate, if kn is too small. For example, in a problem involving
pressing an element onto a contact surface, kn should be greater
EA
than 100 where E is the Youngs modulus, A is the contact
nL
area, L is the element thickness (in the contact direction) and n is
the number of nodes on the contact area. The basic concept is
illustrated in Fig. 4.9-25.

Closest point
on target surface Closest point
that interacts on target surface
with contactor node Target surface that interacts
with contactor node
Contactor
node
Contactor
node
a) Intended interaction of contactor b) Unintended interaction of contactor

node and target surface; contactor node and target surface; contactor is
node is not in contact with target in contact with target with large overlap
Closest point
on target surface
that interacts
with contactor node
DEPTH
c) DEPTH feature used, contactor node

is not in contact with target d) Additional target segments used, contactor
is not in contact with target
Figure 4.9-23: Modeling of a punch
This issue also arises when lower-order elements are used, but
when lower-order elements are used, the variation in the consistent
nodal point forces is much less, so kn can be smaller for the same
accuracy in the element stresses.
Explicit dynamics time step size: In explicit dynamics, the time step
should be smaller than
m
t = 2
kn
This formula is derived from the following considerations.

Consider a single contactor node with mass m and no additional
stiffness or damping, with a velocity normal to the target. If this
node just touches the target at time t t , and penetrates the target
at time t , the node should remain in contact at time t + t . The
choice of t given above satisfies this condition.

Contactor surface, coarse mesh used
Geometric error
Target surface, fine mesh used
Figure 4.9-24: Modeling of a curved target surface

Uniform pressure load
Uniform pressure load
Youngs
modulus E
L Contact area A
Contact forces acting on contactor nodes

are all nearly equal
Contact forces acting on contactor nodes
are nearly equal to consistent forces
a) Soft target surface, kn small, corresponding to pressure load
element stresses are inaccurate
b) Hard target surface, kn large,
element stresses are accurate
Figure 4.9-25: Higher-order elements and rigid target contact
Clearly, decreasing kn will increase the time step t .
A node that is out of contact at time t t , in contact at time t

and out of contact at time t + t is said to have had a contact
reversal (Fig. 4.9-26).

t$tu.
Contactor node at t$t
Contactor node at t#t

t#tu.
Contactor node at t
Figure 4.9-26: Contact reversal due to too large time step in

explicit dynamics
Time step selection in frictional contact: The time step size will
affect the frictional velocities and hence the results. This is
because, in static analysis, the nodal velocities used in the friction
calculations are calculated as the incremental displacements
divided by the time step.
In those steps of the analysis in which friction is important, a

realistic time step should be used.
In those steps of the analysis in which friction is not important, a

large time step can be used, which causes the velocities to be small.
For example, in metal forming analysis, a large time step size can
be used when establishing the blank holder force, and during
springback calculations.
Choice of u f min for frictional contact: The default value of the

minimum sliding velocity u f min is 1E-10. However u f min can be
chosen for optimal convergence. Decreasing u f min can lead to
convergence difficulties.
We recommend that you choose u f min either from experimental

data, or use the largest value of u f min acceptable in your
application.

Explicit dynamics time step size for frictional contact: In explicit

dynamics, the time step should be smaller than
2mu f min
t =
Fn
to prevent reverse sliding. This formula is derived from the finite

difference equation corresponding to explicit time integration,
when applied to a single contactor node with mass m and no
additional stiffness or damping, sticking to the target, but with a
nonzero sticking velocity. If t is larger than the value given
above, the velocity will increase, and eventually the node will slide.
The sliding will then tend to reverse, that is, for a given time
step, the sliding direction will be opposite to the sliding direction in
the previous step (Fig. 4.9-27)
Contactor node .at t Contactor node at t$t
tu t$tu.
f f
Figure 4.9-27: Reverse sliding due to too large time step in explicit
dynamics
2mu f min
Note that when the time step is greater than t = , the
Fn
solution is still stable.
Choice of u DB min for drawbeads: The default value of the

drawbead minimum sliding velocity u DB min is 1E-8. However
u DB min can be chosen for optimal convergence. Decreasing u DB min
can lead to convergence difficulties. We recommend that you
choose u DB min either from experimental data, or use the largest
value of u DB min acceptable in your application.

Explicit dynamics time step size for drawbeads: In explicit

dynamics, the time step should be smaller than
2mDB uDB min

t =
TRMAX
(similar to the time step for friction). Here mDB is the contactor
mass/unit length.
Automatic time step selection in explicit dynamics: When using the

automatic time step selection options in explicit dynamics, the time
step returned from the rigid target contact algorithm is
mi
t = min
i kn
where the minimum is taken over all contactor nodes. Notice that
friction is not considered in the automatic time step selection; this
is because the model remains stable even if the time step is larger
than the friction time step discussed above.
Other user-input parameters:
CGROUP TENS-CONTACT={YES/NO}
This controls the use of the tensile contact feature discussed above.
The default is YES. Tensile contact is not used in explicit
dynamics.
The tensile contact feature is especially useful when shell offsets

are used and/or friction is present.
CGROUP FREE-OVERLAP={YES/NO}
This controls the use of the free overlap feature discussed above.
The default is NO. Free overlap is not used in explicit dynamics.
CGROUP GAP-PUSH

When GAP-PUSH is zero, the feature is not used (the default).

When GAP-PUSH is greater than zero, then, when the overlap is
less than GAP-PUSH, the contact stiffness is adjusted so that the
contactor node is pushed to a gap of GAP-PUSH (in the absence of
additional stiffness on the contactor node.) The GAP-PUSH
feature is not used in explicit dynamics.
The intent of the GAP-PUSH feature is to accelerate the separation

of the contactor and target. Occasionally this can lead to faster
convergence of the equilibrium iterations.
CGROUP TENSILE-FORCE
The maximum tensile force for a node in tensile contact for which
convergence is allowed (default value 0.001). All nodes in tensile
contact must have a tensile force less than this value for the
solution to converge.
CGROUP OSCILLATION-CHECKING
If OSCILLATION-CHECKING = 0, then oscillation checking

(described above) is turned off.
If OSCILLATION-CHECKING =ITE>0, then oscillation checking
is activated after equilibrium iteration ITE. The default is 5.
OSCILLATION-CHECKING is not used in explicit dynamics.
4.9.4 Rigid target contact reports for static and implicit dynamics
The following messages are output at the end of each converged

solution.
Maximum overlap at convergence:
Meaning: Self-explanatory
Recommend: If the maximum overlap is too large, increase kn ;

if the maximum overlap is too small, decrease kn .
Maximum tensile contact gap during iterations for nodes in contact

at convergence:

Meaning: A node that is in contact at the start of the time step

may temporarily move out of contact during the iterations, then
go back into contact before convergence. This report item
reports the maximum contact gap of all such nodes. When the
tensile contact gap is large, then convergence may be difficult.
Recommend: Either reduce the time step or decrease kn to

reduce the tensile contact gap.
Maximum friction velocity at convergence:
Meaning: For nodes in frictional contact, this is the maximum

friction velocity of a node (either sticking or sliding).
Recommend: If the maximum velocity is less than u f min , and

the corresponding node should be sliding, decrease u f min .
Number of nodes in contact, number of nodes in sticking contact,

number of nodes in sliding contact:
Meaning: Self-explanatory. Each node is counted once for each

target surface that the node is in contact with. So a node that is
in contact with two target surfaces simultaneously is counted
twice.
Change of contact status during iterations:
Meaning The number of nodes that switch contact status (not in

contact to in contact, or vice versa), is reported. If there are
many nodes that switch contact status, this may cause
convergence difficulties.
Recommend: Either reduce the time step or decrease kn
In contact at convergence, in tensile contact during iterations.
Meaning: The number of nodes which were in tensile contact

during the iterations (meaning that the nodes were almost out of

contact) and in contact in the converged solution. When there

are many such nodes then convergence may be difficult.
Recommend Either reduce the time step or decrease kn to

reduce the likelihood that nodes go into tensile contact.
Change in frictional contact status during iterations:
Meaning: The number of nodes that change frictional contact

status (from sticking to sliding or vice versa) is reported.
Recommend: If there are many nodes that switch frictional

contact status, reduce the time step or increase u f min .
The following messages are output at the end of each solution that
did not converge.
Maximum change of contact force at end of iterations:
Meaning: The contactor node for which the contact force had
the largest change is output.
Recommend: Examine the model near that contactor node for

hints about why the solution did not converge.
Change of contact status at end of iterations:
Meaning: The number of nodes that are changing contact status

at the end of the iterations.
Recommend: Reduce the time step or decrease kn .
Sliding reversal at end of iterations:
Meaning: The number of nodes that are undergoing sliding

reversals at the end of the iterations.
Recommend: Reduce the time step or increase u f min

Change of target entity at end of iterations.
Meaning: The number of nodes that are oscillating between

different target entities at the end of the iterations.
Recommend: If you do not have oscillation checking turned on,

turn it on. Otherwise refine the target surfaces, or reduce the
time step.
4.9.5 Rigid target contact report for explicit dynamics
The following items are output for each time step in which results
are printed or saved:
Number of nodes in contact, number of nodes in sticking contact,

number of nodes in sliding contact:
Meaning: See the corresponding message in Section 4.9.4.
Maximum overlap since solution start; maximum overlap since last

report:
Recommend: See the corresponding recommendations in

Section 4.9.4
Maximum friction velocity since solution start, maximum friction

velocity since last report:
Recommend: See the corresponding recommendations in

Section 4.9.4.
Contact reversals since solution start, since last report:
Meaning: This is a count of the total number of contact

reversals. Also the number of contact reversals for the node
with the most contact reversals is given, along with the mass of
the node.

Recommend: To reduce the number of contact reversals, either

reduce the time step or decrease kn
Sliding reversals since solution start, since last report:
Meaning: This is a count of the total number of sliding

reversals. Also the number of sliding reversals for the node
with the most sliding reversals is given, along with the mass of
the node.
Recommend: To reduce the number of sliding reversals, either

reduce the time step or increase u f min .
Drawbead reversals since solution start, since last report:
Meaning: This is a count of the total number of drawbead

reversals. Also the number of drawbead reversals for the
drawbead segment with the most drawbead reversals is given.
Recommend: To reduce the number of drawbead reversals,

either reduce the time step or increase u DB min .
4.9.6 Modeling hints and recommendations
In statics and implicit dynamics, for a time step in which contact

is established over a large area, many equilibrium iterations may be
required. This is because the solution cannot converge until the
nodes in and out of contact are determined, and it may take many
equilibrium iterations to determine which nodes are in and out of
contact. An example is shown in Fig. 4.9-28. The ATS cutback
method will not be effective for this time step. Rather, you should
set the maximum number of iterations very large, so that the
program can find the converged solution.
In statics and implicit dynamics, when forming a part that is

relatively thin, setting the plastic-algorithm flag to 2 can allow
the use of larger time steps.

Prescribed displacement
Target surface 1 Contactor surfaces
Target surface 2
a) Solution step before contact with target surface 2
Possible contact in this area, many

equilibrium iterations may be required
Target surface 1
to accurately determine contact here.
Target surface 2
b) Solution step after contact with target surface 2
Figure 4.9-28: Establishment of contact over a large area during a

solution step
The contact search algorithm may take a relatively long time for
the first iteration of the first time step. Similarly, the contact search
algorithm may take a relatively long time for the first iteration of
any time step in which a contact group or contact pair is born.
As the contactor surface is refined, keeping k n constant, the

overlap and contact force will decrease at each contactor node.
Hence kn may need to be adjusted as the mesh is refined. In
general, as the mesh is refined, kn can be decreased in order to
keep the overlap reasonable.

In statics and implicit dynamics, convergence may become

difficult when contactor nodes that were not in contact with the
target suddenly interact with the target. An example is illustrated
in Fig. 4.9-29. Eventually the contactor nodes on the right will
come into contact with the target, and convergence may be
difficult. Alternative ways to model this situation are shown in Fig.
4.9-29.
Contactor node in contact

Contactor node not in contact
Drawing
direction
a) Poor modeling
b) Good modeling
c) Good modeling
Figure 4.9-29: Contactor nodes suddenly coming into contact
Another example is shown in Fig. 4.9-30. In Fig. 4.9-30(a), the top

target surface is flat, and the indicated node suddenly comes into
contact with the top target surface. Convergence is very difficult,
because a very small change in the position of the indicated node
can cause the contact status of that node (and hence the contact
force) to change abruptly. In Fig. 4.9-30(b), the top target surface
has a round corner, and the indicated node gradually comes into
contact with the top target surface. Convergence is easier, because
a very small change of the position of the indicated node results in
only a very small change in the contact force.
In metal forming analysis, this situation in frequently
encountered in the modeling of the blank holder. The modeling is

easiest if the blank holder is modeled as a flat target surface. But

convergence is easier if round corners are added to the blank holder
wherever nodes on the blank are anticipated to contact the blank
holder during drawing.
This node This node

suddenly comes gradually comes
into contact into contact
a) Poor modeling b) Good modeling
Figure 4.9-30: Additional example of nodes suddenly coming into contact
4.9.7 Conversion of models set up using the rigid target algorithm of

ADINA 8.3
The following hints may be useful when running models that were
set up using the rigid target algorithm of ADINA 8.3 and earlier:
The results from the V83 and current rigid target contact
algorithms will usually be quite different:
a) In statics and implicit dynamics, the V83 rigid target

algorithm only determines the state of contact at iteration 0; the
current rigid target algorithm determines the state of contact at
every iteration.
b) The V83 rigid target algorithm only allows contact between a

contactor node and one target surface; the current rigid target
algorithm allows contact between a contactor node and more
than one target surface.
c) In statics and implicit dynamics, the V83 rigid target

algorithm can force ATS cutbacks when it detects tensile forces,
the current rigid target algorithm does not force ATS cutbacks.
d) In explicit dynamics, the basic formulation used by the V83

rigid target algorithm is quite different than the basic

formulation used by the current rigid target algorithm.
In general, in statics and implicit dynamics, the current rigid

target algorithm requires more iterations than the V83 rigid target
algorithm to establish contact. This is because, in the V83 rigid
target algorithm, the state of contact is only determined in iteration
0.
Once contact is established, the current rigid target algorithm

can be used with much larger time steps than the V83 rigid target
algorithm. The excessive penetration issues of the V83 rigid
target algorithm do not exist in the current rigid target algorithm.
Remember that the current rigid target algorithm can

automatically determine the number of buckets (CONTACTPAIR
NX, NY, NZ). So you should check the model to see if these
parameters are specified, and, if they are specified, reset them to
zero.
In the V83 rigid target algorithm, the ATS method is

automatically turned on for statics and implicit dynamics.
However, in the current rigid target algorithm, the ATS method is
not automatically turned on. You may want to explicitly turn on
the ATS method.
The drawbead formulation is quite different in the current rigid

target contact algorithm.
In explicit dynamics, the V83 rigid target algorithm does not

affect the critical time step. But the current rigid target algorithm
can affect the critical time step.
In the V83 rigid target algorithm, the amount of friction is based

on the incremental displacements. In the current rigid target
algorithm, the amount of friction is based on the velocity
(incremental displacements divided by time step). So in frictional
analysis, the time step size will affect the results in the current rigid
target algorithm.


5.1: Introduction
5. Boundary conditions/applied loading/constraint

equations
5.1 Introduction
The objective of this chapter is to present the various options
available in ADINA for the description of boundary conditions,
applied loads and constraint equations.
As discussed in ref. KJB, Section 3.3.2, there are two classes of

boundary conditions: essential boundary conditions, such as
prescribed displacement (and rotation) boundary conditions,
comprise the first class; and natural boundary conditions, such as
applied force and moment boundary conditions, comprise the
second class.
Displacement boundary conditions include nodal degree of

freedom fixities, prescribed displacements and constraint equations.
Force and moment boundary conditions include various types of

applied loading.
All displacement and force boundary conditions can be referred

to the global Cartesian system or to predefined skew systems. For
more information about skew systems, see Ref KJB, Section 4.2.2.
The externally applied load vector used in the governing

equilibrium equations is established using contributions from the
various applied loads (see Chapter 7).
For concentrated loads (Section 5.2) and user-supplied loads
(Section 5.14), the contributions of these nodal loads are directly
assembled into the externally applied load vector.

Chapter 5: Boundary conditions/applied loading/constraint equations
For pressure loading, distributed loading, centrifugal loading

and mass proportional loading, pipe internal pressure loading, and
electromagnetic loading, ADINA first calculates the corresponding
consistent nodal load vectors (consistent in the sense that the
principle of virtual work is used) and then assembles these load
vectors into the externally applied load vector. The evaluations of
the consistent nodal load vectors for the various types of loading
are described in the following sections.
The options of prescribing nodal displacements, temperatures

and temperature gradients are provided to enable you to specify the
time variations of these quantities throughout the solution period.
Temperatures and temperature gradients in ADINA are used in
conjunction with material models which include temperature
effects, see Sections 3.5 to 3.7 and 3.11.
The definition of the time variation of the externally applied

load vector for the various time steps in the solution period depends
on whether automatic step incrementation is used or not.
Time variation of externally applied loads when automatic step

incrementation is not used: Each applied load is associated with a
time function which defines the time variation of the load
throughout the solution period.
In a static analysis in which time-dependent effects (such as

creep, viscoplasticity or friction) are not included in the material
models, time is a "dummy" variable which is used, via the
associated time function of each applied load, to define the load
intensity at a step. Thus, the time step increment directly
establishes the load increments. In the example shown in Fig.
5.1-1, note that the same solution is obtained regardless of the
size of the time step increment.
In a dynamic analysis or if time-dependent effects are

included in the material models in a static analysis, time is used
in a similar way to define the load intensity of an applied load at
a step. However, in these cases, time is a "real" variable
because the time step increment is employed in the actual
integration of the equations of motion in a dynamic analysis,
and in the integration of the element stresses in a creep or

5.1: Introduction
viscoplastic analysis. Hence, in these cases the choice of the

time step increment is no longer arbitrary.
tR
tR tR
200 l 200 l
100 l 2 steps 100 l 2 steps
1 2 t 2 4 t
Run 1: t = 1.0 Run 2: t = 2.0
Note: identical results are obtained in Run 1 and Run 2

for a linear static analysis.
Figure 5.1-1: Example of the definition of load variation using time
Time variation of externally applied loads when automatic step

incrementation is used: Two options are available:
Using the automatic-time-stepping (ATS) procedure (see

Section 7.2), the loads are defined for all times t, 2t,... as for
no automatic step incrementation. In addition, when the
algorithm subdivides a time (load) step, the load vector is
established by linear interpolation of the load vectors at times t
and t + t.
Using the load-displacement-control (LDC) procedure (see

Section 7.2), the applied loads are not associated with any time
function and the time variation of the loads cannot be specified
by the user. The contributions from all the loads are assembled
into a constant load vector denoted as the reference load vector.
During the response calculation, this reference load vector is
scaled proportionally using a load multiplier (in general
different from one step to the next) automatically computed by
the program.

(
A time function is defined as a series of points t , f i ( t ) in )
which t is time and fi ( t ) is the value of time function i at that
time. Between two successive times, ADINA uses linear
interpolation to determine the value of the time function.
ADINA includes special time function multiplying functions

(
that alter the time function value obtained using the t , f i ( t ) input )
data. In the following, let f *( t ) be the value of the time function
( )
obtained from the t , f i ( t ) input data and let f ( t ) be the value of
the time function actually used by ADINA. Then the following
multiplying functions are available:
Constant: f ( t ) = f *( t )
Sinusoidal: f ( t ) = f *( t ) sin (t + ) where is the angular

frequency in degrees/unit time and is the phase angle in
degrees.
Short circuit (type 1): f ( t ) = f *( t ) a + b exp ( t / ) in

which a and b are constants and is the time constant (unit of
time).
Short circuit (type 2):
0
f ( t ) = f *( t ) 2 I ( sin (t + ) + exp ( t / ) sin ( ) )
4
in which 0 is the magnetic permeability in (volt-seconds)/

(meters-amperes), I is the RMS value of the short circuit
current, is the angular frequency in degrees/unit time, is
the phase angle in degrees, is the impedance angle in degrees
and is the time constant (unit of time).

5.1: Introduction
If time functions are used to determine the time variation of the

applied loads, then the activation of the various applied loads can
be delayed through the use of the arrival time option. The arrival
time option does not apply, however, to centrifugal and mass-
proportional loading, see Section 5.4.
The specification of a nonzero arrival time corresponds to a
shifting of the associated time function forward in time. Note that
the time function multiplier is zero for all times t smaller than the
arrival time; see illustrations given in Fig. 5.1-2(a) and (b).
Time functions:
f(t) Input time function Time function shifted
by arrival time
4
2
Time at solution
1 start=0.0
t
2t 4t 6t 8t 10t
-1 Arrival
time Load active
-2
Solution period
-3
Resulting load values for tR = 10f(t):
Time
step 1 2 3 4 5 6 7 8
n tR 0 0 30 40 40 40 0 -10
Figure 5.1-2(a): Example on the use of the arrival time option

Time functions:
f(t) Input time function Time function shifted
by arrival time
4
2
Time at solution
1 start = arrival time
t
Time at 2t 4t 6t 8t 10t
-1 solution
start Load active
-2 Arrival
time Solution period
-3
Resulting load values for tR = 10f(t) and t0 = 2t:
Time
step 1 2 3 4 5 6 7 8
t0+ntR 30 40 40 40 0 -10 -20 0
Figure 5.1-2 (b): (continued)
5.2 Concentrated loads

Concentrated loads are nodal point forces or moments which are
applied at the specified nodes and act in the specified degrees of
freedom.
The direction in which a concentrated load acts depends on the

coordinate system(s) assigned to the node of load application and
upon whether the concentrated load is assigned to translational or
rotational degrees of freedom; see Fig. 5.2-1.

5.1: Introduction
Z Skew system defined

C at node j
B
Y
i j
X
A
Node Degree of freedom Corresponding

(direction) load component
1 X force
2 Y force
i 3 Z force
4 X moment
5 Y moment
6 Z moment
1 A force
2 B force
j
3 C force
4 A moment
5 B moment
6 C moment
(a) Case of nodes with six degrees of freedom
Figure 5.2-1: Reference systems for application of concentrated

loads
Translational degrees of freedom:
If no skew system is defined at the node, then the

concentrated loads applied to these degrees of freedom
correspond to nodal forces acting in the global X, Y, and Z
directions.

If a skew system (A, B, C) is assigned at the node, then the

concentrated loads applied to these degrees of freedom
correspond to nodal forces acting in the A, B and C directions.
Skew system defined

Z C Vmn at node m
Vkn
Vm2
B
Vk2
k m
Y
X Vm
1
Vk1 A
Node Degree of freedom Corresponding

(direction) load component
1 X force
2 Y force
k 3 Z force
4 V1 moment
5 V2 moment
6 (*) Vn moment
1 A force
2 B force
m
3 C force
4 V1 moment
5 V2 moment
6 (*) Vn moment
(*) Note that this degree of freedom is automatically deleted

by the program and moments applied at this degree of
freedom are ignored.
(b) Case of nodes with five degrees of freedom

5.1: Introduction
Rotational degrees of freedom:
In this case a distinction must be made between shell

midsurface nodes using five degrees of freedom and the shell
midsurface nodes using six degrees of freedom.
If six degrees of freedom are used, and if no skew system is

defined at the node, the concentrated loads applied to these
degrees of freedom correspond to nodal moments acting about
the global X, Y, and Z directions. If a skew system (A, B, C) is
assigned at the node, then the concentrated loads correspond to
nodal moments acting about the A, B, and C directions.
If five degrees of freedom are used, the concentrated loads

applied to these degrees of freedom correspond to nodal
moments acting about the V1, V2 and Vn directions of the
midsurface system described in Section 2.7.
Concentrated loads can also be specified as follower loads. In

this case the direction in which a load (force or moment) acts is
given by the vector pointing from an auxiliary node to the node of
load application. This direction is updated during the response
solution by evaluating the most current coordinates of both the
node of the load application and the auxiliary node.
Note that follower moments are not allowed at shell midsurface
nodes using five degrees of freedom.
Follower loads should only be applied in a large displacement

analysis. Equilibrium iterations (see Chapter 7) should, in general,
be performed if follower loads are present.
The direction of a follower load can be established with respect

to a one-dimensional structure (such as a beam structure) using
nonlinear rigid links (see Section 5.15.2). An example is given in
Fig. 5.2-2.
5.3 Pressure and distributed loading

Pressure loading can be applied to the surfaces of the following
types of elements (see Fig. 5.3-1): 2-D solid, 2-D fluid, 3-D solid,
3-D fluid, plate/shell and shell. Considering pressure loads on

2-D elements, the area over which a pressure load acts includes the
element thickness.
Auxiliary node
Rigid link
Node of load
application
k
Thin cantilever
Follower force
at t = 0
a) Configuration at time t=0
Note: t k is the rotation at node k at time t.
Auxiliary node
t
k
Rigid link
Follower force
at time t
b) Configuration at time t
Figure 5.2-2: Example of the use of rigid link to establish

the follower load direction
Distributed loading can be applied along the longitudinal axis of

the following types of elements (see Fig. 5.3-2): beam, iso-beam
(including axisymmetric shells), pipe, shell (along edges).
Distributed loads do not include the element cross-section
dimensions or thickness. The distributed load is applied along the
neutral axis of beam/iso-beam/pipe elements, and along the
midsurface line of shell elements. For example, if a shell element
with original thickness of 2 undergoes uniaxial tension due to a
distributed load applied to an edge of the shell element, the internal

stress is one-half the magnitude of the line load.
For each pressure/distributed load surface specified, a consistent

nodal load vector is calculated to represent the pressure/distributed
loading. The formulations used in the consistent load vector
calculations for the various loading surfaces are given in the
following sections.
Z
Y
X
X Y
a) Two-dimensional element b) Three-dimensional element
Figure 5.3-1: Examples of pressure loading
l
l
l
l
l
l
l Z
Y
X
a) Beam element b) Shell element
Figure 5.3-2: Examples of distributed loading
In a large displacement analysis, the pressure/distributed

loading can be specified as deformation dependent for all element
types. In this case, the calculations of the consistent load vectors
are based on the latest geometry and configuration of the loading
surface.
Deformation dependent loading should only be used in a large

displacement analysis. Equilibrium iterations (see Chapter 7)

should in general be performed if deformation dependent loading is
present.
For pressure loads, you can optionally specify the direction of

loading, as follows:
a) Pressure acts normal to the surface or line to which it is

applied (the default).
b) Use x, y or z-component of a).
c) Pressure represents a traction in the x, y or z directions.
d) Pressure represents a traction tangential to the line (this

option is only valid for 2-D element loadings).
e) Pressure represents a traction load in a user-specified

direction tangent to the surface (this option is only valid for
3-D, plate/shell and shell element loadings).
For distributed loads, you can optionally specify the direction of

loading, as follows:
a) Distributed load acts normal to the line to which it is applied

(the default).
b) Use x, y or z-component of a).
c) Distributed load represents a traction in the x, y or z

directions.
d) Distributed load represents a traction tangential to the line.
5.3.1 Two- and three-dimensional pressure loading
For two- and three-dimensional pressure loading, the consistent

load vector consists of nodal forces acting in the translational
degrees of freedom. The calculation of this load vector is given in
ref. KJB, Section 4.2.1.
Along a 2-D element side, the pressure intensity is assumed to

vary linearly (with respect to the isoparametric coordinate on the

element side). Along a 3-D element face, the pressure intensity is
assumed to vary bilinearly (with respect to the isoparametric
coordinates on the element face).
Deformation-dependent pressure loads that act onto the edges of

plane stress elements do not take into account the change in
element thickness due to in-plane deformations. For example, a
plane stress element that undergoes uniaxial tension due to a
deformation-dependent pressure load has internal stresses larger
than the pressure load by the ratio (original thickness)/( current
thickness).
Load penetration option: If a finite element "dies" due to material

rupture or due to a predefined death time, the pressure loads
applied to that element are transferred to the neighboring "alive"
elements.
In the case when multiple pressure loads are applied to the finite
element model, the total pressure applied inside a cavity created by
"dead" elements is equal to the sum of the pressure loads applied to
the cavity opening divided by the number of pressurized element
faces at the opening (see Figure 5.3-3).

P7
P5 P6
P1 P2 P3 P4
P2 + (P3 + P5 + P7) + (P4 + P6)

P=
3
Figure 5.3-3: Pressure applied inside a domain created

by "dead" elements
Figure 5.3-4 shows some examples of situations arising when

the load penetration option is activated.
Load penetration can be selectively activated for designated
element groups only (thus reducing the computational time
necessary to perform the analysis).
In the case that a part-through crack or void is being created

during the finite element analysis, load penetration should be used
with caution and judgement. You must make sure that the external
pressure loading is consistent at all times of the analysis with the
physics of the problem solved.

P2 P2 P2 P2 P2
P1 P 1 P1 P 1 P1 P1 P1 P1 P1 P1
P2 P2 P2 P2 P2 P = P1 + P2
1
P= (P + P2)
2 1
Figure 5.3-4: Some of the situations arising

with load penetration
Load penetration cannot be used with deformation independent

pressure loading: pressure loading must be deformation dependent.
Note that load penetration is not applicable in case of
dynamic analysis using explicit time integration. Load penetration
is also not applicable when the automatic LDC method is used.
5.3.2 Hermitian (2-node) beam distributed loading
The distributed loading on a beam element is represented by the

equivalent concentrated forces and moments acting at the beam
nodes, see Fig. 5.3-5. These equivalent forces and moments are
equal in magnitude to the fixed-end forces and moments but act in
the opposite directions (the fixed-end forces and moments are the
reaction forces and moments at the nodes when the beam element is
subjected to the distributed load and the nodal displacements and
rotations are enforced to be zero). The consistent load vector for
beam distributed loading consists of these fixed-end forces and
moments.

q2
q1
1l l2
L
(a) Beam distributed loading
L L
F1 = (7q 1 + 3q 2) F2 = (7q 2 + 3q 1)
20 20
l l
L
L2 (3q + 2q ) L2 (3q
M1 = 1 2 M2 = 2 + 2q 1)
60 60
(b) Fixed-end forces/moments representation
Figure 5.3-5: Representation of beam distributed loading
Displacements and stresses in the model are calculated by

representing the actual distributed loading using the consistent load
vector defined above. Hence, the calculated solution corresponds
only to these equivalent concentrated nodal forces and moments,
and may not correspond entirely to beam theory results taking
account of the distributed loading more accurately. However, an
option is available to correct the force and moment results in linear
static analysis to obtain the beam theory solution. This option is
selected by the command MASTER FEFCORR = YES. For more
information on this fixed-end force correction, see Section 2.4.5.7.
5.3.3 Iso-beam and pipe distributed loading
ref. KJB The distributed loading on an iso-beam or pipe element surface

Section 4.2.1 is represented by concentrated forces acting at all nodes on the
element longitudinal axis. These forces are calculated using the

interpolation functions for the nodal displacements as given in ref.

KJB.
Since the nodal translations and rotations are interpolated

independently for the iso-beam and pipe elements, the consistent
load vector for the distributed loading consists of nodal forces only
(no moments).
For iso-beam and pipe elements subjected to distributed loading,

the element nodal forces/moments and stresses calculated in
ADINA are due to the consistent nodal loads.
5.3.4 Plate/shell pressure loading
The pressure loading on a plate/shell element surface is

represented by concentrated forces acting at the three corner nodes.
These forces are calculated assuming a linear interpolation of the
transverse displacements between nodes and hence this consistent
load representation of the pressure loading is only approximate.
5.3.5 Shell loading
The pressure loading on a shell element surface is represented

by concentrated forces acting at the nodes on the element surface.
These forces are calculated using the interpolation functions for the
nodal displacements as given in ref. KJB, Section 4.2.1.
Since the nodal translations and rotations are interpolated
independently for the shell element, the consistent load vector for
the distributed loading consists of nodal forces only (no moments).
Shell elements can also be subjected to a distributed line loading

acting on the edge of the elements (see Fig. 5.3-2). In this case an
auxiliary node needs to be used to define the plane in which the
load is acting.
Deformation-dependent line loads that act onto the edges of
shell elements do not take into account the change in element
thickness due to in-plane deformations.

5.3.6 Contact surface modification of prescribed loads feature
In certain problems, it is desirable for contact surfaces to control

the prescribed loads acting on those surfaces. For example, there
might be a prescribed pressure acting on a contact surface. It might
be desirable for this prescribed pressure to be active only when the
contact surface is not in contact, and for this prescribed pressure to
be inactive when the contact surface is in contact (Fig. 5.3-6).
The "contact surface modification of prescribed loads" feature

provides this control. This feature is off by default. The feature can
be turned on using the PRESSURE-CONTROL parameter of the
CONTACT-CONTROL command.
PRESSURE-CONTROL=YES activates the feature. PRESSURE-

CONTROL=V89 activates the feature using the version 8.9 sealing
criterion (see below), but this option is only provided for
backwards compatibility with version 8.9, and should not be used
otherwise.
Contact surfaces can modify the following prescribed loads:

pressure loads acting on 2D, 3D, plate and shell elements. These
loads must be deformation-dependent.
The contact surfaces that can modify prescribed loads are:

implicit constraint function with "new" contact surfaces, either 2D
or 3D contact, either single or double-sided contact. Only contactor
contact surfaces can modify prescribed loads.

The actual applied pressure at a corner node is equal to the

prescribed pressure (including the time function multiplier)
multiplied by a prescribed load factor. The prescribed pressure is
calculated from the pressure loads acting on 2D, 3D, plate and shell
elements, and again these loads must be deformation-dependent.
The load factor calculation is discussed in more detail below.
In particular, the fluid pressure from an FSI analysis does not
contribute to the prescribed pressure.
The prescribed pressure definition has the following

restrictions:
a) The prescribed pressure must be computed from a single load

definition. In particular, the prescribed pressure cannot be applied
in two separate load definitions with two time functions.
b) The prescribed pressure must be continuous between all element

faces attached to each node. However it is allowed for an element
face to have a pressure load and the adjacent face to have no
pressure load.
The program does not check to see if these restrictions are satisfied.
But if these restrictions are not satisfied, the program will not
compute the adjusted pressure correctly (see below).
The prescribed load factor is based on the contact pressure at the

node, the prescribed pressure at the node, and also on three input
values: the critical pressure, the penetration time and the evacuation
time. These values are input using parameters CRITICAL-
PRESSURE, TIME-PENETRATION, TIME-EVACUATION of
the CONTACT-CONTROL command. The default values of these
parameters are zero.
Sealing and unsealing criteria
In the evaluation of the prescribed load factor, it is useful to

consider the following quantity:
Adjusted pressure = (contact pressure) (unapplied pressure)
in which

Unapplied pressure = (prescribed pressure) (applied

pressure)
In terms of the prescribed load factor,
Applied pressure = (prescribed load factor) (prescribed

pressure)
Unapplied pressure = (1 (prescribed load factor))

(prescribed pressure)
We then can define the following terms:
Unsealing: The adjusted pressure is less than the critical pressure
Sealing: The adjusted pressure is greater than the critical pressure
In order to see that these definitions make sense, consider the

following cases:
1) Contact is established, so that the segment is sealed. There is a

contact pressure and also a prescribed pressure, but the applied
pressure is zero because the segment is sealed. Now suppose that a
small virtual gap is created (like a virtual displacement). In that
case, the prescribed pressure acts on the segment, reducing the
contact pressure to the adjusted pressure. If the adjusted pressure is
greater than the critical pressure, the virtual gap will close,
reestablishing the seal. But if the adjusted pressure is less than the
critical pressure, then the gap will not close and the seal will begin
unsealing.
2) Contact pressure less than the critical pressure, so that the

segment is leaking. Both the prescribed and applied pressures are
nonzero, thus the unapplied pressure is zero and the adjusted
pressure is equal to the contact pressure. Now suppose that in the
next step the contact pressure (and also the adjusted pressure)
becomes greater than the critical pressure. The segment will begin
to seal.
So the adjusted pressure is simply the contact pressure for the case

in which, for whatever reason, the segment hypothetically starts to

leak.
If PRESSURE-CONTROL=V89, then the adjusted pressure is

defined to be the contact pressure. The transition from leaking to
sealing is the same as that given above. But the transition from
sealing to leaking occurs only when the contact pressure itself
becomes less than the critical pressure.
Evaluation of prescribed load factor
At the beginning of the first time step, the prescribed load factor is
1.0.
At the beginning of the remaining time steps, the prescribed load

factor is increased or decreased according to:
Adjusted pressure less than critical pressure. The segment is

unsealing. The prescribed load factor is increased by the value
(delta t)/(penetration time), with a maximum value of prescribed
load factor equal to 1. If the penetration time is equal to zero, the
prescribed load factor is increased to 1.
Adjusted pressure greater than critical pressure. The segment is

sealing. The prescribed load factor is decreased by the value (delta
t)/(evacuation time), with a minimum value of prescribed load
factor equal to 0. If the evacuation time is equal to zero, the
prescribed load factor is decreased to 0.
Example
Figure 5.3-7 schematically shows the unsealing and sealing

processes. For simplicity, only one contact segment is shown, and
the critical pressure is assumed to be zero. The penetration and
evacuation times are greater than zero, so that the unsealing and
sealing take place over several time steps. Also for simplicity, the
deformations in the remainder of the model are held fixed during
the unsealing and sealing (so that the total pressure acting on the
contact segment remains unchanged during the unsealing and
sealing).
In Fig 5.3-7(a), the contact segment is in contact with no

prescribed pressure. In Figs 5.3-7(b) to (e), the prescribed pressure

is increased. Only when the prescribed pressure equals the contact
pressure (Fig 5.3-7(c)) does the applied pressure become nonzero.
As the applied pressure increases, the contact pressure decreases
(so that the total pressure acting on the contact segment is
unchanged). When the applied pressure equals the prescribed
pressure (Fig 5.3-7(e)), the contact pressure is zero and the seal is
leaking.
In Figs 5.3-7(f), contact is reestablished (perhaps due to
deformations acting elsewhere in the model). The contact pressure
is very small, but greater than zero. The applied pressure starts to
decrease. As the applied pressure decreases, the contact pressure
increases further (again, so that the total pressure on the contact
segment is unchanged). Eventually the applied pressure drops to
zero and the seal is sealed.
Notice that the adjusted pressure is zero in Figs 5.3-7(c) to (e),
and also in Figs 5.3-7(f) to (h).


Hints
The prescribed load factor is updated at the beginning of the

time step, and is unchanged throughout the time step. Therefore
small time steps should be used in the analysis.
It is allowed to set the critical pressure to zero.
It is allowed to change the critical contact pressure, penetration

time and evacuation time in a restart analysis.
The prescribed load factors can be listed using the ADINA-

PLOT variable PRESCRIBED_LOAD_FACTOR. The prescribed
load factors are used in the plots of prescribed pressures.
The following points are important to remember:
a) Contact pressure is force/unit length for 2D planar contact and is

force/unit area for 2D axisymmetric and 3D contact.
b) In symmetric contact, the contact pressure is equal to the

expected value if default options are used for the contact and for
the traction calculations. However, if the default options are
changed, then the contact pressure might only be half of the
expected value. See comments in Section 4.4 concerning the
traction calculations in symmetric contact.
c) The prescribed load factor is only evaluated at corner nodes. The

applied pressure at a mid-side node is interpolated from the applied
pressures at the corner nodes, regardless of the value of the contact
pressure at the mid-side node.
d) The contact pressure is the normal part of the contact traction

vector.
e) Contact tractions are computed in an approximate way from the

contact forces.

5.4 Centrifugal, rotational and mass proportional loading

5.4.1 Overview
Centrifugal, rotational, and mass proportional loading can be

used to model the effect of inertia forces that arise from
accelerations to which the structure is subjected:
Centrifugal loading is used for structures that rotate about

one axis of revolution.
Rotational loading is used for structures that rotate about two

axes of revolution.
Mass proportional loading is used for structures that

accelerate only in the translational degrees of freedom, and is
commonly used to model gravity loading and ground excitation.
Centrifugal loading can be applied to the whole model, to

individual element groups, or to element sets. Rotational loading
and mass-proportional loading can only be applied to the whole
model, or to individual element groups.
It is not allowed to have centrifugal loading or rotational

loading applied both to the whole model and to element groups.
It is not allowed to have more than one centrifugal load or

rotational load applied to the whole model.
More than one centrifugal load or rotational load can be applied

to an element group. However, each load acts independently, such
that the angular velocities between the different centrifugal and
rotational loads are not coupled together. This means that certain
inertia forces due to the coupling of the velocities (e.g. gyroscopic
forces) are not accounted for in the analysis. In general, it is
recommended to apply only one centrifugal load or rotational load
to an element group. If a part rotates about two axes, rotational
loading should be applied.
It is not allowed to have both centrifugal loading and rotational

loading applied to the whole model. However, both centrifugal

loading and rotational loading can be present in the model if the

loading is applied to the element groups.
The time function has different meanings in centrifugal and

rotational loading. In centrifugal loading, the time function
describes the time variation of the inertia loads; whereby, in
rotational loading, the time function describes the time variation of
the angular velocities and the angular accelerations.
The mass matrix used in the calculation of centrifugal,

rotational, and mass proportional loading can be lumped or
consistent. Note that the computational effort involved and memory
required is less when a lumped mass matrix is used, but the
resulting solution is, in general, less accurate.
Centrifugal and mass proportional loading can be used in static

analysis, dynamic analysis, and frequency analysis. Rotational
loading can only be used in static analysis and dynamic analysis.
The angular velocities in centrifugal and rotational loading are

not coupled with the nodal velocities in the dynamic response. This
means that certain inertia forces due to the coupling of the
velocities (e.g. gyroscopic forces) are not accounted for in dynamic
analysis. It is recommended to only use centrifugal and rotational
loading in static analysis.
In nonlinear analysis, deformation-dependent centrifugal and

rotational loading is automatically requested for all element types.
In this case, the calculation of the consistent load vector is based

on the current geometry, which adds additional nonlinear terms to
the stiffness matrix and to the load vector. These additional
nonlinear terms account for spin softening/stiffening effects.
In frequency analysis with centrifugal loading, only the spin

softening effect from the centripetal acceleration load is accounted
for, as the nonlinear stiffness matrix contribution associated with
this load is symmetric. The spin softening/stiffening effect from the
angular acceleration load is not accounted for, because the
nonlinear stiffness matrix contribution associated with this load is
skew-symmetric, and ADINA does not support a non-symmetric

eigensolver.
Rotational loading is not supported in frequency analysis.
When the element birth and death option is active (see Section
11.4), the consistent load vector only contains the contributions
from the elements currently alive.
The arrival time option (see the beginning of this chapter) is not
applicable for centrifugal, rotational, and mass proportional
loading.
Centrifugal and rotational loading is not supported for potential-

based elements.
Centrifugal and rotational loads are only approximately

computed for the large displacement beam element. This is because
the centrifugal and rotational loads are calculated assuming that the
beam is straight. The error in the solution is reduced as the beam
element mesh is refined.
5.4.2 Centrifugal loading
Centrifugal loading can be used to model the effect of inertia

forces on a structure that rotates about one axis of revolution. The
consistent load vector for centrifugal loading is computed as
follows (see Fig. 5.4-1):
t
R = [ ( r)
t t
+ t ( t t r ) ] dV (5.4-1)
angular acceleration centripetal acceleration
where t = ALPHA FACTOR f (t ) is the angular
(
acceleration, t = OMEGA FACTOR f ( t ) n is the angular )
t
velocity, r is the radial vector from the axis of rotation to the
node, ALPHA, OMEGA and FACTOR are constants specified in
the load definition, is the density, f(t) is the time function and n
is the unit vector parallel to the axis of rotation.

t

t
7 = angular velocity t
7
t
= angular acceleration
B
Points A and B are not Tangential forces
necessarily nodal points.
Z A
Centrifugal forces
Y
X Structure
Figure 5.4-1: Convention used for centrifugal loading
Note that the time function associated with the centrifugal

loading directly determines the time variation of the inertia forces
and not the angular velocity.
Note that the consistent load vector consists of nodal point
forces applied to all the nodes of the finite element model. But
loads applied to fixed degrees of freedom do not have any effects
in the solution.
In nonlinear analysis, deformation-dependent centrifugal

loading is automatically requested. In this case, the calculation of
the consistent load vector is based on the current geometry, which
adds additional nonlinear terms to the stiffness matrix and to the
load vector. These additional nonlinear contributions are described
in the following. Consider the equilibrium of the finite element
system at time t + t, iteration i. Then
M t +t U (i ) + t +t K (
i 1)
U (i ) = t +t R ( i ) t +t F (
t +t i 1)
(5.4-2)
where M = mass matrix, K = stiffness matrix = K L + K NL0 , U ( )

= acceleration vector, U = incremental displacement vector, R =
external load vector, F = nodal point forces corresponding to
stresses, (i) = iteration i, and t + t = time t + t.

The complete expression of the centripetal acceleration

contribution used by ADINA, including all nonlinear effects is
t +t
( (
R (i ) = ( 0r + t +t U (i 1) + U ( i ) ) dV ))
(5.4-3)
where 0r = vector of initial nodal distances to rotation axis and

= angular velocity vector. From the expression (5.4-3), it can be
seen that a deformation-dependent load is present, given by
t +t
(
K (NL1 ) U (i ) = ( U ( i ) ) dV
i 1
)
and that an additional nonlinear contribution K NL1 to the stiffness
matrix is present, given by
t +t
(
R (NL) = ( t +t U (i 1) ) dV
i
)
The additional stiffness contribution K NL1 is symmetric. Since the
tangent stiffness matrix includes this term, spin softening effects
from the centripetal acceleration contribution is accounted for in
frequency analysis.
The complete expression of the angular acceleration load

contribution used by ADINA, including all nonlinear effects is
t +t i
( (
R ( ) = 0 r + t +t U (
i 1)
)) dV
In this case, a deformation-dependent load is present, given by
t +t i
(
R (NL) = t +t U (
i 1)
) dV
However, ADINA does not include the nonlinear stiffness matrix
contribution from the angular acceleration load. The stiffness
matrix associated with this load is skew-symmetric due to the

single cross-product term. Since the tangent stiffness matrix does

not include this term, the spin stiffening/softening effect from the
angular acceleration load is not accounted for in frequency
analysis.
The correction to the stiffness matrix and the correction to the

loading are made when a deformation dependent loading is
requested.
5.4.3 Rotational loading
Rotational loading can be used to model the effect of inertia

forces on a structure that rotates about two axes of revolution at a
particular snapshot in time. The feature accounts for centripetal,
Coriolis, and angular accelerations, and nonlinear spin
softening/stiffening effects.
When a structure rotates about two axes, the problem is not

steady-state in a Lagrangian sense. In this case, the rotational
inertia forces acting on a material particle change as a function of
time, even if the angular velocity is constant. Hence, rotational
loading can only be used to model the effect of the inertia forces at
a particular snapshot in time.
A reference frame is used to describe the motion of the

structure. The reference frame rotates with angular velocity
frame and angular acceleration frame . The structure rotates with
angular velocity rel and angular acceleration rel relative to this
reference frame (see Fig. 5.4-2).
The position and direction of rel and rel remains fixed with
respect to the reference frame. The total angular velocity of the
structure is tot = frame + rel , and the total angular acceleration
is tot = frame + rel .

axis rotates with

frame reference frame D
rel
frame C rel
B r rel
r frame
A
Z
structure
reference frame Y
Figure 5.4-2: Convention used for rotational loading
In Fig. 5.4-2, points A, B, C, and D are not necessarily nodal

points. These points remain fixed in the reference frame
throughout the analysis.
The consistent load vector for rotational loading is computed as

follows (see Fig. 5.4-2):

t
R= ( t
frame t
r frame + t
rel t
r rel )

Angular Acceleration
+ ( t
frame ( t frame t
r frame ) + t
rel ( t rel t
r rel )) (5.4- 4)
Centripetal Acceleration

+ 2 t
frame ( rel
t t
r rel ) dV

Coriolis Acceleration
where is the density.

The time functions describe the time variation of the angular

velocities and angular accelerations (and not of the inertia forces).
For example, t frame = OMEGA f ( t ) n , where OMEGA is a
constant specified in the REF-FRAME command, f ( t ) is the time
function, and n is the unit vector corresponding to the axis of
revolution of the reference frame. Independent time functions are
used for t frame , t frame , t rel , and t rel .
Rotational loading accounts for angular acceleration, centripetal

acceleration, and Coriolis accelerations. The Coriolis accelerations
give rise to gyroscopic effects.
If the reference frame is stationary, such that

frame = frame = 0 , the rotational load feature applies the same
consistent load vector as the centrifugal load feature.
In nonlinear analysis, deformation-dependent rotational loading

is automatically requested. In this case, the calculation of the
consistent load vector in eq. (5.4-4) is based on the current
geometry, which adds additional nonlinear terms to the stiffness
matrix and to the load vector. The additional nonlinear terms
corresponding to the centripetal acceleration and angular
acceleration forces are discussed in Section 5.4.2. The additional
nonlinear terms corresponding to the Coriolis acceleration forces
can be obtained in a similar manner. A deformation-dependent
load is present for all load contributions, however, only the
stiffness matrix contribution from the centripetal acceleration load
is present. ADINA does not include the stiffness matrix
contribution from the angular acceleration load and the Coriolis
acceleration load, as the stiffness matrix associated with these
loads is, in general, not symmetric.
The additional nonlinear terms for the centripetal acceleration

forces reduce the apparent stiffness of the structure. This is known
as spin softening. However, the additional nonlinear terms for
the Coriolis acceleration can increase or reduce the apparent
stiffness of the structure, depending on frame and rel .

Physically, as the structure deforms under the centrifugal forces,

the radial distance increases which increases the centrifugal force.
However, as the structure deforms under the Coriolis forces, the
deformation may increase or reduce the Coriolis force.
5.4.4 Mass proportional loading
The consistent load vector for mass proportional loading in

direction i is computed using the mass matrix of the entire finite
element system and the specified accelerations (only in the
translational degrees of freedom), as follows:
t
R i = t M di t ai (5.4-5)
where d i is a direction vector with "1" in the portions of the

translational degrees of freedom acting into the direction i and "0"
in the other portions.
The values of the accelerations are specified as follows:
t
ai = MAGNITUDE f ( t ) Ai , i = x, y , z
where MAGNITUDE and Ai are specified in the load definition

and f(t) is the time function.
Note that the consistent load vector includes nodal point forces
applied to all nodes of the finite element model.
In the calculation, the mass coupling term between active and
deleted degrees of freedom is included. This mass coupling can be
clearly seen in the discussion of ground motion loads later in this
section.
Mass proportional loading is commonly used to model gravity

loading. For gravity loading, t ai is the acceleration vector due to
gravity. For example, for the z coordinate in the vertical direction
(increasing z corresponds to movement away from the ground),
enter t az = g , where g is the (positive) acceleration due to
gravity.
Mass-proportional loads are frequently used to model ground

motions. The basis for using mass-proportional loads in modeling
ground motions is given briefly now. The equations of motion for

linear dynamics, not including damping but including ground

motions, are
M11 M12 x1 K11 K12 x1 R1

MT + T = (5.4-6)
12 M 22 x 2 K12 K 22 x 2 R 2
where x1 is the vector of nodal point displacements for nodes not

attached to the ground and x 2 is the vector of nodal point
displacements for nodes attached to the ground. R1 , R 2 are
externally applied forces (for example, concentrated forces).
When the ground motions are the same at all nodes attached to
the ground, x1 = u1 + d1i xgi , x1 = u1 + d1i xgi , x 2 = u 2 + d 2i xgi ,
x 2 = u 2 + d 2i xgi , where u1 is the vector of nodal point displacements
relative to the ground for nodes not attached to the ground and u 2
is the vector of nodal point displacements relative to the ground for
nodes attached to the ground. Clearly, u 2 = 0 . Also d1i is the
direction vector for the nodes not attached to the ground and d 2i is
the direction vector for the nodes attached to the ground, with 1
in the portions of the translational degrees of freedom acting into
the direction i and 0 in the other portions.
The matrix equation of motion becomes
M11 M12 u1 d1i xgi K11 K12 u1 d1i xgi R1

MT + + T + =
12 M 22 0 d 2i xgi K12 K 22 0 d 2i xgi R 2
(5.4-7)
K K d 0 d
Now 11 T
12 1i 1i
d = 0 since the vector di = d corresponds
K
12 K 22 2 i 2i
to a rigid body motion. The matrix equation of motion becomes
M11 M12 u1 K11 K12 u1 R1 M11 M12 d1i xgi

MT + T =
12 M 22 0 K12 K 22 0 R 2 M12
T
M 22 d 2i xgi
(5.4-8)
and therefore the system of equations solved is

M11u1 + K11u1 = R1 M11d1i xgi M12d 2i xgi (5.4-9)
It is seen that the ground acceleration can be applied to the model

as a mass-proportional load, provided that the resulting nodal point
motions are interpreted as motions relative to the ground. The mass
coupling term between active and deleted degrees of freedom
M12d 2i xgi is included.
Please note:
To enter a positive ground acceleration xgi , specify a

negative mass-proportional load ai .
All fixities are relative to the ground. In other words, fixing

a node attaches it to the ground.
All prescribed displacements are relative to the ground.
All single DOF springs are attached to the ground.
Damping can be used. However, concentrated dampers and single

DOF damping spring elements are attached to the ground. Mass-
proportional Rayleigh damping, acts relative to the ground motion.
Although we have illustrated the procedure only for linear
dynamics, the procedure is also valid in nonlinear dynamics.
In a nonlinear static analysis, the nonlinear response due to

gravity loading alone can be obtained first and then the structure
can be subjected to additional concentrated, surface pressure
loading, etc. In this case, the total gravity loading is applied alone
in the first N steps, see Fig. 5.4-3. The time function defines the
load increments with which the gravity loading is applied. Other
concentrated loads, pressure loading, etc. are applied in the load
steps by using either the arrival time option or by the use of another
time function. Hence, in this case during the last (NSTEP-N) steps,
the response of the structure is due to both the maximum gravity
loading and any additional concentrated, pressure loading, etc.

f(t) NSTEP = total number of solution steps

Time function number 1, Time function number 2,
used to apply gravity loading used to apply other loads
t
N t NSTEPt
Figure 5.4-3: Use of gravity loading in nonlinear static
analysis with other types of loading
5.5 Prescribed displacements, velocities & accelerations

In some analyses, it is necessary to prescribe displacements at
specified degrees of freedom. In ADINA, the following
relationship can be specified.
ui = f (t)
where f is a general time function (see Section 5.1) which
prescribes the nodal point displacement ui. Prescribed
displacement conditions can be used in static and dynamic analysis.
Nodal point translations and rotations can be prescribed. The

degree of freedom for which a displacement is prescribed depends
on the coordinate system assigned at the node. The convention
used is exactly the same as the convention used for concentrated
loads, see Section 5.2.
Ovalization and warping modes can be prescribed at the nodes

of pipe-shell elements.
A nodal point can be "fixed" by prescribing a zero displacement

component for all degrees of freedom at this node. This is,
however, different from imposing a fixity boundary condition at
the nodal degrees of freedom because the prescribed degrees of

5.5 Prescribed displacements
freedom are retained in the system matrices (i.e., equation numbers

are assigned) whereas the degrees of freedom at which fixity
conditions are imposed are deleted in the system matrices.
A node which is "fixed" by having zero prescribed

displacements at a step can displace due to nonzero prescribed
displacements at the same degrees of freedom at another step. This
technique can be used to model moving boundaries.
In the Newmark method, the central difference method is used

to calculate velocities and accelerations for nodes with prescribed
degrees of freedom. Note that in the case of unloading from
prescribed displacements, the program does not switch to the
central difference method. Also note that there is no switch to the
central difference method if the Bathe method is used.
Note that prescribed displacements are not recommended on

contactor surfaces (see Chapter 4).
Arrival time option: When the arrival time is specified, the

prescribed displacement is not applied to the node until the arrival
time. Before the arrival time, the node has no load (in other words,
zero prescribed force).
Relative prescribed displacement option: Normally the

specified prescribed displacement is directly applied to the model,
so that the prescribed displacement at the node is equal to the
specified prescribed displacement. However, there is a relative
prescribed displacement option that works as follows. When the
relative prescribed displacement option is active, then the
prescribed displacement at the node is calculated using the
specified prescribed displacement plus a preexisting displacement.
The preexisting displacement is the displacement at the arrival time
(if the arrival time is nonzero), or at the beginning of the current
analysis (which might be a restart analysis).
The time function for the specified prescribed displacement can be

constant or time-dependent.
As a special case, the specified prescribed displacement can be

zero. Then the prescribed displacement is equal to the preexisting
displacement. This option is useful for freezing the motion of a

node, so that before the arrival time, the node has no prescribed
displacement, and after the arrival time, the node has a prescribed
displacement equal to the displacement at the arrival time.
Unloading options: Normally the prescribed displacement

remains throughout the analysis. However, there are unloading
options that allow the prescribed displacement to change into a
prescribed force.
Time unloading: After the user-specified unloading time, the

prescribed displacement changes into a prescribed force. The value
of the prescribed force is calculated as a user-specified unloading
force multiplied by the value of the unloading time function. If the
user-specified unloading force value is zero, than the value of the
prescribed force is calculated as the reaction force at the unloading
time multiplied by the value of the unloading time function.
Force unloading: When the reaction force is larger than the

user-specified prescribed force, the value of the prescribed force is
calculated as the user-specified unloading force multiplied by the
value of the unloading time function. It is an error to enter a user-
specified unloading force value of zero.
Although the options are termed unloading options, it is not

necessary for the prescribed force to actually decrease after the
unloading option changes the prescribed displacement into a
prescribed force.
Typical uses of the unloading options are:
1) Unloading of a formed part. It is then convenient to use time

unloading with a zero user-specified prescribed force. The value of
the unloading time function is then the ratio of the current
prescribed force to the reaction value at the specified unloading
time.
2) Application of a constant blank holding force. Either use time

unloading with a nonzero user-specified unloading force, or use
force unloading with a user-specified unloading force.
3) Unfreezing of a node. Before the unloading time, the

displacement of the node is prescribed (usually to zero), after the

unloading time, the node is free (with zero prescribed force). Use
time unloading with any (zero or nonzero) user-specified unloading
force, and an unloading time function which is always equal to
zero.
The process by which prescribed displacements change into

prescribed forces are illustrated by the following examples.
1) Time unloading with zero user-specified unloading force.

Suppose that the unloading time is 2.5, the solution times are 1.0,
2.0, 3.0, 4.0 etc. The solution for time 3.0 is computed as usual
with a prescribed displacement. The program then saves the
reaction force at time 3.0 for further use. The solutions for times
4.0, 5.0, etc. are computed using a prescribed force calculated as
the reaction force at time 3.0 multiplied by the unloading time
function evaluated at times, 4.0, 5.0, etc.
2) Time unloading with nonzero user-specified unloading force.

Suppose that the unloading time is 2.5, the solution times are 1.0,
2.0, 3.0, 4.0 etc. The solution for time 3.0 is computed as usual
with a prescribed displacement (not with the user-specified
prescribed force). The solutions for times 4.0, 5.0, etc. are
computed using a prescribed force calculated as the user-specified
unloading force multiplied by the unloading time function
evaluated at times 4.0, 5.0, etc.
3) Force unloading. Suppose that the user-specified unloading force

is 1000.0. At time 2.0, the solution is obtained using a prescribed
displacement, and the reaction force is 700.0. At time 3.0, the
solution is obtained using a prescribed displacement, and the
reaction force is 1200.0. Since the reaction force is larger than the
user-specified unloading force, the solutions for the successive
times 4.0, 5.0, etc. are computed using prescribed forces as follows.
The solutions for times 4.0, 5.0, etc are computed using a
prescribed force calculated as the user-specified unloading force
(1000.0) multiplied by the unloading time function evaluated at
times 4.0, 5.0, etc.
Prescribed velocities and accelerations can only be used in

dynamic direct time integration solutions using the following
methods:

- Newmark
- Bathe method
- explicit time integration
Prescribed velocities and accelerations can be applied in the

same way as prescribed displacements, except that:
- the unloading option cannot be used with prescribed

velocities and accelerations
- the relative prescribed option is not applicable to prescribed

velocities and accelerations
Prescribed rotations can be applied to the rotational degrees of

freedom of large displacement formulation structural elements
(such as Hermitian beam elements), and also to the master
rotational degrees of freedom of large displacement rigid links. The
formula used is now discussed.
In large displacement analysis, the increment in prescribed
rotation is computed as = t +t t , where t , t +t are the
prescribed rotations at times t , t + t . The components of the
increment in prescribed rotation are = ( x , y , z ) .
Consider an incremental rotation with components
= ( x , y , z ) applied to the master node of a large
displacement rigid link with end coordinates t x m , t x s . ("m" =
master, "s" = slave). The vector connecting these coordinates is
t
x sm = t x s t x m . When this incremental rotation is applied to the
master node of the rigid link, the vector t x sm changes to

sin
t +t sin 1 2 S2
x sm = Q() t x sm , where Q() = I + S+
2
2

0 z y

and S = z 0 x , = x2 + y2 + 32 . The
y x 0

above formula can be found in ref KJB, p 580, exercise 6.56.
This formula gives the exact answer for the case when the
incremental rotations are finite. Physically, the incremental
rotations are interpreted as a vector, with the direction of the vector
interpreted as the axis of rotation, and the magnitude of the vector
interpreted as the amount of rotation about that axis.
The same formula is applied to the end-node triad of a beam
element local node, and to the director vector of a shell element
local node.
Calculation of accelerations and reactions for nodes with

prescribed displacements or velocities
In dynamic analysis, if the prescribed displacements or

prescribed velocities vary nonlinearly in time, the accelerations at
the prescribed DOF are very sensitive to the number of time
function points used to specify the prescribed motion. The reason is
that the accelerations are computed from the second time derivative
of displacement, so any error in displacement is magnified by the
differentiation.
To compute the accelerations as accurately as possible there

should be a time function point corresponding to every solution
time, including the solution times corresponding to sub-steps in the
Bathe and Noh-Bathe methods, and including the solution times
corresponding to ATS cutbacks. For example, in an analysis in
which the Bathe method is used, and assuming that there are no
ATS cutbacks, there should be a time function point for all times
t + t as well as for times t + t .
Sometimes it is not possible to know in advance the solution

times for which ADINA evaluates the time function, due to
possible ATS cutbacks during the solution. Also the solution times
used by ADINA are sometimes not evenly spaced, for example in
linear analysis with the Bathe method, in which = 2 2 .

If there is no time function point at a given solution time,

ADINA uses linear interpolation in time between the two
bracketing times at which there are time function points.
Depending upon the actual (mathematical) shape of the time
function, this linear interpolation may or may not cause
inaccuracies in the accelerations. For example, if the mathematical
shape of the time function is linear at each solution time, then the
linear interpolation does not cause any inaccuracy. However, if the
time function is varying nonlinearly, the linear interpolation can
cause unwanted inaccuracies.
The most general way to avoid the inaccuracies is to sample the

(mathematical) time function using more than one time function
point per solution step. For example, supposing that the solution is
to be obtained at times 0.01, 0.02, etc., the time function could be
sampled using a much smaller time step size, e.g. 0.001, 0.002, etc.
The speed of the program solution is not affected by the number of
time function points in the time function, so there is no drawback in
specifying a large number of time function points.
Reasonable guidelines are to use at least 2 time function points

for each time step when using the Bathe method in nonlinear
analysis (with the default = 0.5 ), and at least 20 time function
points for each time step when using the Bathe method in linear
analysis (since = 2 2 by default in linear analysis), the
additional points ensure that at least one point is close to the
solution time.
The reactions at nodes with prescribed displacements or

velocities can also be inaccurate, for the same reasons as described
above.
In some cases, inaccurate reactions can be detected by

computing the energies for the system (see Section 11.7). When the
reactions are inaccurate, the energies do not balance.
The solutions at other nodes (nodes with free DOFs) are not
affected by the above issues.

5.6 Prescribed temperatures and temperature gradients

If temperature-dependent material models are used in the
analysis, then the nodal point temperatures for all nodes must be
defined at all solution steps.
Nodal temperature gradients must be defined if shell elements

or axisymmetric shell elements are present and temperature-
dependent material models are used in the analysis. Nodal
temperature gradients should only be assigned to shell midsurface
nodes and/or axisymmetric shell nodes. The direction of the
temperature gradient at a node is the same as the direction of the
normal vector (or the midsurface director vector) at the node, see
Section 2.7.1.
Temperature data can be obtained from an ADINA-T analysis in

which the same finite element mesh is used and thus the number of
nodal points, the number of elements, etc., are the same as in the
ADINA analysis. In this case ADINA-T directly creates a file
containing the temperatures. This file is written as an unformatted
FORTRAN file with the following line:
WRITE (56) TIME,(TEMP(I),I=1,NUMNP)
See Chapter 8 of the ADINA-T Theory and Modeling Guide for

more information.
Temperature data can also be obtained from an ADINA-T or

ADINA-F analysis in which a different finite element mesh is used.
In this case, ADINA-T and ADINA-F create mapping files, which
are then read by the AUI in order to create temperature files for the
ADINA analysis.
Also, the nodal temperatures for all solution steps can be

defined via direct input, in which case the time variations of the
temperatures are specified by time functions. In the following, we
describe how this is done using the AUI.
The formulas used to determine the prescribed nodal
temperature at node i are

i = TLOAD , if no temperature is prescribed at the node

t
and
N
1
t
i =
N
f (t ) F
j =1
j i
j
In these formulas, TLOAD is the overall prescribed temperature

(also referred to as the default prescribed temperature or the
reference prescribed temperature), N = number of temperature load
applications at the node, f j is the time function associated with
load application j (not necessarily time function number j) and Fi j
is the multiplying factor associated with load application j.
If the arrival time option is used in the nodal temperature

definition, then the time function value for time tARTM is zero
and hence a thermal strain will result if the initial temperature is
nonzero. In order to obtain a zero thermal strain prior to the arrival
time, multiple temperature data sets can be assigned to the same
nodal points. For example, when temperatures are prescribed
directly to the nodes (not through geometry), this can be done as
shown in Fig. 5.6-1.

Temperature Temperature

t * t
-0

0 Time 0 t = ARTM Time

Solution period
(a) Prescribed temperature set 1 (b) Prescribed temperature set 2
Temperature It is assumed that the prescribed

temperatures are applied directly
to the node (not through geometry).

0 t = ARTM Time
(c) Combined temperature variation (cases (a) and (b) are added)
Figure 5.6-1: Use of the arrival time option for prescribed

temperatures
If temperatures are obtained both from a separate file and from

direct input, the final temperatures used in the analysis are obtained
by adding the temperatures from the file and the direct input.
Temperature gradients at nodes can be read from a file and/or

from direct input in a way analogous to the nodal temperature
input.
The input of prescribed temperatures and temperature gradients

is completely separate from the input of initial temperatures and
temperature gradients. See Section 11.6 for the input of initial
temperatures and temperature gradients.
It should also be noted that when using higher-order elements,

the temperatures can be significantly different within the element
than at the nodal points. For example, the temperature can be
negative at points within an element, although the nodal point
temperatures are all non-negative. This is illustrated in Fig. 5.6-2
(see element 1). This observation can be important when
performing an analysis with temperature-dependent material
properties.

Linear elements:

1
0 l l l l l l l
x
El. 1 El. 2 El. 3 El. 4 El. 5 El. 6
Quadratic elements:
1
8 0
0 l l l l l l l
x
El. 1 El. 2 El. 3
Figure 5.6-2: Interpolation of temperature boundary

conditions
5.7 Pipe internal pressure data

If pipe elements are used, then pipe internal pressures can be
input as a load on the structure (see Section 2.8.2). The pipe
internal pressure is a nodal point variable which must be defined
for all nodes throughout the solution period.
The time variation of the nodal values of the internal pressure is

specified by the use of time functions.
Pipe internal pressures can also be read from a separate user-

supplied file. The format of this file is analogous to the file format
used for reading nodal temperatures (see Section 5.6).
If pipe internal pressures are specified by time functions and

also read from a file, the final pipe internal pressures are obtained

by adding the separate contributions.
5.8 Electromagnetic loading

The interaction effects of short-circuit currents in cables and
beams can be modeled with ADINA using the electromagnetic
loading option. Cables and beams can be modeled with truss and
beam elements.
Considering two electric conductors traversed by short-circuit

currents i1(t) and i2(t) and placed in a medium of magnetic
permeability 0, each differential element dr2 of one conductor
exerts on a differential element dr1 of the other conductor a
magnetic force given by Grassmans formula (see Fig. 5.8-1):
0 dr1 ( dr2 r21 )

dF21 = i1 ( t ) i2 ( t )
4 r21
3
where r21 is the vector between the middle of the differential

elements dr1 and dr2. Integration of dF21 along a finite element
yields the total electromagnetic force on the element, half of which
is assigned to each end node.
When entering this loading in the AUI, it is necessary to apply

the load onto each of the electric conductors. For the load on cable
0
k, the time function value should equal ik ( t ) .
4
3-point Gauss integration is used for all numerical integrations.

Cable 1 i1(t)
dF21
dr1
r21
Cable 2
dr2
i2(t)
Figure 5.8-1: Conventions for electromagnetic loading
5.9 Poreflow loads

The flux of pore fluids in porous media can be specified using
poreflow loads. See Section 3.12 for more information about
analysis of porous media.
The flux can either be concentrated (that is, applied directly to

nodes), or distributed (that is, applied to the boundaries of 2-D and
3-D elements).
5.10 Phiflux loads

The flux of fluids in potential-based fluid elements can be
specified using phiflux loads. See Section 2.11 for more
information about the potential-based fluid elements and Section
2.11.14 for more information about loads applied to potential-based
fluid elements.
The flux can either be concentrated (that is, applied directly to

nodes), or distributed (that is, applied to the boundaries of 2-D and
3-D elements).
Positive mass flux is assumed to represent mass flowing into the

fluid domain.

5.11 Contact slip loads

The contact slip loading enables you to add a tangential or slip
velocity component between contacting surfaces (see Figure 5.11-
1). The contact slip is defined by an axis of rotation, an angular
velocity and a time function. The total slip velocity between two
contacting surfaces is the vectorial sum of the actual slip resulting
from the relative displacement between the surfaces and this
contact slip load. The total slip velocity is used to evaluate stick
and slip and is also used to determine the amount of heat generated
due to friction. In 2-D analysis, the slip velocity component can be
out-of-plane (i.e. in the x-direction). Note that in static analysis the
tangential velocity is equal to the tangential displacement divided
by the time variable.
Total
velocity slip
7 7r Relative velocity
between contact
surfaces
r
Contact surface
Node of
contact surface
Axis of contact
slip rotation
Figure 5.11-1: Contact slip loading
Contact slip loading can be applied to 2-D and 3-D contact

surfaces.
5.12 Surface tension boundary
A constant surface tension value can be applied to the
boundaries of 2-D and 3-D elements as shown in Figure 5.12-1. A

similar boundary condition for fluids exists in ADINA-F.
3-D surface
T T
T
T
Solid
domain T
2-D line
Figure 5.12-1: Surface tension applied to 2-D and

3-D boundaries
In 2-D analysis, the surface tension boundary can be applied to

geometry lines or element edges. In 3-D analysis, the surface
tension boundary can be applied to geometry surfaces or element
faces.
The surface tension boundary is converted by the AUI into

equivalent ADINA elements when the ADINA data file is
generated. The ADINA elements have an initial stress equal to the
magnitude of the surface tension, T. In 2-D analysis, isobeam
elements are used. In 3-D analysis, 2-D generalized plane stress
elements are used. When post-processing the results, the AUI
automatically converts the ADINA elements back into a surface
tension boundary.
The surface tension option can be used with the small

large displacement/large strain formulations. When large
displacements are selected, the surface tension is treated as a true
(Cauchy) stress.
5.13 Initial load calculations

Considering temperature, temperature gradient and pipe internal
pressure loads, see Section 11.6 where the initial conditions are
discussed.

Considering all other externally applied loading, the loads

corresponding to time TSTART are not used and not calculated,
except when the ATS method is employed. When the ATS method
is employed, the loads are calculated either directly from the time
functions, or, in the case of fluid-structure interaction with
ADINA-FSI, from the initial displacements, initial velocities, initial
accelerations and initial strains.
5.14 User-supplied loads

An option for user-supplied loads is available in ADINA to
allow the modeling of very general types of loading. For example,
loads can be defined as functions of nodal coordinates,
displacements, velocities, accelerations, temperatures, times, etc.
In order to specify the loads, you must provide the desired

definitions of the loads in the form of FORTRAN coding which is
inserted in the specially provided subroutines in ADINA.
The user-supplied loads applied at a node can be a function of

the various nodal quantities at the same node, or they can be a
function of nodal quantities at other nodes, as described in detail
below.
The presence of user-supplied loads requires in general the use

of equilibrium iterations in the response solution, see Section 7.2.
5.14.2 Usage of the user-supplied loads
User-supplied loads can be specified in static or dynamic, linear

or nonlinear analysis. However, the definition of the loads must be
consistent with the type of analysis; e.g., the loads cannot be a
function of the velocities in a static analysis.
User-supplied loads cannot be specified if the LDC method is

employed (see Section 7.2.1 for information about the LDC
method). In addition, user-supplied loads cannot be applied to any
substructures; however, these loads can be applied to the main

structure to which substructures are coupled.
During the step-by-step response calculation, the user-supplied

loads are evaluated in each time step and in each iteration. These
loads are added to the effective load vector normally established in
the response calculation for the time steps.
Evaluation of the user-supplied loads: You must modify the

coding in the user-supplied subroutines IUSER and USERSL in
order to select the nodes whose nodal variables are used in the load
calculation, to select the nodes to which the loads are applied, and
to calculate the loads.
Sample coding is provided in file ovl170u.f. This coding
corresponds to the example given in Section 5.14.3 below.
You enter some control parameters in the AUI in the

specification of the user-supplied loading (in the command-line
interface, the command is APPLY USER-SPECIFIED-LOADS). If
loads calculated at a given node M depend only upon variables at
the node M, set parameter NODE-DEPENDENCE to 1; if loads
calculated at a given node M depend upon variables at other nodes,
set parameter NODE-DEPENDENCE to 2.
You also can define integer and floating point constants in the
AUI for use in the user-supplied loading subroutines:
NRCONS: Number of floating point constants (maximum of

100).
NICONS: Number of integer constants (maximum of 100).
RCONS(NRCONS): Floating point constants. In subroutine

USERSL, this corresponds to vector ULDATA(100).
ICONS(NICONS): Integer constants. In subroutine USERSL,

this corresponds to vector KULINT(100).
NODE-DEPENDENCE=1: If NODE-DEPENDENCE=1, the

loads applied at a node M can depend only on nodal variables at the
same node.

5.14: User-supplied loads
Node selection using subroutine IUSER: Subroutine IUSER is

called once for each nodal point each time the effective load
vector is calculated. In this case MFLAG should be set to 1
only for nodes at which loads are to be applied.
Load calculation using subroutine USERSL: Subroutine

USERSL is called once for each node selected by subroutine
IUSER each time the effective load vector is calculated.
Note that for each node M, only the non-zero components of
the new additional load vector need to be calculated. ADINA
distinguishes internally independent, dependent or fixed degrees
of freedom and performs the proper summation of the new
additional load vector to the effective load vector.
NODE-DEPENDENCE=2: If NODE-DEPENDENCE=2, then the

loads applied at a node M depend on the nodal variables at a
number of other nodes N.
Node selection using subroutine IUSER: Two flags, MFLAG

and NFLAG, are used in subroutine IUSER to select the
appropriate nodes for the load calculation, as schematically
shown.
DO 100 M=1,NUMNP
DO 200 N=1,NUMNP
.
.
.
CALL IUSER(M,N,NUMNP,MFLAG,NFLAG)
IF (MFLAG.EQ.0) GO TO 100
IF (NFLAG.EQ.0) GO TO 200
.
.
.
CALL USERSL(...)
.
.
.
200 CONTINUE
100 CONTINUE

Note that subroutine USERSL is only called for nodes selected by

the flags set in subroutine IUSER.
Load calculation using subroutine USERSL: Each time the

effective load vector is calculated, for each node M at which the
user-supplied loads are applied, subroutine USERSL is called once
for N=N1,N2 ..., NN where N1, N2,..., NN are arranged in
ascending order and correspond to the nodes whose nodal variables
are used to calculate the loads at node M. Nodes N1, N2,..., NN
are specified in subroutine IUSER as described above.
Note that for the load calculations at node M, the required nodal
variables must be stored in auxiliary working arrays when
N.LT.NN, and the actual load computation must be performed only
when N.EQ.NN. The arrays WA(6), WB(6), ... WF(6) are
provided for this purpose.
Note that for each node M, only the nonzero components of the
additional load vector need to be calculated. ADINA distinguishes
internally independent, dependent or fixed degrees of freedom and
performs proper summation of the new additional load vector to the
effective load vector.
5.14.3 Example: Hydrodynamic forces
For a cylinder moving with velocity U c (t ) and acceleration

U c (t ) in the direction of an oncoming fluid stream of velocity
U f (x, t ) , the total hydrodynamic force per unit length due to drag
and inertial effects, excluding diffraction effects, can be
approximated by Morisons general formula:
Fc = 0.5 D CD ( U f U c ) U f U c
U f U f
+0.25 D 2 (1 + C A ) + ( U f U c )i (5.14-1)
t x
0.25 D 2 C A U c
where is the density of the fluid, D is the diameter of the

cylinder and CD and C A are the drag and added mass coefficients

of the cylinder.
In the right-hand-side of (5.14-1), the first term represents the
form drag, the second term corresponds to inertial forces due to the
local and convective accelerations of the fluid flowing around the
cylinder and the third term corresponds to the inertial force due to
the motion of the cylinder.
Morisons equation is commonly used for the analysis of
submerged and partially submerged structures such as offshore
rigs, risers, underwater pipes and cables.
More information about Morisons equation and about the drag
and inertia coefficients CD and C A (which depend upon many
parameters, including the section shape and surface roughness of
the cylinder and the Reynolds number) can be found in the
ref. Sarpkaya, T. and Isaacson, M., Mechanics of Wave

Forces on Offshore Structures, Van Nostrand Reinhold
Co., New York, 1981.
In ADINA, the user-supplied load interface has been used to

implement the calculation of the Morison forces in the case of a
sphere suspended at the bottom of a cable, and towed in a still fluid
U f (x, t ) = 0 (see Figure 5.14-1). The model must consist of one
line of elements, with possibly concentrated masses and/or dampers
at the nodes. The nodes must be numbered 1, 2, , N. At the
bottom node N, a sphere generating additional drag and inertia
forces can be added. The following items can be specified: the
length of the cable, the fluid density, the diameter of the sphere, the
cable drag and added mass coefficients and the sphere drag and
added mass coefficients.

Tow speed
Water surface
X
Hydrodynamic drag
Towing cable and inertia forces
Towed sphere
(1) (2)
(1) Initial geometry
(2) Structure submitted to gravity and hydrodynamic forces
Figure 5.14-1: Usage of the user-supplied load subroutines

for the analysis of a sphere towed underwater
The hydrodynamic force on a unit length of the cylinder

(velocity U c , acceleration U c ) is calculated using
Fc = 0.5 D CD U c U c 0.25 D 2 C A U c
where is the density of the fluid, D is the diameter of the

cylinder and CD and C A are the drag and added mass coefficients
of the cylinder.
The hydrodynamic force on the sphere (velocity Us ,

acceleration U s ) is calculated using
2 3
D 4 D
Fs = 0.5 CD U s U s C A U s
2 3 2

where is the density of the fluid, D is the diameter of the sphere

and CD and C A are the drag and added mass coefficients of the
sphere.
Either a static or a dynamic analysis can be performed with

ADINA. In a static analysis, the velocity at the top of the cable
must be entered.
Verification problem B.130 in the ADINA Verification Manual

demonstrates the usage of this user-supplied load.
5.15 Constraint equations

ADINA can impose two basic types of constraints between

nodal degrees of freedom. The first type has the form
U k = jU j (5.15-1)
j
where Uk is a dependent (or slave) degree of freedom which is

controlled by multiple independent (master) Uj degrees of freedom
via factors j. Alternatively, a general constraint can be defined as
U
j
j j =0 (5.15-2)
where none of the Uj degrees of freedom are made dependent

(there are no slave degrees of freedom). In this case, the constraint
is imposed using Lagrange Multipliers.
Note that each constraint of the first type reduces the number of
independent degrees of freedom by one, while each general
constraint of the second type increases the number of degrees of
freedom by one (by adding a Lagrange Multiplier). Hence, the first
type should be used whenever possible. In some cases, however,
one cannot easily express a constraint in the form of Eq. (5.15-1)
where a slave degree of freedom is only related to master degrees

of freedom. See the example below for an illustration.
U 2( z ) = U1( z ) + 31( x ) valid constraint

U 3( z ) = U 2( z ) + U1( z ) 4 2( x ) invalid constraint

This DOF is already dependent
Note that in the above example, the problem with the second
constraint could be resolved by manually applying the first
constraint to obtain
U 3( z ) = 2U1( z ) + 31( x ) 4 2( x )
The mesh glueing feature (Section 5.16) automatically produces

general constraints which are expressed in the form of Eq. (5.15-2).
Displacement constraints can be applied to static and dynamic

analyses.
The basic constraints of Eq. (5.15-1) are only approximately

satisfied in an explicit analysis, since imposing the constraint
exactly requires a non-diagonal mass matrix.
General constraints cannot be used in explicit analysis.
The following comments apply to basic constraints of Eq. (5.15-

1) in which the slave degree of freedom is constrained to more than
one master degree of freedom.
- If a consistent mass matrix is used and concentrated masses

are applied to constrained nodes, then the off-diagonal terms are
included in the global mass matrix.
- Similarly, if concentrated dampers are applied to

constrained nodes, then the off-diagonal terms are included
in the global damping matrix.

5.15: Constraint equations
5.15.2 Rigid links
Rigid links are special constraint equations established

automatically by the program between two nodes a master node
and a slave node.
As the nodes displace due to deformation, the slave node is
constrained to translate and rotate such that the distance between
the master node and the slave node remains constant, and that the
rotations at the slave node are the same as the corresponding
rotations at the master node.
Rigid links can be kinematically linear or nonlinear. For linear

rigid links, the constraint equations are unchanged throughout the
solution and hence they are applicable for small displacement
analysis.
In a large displacement analysis, nonlinear rigid links should be
used recognizing that the effects of large displacements are taken
into consideration in updating the constraint equations for the
master and slave nodes. Hence, the coefficients in Eq. (5.15-1) are
no longer constant.
It is allowed to select the degrees of freedom that are selected

by a rigid link. By default, all master degrees of freedom are
constrained.
Rigid links are useful in the modeling of stiffened shells, for

example, in coupling iso-beam elements (see Section 2.5) to shell
elements (see Section 2.7).
The internal implementation of the tied contact feature

discussed in Section 4.1.1 is conceptually similar to creating rigid
links between the contact surfaces.
When a rigid link that constrains rotational degrees of freedom

is attached to a node of a shell element, 6 degrees of freedom must
be used for the node.
The rotation around the shell node director vector may have
zero stiffness, depending upon the modeling of the other end of the
rigid link (see Figure 5.15-1). There is an option to assign stiffness
to those shell node rotational degrees of freedom with zero stiffness
that are attached to rigid links.

Zero stiffness
for director vector
rotational DOF
Rigid links
Nonzero stiffness
for director vector
rotational DOF
Shell elements
Figure 5.15-1: Rigid links and shell elements
You can use fixities on nodes connected to rigid links, provided

that the fixities not prevent the rigid link from maintaining constant
distance between the nodes.
Nonlinear rigid links cannot be used in conjunction with the

Wilson theta method of time integration.
When a large displacement rigid link is attached to a linear

element or a deformation-independent load, the initial constraint
equations are used.
Rigid links and continuum elements: Care must be used when

connecting continuum elements (2-D and 3-D solid elements) with
rigid links. In order for the rigid link to rotate, the master DOF of
the rigid link must have its rotational DOFs free.
This situation is illustrated in Figure 5.15-2. Two groups of 2-D

solid elements are connected together with large displacement rigid
links. If the x rotational DOFs are fixed, then when the lower
element is rotated, the rigid links maintain both their length and
their angle, and the upper element group distorts as shown in Fig
5.15-2(b). If the x rotational DOFs are free, then when the lower
element is rotated, the rigid links are free to rotate and the upper
element group can also rotate without distortion as shown in Fig
5.15-2(c).

5.15: Constraint equations
By default, the AUI does not fix the rotational DOFs of all nodes
connected to continuum elements. Thus, by default, the model in
Figure 5.15-2 will behave as shown in Fig 5.15-2(c), as expected.
However, this default can be changed using the command
parameter
ADINA FIXBOUNDARY=V89
or
KINEMATICS RIGIDLINK-6DOF=V89
With this option, the AUI fixes the rotational DOFs of nodes
connected to continuum elements and rigid links, and the model in
Figure 5.15-2 will behave as shown in Fig 5.15-2(b). This option is
recommended for use only for backwards compatibility.

S S M = Master node
S = Slave node
M M
M S
Rigid link
a) Two element groups connected with rigid links
Elements must distort
b) Master rotational DOFs are fixed, rigid links cannot rotate
Elements remain undistorted
c) Master rotational DOFs are free, rigid links can rotate
Figure 5.15-2: Rigid links used with continuum elements

5.16: Mesh glueing
5.16 Mesh glueing

The glueing feature is used to attach two surfaces (or lines in
2-D) together. These two surfaces usually involve different finite
element meshes (see Fig. 5.16-1). The glueing procedure should
lead to smooth transition of displacements and tractions between
the glued surfaces. This feature is useful for several applications:
When a fine mesh is desired in a certain region and coarser

meshes are desired in other regions.
When different regions are meshed independently with

unstructured free meshes.
When different regions are meshed with different element

types (such as a tetrahedral mesh attached to a brick mesh).
Mesh glueing along dashed lines
Figure 5.16-1. Examples requiring mesh glueing
The proper glueing constraint between the two surfaces can be

expressed as

(u1 u 2 )d = 0 (5.16-1)

where u1 is the displacement of the first glued surface, u2 is the

displacement of the second surface and is the Lagrange Multiplier
field imposing the constraint.
One of the glued surfaces is designated as the master and the

other as the slave. The Lagrange Multiplier field involves nodal
degrees of freedom at the nodes of the slave surface, and the
integration is also performed over the slave surface. Hence Eq.
(5.16-1) becomes
S
S (u M u S )d S = 0 (5.16-2)
The accurate integration of Eq. (5.16-2) is not trivial since the

displacements uM and uS are generally interpolated over different
domains. This integration is performed by the AUI and results in
multiple general constraints (see Section 5.15.1).
Mesh glueing is not available in explicit analysis.
Mesh glueing is available for the ADINA heat flow solution

options.
Glueing is superior to tied contact and should be used in its

place whenever applicable.
Only 2-D solid elements can be used in the gluing of 2-D lines.
The glued element side can have linear or quadratic displacement
interpolation. Truss, beam and pipe elements are not supported.
Only 3-D solid elements, shell elements and 3-D plane stress
(membrane) elements can be used in the gluing of 3-D surfaces.
The glued element faces can be triangles or quads, and they can
have linear or quadratic displacement interpolation.
For shell elements, only the translational degrees of freedom are

glued, the rotational degrees of freedom are not glued. Also, only
surface-to- surface glueing is supported. Edge-to-surface gluing is
not supported.
It is allowed to glue 3D-solid elements to shell element surfaces.

5.16: Mesh glueing
Nodes on glued surfaces (both master or slave) cannot have

dependent translation degrees of freedom. Therefore, they cannot
be slaves in rigid links or constraint equations involving
translations.
If one glue surface is smaller than the other, the smaller surface
should preferably be the slave. However, the glueing will also
work if the smaller surface is the master. The two glued surfaces
can also be partially overlapping. These cases are shown in Figure
5.16-2.
Master surface
Slave surface
Slave surface
Master surface
Glue surface 2
Glue surface 1
Figure 5.16-2: Examples of master and slave glue surfaces
The two glued surfaces should ideally be smooth surfaces (no

sharp corners). If corners exist it is better to create multiple glued
meshes, as shown in Fig. 5.16-3.
The Lagrange Multiplier field is modified at nodes where

multiple glue meshes intersect. These nodes are called cross-over

points.
One glue-mesh Cross-over points
One glue-mesh
Four glue-meshes
(one for each side)
Figure 5.16-3. Glueing non-smooth surfaces
If the two surfaces have different mesh densities, either one can
be used as slave. However it is recommended that the finer mesh
surface be the slave surface. Using the finer meshed surface as a
slave will produce more equations, since the Lagrange multipliers
degrees of freedom are on the nodes of the slave surface.
The master glue surface can be enlarged beyond its geometric

bounds, so that the slave points that project slightly outside the
master can still be considered glued. This is done via the
EXTENSION parameter in the GLUEMESH command.
The birth and death time option can be applied to glued

surfaces.
If the independent parts of the model have rigid body motions

(not considering the effect of glueing), the equation solver might
stop due to zero pivots. In this case, stiffness stabilization can be
employed (see Section 11.15).
The mesh glue feature admits rigid body translations. Therefore,

when the model is rigidly translated, no spurious stresses and
strains are generated, and force equilibrium is satisfied between the
glued surfaces.

5.16: Mesh glueing
The mesh glue feature only admits rigid body rotations when
the two glued surfaces exactly coincide, that is, when the two glued
surfaces are flat. Curved surfaces do not exactly coincide due to the
discretization error, as illustrated in Figure 5.16-4.
Figure 5.16-4. Glued surfaces do not exactly coincide when the

surfaces are curved
As a result, if the glued surfaces are curved, moment equilibrium is

not satisfied between the glued surfaces, spurious stresses and
strains are generated when the model rotates, and rotational
motions are resisted, just as if the glued surfaces are attached to a
grounded rotational spring. The stiffer the glued surfaces, the stiffer
the rotational spring. This effect is less severe when the coarser
meshed surface is the slave surface of the glue mesh pair because in
this case the surfaces are not as tightly glued (fewer glueing
constraints are created), and hence the distortional motion imposed
by the mesh glue is less severe.
Figure 5.16-5 shows the stress results for a large displacement

static analysis where a steel cylinder of radius 1mm is rotated by 10
degrees by prescribing the rotation on the outside surface of the
cylinder. Other than the prescribed rotation, the cylinder is
completely free. We observe that the exact solution is obtained
when there is no mesh glue and when the glued surfaces are flat,
but spurious stresses are generated when the glued surfaces are
curved, especially when the finer meshed surface is the slave
surface in the mesh glue pair.

Stress = 0 MPa (0% Error) Stress = 0 MPa (0% Error)
No mesh glue Flat glued surfaces
Stress = 8,914 MPa Stress = 469 MPa
Curved glued surfaces. Curved glued surfaces.

Outside cylinder (finer meshed Inside cylinder (coarser meshed
surface) is the glue slave surface) is the glue slave
Figure 5.16-5. Stress results when cylinder is rotated by 10 degrees

6.1 Linearized buckling analysis
6. Eigenvalue problems
ref. KJB The two types of analyses in ADINA for which the solution is
Section 6.8.2 obtained by solving the corresponding generalized eigenvalue
and Chapter 10 problem are:
Linearized buckling analysis

Frequency analysis
These two types of analyses are discussed in the following sections.

The goal of linearized buckling analysis is to estimate a load level

such that the tangent stiffness matrix corresponding to the load
level has a zero eigenvalue.
A brief description of the theory of linearized buckling analysis
is now given. Further information is given in the reference:
ref. K.J. Bathe and E.N. Dvorkin, "On the Automatic

Solution of Nonlinear Finite Element Equations,"
Computers & Structures, Vol. 17, No. 5-6, pp. 871-879,
1983.
ref. KJB The solution in nonlinear static analysis is considered a function

Section 6.1 of time:
t
R is the external load vector at time t ,
t
F is the internal load vector at time t ,
and at equilibrium t R t F = 0 . The tangent stiffness matrix t

is also considered a function of time.
The external load vector t R is assumed to be a function of the
load factor . The linearized buckling analysis solves for .
ADINA has two formulations for linearized buckling analysis:
classical formulation and secant formulation. These formulations
differ in how the external load vector t R depends upon the load

Chapter 6: Eigenvalue problems
factor .
Classical formulation)
For the classical formulation, we assume that
t
R = t1 R (6.1-1)
in which time t1 is the time at the first time step. In other words,
the load vector is assumed to be proportional to the load factor .
We also assume that the stiffness matrix can be broken into
linear and nonlinear parts:
t
= t L + t NL (6.1-2)
and furthermore that the linear part is not time-dependent and the
nonlinear part is proportional to , so that
t
= L + t NL (6.1-3)
Now we seek values of so that the determinant of t , as

constructed using (6.1-3), is equal to 0:
det( L + i NL ) = 0 (6.1-4)
i 1
Equation (6.1-4) can be rewritten using the definition i = :
i
det( t1 i ( t1 t1 NL )) = 0 (6.1-5)
Equation (6.1-5) corresponds to the eigenproblem
t1
i = i ( t1 t1 NL )i (6.1-6)
where i is the buckling mode shape for mode i .

It is clear from the above that the classical buckling formulation

assumes equation (6.1-3). However, although equation (6.1-3)

might be true for certain specialized problems, equation (6.1-3) is
certainly not true in general. On the other hand, if t1 R is chosen
such that i 1 , then i 0 and equation (6.1-5) implies that
det( t1 ) 0 , which means that the load level i 1 corresponds
to buckling. Therefore, the closer t1 R is to the actual buckling
load, the more accurate the classical buckling formulation will be.
Secant formulation:
In the secant formulation, we assume
t
R = t0 R + ( t1 R t0 R ) (6.1-7)
where time t0 is the time at the start of the current analysis and
time t1 is the time at the first time step of the current analysis.
When the current analysis is not a restart analysis, 0 R = 0 , but
t
when the current analysis is a restart analysis, 0 R 0 .

t
We also assume
t
K = t0 K + ( t1 K t0 K ) (6.1-8)
Now we seek values of so that the determinant of t , as

constructed using (6.1-8), is equal to 0:
det( t0 K + i ( t1 K t0 K )) = 0 (6.1-9)
i 1
Again, substituting the definition i = , equation (6.1-9)
i
becomes
det( t1 K i t0 K ) = 0 (6.1-10)
which corresponds to the eigenproblem

t1
i = i t0 i (6.1-11)
We see that the secant formulation is less restrictive than the

classical formulation, since the assumptions of equations (6.1-7)
and (6.1-8) become better as t0 and t1 approach t .
As with the classical formulation, if t1 R is chosen such that
i 1 , then i 0 and (6.1-10) implies that det( t1 ) 0 , which
means that the load level at time t1 corresponds to buckling,
regardless of the choice of t0 .
Comments
The classical formulation is the default.
The classical buckling formulation can compute classical

buckling loads more accurately than the secant formulation,
especially in problems that include initial strains.
Note that linearized buckling analysis must be specified in

ADINA as a static analysis. In addition, only the subspace iteration
method can be used in the eigenproblem solution.
In order to perform a linearized buckling analysis at step n,

where n > 1, the ADINA restart option can be employed. In the
first run, the nonlinear static response is solved for n - 1 steps. In
the restart run, a linearized buckling analysis can be specified to
evaluate the buckling response linearized at step n.
In the eigenvalue extraction, convergence of the eigenvalues i

is measured using
i( k +1) i( k )
SSTOL; i = 1, 2,..., N
i( k +1)
where N is the number of eigenmodes requested, k is the iteration

counter and SSTOL is a user-specified convergence tolerance.
If termination of solution occurs without attaining convergence,

you should rerun the analysis using:
(a) a larger load level for the first time step; and
(b) a larger number of iteration vectors.
A Sturm sequence check is always performed after the subspace

iterations have converged to verify that indeed the smallest
required eigenvalues and the corresponding mode shapes have been
calculated. If this check fails, the program prints an approximation
(p) to the actual critical load factor (1), where 1 < p.
In order to perform a successful analysis, you should rerun the
analysis using (a) and (b) above.
The eigenvalues are extracted in numerically ascending

sequence. If the following eigenvalues are obtained from the
eigenvalue solution:
1 2 ... M < 1 < M +1 ... N
where N is the number of buckling modes requested, then the

following inequalities hold for the critical load factors:
1 < 1 2 ... M
and
M +1 M + 2 ... N < 0
The 1 value obtained should satisfy 1 < 1 < 500 . If this

condition is violated, the program issues an error message and
stops the execution, and you must adjust the applied load level for
the first time step to successfully complete the analysis.
At the completion of the eigenvalue solution, an error measure

for each eigenvalue and eigenvector is calculated using
1
Error ( k ) T ( k ) 2
measure ( qi ) qi
for = 1
mode i ( i )

(k ) 2


where q i( k ) is the vector of the matrix Q k corresponding to i( k )

(see ref. KJB).
With either the classical or secant formulations, the buckling

load should be checked by increasing the load level until 1 1 .
Even with 1 1 , the buckling load might not be correct if the

stiffness matrix is not tangent. It is assumed throughout the
linearized buckling analysis derivations that the stiffness matrix
computed by ADINA is in fact a tangent stiffness matrix. It is also
assumed that the solution tolerances are tight enough that
equilibrium is in fact reached.
6.1.2 Introduction of geometric imperfections
The buckling modes can be used as an initial geometry

perturbation in a subsequent analysis. In a collapse analysis, the
collapse loads can significantly depend on the geometric
imperfections of the structure. A natural way to introduce
appropriate geometric imperfections in a finite element analysis is
to first perform a linearized buckling analysis and then
superimpose on the "perfect" geometry of the finite element model
a coordinate perturbation proportional to the buckling mode shapes.
A collapse analysis can then be performed preferably using the
LDC method on the perturbed model. See Section 11.16 for more
details on geometric imperfections.
The buckling loads should ideally be confirmed by imposing

geometrical imperfections as described above, then running an
LDC analysis. Verification problems B.43 and B.44, and primer
problem 50, show the technique.
6.2 Frequency analysis

The natural frequencies and mode shapes of vibration of the
structural system are calculated using

K i = i2 M i
where K is the system stiffness matrix corresponding to the time of

solution start, M is the system mass matrix corresponding to the
time of solution start, i and i are the angular frequency and
mode shape, respectively, for mode i. Note that the frequencies are
extracted in the eigenvalue solution in numerically ascending
sequence. The eigenvectors are M-orthonormal, that is
iT M i = 1
The effects of prescribed displacements are not included in

K.
Frequencies and mode shapes can be computed in nonlinear

analysis. The nonlinear analysis may contain geometric and
material nonlinearities including contact, and the frequencies
and mode shapes can be obtained at any solution step of the
nonlinear analysis. The frequency solution corresponds to the
eigensolutions of the tangent stiffness matrix and the mass
matrix in the current equilibrium configuration.
Physically, the frequencies and mode shapes in nonlinear
analysis represent the response of the nonlinear structure to an
infinitesimal perturbation about the equilibrium state. That is, it
is the response when the nonlinear structure is pinged about
its deformed configuration and allowed to undergo undamped
vibrations.
To obtain the frequencies and mode shapes of a nonlinear

structure, the analysis must be performed in two runs. In the
first run, the structure is loaded in a nonlinear static or dynamic
analysis. This run can contain one or more solution steps. In the
second run, you restart from the last solution of the first run (or
use ANALYSIS-SWITCH), then perform a frequency analysis.
The frequency analysis is based on the tangent stiffness

matrix and mass matrix at the last converged state. Therefore,
the stiffness and mass matrices include all geometric and
material nonlinearities corresponding to the end of the nonlinear

static or dynamic analysis.
When a frequency analysis is performed on a nonlinear

structure, only those effects that are included in the tangent
stiffness matrix are accounted for, such as:
Geometric nonlinear stiffening effects, e.g. stress stiffening

and centrifugal spin softening.
Contact.
Mesh glue.
The following effects are not included in the tangent stiffness

matrix, and hence are not accounted for in the frequency analysis:
Stiffness contribution due to deformation-dependent pressure
and concentrated follower force loading for large
displacement analysis.
Spin softening effects and the damping terms due to Coriolis

loading. This requires a non-symmetric complex eigen-
solver which is not supported in ADINA.
Prescribed displacements. A stiff spring (1E30 times stiffer

than any DOF in the assemblage) with a load to deform the
spring by the prescribed displacement amount should be
used instead.
When contact is included in frequency analysis, certain

contactor node degrees of freedom are assumed fixed (glued) to the
target segment:
For frictionless contact:
- Contactor nodes whose contact status is OPEN are assumed

free.
- Contactor nodes whose contact status is CLOSED have their

normal DOF fixed (glued) to the target segment in frequency
analysis. The tangential DOF are assumed free (slipping
contact).

For frictional contact:
- Contactor nodes whose contact status is OPEN are assumed

free.
- Contactor nodes whose contact status is STICKING (sliding

velocity less than or equal to EPST) have all their DOF fixed
to the target segment in frequency analysis. In this case, the
contactor node is assumed glued to the target segment.
- Contactor nodes whose contact status is SLIPPING (sliding

velocity greater than EPST) only have their normal DOF
fixed to the target segment in frequency analysis. The
tangential DOF are assumed free (slipping contact).
In frictional contact frequency analysis, if all contactor nodes

should be fixed, the load should be held constant for several time
steps until the contact status is STICKING for all contactor nodes.
Recall that EPST is inversely proportional to the relaxation time for
frictional contact. Hence, the larger EPST, the quicker the contact
status will change to STICKING.
Contact compliance is accounted for in frequency analysis. This

lowers the natural frequencies because the stiffness is reduced.
At the completion of the eigenvalue solution, an error measure

for each eigenvalue and eigenvector is calculated using
1
Error ( k ) T ( k ) 2
measure ( qi ) qi
for = 1
mode i ( i )

(k ) 2

where q i( k ) is the vector of the matrix Q k corresponding to i( k )

(see ref. KJB).
If there are frequencies that are numerically zero, the

eigensolution procedures can still calculate the frequencies and
corresponding mode shapes. Frequencies that are numerically zero
are present in the model if all rigid body modes have not been

suppressed, or if there are input or modeling errors. The number of

rigid body modes in the model is equal to the number of zero
frequencies.
In order to obtain a frequency solution when there are zero
frequencies, you must activate the "allow rigid body modes"
option, otherwise ADINA will stop. This option is also useful if
any part of the model would be insufficiently supported (i.e., have
rigid body modes) if all contact, mesh glueing and generalized
constraints are removed.
When the rigid body mode shift option is used, you can enter
the rigid body mode shift value (which must be negative). If you
instead use the default shift value of 0.0, ADINA calculates it using
the formula
k
RBMSH = f min ii
i =1,..., N m
ii
-3
where f = -10 for the determinant search method or if contact is
-7
present, f = -10 otherwise and N is the number of mass degrees
of freedom in the finite element model.
The frequencies and mode shapes of vibration can be calculated

using the following methods: determinant search, subspace
iteration or Lanczos iteration.
Frequency analysis including potential-based fluid elements can

be performed using only the determinant search or Lanczos
iteration methods. In addition, the frequency analysis cannot have
potential-based fluid elements together with contact, glueing, or
generalized constraints.
6.2.1 Determinant search method
ref. KJB
The use of the determinant search method is usually effective
Section 11.4 when
A narrow banded finite element system is considered.
The total stiffness matrix can be processed in-core.
The determinant search method calculates the smallest

frequencies and corresponding mode shapes of the system.
6.2.2 Subspace iteration method
ref. KJB The Bathe subspace iteration method used in ADINA contains
Section 11.6 all the capabilities of the original basic subspace iteration method
described in
ref. K.J. Bathe, "Solution Methods of Large Generalized

Eigenvalue Problems in Structural Engineering," Report
UC SESM 71-20, Univ. of California, Berkeley, 1971.
plus the acceleration techniques developed more recently:
ref. K.J. Bathe and S. Ramaswamy, "An Accelerated

Subspace Iteration Method," J. Computer Meth. in
Applied Mech. and Eng., Vol. 23, pp. 313-331, 1980.
When the acceleration techniques are employed, the method can

be significantly more effective. These techniques consist of
The use of the Lanczos method to generate the starting

subspace
Vector over-relaxation in the vector iterations
The default number of iteration vectors, q, used is max{2p,

p+8} when p frequencies and associated mode shapes are to be
calculated. For experiences with the method see:
ref. K. J. Bathe, The Subspace Iteration Method

Revisited, Computers & Structures,
j.compstruct.2012.06.02
The subspace iteration can be employed to calculate the smallest

frequencies and corresponding mode shapes, or the frequencies
(and mode shapes) within a certain prescribed interval.
If q iteration vectors are used and there are only q mass degrees
ref. KJB of freedom, the subspace iteration method corresponds to a static
Sections 10.3.1 condensation (Guyan reduction) and subsequent solution of the
10.3.2 and
11.6.3

eigenvalues.
This analysis feature may be useful when you assign zero mass
density to the material, and account for the mass by means of
concentrated masses corresponding to a few degrees of freedom
only. However, note that such mass lumping requires much
experience with the problem to be solved and the finite element
model used.
ref. KJB The standard Lanczos method calculates only approximations to

Section 11.5 the eigenvalues and eigenvectors, and for this reason it is here only
used to generate the starting subspace. If the Lanczos method is
employed, it is usually effective to use a number of iteration
vectors q considerably larger than the number of frequencies
required p, i.e., use q = 2p, with q still much smaller than the total
number of frequencies of the finite element system.
ADINA can perform a Sturm sequence check after the subspace

iterations have converged to verify that the smallest required
eigenvalues and corresponding mode shapes have been calculated.
The cost of the Sturm sequence check is approximately equal to the
cost of a static solution.
6.2.3 Lanczos iteration method
ref. KJB The Lanczos method can be used to calculate the smallest
Section 11.5 frequencies and corresponding mode shapes. The method is very
effective for large problems.
When potential-based fluid elements are present, the Lanczos

method should be used. (For small problems, the determinant
search method can also be used; for large problems, the
determinant search method is ineffective.)
The convergence of the iteration is assessed by the criterion
i( k 1) i( k )
; i = 1,..., N
i( k 1)
where N is the number of eigenmodes requested, k is the Lanczos

iteration stage counter and the value = 109 is used. See ref.

KJB, Section 11.5 for all notation not described here.
After convergence, the error bound q sqi and the physical

K i i2 M i
error estimate are calculated.
K i
ADINA can perform a Sturm sequence check after the Lanczos

When ADINA performs a Sturm sequence check, ADINA also

calculates the M-orthogonal checks
DII = max ( iT ) t0 M ( i ) 1 , DIJ = max ( iT ) t0 M ( j )

1 i n 1 i , j n
i j
to verify that the eigenvectors are M-orthogonal.
The sparse solver is always used as the equation solver in the

Lanczos iteration method.
For potential-based fluid elements, when a relatively large

number of frequencies needs to be calculated, it can be effective,
and sometimes necessary, to use the shifting procedure. The option
to use the shifting procedure is controlled by the parameters
NSHIFT and NSHIFT-BLOCK in the FREQUENCIES command.
ADINA can perform a Sturm sequence check after the Lanczos

For models with potential-based fluid elements, the Sturm
sequence check is always performed.
For potential-based elements, the error measures printed in the

output file should always be checked to make sure that they are not
large. The use of parameter RSHIFT in the FREQUENCIES
command can sometimes be useful to obtain solutions with small
physical error norms, for example RSHIFT=-0.005.

The Lanczos method cannot be used in conjunction with

potential-based fluid elements and any of the following: contact,
glueing or generalized constraints.
6.2.4 Modal stresses
Modal strains, stresses, forces and reactions can be requested in

ADINA. To calculate these quantities, small displacements and
small strains are assumed. Modal displacements and modal stresses
are therefore proportional.
A frequency analysis must be performed first to obtain the

modal displacements. This analysis can follow a materially or
geometrically nonlinear analysis.
In linear analysis, the modal stresses are directly calculated

from the M-orthonormal modal displacements i , where
iT M i = 1 .
In materially or geometrically nonlinear analysis, the modal

stresses for mode i (1 i N) are calculated using the
approximation
m
=
( (U ) )
i
m
i
t
i
i
( )
where i U im are the stresses computed from the perturbed
displacement vector U im = t U + i i , t is the stress vector at
time t, and i is a perturbation factor, controlled by parameter
DUSIZE as follows.
DUSIZE
If DUSIZE is not equal to 0, then i = ; this choice
i
implies U im t U = DUSIZE . Here the symbol is the norm of
the enclosed vector only including translations. DUSIZE is seen to
be the size of the displacement perturbation.
If DUSIZE is equal to 0, then ADINA computes

t
U
i = 0.001 t
where U is the total displacement vector at time t
i
(last step of analysis before frequency analysis). This choice
implies U im t U = 0.001 t U , which is usually quite
reasonable. In the special case in which the total displacement
vector is zero (for example, if the modal stress calculation is
requested at the start of the analysis), then ADINA sets i to 0.001.
The choice i = 0.001 can lead to very large or very small
0.001
displacement perturbations because then U im t U = and,
i
depending upon the units used, i can be very large or very
small. So we recommend that you specify DUSIZE when the total
displacement vector is zero.


7.1 Linear static analysis
7. Static and implicit dynamic analysis

This chapter presents the formulations and algorithms used to solve
static and dynamic problems using ADINA. This includes
convergence checking and the available solvers.
Information about the progress of the solution are always output to

the .out file. A shorter summarized output is provided in the .msg
file or the dialog box.

ref. KJB In linear analysis using ADINA, the finite element system
Sections 8.2.1, equilibrium equations to be solved are
8.2.2 and 8.2.3
KU = R
For the solution of these equations, the following solvers are

available: sparse solver, iterative solver, iterative multigrid solver,
3D-iterative solver, skyline solver. The sparse solver is the default.
The equation solvers assume that the system stiffness matrix is

symmetric.
The equation solvers assume that the system stiffness matrix is

positive definite. This requirement can be summarized as follows:
The Rayleigh quotient
T K
() = T

must be greater than zero for any displacement vector . Since

( ) is equal to twice the strain energy stored in the system (for
T = 1 ), this is equivalent to the requirement that the strain
energy stored in the finite element system when subjected to any
displacement vector must be greater than zero.
Hence, the finite element system must be properly supported,

Chapter 7: Static and implicit dynamic analysis
so that the system cannot undergo any rigid body displacements

or rotations.
It also follows that no part of the total finite element system

must represent a mechanism, see ref. KJB, Fig. 8.7, p. 704, for
such a case.
Nodal point degrees of freedom for which there is no

stiffness must have been deleted. A degree of freedom does not
carry any stiffness if all of the elements connected to the nodal
point do not carry stiffness into that degree of freedom. In this
case the degree of freedom must be deleted using any one of the
boundary conditions.
Note that nodal degrees of freedom which are not connected
to any elements and are not used as slave nodes in constraint
equations are (by default) automatically deleted by the AUI.
The elements joining into a nodal point and contributing to

the stiffness of the nodal point degrees of freedom must all be
defined properly and in a physically correct manner. For
example, if there are program input errors that yield a zero or
negative Young's modulus or an incorrect nodal point
numbering for an element, the stiffness at a system degree of
freedom may be zero or negative.
7.1.1 Sparse and skyline solvers
The main direct solution method in ADINA is a sparse matrix

solver. An active column solver based on the Gauss elimination
procedure (skyline direct solver) is also available, but should only
be used for small problems.
The ADINA sparse matrix solver has been proven to yield

drastic reductions in solution times, being one to two orders of
magnitude faster than the active column solver. In addition, disk
and overall memory requirements are reduced.
The ADINA sparse matrix solver is very reliable and robust and
is generally to be used for most problems in ADINA. The ADINA
sparse matrix solver is the default solver.

The sparse solver allocates memory outside that allocated by

ADINA. Therefore the total memory used is the sum of the
memory reserved by ADINA plus that dynamically allocated by the
sparse solver code.
The sparse solver can be used both in-core and out-of-core. It

should be noted that it is more efficient to run an out-of-core sparse
solver using real (physical) memory than it is to run an in-core
sparse solver using virtual memory. Therefore, if real memory is
limited, we recommend the following guidelines:
1) The memory allocation MTOT for ADINA should be

specified just high enough to allow ADINA to use IOPTIM=3
(refer to the output file for the printout of the IOPTIM value).
2) In the AUI, you should provide a memory limit for the sparse
solver using the following formula
MAXSOLMEM 0.85 x (real memory of your

computer) - MTOT
in which MAXSOLMEM, (real memory) and MTOT are

measured in MW (or MB).
When a non-positive definite stiffness matrix (i.e. one with a

zero or negative diagonal element) is encountered during solution,
ADINA may stop or continue, according to the following rules:
If a diagonal element is exactly equal to 0.0, ADINA stops

unless
- The equation number corresponding to the zero diagonal

element is attached to an inactive element (element
birth/death option is used).
- You have requested that ADINA continue execution (see

below).
If the value of a diagonal element is smaller than 10-12 but

not equal to zero, or the value of a diagonal element is negative,
ADINA stops unless one of the following options is used:

- Automatic load-displacement (LDC)

- Automatic time-stepping (ATS)
- Potential-based fluid elements
- Contact analysis
or unless you have requested that ADINA continue execution.
When ADINA stops, ADINA prints informational messages for

the zero or negative diagonal elements, up to 25% of the number of
equations.
When ADINA continues execution, ADINA assigns a very

large number to the diagonal element, effectively attaching a very
stiff spring to that degree of freedom.
You can request that ADINA continue execution using the

control parameter Continue Even When Non-Positive Definite
Stiffness Matrix Encountered in the AUI ControlSolution
Process dialog box.
Note that the stiffness matrix can be non-positive definite due to

a modeling error, for example if the model is not sufficiently
restrained in static analysis. In this case the results obtained can be
misleading.
7.1.2 Iterative and multigrid solvers
In the analysis of large problems, the amount of storage required

by a direct solution method may be too large for the available
computer resources. For such problems, the use of the iterative
method of solution is necessary.
Iterative solution methods related to those used in ADINA are

described in ref. KJB, Section 8.3.2.
The iterative solver can be used with all solution options of

ADINA, in linear and nonlinear, static and dynamic analysis
(except for frequency and linearized buckling analysis). The
conditioning of the coefficient matrix is generally significantly

better in dynamic analysis because of the effects of the mass matrix

(inertia effects) in the coefficient matrix.
The iterative solver cannot recognize that the stiffness matrix

has a zero pivot. If the stiffness matrix is not positive definite, the
iteration will continue without reaching convergence.
The main practical differences between the use of the direct

solver and the iterative solver are as follows:
The direct solver executes a predetermined number of

operations after which the solution is obtained.
The iterative solver performs a predetermined number of

operations per iteration, but the number of iterations is not
known beforehand. The number of iterations depends on the
condition number of the coefficient matrix: the higher the
condition number, the more iterations are needed.
The number of iterations required varies from a few hundred
to a few thousand. In general, the solution of bulky 3-D solids
requires significantly less iteration than the solution of thin
structural configurations.
Regarding the convergence of the iterative method, assume that

the system of equations to be solved is Ax = b , where x ( k ) is the
approximate solution at iteration k. and the residual vector is
r ( k ) = b Ax ( k ) . The iterative method is converged when either
r(k )
r (k )
EPSIA and 2
EPSIB
2 b 2
or
r(k ) EPSII
2
Note that EPSII should be less than or equal to EPSIA. The

defaults are EPSIA=1E-6, EPSIB=1E-4, EPSII=1E-8. However,

for nonlinear analysis with equilibrium iterations, looser tolerances
can be used, e.g. EPSIA=1E-4, EPSIB=1E-3, EPSII=1E-5.
Difficulties might be encountered when the matrix A is ill-

conditioned. For example, static shell problems are difficult to
solve because shell structures are much stiffer in membrane action
than in bending action. For such cases, a shift factor SHIFT can be
effective, with SHIFT > 1.0, e.g. SHIFT=1.02. SHIFT is used to
scale the off-diagonal entries of A by 1/ SHIFT in order to make
the preconditioner more diagonally dominant.
ADINA also includes a multigrid solver. The multigrid solver

is intended for use with very large systems of equations. The
multigrid solver is efficiently used when the model consists of 3-D
tetrahedral solid elements that were generated using free-form
meshing techniques.
7.1.3 3D-iterative solver
The 3D-iterative solver has been developed to efficiently solve

large models (i.e. problems with more than 500,000 equations)
containing mainly higher order 3-D solid elements.
In addition to the higher order 3-D solid elements, the models

can contain other elements available in the program (e.g., shells,
rods, beams, rebars, etc.), including contact conditions.
The 3D-iterative solver is effective in linear or nonlinear static

analysis and in nonlinear dynamic analysis. For linear dynamic
analysis, the sparse solver is usually more effective.
The 3D-iterative solver, like all iterative solvers, performs a

number of iterations until convergence is reached.
In linear analysis, the 3D-iterative solver convergence tolerance

affects the accuracy of the solution. If the convergence tolerance
(EPSN parameter of the SOLVER ITERATIVE command) is too
loose, inaccurate results are obtained, and if the tolerance is too
tight, much computational effort is spent to obtain needless
accuracy.

In nonlinear analysis, the 3D-iterative solver is used to solve the

linearized equations in each Newton-Raphson equilibrium iteration.
Hence, in this case, the 3D-iterative solver convergence tolerance
only affects the convergence rate, not the accuracy of the solution.
The accuracy of the solution is specified by the equilibrium
iteration tolerances (using the TOLERANCES ITERATION
command).
The benefits of the 3D-iterative solver are:
When the 3D-iterative solver is used, the solution time and

memory requirements scale approximately linearly with the
number of equations. When the sparse solver is used, the
solution time and memory requirements scale approximately
quadratically with the number of equations. Therefore, the 3D-
iterative solver allows the solution of very large problems.
The 3D-iterative solver is significantly more stable than the

multigrid iterative solver; hence, the 3D-iterative solver should
always be used.
The limitations of the 3D-iterative solver are:
The 3D-iterative solver convergence tolerance (EPSN) must

be appropriately set.
The 3D-iterative solver does not scale well for shared
memory parallel processing. However, the sparse matrix solver
scales well to about 8-cores.
Nearly incompressible hyperelastic materials may slow
down the convergence of the 3D-iterative solver. For these
material models, the bulk modulus should be restricted to a
value corresponding to = 0.49, instead of the default 0.499
(see Eq. 3.8-13)
Convergence control in the 3D-iterative solver is as follows:
Considering the linearized equation Ax = b, let xi be its

approximate solution at inner-iteration i and r i (= b Ax i ) be its
corresponding residual. The convergence in the iterative solver is

said obtained if any one of the following criteria is satisfied
b or r i or x i x i1
0.001b i <3
or equation residual: r i rscale
b i3
0.001v i <3
or solution residual: x i x i1 xscale
v i 3
x = 1 1 xi
n
The solution norm is
n
In the above, 1016 , the equation scale rscale = b , and the
variable scale
{
max , 1 x1 + x 2 + x 3
3( ) } for linear problems
xscale =

{
max , 3 ( x + x + x ) , x
1 1 2 3
} for nonlinear problems
where x = t +tU tU , x i = U ( i ) is the current solution

increment in the Newton-Raphson iteration i, b = EPSIB of the
-6 -3
SOLVER ITERATIVE command, and = min(10 , b10 ).
The equation residual (EQ) and the solution residual (VAR) are
available in the .out file.
7.2 Nonlinear static analysis

In nonlinear static analysis the equilibrium equations to be
ref. KJB
Section 8.4 solved are:
t +t
R t +t F = 0
t+t
where R is the vector of externally applied nodal loads at time
t+t
(load) step t+t, and F is the force vector equivalent (in the
virtual work sense) to the element stresses at time t+t.

The nonlinearity may come from the material properties, the

kinematic assumptions, the use of contact surfaces, or the use of
special features such as the element birth/death option.
The solution to the static equilibrium equations can be obtained

in ADINA using
Load/displacement incrementation without iteration
Modified Newton iterations, with or without line searches
The BFGS (Broyden-Fletcher-Goldfarb-Shanno) method
Full Newton iterations, with or without line searches
Automatic step incrementation (automatic-time-stepping and

load-displacement-control methods)
These methods are described in detail in the following sections and

also in Sections 6.1 and 8.4 of ref. KJB.
The same equation solvers are used in nonlinear analysis as in

linear analysis. In addition, a non-symmetric sparse solver is
available.
The automatic time stepping algorithms do not require the

stiffness matrix to be positive definite, thus allowing for the solving
of bifurcation problems.
A special case is reached when the element birth or death option

is used for some elements. In this case the equation solver allows
that during the solution a degree of freedom may not carry any
stiffness at a particular time, because the adjoining element(s) may
not have been born yet (or may have died already). Note that the
program does not check if any load is applied in that degree of
freedom, and that the appropriate time functions should be used to
generate/delete that load according to the element birth or death.
The stiffness stabilization feature can be used to treat some

nonlinear static problems involving a non-positive definite stiffness

matrix. See section 11.14 for details.
7.2.1 Solution of incremental nonlinear static equations
Load/displacement incrementation without iteration: The

following algorithm is used if equilibrium iteration is not
performed:

KU = t +t R t F
t +t
U =tU + U

where K is the tangent stiffness matrix at time , where t;
t+t
R is the externally applied load vector at time t+t; tF is the
consistent nodal force vector corresponding to the element stresses
t
due to the displacement vector U, and U is the incremental
displacement vector.

Note that K is the stiffness matrix corresponding to the last
stiffness reformation. Stiffness reformations are performed only at
the solution steps that you specify.
Since equilibrium iteration is not performed, the solution in
general does not satisfy accurately nodal point equilibrium. Hence,
in practice this method should only be used with a sufficiently
small time step increment so that divergence from the equilibrium
solution is small. Whether nodal point equilibrium is sufficiently
satisfied can be approximately checked by performing equilibrium
iterations at the final solution step and noting how many iterations
are required.
Modified Newton iteration: In modified Newton iteration the

following algorithms are used:
Without line search

K U (i ) = t +t R t +t F (i 1)
t +t
U (i ) = t +t U (i 1) + U (i )
(7.2-1a,b)

With line search

K U (i ) = t +t R t +t F (i 1)
(7.2-1c,d)
t +t
U (i ) = t +t U (i 1) + (i ) U (i )
t+t
where K is the tangent stiffness matrix at time , where t; R
t+t (i-1)
is the externally applied load vector at time t + t; F is the
consistent nodal force vector corresponding to the element stresses
+t (i-1) (i)
due to the displacement vector t U ; U is the incremental
(i)
displacement vector in iteration (i) and is an acceleration factor
obtained from line search.

Note that K is the stiffness matrix corresponding to the last
stiffness reformation. Stiffness reformations are performed only at
the solution steps that you specify.
Modified Newton iterations are performed only at the solution

steps that you specify.
ref. KJB BFGS method: In the BFGS matrix update method, the following
Section 8.4.2 algorithm is used:
t +t
K *(i 1) U ( i ) = t +t R t +t F (i 1)
t +t
U (i ) = t +t U ( i 1) + (i ) U (i )
t+t *(i-1)
where K is an updated stiffness matrix (based on the
iteration history). The definitions of the other variables are as for
Eq. (7.2-1).
t+t *(i-1)
Note that K is not explicitly formed, but instead the
inverse of the stiffness matrix is updated using vector products to
provide a secant approximation to the stiffness matrix in successive
iterations.
BFGS matrix updates are performed only at the solution steps
that you specify.
Full Newton iterations: In the full Newton iteration method, the

algorithms employed without and with line searches are:

Without line search
t +t
K (i 1) U ( i ) = t +t R t +t F (i 1)
t +t
U (i ) = t +t U ( i 1) + U ( i )
With line search
t +t
K (i 1) U ( i ) = t +t R t +t F (i 1)
t +t
U (i ) = t +t U ( i 1) + (i ) U (i )
where t +t K ( i 1) is the tangent stiffness matrix based on the

solution calculated at the end of iteration (i - 1) at time t+t.
The definitions of the other variables are as for Eq. (7.2-1).
Note that when the full Newton iteration method is employed a
new stiffness matrix is always formed at the beginning of each new
load step and in each iteration.
An upper bound for the incremental displacements in U can be

set by the user. If the largest increment displacement component
exceeds the limiting value, the whole incremental displacement
vector is scaled down to satisfy the upper bound.
This feature is useful for problems where one or more iterations
can produce unrealistically large incremental displacements. This
may happen, for example, if a load is applied to contacting bodies
before contact is properly established, or in the first unloading steps
after a material has undergone plastic deformation.
7.2.2 Line search
When the line search feature is activated, the incremental

displacements obtained from the solver are modified as follows
t +t
U (i ) = t +t U (i 1) + (i ) U (i )
where is a scaling factor obtained from a line search in the

direction U(i) in order to reduce out-of-balance residuals,
according to the following criterion

T
U (i ) t +t R t +t F (i )
T
STOL (7.2-2)
U (i ) t +t R t +t F (i 1)
where STOL is a user-input line search convergence tolerance, and

t+t (i)
F is calculated using the total displacement vector t+tU(i).
The magnitude of is also governed by the following bounds
LSLOWER LSUPPER (7.2-3)
where LSUPPER and LSLOWER are user-input parameters.
The incremental displacements are not modified (i.e., = 1) if no

suitable line search parameter satisfying Equations (7.2-2) and
(7.2-3) is found within a reasonable number of line search
iterations, or if the unbalanced energy falls below a certain user-
specified energy threshold ENLSTH.
Line search is off by default except for the BFGS method. It is

useful for problems involving plasticity, as well as large
displacement problems involving beams and shells. It is also
helpful in many contact problems. In the case of contact problems,
it is sometimes better to set LSUPPER to 1.0 so that the line search
only scales down displacements.
The effect of line search is more prominent when the current

displacements are far from the converged solution. This usually
happens in the first few iterations of a time step, or when a major
change occurs in the model, due for example, to contact
initiation/separation, or the onset of plasticity.
Note that line search increases the computational time for each
iteration. Most of the extra time goes towards the evaluation of
t+t (i)
F in Equation (7.2-2). However, for the types of problems
mentioned above the reduction in the number of iterations and the
ability to use bigger time steps leads to an overall reduction in
solution time.

7.2.3 Low speed dynamics feature
A low speed dynamics option is available for static analysis

(can only be used with ATS or TLA/TLA-S features). Low speed
dynamics is a special technique developed to overcome
convergence difficulties in collapse, post-collapse and certain
contact problems.
In essence, this feature includes dynamics effects in an
otherwise static problem. ADINA solves
M t +t U (i ) + C t +t U ( i ) + t +t K (i 1) U (i ) = t +t R t +t F (i 1) (7.2-4)
where M is the mass matrix and is a mass scaling factor that can
vary from 0.0 to 1.0 (default 1.0), to partially account for the
dynamic inertia effect . The C matrix is evaluated using
C=K
where is a user-specified parameter (default 10-4) and K is the

(initial) total stiffness matrix (corresponding to zero initial
displacements). For details on this dynamic equation refer to
Section 7.4.
For low speed dynamics with the group-based damping option,

the standard Rayleigh damping parameters and are used to
evaluate the C matrix, i.e.,
C = M+ K
When the low-speed dynamics option is used with ATS, the

time step size will influence the results. It is recommended that the
time step size be at least 105 . Or, it is recommended that the
loads be held constant for a period of time of at least 10 so that
the dynamic effects die out.
Notice that when C = K , rigid body motions are not damped

out. In order to achieve damping of rigid-body motions,
C = M + K must be used.

Note that mass effects may not be needed in the low speed
dynamic analysis. In this case, set the parameter of Eq. (7.2-4) to
zero, or, alternatively, set the material densities to zero. That way,
only structural damping effects will be present in the otherwise
static analysis.
When there are potential-based fluid elements in the model, the

low-speed dynamics option should only be used with the group-
based damping option, and there should be no damping assigned to
the potential-based fluid element groups.
ADINA outputs the damping and inertia force norms at every

time step (see Section 7.2.5 for more information on these norms).
The damping and inertia forces should be compared with the
external forces in order to ensure that they are not excessive.
7.2.4 Automatic-Time-Stepping (ATS) method
The automatic-time-stepping (ATS) method controls the time

step size in order to obtain a converged solution. If there in no
convergence with the user-specified time step, the program
automatically subdivides the time step until it reaches convergence.
In some cases, the time step size may be increased to accelerate the
solution.
The user sets the division factor that ADINA uses to subdivide
the time step when there is no convergence. Successive
subdivisions can be performed, if necessary, provided that the time
step size is not smaller than a minimum value. This minimum value
is set as the original step size divided by a scaling factor provided
by the user.
Note that the loads at any intermediate time instant created by

the ATS method are recalculated based on the current value of the
time functions.
The solution output is only furnished at the user-specified times,

except when the solution is abandoned due to too many time step
subdivisions without convergence. In this case, the solution output
is also given for the last converged time instant.

There are three options for controlling the time step size once
convergence is reached after ATS subdivisions. You can select any
of these options, or request that ADINA does the selection.
1. Use the time increment that gave convergence
In this case, the program continues the analysis using the last
converged time step. Once the end of the user-specified time
step is reached, the program may increase the time step based on
the iteration history. This option is the default in analyses
without contact. Corresponding command input: AUTOMATIC
TIME-STEPPING RESTORE=NO
2. Return to original time step size
In this case, the program continues the analysis using the
original user-provided time increment. This option is the default
when contact is present. Corresponding command input:
AUTOMATIC TIME-STEPPING RESTORE=YES
3. Proceed through user-defined time points
In this case, the program sets the time step size such that the
final time is that initially provided by the user. Hence, the
analysis always proceeds through all user-specified time steps.
Corresponding command input: AUTOMATIC TIME-
STEPPING RESTORE=ORIGINAL
The program automatically switches to option 3 if, when option 2
is specified, one of the following features is used: Bathes
composite method, substructuring, FSI, TMC, or mapping.
The ATS method can also increase the time step beyond the
user-specified value if the iteration history shows that such an
increase is useful. This is only possible in a static analysis without
low-speed dynamics. The increase in time step cannot be larger
than a user-specified factor. Due to this increase, the analysis may
be completed in fewer time steps than requested. This time step
increase is only possible when the ATS subdivision is set to
Return to original step size. Note that if time step blocks are
used, then the ATS method does not increase the time step size.
Following is an example to illustrate the basic options of ATS

subdivisions. Assume we are in load step 15 of a particular problem
with initial time t = 15.0 and a time step of 1.0. The solution does
not converge and the time step is set to 0.5 (assuming a time step

division factor of 2.0). If that too does not converge, the time step
is set to 0.25. If that converges, the results are not yet saved.
Another sub-step is performed for load step 15. The size of this
step depends on which of the three options above is selected:
In option 1, the next sub-step will use a time step size of
0.25. Two other sub-steps will be performed within load step 15
both of size 0.25 (assuming they all converge).
In option 2, the next sub-step will use a time step size of 1.0.
If this converges load step 16 starts with t = 16.25.
In option 3, the sub-step will use a time step size of 0.75. If
this converges load step 16 starts with t = 16.0.
Note that while options 1 and 2 may result in outputted solution

times that are different from those initially specified by the user,
there are certain time values that are not skipped. These are the
time values at the end of time step blocks. In this case, the time
step size is reduced such that the solution time at the end of the
block is satisfied. The program determines these time step blocks
based on the time step pattern input by the user. The final solution
time is always assumed to be an end of a time step block.
Option 2 is useful for contact because of the highly nonlinear

response (sudden changes in stiffness) that occurs when a nodes
comes into contact, or is released from contact, or even moves from
one contact segment to another. A small step size may sometimes
be needed only in the vicinity of this contact incident. Once contact
is established/released the problem is less nonlinear and the
original time step size can be used.
Whenever the ATS method is used, the program always checks

if a time function point could be skipped for a time step used, and
makes an appropriate adjustment to the magnitude of the time step
if necessary. Let us consider the time function shown in Fig. 7.2-1.

F(t)
1.0

0 0.5 1.0 1.5 t
Figure 7.2-1: Time function illustrating ATS program checking
With a time step equal to 0.2, assuming that convergence is

obtained, the program will provide solutions for times t = 0.2 and t
= 0.4. If the time step of 0.2 is used again, then the maximum time
function value at time t = 0.5 will be missed. Therefore, the
program uses two substeps of value 0.1 to get to time t = 0.6.
The above option is not available for substructures. In the case

of substructures, a constant time stepping is used (ATS is only
active when there is no convergence).
For dynamic nonlinear analysis, the ATS method can also be

used to obtain a more accurate solution. The accuracy checking is
performed after iteration convergence in each step, using the
accuracy criterion
t t
t+ t+
2
U im 2
U ih DISTOL
max
in which i represents each translational degree of freedom,
t+
t
2
U im =
(U t
i + t +tU i )
2
t
t t t 2 t
( U i + t +tU i )
t+
2
U ih = U i +
t
Ui +
2 16
t
t+ t
and 2
U ih is the displacement evaluated at time t + .
2

Note that all static and dynamic nonlinear analyses, except those
only possible with the load-displacement-control (LDC) method,
can be carried out with the ATS method.
The ATS method cannot be used in linearized buckling analysis,

mode superposition analysis, response spectrum analysis or
dynamic analysis with the Wilson theta method.
7.2.5 Total Load Application (TLA) method and Stabilized TLA (TLA-S)
method
The Total Load Application method is useful for nonlinear static

analysis problems where all applied loads do not require the user to
explicitly specify the time step sequence. In this case, the user
applies the full load value and ADINA automatically applies the
load through a ramp time function, and increases or reduces the
time step size depending on how well the solution converges. This
method cannot be used in dynamic analysis.
The TLA method automatically activates the following features

that are suitable for this type of uniform loading static problems:
- The first time step has a size of 1/50 of the total time.
- Maximum number of equilibrium iterations is 30.
- Line search is used.
- Limiting incremental displacements per iteration is set to
5% of the largest model dimension.
- The maximum number of time step subdivisions is set to 64.
- The time step cannot be increased more than 16 times its
initial size.
Most of the TLA settings listed above can be modified by the user
(via the AUTOMATIC TOTAL-LOAD-APPLICATION
command).
The Stabilized TLA method (TLA-S) is identical to the regular

TLA method with the addition of various stabilizing features to
create a more stable system and aid convergence. The sources of
stabilization are low speed dynamics which adds inertia and

stiffness proportional damping (see Section 7.2.3), contact damping

(see Section 4.7.6), and stiffness stabilization (see Section 11.14).
Indicators are provided in the output file after each converged
solution to show if the forces due to the various stabilization effects
are excessive.
The following solution indicators are printed:
- external forces
I EF = R iU
where R is an external load vector, and U is a

displacement vector in a given step.
- damping forces
I DF = R D iU
where RD is a damping force vector.
- bolt forces
I BOLT = R BOLT iD BOLT
where RBOLT are bolt forces and DBOLT are bolt

shortenings.
- inertia forces
I I = R I iU
where RI are inertia forces.
- contact damping forces
I CD = R CD iU
where RCD are contact damping forces.

- stiffness stabilization
( )
I SST = K iU iU
where K is the extracted stabilization part from the stiffness

matrix.
- drilling forces (applicable to shell elements only)
I DR = R DR iU
where RDR are drilling forces.
Note that during bolt iterations, all indicators are referenced against
the bolt force indicator. For step-by-step solutions, external force
indicators are used for references.
The TLA-S method can serve several purposes. If the stabilization

indicators are all within reasonable bounds, typically less than 1%
of the external force indicator, then the TLA-S solution may be
reasonably accurate. However, even when the indicators are large,
the TLA-S method provides a useful approximate solution that can
frequently be used to detect modeling errors such as incorrect
contact definition, applied load, or boundary conditions.
The following features cannot be used with the TLA and TLA-S
methods:
- All materials with creep effects (applies to elastic and
plastic creep materials)
- All materials with viscosity effects (visco-elastic material
and Anand solder material)
- User-supplied materials
- Rigid target contact
7.2.6 Load-Displacement-Control (LDC) method
The load-displacement-control (LDC) method (arc length

method) can be used to solve for the nonlinear equilibrium path of
a model until its collapse. If desired, the post-collapse response of
the model can also be calculated. The main feature of the method

is that the level of the externally applied loads is adjusted

automatically by the program.
The LDC method can only be used in nonlinear static analysis
in which there are no temperature, strain-rate, pipe internal pressure
or creep effects. The LDC method can be used in contact problems.
The formulations used in the LDC method in ADINA are

described in ref. KJB Section 8.4.3 and in the following reference:
ref. Bathe, K.J. and Dvorkin, E.N., "On the Automatic

Solution of Nonlinear Finite Element Equations," J.
Computers and Structures, Vol. 17, No. 5-6, pp. 871-
879, 1983.
The LDC method is activated via AUTOMATIC=LDC in the

MASTER command. An enforced displacement on a user-
specified degree of freedom is used to evaluate the initial load
vector, and analysis continues until a specified displacement is
reached at a certain node, or a critical point on the equilibrium path
is reached. The LDC method may also be used as a restart analysis
following a static or dynamic analysis. Variants of the LDC
method are commonly referred to as arc-length methods.
The equations employed in the equilibrium iterations are
t +t
K (i 1) U (i ) = ( t +t ( i 1) + ( i ) ) R + R P t +t F (i 1)
t +t
U ( i ) = t +t U (i 1) + U (i ) (7.2-5)
f ( (i ) , U ( i ) ) = 0
where
t +t
K ( i 1) = tangent stiffness matrix at time t+t, end of
iteration (i-1)
R = constant reference load vector
R P = load vector from previous solution run

t +t ( i 1)
= load scaling factor (used on R) at the end of
iteration (i-1) at time t+t

( i ) = increment in the load scaling factor in iteration

(i)
The quantities t +t F ( i 1) and U ( i ) are as defined for Eq. (7.2-1).

Note that in Eq. (7.2-5), the equation f = 0 is used to constrain
the length of the load step. The constant spherical arc length
constraint method is usually used and the constant increment of
external work method is used if the arc length method has
difficulties to converge.
The reference load vector R is evaluated from all the
mechanical loads (except user-supplied and prescribed
displacement loadings).
In a restart LDC analysis, the load vector from the previous

solution run R P may be held constant. This is achieved by
requesting, in the first run, that the load vector be transferred to the
restart file. This feature is useful, for example, to model varying
loads in the presence of a constant gravity load. In the first run,
enter only the gravity loading and request the load transfer. Then,
in the restart run, add the varying loads using the LDC method.
To start the LDC method, the load multiplier for the first step
(t), (used to obtain the corresponding load vector t R) is
calculated using a user-specified prescribed displacement. This
prescribed displacement (see Fig. 7.2-1) acts in the specified
direction at the specified node. The direction of the displacement is
given by its sign.
Note that the input for the initial prescribed displacement (in
particular whether it is positive or negative) is critical in
establishing successive equilibrium positions using the LDC
method.
As an example, two entirely different solution paths will be
obtained for the same model shown in Fig. 7.2-1 if initial
displacements of different signs are prescribed for the first solution
step.
After the first step, the program automatically traces the

nonlinear response by scaling the external load vector R
proportionally, subject to the constraint of Eq. (7.2-5), so that at

t (i)
any discrete time t in iteration (i), the external load vector is R.
t
The scaling of the reference load vector using is analogous to
the scaling of the applied loads R using a user predefined time
function when the LDC method is not used (see Chapter 5). In the
case of the LDC method, the scaling function is determined
internally by the program instead of being user-specified.
R
0
Y
Reference load = R, actual load at time t = t%R

Prescribed displacement for first step = 0, displacement at time t = t
a) Model considered
t%R 0 specified positive
t
0 specified negative
b) Equilibrium paths
Figure 7.2-2: Example of the dependence of solution path on the

displacement prescribed in the first step for the LDC
method
The converged displacement must satisfy the following relation:
U 2 100 t
U
2
where U = t +t U t U is the incremental displacement vector for

the current step, is a displacement convergence input factor and

t
U is the displacement vector obtained in the first step. If the
above inequality is not satisfied, an internal restart of the iteration
for the current step is performed by the program.
The LDC solution terminates normally when any one of the

following conditions is satisfied:
The maximum displacement that you specify is reached.
A critical point on the equilibrium path has been passed and

you request solution termination.
The number of converged solution steps is reached.
The incremental solution has been attempted a number of

times as specified by input (default is ten) from a calculated
equilibrium configuration using different strategies but each
time the solution has failed to converge within the number of
allowed iterations.
7.2.7 Convergence criteria for equilibrium iterations
The following convergence criteria can be specified in ADINA:
energy only,
energy and force/moment,
energy and translation/rotation,
force/moment only, and
translation/rotation only.
If contact surfaces are defined in an analysis, then the contact

force convergence criterion is always used in addition to the above
criteria (see Chapter 4).
The values of all convergence norms, whether used or not, are

provided the .out file. For more details on the .out file format see
Section 7.2.9.

When ADINA runs in interactive mode, graphical output of all

convergence norms is also provided.
LDC method not used
If the LDC method is not used, the convergence in equilibrium

iterations is reached when the following inequalities are satisfied:
Energy convergence criterion: For all degrees of freedom
T
U (i ) t +t R t +t F (i 1)
T
ETOL (7.2-6)
U (1) t +t R t F
where ETOL is a user-specified energy convergence tolerance.
Force and moment convergence criteria: For the translational

degrees of freedom
t +t
R t +t F (i 1)
2
RTOL
RNORM
For the rotational degrees of freedom
t +t
R t +t F (i 1)
2
RTOL
RMNORM
where RTOL is a user-specified force convergence tolerance,

RNORM and RMNORM are user-specified reference force and
moment norms. If left undefined the program automatically
determines RNORM and RMNORM during execution.
Translation/rotation convergence criteria: For the translational

degrees of freedom
U (i )
2
DTOL
DNORM

U (i )
2
DTOL
DMNORM
where DTOL is a user-specified force convergence tolerance,

DNORM and DMNORM are user-specified reference
displacement and rotation norms. If left undefined the program
automatically determines DNORM and DMNORM during
execution.
Note that in each of these convergence criteria the residual norm

is measured against a user-specified maximum residual value; for
example, the force criterion could be interpreted as
(norm of out-of-balance loads) RTOL RNORM
where RTOL RNORM is equal to the user-specified maximum

allowed out-of-balance load.
Note also that these convergence criteria are used in each
subdivision of time or load step when the ATS method of
automatic step incrementation is used.
If contact surface groups are present in the analysis the

following additional criterion is always used in measuring
convergence
(
max R c(i 1) R (ci 2) , (i ) ( i 1)
2 2
) RCTOL
(
max R c(i 2) , RCONSM
2
)
where R (ci 1) is the contact force vector at the end of iteration
( i 1) , (i ) is the incremental Lagrange multiplier vector at the
end of iteration i., RCONSM is a reference contact force level used
to prevent possible division by zero and RCTOL is a user-specified
contact force convergence tolerance.
Non-convergence: Convergence might not occur when the

maximum number of iterations is reached or when the solution is
diverging.

If the specified convergence criteria are not satisfied within the

allowed number of iterations, but the solution is not diverging, the
following can be attempted:
Check the model according to the guidelines in Section

7.2.6.
Use a smaller time step.
Increase the number of allowable iterations.
Change the ATS parameters.
Change convergence tolerances. In most cases, looser

tolerances help. However, in some problems, tighter tolerances
help by not allowing approximate solutions that could
potentially prevent convergence in future time steps.
Change line search. Some problems, such as those involving

plasticity, perform better with line search.
Change contact settings. The optimal contact settings and

features depend on the model. See Chapter 4 for more details.
Divergence of solution terminates the iteration process before

the maximum number of iterations is reached. It is sometimes
detected when the energy convergence ratio in Eq. (7.2-6) becomes
unacceptably large, or when the excessive displacements lead to
distorted elements and negative Jacobians. The program will
repeat the iteration number at which divergence occurs. In this
case, the following can be attempted:
Check the model according to the guidelines in Section

7.2.6.
Use a smaller time step.
Make sure there are sufficient constraints to remove rigid body
modes from all components in the model. Presence of rigid body
modes usually results in a large ratio of maximum to minimum
pivot during factorization (with sparse solver).

LDC method used
Convergence in the equilibrium iterations is reached when the

following inequalities are satisfied:
T
U (i ) t +t (i 1) R t +t F (i 1)
T
ETOL
U (1) (1) R
where ETOL is a user-specified energy convergence tolerance.

degrees of freedom
t +t
(i 1) R t +t F ( i 1) 2
RTOL
RNORM
t +t
(i 1) R t +t F ( i 1) 2
RTOL
RMNORM
where RTOL is a user-specified force convergence tolerance,

RNORM and RMNORM are user-specified reference force and
moment norms. If left undefined the program automatically
determines RNORM and RMNORM during execution.
The translation/rotation convergence criteria, and the contact

convergence criterion, are the same as when the LDC method is not
used, see above.
Nonconvergence: If convergence has not been reached from an

established equilibrium configuration after the maximum restart
attempts, the program saves the required restart information and
program execution is terminated.

The solution can be continued by performing a restart run. Note

that in this case the LDC method must be used in the restart run. A
different value for the initial displacement can be prescribed at a
different nodal point in the first step of the restart run. The
prescribed initial displacement then corresponds to a displacement
increment from the last converged equilibrium position, that is, at
the time of solution start for the restart analysis.
7.2.8 Selection of incremental solution method
This section gives recommendations on which incremental

solution method to use for a given analysis.
Every nonlinear analysis should be preceded by a linear

analysis, if only to check that the model has been set up correctly.
The linear analysis results will tell whether proper boundary
conditions are imposed, all degrees of freedom without stiffness
have been deleted, and so on (see Section 7.1), and the finite
element mesh is adequate.
In practice, it may be best to set up the complete nonlinear
model, but only perform a one load step analysis without
equilibrium iterations. The displacement results obtained from this
solution are linear analysis results and can frequently be used to
establish a reasonable load incrementation for the nonlinear
solution.
If the use of a sufficient number of load steps and equilibrium

iterations with tight convergence tolerances at each load step is
considered to yield an accurate solution of the model, then the
basic aim is to obtain a response prediction close to this accurate
one at as small a solution cost as possible.
If you are not at all familiar with the nonlinear response of the
structural model considered, it may be advisable to first perform a
nonlinear analysis without equilibrium iterations, but using a
sufficiently large number of load steps. The calculated response is
then not necessarily very accurate, but at least you now have a
good idea of the nonlinear response of the model. The next
objective must be to ensure that the analysis results are close
enough to the accurate solution of the model. This can be achieved
by performing the analysis again with more load increments and by

the use of equilibrium iterations.
If the cost of solution without equilibrium iterations seems

reasonable and reasonable nonlinearities per step were observed, it
is usually best to perform the next analysis with equilibrium
iterations using the default iteration technique, the full Newton
method, and default tolerances of the program.
It is helpful to know if the model softens or hardens under

increasing load. Structures can soften due to the spread of
plasticity, and they can soften or harden due to geometric nonlinear
effects. Contact usually leads to hardening. Fig. 7.2-3 shows some
examples.
Displacement controlled loading generally converges faster than

force controlled loading. For example, in the model shown in Fig.
7.2-3(b), applying an increasing tip displacement to the beam will
converge faster than an applied load P, and both will follow the
same load displacement curve. For the model in Fig. 7.2-3(c) force
control would fail past the local maximum on the load-
displacement curve. Displacement controlled loading (apply an
increasing ) would work in this case. Note that this case is also
suitable for the LDC method.
The use of the full Newton iteration method and a reasonable

load incrementation is frequently sufficient to ensure that an
accurate solution of the response of the model will be obtained.
However, considerations for the use of the other solution
techniques available in ADINA are given below.
The modified Newton iteration is less expensive (per

iteration) than the BFGS method, which is, in turn, less
expensive per iteration than the full Newton iteration.
However, the more expensive method (per iteration) may
require less iterations to achieve convergence. Line searches
help in achieving convergence but also add to the cost per
iteration.


(a) Softening problem. Materially-nonlinear-only analysis,

elasto-plastic analysis of a cylinder
P
P
(b) Stiffening problem. Large displacement nonlinear elastic

analysis of a cantilever
P
P
(c) Softening/stiffening problem. Large displacement analysis of a

thin arch
Figure 7.2-3: Different types of nonlinear analyses

Of the methods available with line searches, modified

Newton iteration should be used for mild nonlinearities; the
BFGS method is effective with somewhat larger nonlinearities,
and full Newton iteration is most powerful. Hence, when
convergence difficulties are encountered full Newton iteration
with line searches should be used.
The ATS method can be used with the modified Newton

iteration, the BFGS method and the full Newton method. If a
reasonable number of time (or load) steps is used for the nonlinear
response, then the use of the ATS method will result in almost the
same "iteration path" as when not using the method. Namely, no
step subdividing will be performed if convergence is always
directly reached at the user-specified load levels.
Hence, in general, if you want to use the most convenient method,
it is recommended to use a reasonable number of load steps with
the default iteration scheme (full Newton method), with default
tolerances and the ATS method. However, a less expensive
solution can frequently be reached by use of the other available
iteration methods.
The LDC method is very useful if collapse of the structure

occurs during the (static) solution. However, the use of this
algorithm can be very costly. The LDC method is recommended if
you do not want to specify the load increments for the solution
period and computational costs are not of primary concern. The
LDC method allows also the calculation of the post-collapse
response. Note, however, that the displacement solution at a
specific load level cannot be obtained using the LDC method
because the load increments are calculated by the program.
Note that usually it is quite adequate to employ the energy

convergence tolerance only. The need to use the force/moment
criteria and/or translation/rotation criteria arises when the energy
convergence is not tight (small) enough. In addition, there exist
special loading conditions under which the denominator of the
inequality (7.2-6) in Section 7.2.7 becomes small and hence the
inequality is difficult to satisfy.

942
7.2.9 Example
STEP NUMBER = 4 ( TIME STEP = 0.20000E-03 SOLUTION TIME = 0.80000E-03 )

DIAG ELEMENT (WITH MAX ABS VALUE) OF THE FACTORIZED MATRIX = 0.25547E+10 NODE = 740 DOF = Z-TRANSLATION
DIAG ELEMENT (WITH MIN ABS VALUE) OF THE FACTORIZED MATRIX = 0.58413E+02 NODE = 161 DOF = X-ROTATION
I T E R A T I O N
METHOD STEP-NUMBER SUBINCREMENT TIME STEP SOLUTION TIME INITIAL ENERGY
*ATS* 4 1 0.200000E-03 0.800000E-03 0.223798E+04
SOLUTION TIME (SECONDS)

I N T E R M E D I A T E P R I N T O U T D U R I N G E Q U I L I B R I U M I T E R A T I O N S
T I M E
OUT-OF- NORM OF CONVERGENCE RATIOS CONVERGENCE RATIOS OUT-OF-BALANCE LOAD
BALANCE OUT-OF-BALANCE NORM OF INCREMENTAL FOR OUT-OF-BALANCE FOR INCREMENTAL VECTOR CALCULATION
ENERGY FORCE MOMENT DISP. ROTN. CFORCE ENERGY FORCE DISP. CFORCE BETA RATIO
NODE-DOF NODE-DOF NODE-DOF NODE-DOF CFNORM MOMENT ROTN. (ITERNS)
MAX VALUE MAX VALUE MAX VALUE MAX VALUE
L O G
COMPARE WITH COMPARE WITH
ETOL RTOL DTOL RCTOL
1.00E-06 1.00E-02 (NOT USED) 1.00E-03
ITE= 0 2.24E+03 2.32E+06 3.27E-03 4.79E-03 7.45E-01 4.74E+03 1.00E+00 2.32E+05 6.92E-02 1.15E+00
740-Z 319-Y 159-Z 239-Y 4.11E+03 3.27E-04 8.63E-01
NUMBER OF SUBINCREMENTS IN THIS TIME STEP =

-2.32E+06 -6.22E-04 -1.14E-03 1.64E-01
ITE= 1 3.95E+00 1.96E+04 9.93E+00 6.16E-04 2.03E-01 4.74E+03 1.77E-03 1.96E+03 8.90E-03 9.01E-01 9.99E-01 -5.95E-08
1
239-X 39-Y 239-X 279-Y 5.26E+03 9.93E-01 2.35E-01 ( 3)
4.14E+03 -2.37E+00 1.52E-04 6.71E-02
STIFFNESS REFORMED FOR EVERY ITERATION OF THIS STEP

ITE= 2 2.67E-01 5.89E+03 4.45E+00 1.29E-04 5.98E-02 4.54E+02 1.19E-04 5.89E+02 1.86E-03 9.06E-02 4.84E-01 2.85E-06
740-Z 199-Y 42-X 279-Y 5.01E+03 4.45E-01 6.92E-02 ( 5)
-1.35E+03 -1.16E+00 -1.67E-05 -1.95E-02

ITE= 3 4.28E-03 1.00E+03 5.20E-01 2.45E-05 7.30E-03 4.46E+02 1.91E-06 1.00E+02 3.53E-04 9.67E-02 9.81E-01 -3.45E-05
39-Z 479-Y 39-X 79-Y 4.61E+03 5.20E-02 8.45E-03 ( 3)
-2.80E+02 1.72E-01 3.53E-06 -1.86E-03
. . . . . . . . . . . . . . . . . . . =
ITE= 4 2.48E-06 1.77E+01 1.35E-02 3.47E-07 1.71E-04 3.12E+01 1.11E-09 1.77E+00 5.01E-06 6.80E-03 1.00E+00 8.78E-03
Figure 7.2-4: Example of iteration history printout

39-Z 239-Y 159-Z 119-Y 4.58E+03 1.35E-03 1.98E-04 ( 1)
PERCENT OF TIME SPENT IN SOLUTION OF EQUATIONS . . . . . . . . . . =

PERCENT OF TIME SPENT IN LOAD VECTOR/STIFFNESS MATRIX CALCULATION . =
PERCENT OF TIME SPENT IN LINE SEARCHING . . . . . . . . . . . . . . =
-5.76E+00 -5.12E-03 -1.16E-07 -5.42E-05
ITE= 5 8.61E-10 1.51E-01 1.53E-04 1.14E-08 6.76E-06 1.99E-01 3.85E-13 1.51E-02 1.65E-07 4.35E-05 1.00E+00 1.54E-03
79-Z 119-Y 159-Z 119-Y 4.58E+03 1.53E-05 7.83E-06 ( 1)
5.41E-02 7.55E-05 3.59E-09 2.76E-06
6.43
39.57
54.00
5.13
ITE= 6 1.87E-12 5.22E-03 9.02E-06 4.84E-10 2.62E-07 2.64E-02 8.35E-16 5.22E-04 6.99E-09 5.76E-06 1.00E+00 -2.39E-02
119-Z 119-Y 159-Z 119-Y 4.58E+03 9.02E-07 3.03E-07 ( 1)
2.05E-03 -4.21E-06 -1.80E-10 -1.30E-07
6 EQUILIBRIUM ITERATIONS PERFORMED IN THIS TIME STEP TO REESTABLISH EQUILIBRIUM
ADINA Structures Theory and Modeling Guide

This section presents a worked example that illustrates the

nonlinear iteration and convergence concepts previously discussed.
Figure 7.2-4 shows the iteration history for a load step. The
standard Newton method with line searches is used with the energy
and force convergence criterion (ETOL=110-6, RTOL=0.01,
RNORM=10.0, RMNORM =10.0). For contact RCTOL=110-3
and the reference contact force RCONSM=0.01. For line search
STOL=0.01.
Row ITE=0: This row shows the result of the initial iteration
called iteration 0. For this iteration, the program performs the
following steps:
t +t
Compute U (0) = t U .
t +t t +t t +t
Compute F (0) and K (0) using U (0) .
Compute the out-of-balance force vector t +t R t +t F (0) . Only

considering translational degrees of freedom, the norm of the
out-of-balance force vector is t +t R t +t F (0) = 2.32106 and
2
the largest magnitude in the out-of-balance force vector is
-2.32106 at the Z translational degree of freedom of node 740.
Only considering rotational degrees of freedom, the norm of the
out-of-balance force vector is t +t R t +t F (0) = 3.2710-3
2
and the largest magnitude in the out-of-balance force vector is
-6.2210-4 at the Y rotational degree of freedom of node 319.
Compute U (0) using t +t K (0) U (0) =t +t R t +t F (0) . Only

incremental displacement vector is U (0) = 4.7910-3 and
2
the largest magnitude in the incremental displacement vector is
1.1410-3 at the Z displacement of node 159. Only
considering rotational degrees of freedom, the norm of the
2
1.6410-1 at the Y rotation of node 239.
Computes the out-of-balance energy

U (0)
T
( t +t
R t +t F (0) ) = 2.24103.
Compute the norm of the change in contact forces

CFORCE=4.74103, and the norm of the contact forces
CFNORM=4.11103.
Compute the energy convergence criterion

T
U (0) t +t R t +t F (0)
T
= 1.00
U (0) t +t R t F
Compute the force and moment convergence criteria

t +t
R t +t F (0)
2
= 2.32105 and
RNORM
t +t
R t +t F (0)
2
= 3.2710-4.
RMNORM
Compute the contact convergence criterion
CFORCE
= 1.15100
max(CFNORM,RCONSM)
The energy convergence criterion is greater than ETOL, the force

convergence criterion is greater than RTOL and the contact
convergence criterion is greater than RCTOL. Therefore,
convergence is not satisfied.
Note that the displacement and rotation norms are also substituted
into the displacement convergence criterion, which results in
U (0)
2
= 6.9210-2 and
DNORM
U (0)
2
= 8.6310-1.
DMNORM

Since DNORM and DMNORM are not provided by the user, they
are automatically estimated by the program. The above
displacement convergence values however are not used in
determining convergence.
Row ITE=1: This row shows the results of the first equilibrium
iteration. In this iteration, the program performs the following
steps:
t +t t +t
Compute U (1) = U (0) + U (0) , t +t F (1) and the line
T
U (0) t +t R t +t F (1)
search ratio T
. This ratio turns out to
U (0) t +t R t +t F (0)
be greater than STOL=0.01, so line search is performed for 3
steps and ends up with a line search energy ratio of 5.9510-8
which is less than STOL, corresponding to (1) = 0.999 .
Compute t +t
U (1) = t +t
U (0) + (1) U (0) .
t +t t +t t +t
Compute F (1) using U (1) , and compute K (1) .
Note that the stiffness matrix is updated since the standard
Newton method is used.
Compute the out-of-balance force vector t +t R t +t F (1) . Only

out-of-balance force vector is t +t R t +t F (1) = 1.96104
2
and the largest magnitude in the out-of-balance force vector is
4.14103 at node 239 (X translation). Only considering
rotational degrees of freedom, the norm of the out of balance
force vector is t +t R t +t F (1) = 9.93100 and the largest
2
magnitude in the out-of-balance force vector is 2.37100 at

node 39 (Y rotation).
Compute U (1) using t +t K (1) U (1) =t +t R t +t F (1) . Only
2

1.5210-4 at node 239 (X translation). Only considering

rotational degrees of freedom, the norm of the incremental
displacement vector is U (1) = 2.0310-1 and the largest
2
magnitude in the incremental displacement vector is 6.7110-2

at node 279 (Y rotation).
Compute CFORCE=4.7410 3 and CFNORM=5.2610 3.
Compute the out-of-balance energy

U (1)
T
( t +t
R t +t F (1) ) = 3.95100.
Compute the energy convergence criterion

T
U (1) t +t R t +t F (1)
= 1.7710-3.
(1)T
U R F
t +t t
Compute the force and moment convergence criteria

t +t
R t +t F (1)
2
= 1.96103 and
RNORM
t +t
R t +t F (1)
2
= 9.9310-1.
RMNORM
Compute the contact convergence criterion
CFORCE
= 9.0110-1
max(CFNORM,RCONSM)
The energy convergence criterion is greater than

ETOL, the force convergence criterion is greater than RTOL, the
moment convergence criterion is greater than RTOL, and the
contact convergence criterion is greater than RCTOL. Therefore,
The displacement convergence criterion is also evaluated for

informational purposes.
Row ITE=2: This row shows the results from the second
equilibrium iteration. This row is interpreted exactly as is row
ITE=1. The line search factor in this case is 4.8410-1 obtained in

5 line search iterations.
Again, none of the convergence criteria is satisfied. However, they

are all getting smaller.
Row ITE=3: This row shows the results from the third equilibrium
iteration. This row is interpreted exactly as is row ITE=2.
Again, none of the convergence criteria is satisfied.
Row ITE=4: This row shows the results from the fourth
equilibrium iteration. In this case, the previous increment of
displacement from the solver U (3) satisfies the line search energy
tolerance STOL so no line search is performed.
t +t
U (4) = t +t U (3) + U (3)
At the end of the that equilibrium iteration, the energy convergence

criterion is
T
U (4) t +t R t +t F (4)
= 1.1110-9 which is less than
(4)T
U R F
t +t t
ETOL.The force convergence value is 1.77100 and the moment

convergence value is 1.3510-3. The contact convergence value is
6.8010-3.
Two of these four criteria (force and contact) are not satisfied, so
Row ITE=5: The row shows the results for the fifth equilibrium
iteration. Contact convergence criterion is now also satisfied
leaving only force convergence unsatisfied. The solution
continues.
Row ITE=6: The row shows the results from the sixth equilibrium
iteration. In this case, all convergence criteria are satisfied and
convergence is reached.

7.3 Linear dynamic analysis

The following procedures are available in ADINA for solution
of the finite element equations in a linear dynamic analysis:
Step-by-step direct integration

Explicit time integration using the central difference method
(described in Chapter 8)
Implicit time integration using the Newmark method, the
Wilson theta method, or the Bathe time integration method.
Time history mode superposition analysis
Implicit time integration using the Newmark method

(trapezoidal rule)
In this case, the eigenproblem for the vibration of the structural

system is solved for the natural frequencies and mode shapes (see
Chapter 6 for details).
The notation given below is used in the following sections in

the descriptions of the equilibrium equations:
M = constant mass matrix

C = constant damping matrix
K = constant stiffness matrix
t
R, t +t R = external load vector applied at time t, t+t
t
F = nodal point force vector equivalent to the element
stresses at time t
t
U, t +t U = vectors of nodal point velocities at time t, t+t
t
U, t +t U = vectors of nodal point accelerations at time t, t+t
U = vector of nodal point displacement increments from
time t to time t+t, i.e., U = t+tU - tU.
7.3.1 Step-by-step implicit time integration
The governing equilibrium equation at time t+t is given by
M t +t U + C t +t U + K t +t U = t +t R (7.3-1)

The procedures used in the time integration of the governing

equations for dynamic analysis are summarized in ref. KJB,
Chapter 9.
The Newmark method and the Wilson theta method are

summarized in ref. KJB., Sections 9.2.3 and 9.2.4.
The time integration of the governing equations can use the

Newmark method or the Bathe time integration method. The
Newmark method is explained in ref. KJB., Section 9.2.4, and the
Bathe method is explained in the following paper.
ref. K.J. Bathe, Conserving Energy and Momentum in

Nonlinear Dynamics: A Simple Implicit Time
Integration Scheme J. Computers and Structures, Vol.
85, Issue 7-8, pp. 437-445. (2007)
In the Bathe method, the time increment t is divided into two,

the displacements, velocities, and accelerations are solved for at a
time t + t, where (0,1) using the standard Newmark method.
The parameter is always set to 2 2 0.5858 to keep the same
effective stiffness matrix in the two substeps and to avoid
recalculating that matrix and refactorization. In the second substep
a 3-point Euler backward method is used to solve for the
displacements, velocities and accelerations at time t + t using the
results at time t and t + t .
Note that, for a given step size, the Bathe scheme is about twice
as expensive as the Newmark method due to the extra solution step
at time t + t.
The following assumptions are used in the Newmark method:
t +t
U = t U + (1 ) t U + t +t
U t (7.3-2)
1
t +t
U = t U + t U t + t U + t +t
U t 2 (7.3-3)
2
where and are the Newmark time integration parameters.

This transforms Eq. (7.3-1) to
t +t U = t +t R
K (7.3-4)
where
=K+a M+aC
K (7.3-5)
0 1
=t +t R + M ( a t U + a t U + a t U ) + C ( a t U + a t U + a t U )
R 0 2 3 1 4 5
(7.3-6)
and where a0, a1, ... , a5 are integration constants for the Newmark
method (see Ref. KJB, Section 9.2.4).
A similar procedure can be followed for the Bathe time
integration scheme.
ref. KJB The trapezoidal rule (also called the constant-average-

Sections 9.2.4 acceleration method of Newmark) obtained by using = 0.5, =
and 9.4.4
0.25 is recommended for linear dynamic analysis (when the
Newmark method is used).
The trapezoidal rule has the following characteristics:
It is an implicit integration method, meaning that equilibrium

of the system is considered at time t+t to obtain the solution at
time t+t.
It is unconditionally stable for linear analysis. Hence, the

time step size t is selected based on accuracy considerations
only, see ref. KJB, Section 9.4.4.
The Newmark method is in general stable when the following

constraints are satisfied: 0.5, 0.25( + 0.5) 2 .
The Newmark method can be effective in wave propagation

problems, but only if the finite element system has a narrow
bandwidth. In this case the use of higher-order elements with
consistent mass idealization can be a good choice.

The Newmark method is usually more effective for structural

vibration problems. In these analyses, the use of higher-order
elements, just as in static analysis, and the use of a consistent mass
discretization can be effective.
The step-by-step direct integration solution of a vibration

problem is frequently more effective than a mode superposition
analysis when the response need only be calculated over a few time
steps, say less than 100 steps.
The time step increment (t) recommended for dynamic

analysis with the trapezoidal rule is given by co t 0.20
where co is the highest frequency of interest in the dynamic
response.
In FSI analysis with ADINA-FSI, the time integration is

controlled by the ADINA-F models parameter, and the input
for the ADINA models Newmark constants is ignored.
Whether the mass and damping matrices are diagonal or banded

(lumped or consistent discretization), the solution always requires
that a coefficient matrix be assembled and factorized.
7.3.2 Time history by mode superposition

ref. KJB
Section 9.3 In mode superposition analysis, the governing finite element
equations are solved using the transformation
s
U = i xi
i=r
where the i , i = r ,..., s are the mode shapes calculated in a

frequency solution, and the xi are the corresponding unknown
generalized displacements. The xi are calculated by solving the
decoupled modal equations
xi + 2ii xi + i2 xi = ri
where i is the critical damping ratio corresponding to the

frequency i , and ri = iT R . The Newmark method (trapezoidal

rule) is used for the time integration.
When R includes mass-proportional loads, the mass-

proportional load calculation does not include the mass coupling
between active and deleted degrees of freedom.
In ADINA, either the mode shapes corresponding to the

smallest frequencies or the frequencies in a band can be used (see
Chapter 6).
Using ADINA, either no damping is specified, or modal

damping is specified, in which case the values of i , i = r ,..., s can
be all different. Concentrated dampers are ignored.
Mode superposition is effective when the time integration has to

be carried out over many time steps (e.g., earthquake loading), and
the cost of calculating the required frequencies and mode shapes is
reasonable.
Prescribed displacement loads, prescribed velocities, and

prescribed accelerations can be employed in mode superposition
analysis.
Static correction can be applied in linear mode superposition

analysis. The static correction is calculated as follows:
p
R = R ri ( M ii )
i =1
where
p is the number of calculated mode shapes;
M ii is calculated and stored on disk at the beginning of mode

superposition solutions;
ri = Ti iR is calculated during modal solutions and stored for

every requested static correction step; and
R is the external load vector
We then solve
K iU ( t ) = R ( t )
If static correction is used, static correction responses follow

mode superposition responses (e.g., in graphs) for the same time
spans. Note that static correction solutions can be requested for a
different (smaller) number of steps than the number of steps used in
mode superposition solutions.
7.3.3 Lumped and consistent damping; Rayleigh damping
For a lumped damping matrix, contributions come from

concentrated nodal dampers and/or lumped general element
dampers and/or lumped spring element dampers. Direct time
integration must be employed.
ref. KJB For a consistent damping matrix, contributions come from

Section 9.3.3 Rayleigh damping and/or consistent general element damping
and/or consistent spring element damping and/or concentrated
nodal dampers. Direct time integration must be employed.
Explicit time integration cannot be used. Substructuring cannot be
used.
If Rayleigh damping is specified, the contributions of the

(
following matrix C Rayleigh ) are added to the total system damping
matrix C described in Section 7.3:
C Rayleigh = M + K
where M is the total system mass matrix which can be lumped or

consistent, and K is the (initial) total system stiffness matrix
(corresponding to zero initial displacements).
Different Rayleigh damping factors can be input for different

element groups.
Note that C Rayleigh is in general a consistent damping matrix

( C Rayleigh is diagonal if is zero and a lumped mass matrix is
used).
If the Rayleigh damping option UPDATE is set to STEP or

MASS, then the damping matrix is updated either every time step
or when the mass matrix is changed; otherwise, C Rayleigh is
constant throughout the solution and it is formed only once in
ADINA before the step-by-step solution of the equilibrium
equations.
See Ref. KJB, Section 9.3.3, for information about selecting the
Rayleigh damping constants , . In the modal basis, the damping
ratio can be written as
i
i = +
2i 2
where i is the damping ratio for mode i . It is seen that tends

to damp lower modes and tends to damp higher modes.
Tp
If is not used, it is seen that a value of = will overdamp

all motions with periods smaller than Tp . Hence motions with
Tp
periods smaller than Tp can be suppressed by choosing = .

This may be of interest when using damping to suppress numerical
oscillations.
The above comments apply only when the stiffness matrix does
not change significantly during the analysis, however.
The above formula and comments do not apply when potential
based elements are used.
If the potential-based fluid elements are used in the model, the

matrices M and K also include the fluid elements contributions
(see Section 2.11.2). However, the fluid element contributions to

M and K are not included in C Rayleigh .
7.3.4 Modal damping ratios based on strain energy proportional damping
Objective
To enable the user to calculate appropriate values of modal

damping ratios i in transient modal analysis, based on strain
energy proportional damping ratios hi. These damping ratios can be
either constant over the whole structure, or locally varying between
components (element groups).
In damped modal analysis in ADINA, the following system of

equations is solved
xi (t ) + 2i i xi (t ) + i2 xi (t ) = ri (t ) i = 1, 2, ..., p
where i is the mode number, ranging from 1 to p, i is the natural

frequency and i is the modal damping ratio. The modal equations
are all decoupled. For more details, see Section 7.3.2 of the
ADINA-TMG, or Section 9.3.3 of the Finite Element Procedures
book.
If the user requires a different form of damping, such as strain

energy damping, then that damping form must be converted to
equivalent modal damping ratios. A standard damped modal
analysis is then performed.
Two types of strain energy proportional damping will be allowed

by ADINA. In the first type, energy damping ratios applicable to
the whole structure are input. In the second type, the energy
damping ratios are allowed to vary between different element
groups.
Global strain energy damping
The energy damping ratio hi for strain energy proportional damping

is defined for each mode i as

Ei
hi = (7.3-7)
Ei
where Ei is the energy dissipated in the i-th mode, and Ei is the

total strain energy for that mode. In axisymmetric analysis, this
energy damping ratio provides the energy dissipated per unit
radian.
The total strain energy for mode i in modal analysis is proportional

to the projected stiffness matrix
Ei = 12 Ti K i = 12 i2 ; Ti M i = 1 (7.3-8)
where i is mode shape i based on an undamped system and K and

M are the global stiffness and mass matrices.
Local strain energy damping
If the strain energy damping ratios are not uniform throughout the
structure, ADINA will allow each sub-structure or component
(represented as element group in ADINA) to have its own local
energy damping ratios
Ei | j
hi | j = (7.3-9)
Ei | j
where Ei|j and Ei|j are the dissipated and total strain energy for
mode i for element group j.
The strain energies in Eqn. (7.3-9) can be expressed as
Ei | j = 12 Ti K j i
where i is undamped mode shape i for the nodes in element group

j and Kj is the stiffness matrix for element group j. i is a subset of
mode shape i for the whole structure.
ADINA allows each element group to have its own input of local

energy damping ratios hi|j and then calculates the global energy
damping ratios using
E
j =1
i|j hi | j
hi = N
E
j =1
i|j
where N is the number of element groups. This calculation is

performed before the damped modal analysis, using the mode
shapes
T
i K j i hi | j
1 N

j =1
hi = = T
i K j i hi | j (7.3-10)
N
i2
j =1
T
i K ji j =1
Once the global strain energy damping ratios hi are available, the
modal damping ratio for each mode is set by ADINA as
hi
i = (7.3-11)
2
A standard damped modal analysis is then performed.
Notes
It is the users responsibility to provide proper values of strain

energy damping ratios, hi in the case of global parameters, or hi|j in
case of local damping.
In the case of local strain energy damping, the AUI will provide
means for the user to apply the damping ratios to multiple element
groups simultaneously, instead of each element group separately.
Equation (7.3-11) is applicable for small damping ratios. It can be

derived, for a single degree of freedom model, by comparing the

rate of decay resulting from the damping ratio i, and that from the
strain energy damping ratio hi. For viscous damping, the following
relationship holds for small amount of damping

= (7.3-12)
+ 4 2
2
where is the logarithmic decrement
x0
= ln( )
x1
which can be expressed in terms of the strain energy damping

ratio, as follows
E1 E (1 h) 2
= 12 ln( ) = 12 ln( 0 ) = ln(1 h) (7.3-13)
E0 E0
Substituting Eqn. (7.3-13) into Eqn. (7.3-12) and assuming small

values of and h, results in Eqn. (7.3-11).
AUI Input
Modal damping ratios are calculated if one of the modal analysis

options is specified (e.g., MODAL-TRANSIENT, MODAL-
PARTICIPATION-FACTORS, etc.) and if the following command
is used:
MODAL-DAMPING OPTION=LOCAL EGROUP=...
This command can be used for every element group. If EGROUP =

0 is specified then the damping factors are taken for all groups not
listed in the MODAL-DAMPING command input. Note that if
MODAL-TRANSIENT analysis is specified, then modal damping
factors are taken according to Eqn. (7.3-11).
Modal damping ratios are listed in the output (.out) file.
Ref. K. Kawashima, H. Nagashima and H. Iwasaki,

"Evaluation of Modal Damping Ratio Based on Strain

Energy Proportional Damping Method", Journal of
Structural Engineering, Vol 40A, 1994.
7.4 Nonlinear dynamic analysis

In nonlinear dynamic analysis the solution of the finite element
ref. KJB
Section 9.5 equations is usually obtained by direct integration procedures. The
following major techniques can be employed with ADINA:
Explicit integration (described in Chapter 8)
Implicit integration (using the Newmark method or the Bathe

time integration method)
Some considerations on the choice of the time integration

method are given in Section 7.5.
For certain types of problems it can be effective to use mode

superposition for the solution of the nonlinear response. In this
case, the finite element equilibrium equations are established in the
selected mode shapes and solved by using the Newmark method
(see Section 7.4.2).
It can also be effective to use substructuring for the solution of

the equations (see Section 10.1).
The use of Rayleigh damping is the same as described in

Section 7.3.3. Note that the Rayleigh damping matrix remains
constant throughout the time integration if option UPDATE=none
is usesd in the RAYLEIGH-DAMPING command. If UPDATE=
step or mass is used, then the damping matrix is updated according
to the above requests.
7.4.1 Step-by-step implicit time integration
In nonlinear analysis the incremental finite element equilibrium

equations used are, in implicit time integration (without
equilibrium iterations),

M t +t U + C t +t U + t K U = t +t R t F (7.4-1)
If equilibrium iterations are performed, the governing equations

are
M t +t U (i ) + C t +t U (i ) + t +t K U ( i ) = t +t R t +t F (i 1) (7.4-2)
where t +t U ( i ) , t +t U ( i ) , t +t U ( i 1) + U ( i ) are the approximations

to the accelerations, velocities, and displacements obtained in
iteration (i) respectively.
t+t (i-1)
The vector of nodal point forces F is equivalent to the
element stresses in the configuration corresponding to the
t+t (i-1)
displacements U .
The Newmark method is summarized in ref. KJB, Section

9.2.4. The trapezoidal rule is obtained by using = 0.5 and
= 0.25 and is recommended if the Newmark method is used.
In the Bathe method the time increment t is divided into two

substeps. In the first substep, the displacements, velocities,
accelerations are solved for at a time t + t, where (0,1) using
the trapezoidal rule. In the second substep, a 3-point Euler
backward method is used to solve for the displacements, velocities,
accelerations at time t + t, using the results at both time t and t +
t.
ref. K. J. Bathe, Conserving Energy and Momentum in

Nonlinear Dynamics: A Simple Implicit Time
Integration Scheme, Computers & Structures, Vol. 85,
pp. 437-445, 2007.
For nonlinear solutions, is a user input parameter with a default

value equal to 0.5. Note that for linear solutions is set to 2 2
and cannot be overwritten by an input.
For large deformation problems, the Newmark method can

become unstable, while the Bathe scheme remains stable.However,

for a given time step size, the scheme is about twice as expensive
computationally as the Newmark method, due to the extra solution
step at time t + t, if twice the number of equilibrium iterations in
nonlinear analysis are performed. However, fewer iterations are
generally used and/ or a larger time step can be employed.
Some properties of the Bathe method are discussed in
ref. K. J. Bathe and G. Noh, Insight into an Implicit Time

Integration Scheme for Structural Dynamics,
Computers & Structures, Vol. 98-99, pp. 1-6, 2012.
ref. G. Noh, S. Ham and K.J. Bathe, Performance of an

Implicit Time Integration Scheme in the Analysis of
Wave Propagations, Computers & Structures, Vol. 123,
pp. 93-105, 2013.
The dynamic equilibrium equations are solved by use of the

iterative procedures also used in static analysis, including the ATS
method. Line search is also applicable to dynamic analysis.
However, the LDC method cannot be used (see Section 7.2.1).
The energy and force/moment convergence criteria used in

nonlinear dynamic analysis are:
T
U (i ) t +t R M t +t U (i 1) C t +t U (i 1) t +t F (i 1)
T
ETOL
U (1) t +t R M t +t U (0) C t +t U (0) t F

degrees of freedom
t +t
R M t +t U (i 1) C t +t U ( i 1) t +t F (i 1)
2
RTOL
RNORM

t +t
R M t +t U (i 1) C t +t U ( i 1) t +t F (i 1)
2
RTOL
RMNORM
The other convergence criteria and the notation and considerations

for the use of the convergence criteria are the same as in nonlinear
static analysis; see Section 7.2.2.
7.4.2 Time history by mode superposition
A nonlinear dynamic problem is solved effectively by the mode

ref. KJB
Section 9.5.3 superposition procedure if the response can be adequately
represented by the superposition of the response in a few selected
mode shapes.
The mode shapes result from a linearization about a reference

configuration and are not updated during the mode superposition
analysis. The reference configuration can be based on a restart
analysis.
The mode shapes can be calculated at the start of the nonlinear

mode superposition analysis or read from a prior frequency
analysis file.
The equations solved in ADINA to obtain the solution of the

nonlinear dynamic equations by mode superposition are presented
on pp. 828-829 of ref. KJB. In ADINA, throughout the time
integration, the eigenmodes corresponding to the time at solution
start are employed, and the modal equations are integrated using
the Newmark method.
If desired, nonlinear mode superposition analysis can be

performed in stages. Each restart stage can start by recalculating
the eigenmodes based on the last available solution.
If the model during the mode superposition analysis experiences

significant further nonlinearity the mode shapes of the reference
configuration may no longer be suitable. This happens, for
example, if more nodes come into or go out of contact, or if

frictional states change between stick and slip. For such problems,
a non-modal nonlinear dynamic analysis should be performed.
It is the users responsibility to ensure that sufficient mode

shapes are used, and that they are applicable throughout the
nonlinear dynamic simulation.
Some applications using the mode superposition method are

given in the following reference:

Analysis Using Substructuring and Mode
Superposition," J. Computers and Structures, Vol. 13,
pp. 699-707, 1981.
7.4.3 Global mass matrix
The global mass matrix used in nonlinear dynamic analysis is the

same as the global mass matrix used in linear dynamic analysis.
Unlike the global stiffness matrix, which is typically recalculated

during each equilibrium iteration, the global mass matrix is
recalculated only for the following case:
Element birth-death: only the elements that are currently alive

contribute to the global mass matrix. (This includes the case of
elements that have ruptured; ruptured elements also do not
contribute to the global mass matrix.)
In the following discussion, we consider an analysis without

element birth-death. In this case, the global mass matrix is
calculated only once, at the beginning of the analysis.
For continuum elements, in which the unknowns are the

displacement degrees of freedom, the mass matrix does not change
due to large deformations (ref KJB, p 542). Hence, in this case, it is
theoretically justified to not recalculate the global mass matrix.
However, it should be recognized that there are some situations in

which, theoretically, the global mass matrix changes during the
analysis. These situations are all associated with rotational degrees

of freedom.
Structural elements with rotational degrees of freedom,

consistent mass matrix. The effect is most pronounced when the
element cross-section or thickness is large, and is also most
pronounced for the Hermitian beam element.
Rigid links and constraint equations with rotational degrees of

freedom
Concentrated rotational masses with different inertias in

different rotational directions
Typically the effect of neglecting the time-varying parts of the

global mass matrix is small. However, for certain analyses in which
there are large rotations, the effect can be significant.
7.5 Choosing between implicit and explicit formulations

The main criterion governing the selection of the implicit or
explicit formulations is the time scale of the solution.
The implicit method can use much larger time steps since it is
ref. KJB
Section 9.2 unconditionally stable. However, it involves the assembly and
solution of a system of equations, and it is iterative. Therefore, the
computational time per load step is relatively high. The explicit
method uses much smaller time steps since it is conditionally
stable, meaning that the time step for the solution has to be less
than a certain critical time step, which depends on the smallest
element size and the material properties. However, it involves no
matrix solution and is non-iterative. Therefore, the computational
time per load step is relatively low.
For both linear and nonlinear static problems, the implicit

method is the only option.
For low-speed dynamic problems, the solution time spans a

period of time considerably longer than the time it takes the wave
to propagate through an element. The solution in this case is
dominated by the lower frequencies of the structure. This class of

7.5 Choosing between implicit and explicit formulations
problems covers most structural dynamics problems, certain metal

forming problems, crush analysis, earthquake response and
biomedical problems. When the explicit method is used for such
problems the resulting number of time steps will be excessive,
unless mass-scaling is applied, or the loads are artificially applied
over a shorter time frame. No such modifications are needed in the
implicit method. Hence, the implicit method is the optimal choice.
For high-speed dynamic problems, the solution time is

comparable to the time required for the wave to propagate through
the structure. This class of problems covers most wave propagation
problems, explosives problems, and high-speed impact problems.
For these problems, the number of steps required with the explicit
method is not excessive. If the implicit method uses a similar time
step it will be much slower and if it uses a much larger time step it
will introduce other solution errors since it will not be capturing the
pertinent features of the solution (but it will remain stable). Hence,
the explicit method is the optimal choice.
A large number of dynamics problems cannot be fully classified

as either low-speed or high-speed dynamic. This includes many
crash problems, drop tests and metal forming problems. For these
problems both solution methods are comparable. However,
whenever possible (when the time step is relatively large and there
are no convergence difficulties) we recommend the use of the
implicit solution method.
Note that the explicit solution provided in ADINA does not use
reduced integration with hour-glassing. This technique reduces the
computational time per load step. However, it can have detrimental
effect on the accuracy and reliability of the solution.
Since the explicit time step size depends on the length of the
smallest element, one excessively small element will reduce the
stable time step for the whole model. Mass-scaling can be applied
to these small elements to increase their stable time step. The
implicit method is not sensitive to such small elements.
Since the explicit time step size depends on the material

properties, a nearly incompressible material will also significantly
reduce the stable time step. The compressibility of the material can

be increased in explicit analysis to achieve a more acceptable

solution time. The implicit method is not as sensitive to highly
incompressible materials (provided that a mixed formulation is
used).
Higher order elements such as the 10-node tetrahedral, 20 and

27 node brick elements are only available in implicit analysis. They
are not used in explicit analysis because no suitable mass-lumping
technique is available for these elements.
Model nonlinearity is another criterion influencing the choice
between implicit and explicit solutions. As the level of nonlinearity
increases, the implicit method requires more time steps, and each
time step may require more iterations to converge. In some cases,
no convergence is reached. The explicit method however, is less
sensitive to the level of nonlinearity.
Note that when the implicit method fails it is usually due to non-
convergence within a time step, while when the explicit method
fails it is usually due to a diverging solution.
The memory requirements is another factor. For the same mesh,

the explicit method requires less memory since it does not store a
stiffness matrix and does not require a solver. This can be
significant for very large problems.
Since ADINA handles both implicit and explicit analysis with

very similar inputs, the user can in many cases restart from one
analysis type to the other. This capability can be used, for example,
to perform implicit springback analysis following an explicit metal
forming simulation, or to perform an explicit analysis following the
implicit application of a gravity load.
It can also be used to overcome certain convergence difficulties in

implicit analyses. A restart from the last converged implicit
solution to explicit can be performed, then, once that stage is
passed, another restart from explicit to implicit can be performed to
proceed with the rest of the solution.
7.6 Tracking solution progress

Important model parameters such as the memory used by the
model, memory used by the solver, number of degrees of freedom,

7.6 Tracking soluiton progress
solution times, warning messages, and error messages are all

provided in the .out file.
Detailed iteration by iteration convergence information is also

written to the .out file as illustrated in the example of Section 7.2.9.
The program outputs a more summarized time step information

to the .msg file. This output focuses on the time steps and the ATS
history.
7.6.1 .rto file
Several solution control parameters can be modified during

execution by creating a runtime option file with extension .rto. The
.rto file can be created either by hand, using the text editor of
your choice, or by the AUI, using the RTOFILE command.
The syntax of the .rto file is now described.
The .rto file can contain an arbitrary number of lines. Each line in
the .rto file is a string of up to 80 characters. A line beginning with
a * or # is a comment line. Any text following // is a comment.
All non-comment lines or text are converted to upper case.
Each line of the .rto file sets a parameter. Parameters are set using
syntax such as
DTOL=0.234
ETOL=1E-5
Allowable parameters are given in Table 7-6.
ADINA reads the .rto file at the start of program execution, and
also before each equilibrium iteration. It is allowed to modify the
.rto file during program execution, for example to reset the
convergence tolerances during equilibrium iterations, or to save the
current solution, then stop. Use your favorite text editor to modify
the .rto file.
Table 7-6: Parameters that can be set in the .rto file

Parameter Description Allowable values

name
DMNORM Reference value for Real, > 0.0
displacement norm,
rotations
DNORM Reference value for Real, > 0.0
displacement norm,
translations
DTOL Tolerance for displacement Real, > 0.0
norm
ETOL Tolerance for energy norm Real, > 0.0
MAXDISP Limiting incremental Real, > 0.0
displacement during
MAXITE Maximum number of Integer, > 0
iterations allowed during
MINITE Minimum number of Integer, > 0
iterations performed during
RCONSM Reference contact force Real, > 0.0
level
RCTOL Contact force convergence Real, > 0.0
tolerance
RMNORM Reference value for force Real, > 0.0
norm, moments
RNORM Reference value for force Real, > 0.0
norm, forces
RTOL Tolerance for force norm Real, > 0.0
STOL Tolerance for line searching Real, > 0.0
STOP Saves the previously Either the string
converged solution in the YES or the integer
porthole file, then stops 1. E.g.,
program execution. STOP=YES, or
STOP=1
7.7 Solution monitoring
This feature enables the user to monitor key solution variables
during analysis (without loading the porthole file). Up to 8 values
can be monitored simultaneously. The program outputs the value of

7.7 Solution monitoring
the monitored quantity, and the location where it reaches its

extreme value. This output automatically goes to the .out file when
solution monitoring is activated.
Three types of quantities can be monitored: (i) stress or strain

variables (from 2D or 3D elements only) extrapolated to nodes and
averaged over the whole model, (ii) same as the previous but
averaged over a specific element group, and (iii) nodal variables
(displacement, velocity or acceleration) computed over the whole
model.
For stress and strain variables, all components can be

monitored, as well as the principal values and the effective value
(for stress only). For nodal variables, all components of
displacements and rotations can be monitored, as well as
displacement and rotation magnitudes. For dynamic analysis, all
components and magnitudes of velocities and accelerations (and
their rotational equivalents) can also be monitored.
The monitored values can also be used to stop the analysis, if

they reach a critical value.
In static and implicit dynamics, solution monitoring is

performed after every converged time step or sub-step. In explicit
dynamic analysis, it is performed every fixed number of time steps,
as set by the user.
Currently, solution monitoring cannot be used with the

following features: cyclic symmetry, FSI, TMC, Frequency
analysis, modal superposition, harmonic analyses, op2 file output,
Ensight file output, substructures, direct skyline solver, iterative
and multigrid solvers.


8.1: Formulation
8. Explicit dynamic analysis

This chapter presents the formulations and algorithms used to solve explicit dynamic
problems, including time step calculation. A summary of the elements and material
properties available for explicit analysis are listed in Table 8-1. More details are
provided in the appropriate sections in Chapters 2 and 3.

Chapter 8: Explicit dynamic analysis
Table 8-1: Elements and materials available in explicit time integration
Rubber-like ***
Drucker-Prag.
Thermo-plast.
Nonlin.-elast.
Gurson-plast.
Thermo-orth.
Plastic-cyclic
Plastic-bilin.
Plastic-mult.
Elastic-orth.
Thermo-iso.
DF Concrete
Plastic-orth.
Concrete &
Element type Nodes
Gasket**
Ilyushin
Elastic
Fluid
Truss 2 9 9 9 9 9 9 9
Beam 2 9 9 9
Isobeam 2 9 9 9 9 9
Plate 3 9 9 9
Shell (MITC) 3/4/9 9 9 9 9 9 9 9 9 9
3D-Shell 3/4 9 9 9
Solid (2D, displ.) 3/4/6/9 9 9 9 9 9 9 9 9 9 9 9 9 9
Solid (2D, u/p) 3/4/6/9 9 9 9 9 9 9 9 9 9 9 9
Solid (2D, inc.mod) 3*/4 9 9 9 9 9 9 9 9 9 9
4/5/6/8
Solid (3D, displ.) 9 9 9 9 9 9 9 9 9 9 9 9 9
/10/27
4/5/6/8
Solid (3D, u/p) 9 9 9 9 9 9 9 9 9 9 9
/10/27
Solid (3D, inc.mod) 4*/8 9 9 9 9 9 9 9 9 9 9
Disp.fluid 2-D 4
9
Disp.fluid 3-D 8
In addition, linear and nonlinear spring elements are available.
*) 3-node triangular 2-D and 4-node 3-D tetrahedral elements do not use incompatible modes.
**) No critical time step checking for gasket materials
***) The rubber-like materials supported in explicit analysis are the Mooney-Rivlin, Ogden, Arruda-
Boyce, eight-chain and Sussman-Bathe materials. Viscoelastic, Mullins and orthotropic effects are not
supported.

8.1: Formulation
Table 8-2 lists more element types not available in explicit

analysis.
Table 8-2: Elements types not available in explicit analysis
potential-based fluid elements

porous media
pipe elements
general elements
user-supplied elements
shell elements not classified as new MITC elements
(e.g., layered shells, shells with composite failure criteria, MITC8,
isoparametric variable node shells, etc.)
Table 8-3 lists the ADINA options not available in explicit analysis.
Table 8-3: ADINA options not available in explicit analysis
cyclic symmetry
substructuring
mapping
zooming
Thermo-mechanical coupling (TMC)
Fluid-structure interaction (FSI)
consistent mass matrix
consistent Rayleigh damping using the central difference method
temperatures provided from file
load penetration (for ruptured elements)
general constraints
mesh glueing
8.1 Formulation
One of the two following explicit time integration methods can be
used:
The central difference method (CDM - see ref. KJB,

Section 9.2.1)
The Noh-Bathe method (see reference below)
ref G. Noh and K.J. Bathe, "An explicit time integration

scheme for the analysis of wave propagations",
Computers & Structures, 2013
8.1.1 Central difference method (CDM)
1 t t
t
U=
t 2
( U 2 t U + t +t U ) (8.1-1)
and the velocity is calculated using
1
t
U=
2t
( t t U + t +t U ) (8.1-2)
The governing equilibrium equation at time t is given by
M tU + C t U = tR tF (8.1-3)
Substituting the relations for t U and t U in Eq. (8.1-1) and (8.1-2),

respectively, into Eq. (8.1-3), we obtain
1 1 t +t 2 1 1 t t
2 M+ C U = tR tF + 2 M tU 2 M + C U
t 2t t t 2 t
(8.1-4)
t +t
from which we can solve for U.
ref. KJB The central difference method has the following characteristics:
Sections 9.2.1,
9.4 and 9.5.1 It is an explicit integration method, meaning that equilibrium
of the finite element system is considered at time t to obtain the
solution at time t+t.
When the mass and damping matrices are diagonal, no

8.1: Formulation
coefficient matrix needs to be factorized, see ref. KJB, p. 772.

The use of the central difference method is only effective when
this condition is satisfied. Therefore, only lumped mass can be
used. Also damping can only be mass-proportional.
No degree of freedom should have zero mass. This will lead

to a singularity in the calculation of displacements according to
Eq. 8.1-4, and will also result in a zero stable time step.
The central difference method is conditionally stable. The

time step size t is governed by the following criterion
TNmin
t tCR =

where tCR is the critical time step size, and TNmin is the
smallest period in the finite element mesh.
The central difference method is most effective when low-order

elements are employed. Hence quadratic 3-D solid and shell
elements are not allowed.
The time step can be specified by the user, or calculated

automatically. When the user specifies the time, the program does
not perform any stability checking. It is the users responsibility, in
this case, to ensure that an appropriate stable time step is used.
When automatic time step calculation is selected, the

TIMESTEP entry is only used to determine the number of nominal
time steps and the frequency of output of results. The stable time
step is used instead of the value in TIMESTEP (unless the value in
TIMESTEP is smaller).
For example, if the user requests 12 steps of size 1.0 with
porthole output every 4 steps, there will be 12 nominal steps each
of size 1.0. If the stable time step is smaller than 1.0 it will be used
instead and results will be saved as soon as the solution time
exceeds 4.0, 8.0 and exactly at 12.0 since it is the last step of the
analysis.

8.1.2 Noh-Bathe method
The Noh-Bathe method is a second order accurate explicit time

integration scheme, which can be used with or without stiffness and
mass based Rayleigh damping. The method is stable for time steps
tcr
t
p
where tcr is define the same way as for the CDM method and p
is a parameter, p (0,1) . For a given time step t , the program
uses two sub-steps with time step sizes pt and (1 p )t .
For the first sub-step we use the following procedure:
- we calculate displacements and effective loads at time t + pt :
t + p t
U = t U + a0 t U + a1 t U
t + p t
R = (1 p ) t R + p t +t R
t + p t t + p t
R= R t + pt F C( t U + a0 t U)
t + p t
where F is an element internal force vector.
- we then solve for accelerations at time t + pt :
M t + pt U = t + pt
R
and we can calculate velocities at time t + pt :
t + p t
U = t U + a2 ( t U + t + pt U)
For the second sub-step, we calculate displacements for time

t + t and the corresponding load vector:

8.1: Formulation
t +t t + pt
U= U + a3 t + pt U + a4 t + pt U
t +t t +t
R= R t +t F C( t + pt U + a3 t + pt U)
- we can then solve for accelerations at time t + t :
M t +t U = t +t
R
and we can calculate velocities at t + t :
t +t t + pt
U= U + a3 t U + a6 t + pt U + a7 t +t U
where:
a0 = pt a3 = (1 p)t
a1 = 12 ( pt ) 2 a4 = 12 ((1 p)t ) 2
a2 = 12 a0 a5 = q0 a3
a6 = ( 12 + q1 )a3
a7 = q2 a3
where
1 2 p
q1 =
2 p(1 p)
q2 = 12 pq1
q0 = q1 q2 + 12
In some cases (e.g., when post-impact calculations are used for

the kinematic constraint contact algorithm or if prescribed
velocities are used), we need to calculate velocities from
displacements and accelerations from velocities:
- for t + pt step we have:

t + p t
U = a8 ( t + pt U t U) t U
t + p t
U = a8 ( t + pt U t U) t U
- for t + t step we have:
t +t
U = a9 ( t +t U t + pt U) t + pt U
t +t
U = a9 ( t +t U t + pt U) t + pt U
where
2
a8 =
pt
2
a9 =
(1 p )t
8.1.3 Mass matrix
The construction of the lumped mass matrix depends on the

type of element used. Details are provided in the appropriate
section in Chapter 2.
For elements with translational degrees of freedom only, the total

mass of the element is divided equally among its nodes. For
elements with rotational masses (beam and shell elements), the
lumping procedure is element dependent.
Note that the lumping of rotational degrees of freedom is slightly

different in implicit and explicit analysis. The rotational masses in
explicit analysis are sometimes scaled up so that they do not affect
the elements critical time step.
8.1.4 Damping central difference method
Damping can be added directly to the model through Rayleigh

damping. Additional indirect damping results from plasticity,
friction and rate dependent penalty contact.

8.1: Formulation
Only mass-proportional Rayleigh damping is available in

explicit analysis. Hence, the damping matrix C in Eq. 8.1-3 is set
to:
C Rayleigh = M
where M is the total lumped mass matrix. See Ref. KJB, Section
9.3.3, for information about selecting the Rayleigh damping
constant .
8.1.5 Damping Noh-Bathe method
For the Noh-Bathe method, stiffness and mass proportional

Rayleigh damping can be used:
C = M + K
where M is the total lumped mass matrix and K is the initial

stiffness matrix.
It should be noted that excessive damping can affect the critical

time step for Noh-Bathe method, therefore it is important to
properly calculate , parameters. One way to do it is to
calculate these parameters based on the lowest frequency 1 and
the highest frequency 2 using a damping ratio formulation shown
in ref. KJB, page 797. The lowest frequency 1 can be calculated
by the program in a separate run. The highest frequency 2 can be
obtained from the critical time, i.e.,
2
2 =
tcr
We can then use two equations from which and can be

calculated:

+ 12
= 1
21
+ 22
= 2
22
where 1 and 2 are damping ratios. It has been observed that

using in the above equations damping ratios in the order of 1 to 2
percent gives a reasonable amount of damping.
8.2 Stability
ref. KJB The stable time step for a single degree of freedom with central
Section 9.4.2 difference time integration is
TN 2
tCR = =
N
The stable time step for a finite element assembly is
TN min 2 2
t tCR = =
N max E max
where N max is the highest natural frequency of the system, which

is bound by the highest natural frequency of all individual elements
in a model Emax (see Ref. KJB, Example 9.13, p. 815).
For the Noh-Bathe method, the stable time step is:
tcr
t
p
When automatic time step is selected, the time step size is

determined for both, CDM and NB methods, according to the
following relationship

8.2: Stability
2
t = K t E min = K (8.2-1)
E max
where K is a factor that scales the time step.
For most element types the critical time step can be expressed in
terms of a characteristic length and a material wave speed
L
t E = (8.2-2)
c
where the definition of the length L and the wave speed c depend
on the element and material type. For all elastic-plastic materials
the elastic wave is used. This condition is used in the program
instead of actually evaluating the natural frequency in Eq. (8.2-1).
Note that the critical time step calculated for all elements is only
an estimate. For some elements and material combinations it is
exact, and for others it is slightly conservative. However, it may
not be small enough for excessively distorted elements (3-D solid
and shells), and it will therefore need scaling using K factor in Eq
(8.2-1).
The time step also changes with deformation, due to the change
in the geometry of the elements and the change in the wave speed
through the element (resulting from a change in the material
properties).
Truss and cable elements

The critical time step for a 2-node truss element is
L
t E =
c
where L is the length of the element, and c is the wave speed

through the element
E
c=


Beam and iso-beam elements

The critical time step for the beam element is
L 12 I
t E = / 1+ 2
c AL
where L is the length of the element, A is the element area, I is the

largest moment of inertia, and c is the wave speed through the
element
E
c=

Shell elements
The critical time step for shell elements is
L
t E =
c
where L is a characteristic length of the element based on its area

and the length of its sides, and c is the planar wave speed through
the shell, which for linear isotropic elastic materials is
E
c=
(1 2 )
The critical time step estimated here is only approximate, and may
be too large for excessively distorted shell elements.
3-D solid elements

The critical time step for the 3-D solid elements is
L
t E =
c
where L is a characteristic length of the element, based on its

volume and the area of its sides, and c is the wave speed through
the element. For linear isotropic elastic materials c is given as

8.2: Stability
E (1 )
c=
(1 + )(1 2 )
The critical time step estimated here is only approximate, and may
be too large for excessively distorted 3-D solid elements.
Spring elements
The critical time step for a spring element is
2 M 1M 2
t E = =2
N K (M1 + M 2 )
where M1 and M2 are the masses of the two spring nodes and K is
its stiffness. Massless springs are not taken into account in the
calculation of the stable time step.
Rigid links
These elements are perfectly rigid and therefore do not affect the
stability of explicit analysis.
8.3 Time step management

The stable time step size has a major influence on the total
simulation time. Since this time step is determined based on the
highest eigenvalue of the smallest element, a single small or
excessively distorted element could considerably increase the
solution time, even if this element is not relevant to the full model.
Note that the element having the smallest critical time step size is
always provided in the output file.
Ideally, all elements should have similar critical time steps. If

the material properties are uniform throughout the model this
means that elements should approximately have the same lengths
(see Eq. 8.2-2).
The evaluation of the critical time step for each element takes

some computational time. Therefore, it does not need to be

performed at every time step. The parameter NCRSTEP in the
ANALYSIS DYNAMIC command determines how frequently the
critical time step is reevaluated.
The time step size for explicit analysis can be unduly small for a
realistic solution time. Three features are provided to deal with this
problem.
A global mass scaling variable can be applied to all elements in

the model (the MASS-SCALE parameter in the ANALYSIS
DYNAMIC command). This scale factor is applied to the densities
of all elements, except scalar elements where it is applied directly
to their mass.
Mass scaling could also be applied to elements whose

automatically calculated initial time step is below a certain value
(DTMIN1 parameter in ANALYSIS DYNAMIC). A mass scale
factor is then applied to these elements to make their time steps
reach DTMIN1. The mass scaling ratio is then held constant for the
duration of the analysis. This option is not used when the time step
size is user-specified.
Elements with automatically calculated time step smaller than a

specified value (DTMIN2 parameter in ANALYSIS DYNAMIC)
can be completely removed from the model. This parameter is
useful for extremely small or distorted elements that do not affect
the rest of the model.
The three parameters explained above (NCRSTEP, DTMIN1

and DTMIN2) should all be used with great care to ensure that the
accuracy of the analysis is not significantly compromised.
8.4 Tracking solution progress

Important model parameters such as the memory used by the
model, number of degrees of freedom, solution times, minimum
stable time step, warning and error messages are all provided in the
.out file.

8.2: Stability
The program outputs a more summarized time step information

to the .msg file.
For the Noh-Bathe method, the program outputs solution results

for t + t time steps only.


9.1 Response spectrum analysis
9. Frequency domain analysis

The ADINA system includes several capabilities for the
characterization of a structural response using frequency domain
analysis.
The dynamic response of a structure submitted to a given

excitation described by a response spectrum (ground or support
motions typical of earthquakes or shocks) can be investigated
(see Section 9.1).
A Fourier analysis of the time history of a point in the

structure can be performed (see Section 9.2).
Structural response to harmonic (periodic) vibrations or to

random vibrations, whether from a base motion or from applied
forces, can be analyzed (see Sections 9.3 and 9.4).
The response of an SDOF system connected to the finite

element model can also be investigated (see Section 9.5).
Much of the frequency domain analysis is performed by the AUI

during post-processing.

Consider the response of a finite element system to an
earthquake, with a ground acceleration ak, into direction k. We
assume that all supports of the system are accelerated
simultaneously with ak. The governing equilibrium equations are
(see ref. KJB)
Mu + Cu + Ku = Md k ak (t ) (9.1-1)
where dk is a direction vector with "1" in the portions of the

translational degrees of freedom acting into the direction k and "0"
in the other portions, and u is the vector of relative displacements
of the finite element system to the ground motion.

Chapter 9: Frequency domain analysis
If we perform the mode superposition analysis, the governing

decoupled equations are
xi + 2ii xi + i2 xi = ik ak (t ) (9.1-2)
where ik are the modal participation factors for ground motion

loading (see equation (9.1-7) below). The response solution is
t
1 ii ( t ) 1 2 ( t ) a ( ) d
2
xik (t ) = ik e sin
i i
k
i 1 i 0
(9.1-3)
or
xik (t ) = ik Dik (9.1-4)
If we note that only the lowest modes (up to a cut-off frequency

u ) contribute to the response, we obtain for the complete
response of the system,
u = ik Dik i + u r (9.1-5)
i u
= D
i u
k
i i
k
i + r (9.1-6)
where ur and r are the residual responses due to summing the

modal contributions only up to frequency u .
The modal participation factors are calculated as follows by

ADINA:
ik = T M d k
i
(9.1-7)
where i is the mode shape of mode i and M is the system mass

matrix. Only the active degrees of freedom are included in this
calculation. In particular, any mass coupling between active and

deleted degrees of freedom is ignored.
The modal participation factors have the following properties:
1)
all modes
ik = d k
i
(9.1-8)
This formula is used in deriving the residual (static correction)

formulas given below (see equation (9.1-29)), and also is used in
deriving the floor response spectrum formulas in Section 9.5.
( ) = (d ) k 2 T
2) i k M dk (9.1-9)
all modes
This formula states that the sum of the squares of the modal
participation factors for direction k equals the structural mass for
direction k, excluding mass associated with fixed degrees of
freedom of the structure. Hence the modal mass for mode i,
( )
2
direction k, is ik .
In the response spectrum analysis, the actual dynamic response

is estimated in the following manner.
Response not including residual terms: Let S a (i , i ) be the

design response spectrum acceleration, then the maximum
accelerations for mode i are
( ui )max = ik Sa(i , i ) i
(9.1-10)
The maximum relative displacements for mode i are
1
( ui )max = ik Sa (i , i ) (9.1-11)
i
2 i
The maximum stresses for mode i are

1
( i )max = ik Sa (i , i ) (9.1-12)
i2 i
The maximum element forces and section results for mode i are
1
( Fi )max = ik Sa (i , i ) Fi (9.1-13)
i
2
The maximum reactions for mode i are
1
( R i )max = ik Sa (i , i ) R i (9.1-14)
i
2
Let Mi represent any maximum modal response component for

mode i. Then to obtain the modal combination response for the
selected component considering all modes with circular frequencies
less than or equal to u , the following modal combinations are
available:
SRSS method (Square Root of the Sum of the Squares):
1
2
2
M SUM = (Mi ) (9.1-15)
i u
Absolute sum method:
M SUM =
i u
Mi (9.1-16)
Ten percent method:
1
2

= (Mi ) + 2 MiM j
2
M SUM (9.1-17)
i u i u
j u

The second summation is done on all i and j modes having

frequencies which are closely spaced, i.e., for which, with
i< ju,
j i
0.1
i
Double sum method, DSC (Discrete Singular Convolution)

method:
1
2
M SUM = M r M s rs (9.1-18)
r u s u
where
1
2
rs = 1 + r s

rr + ss
and
1
r = r (1 r2 ) 2
2
r = xr +
t d r
and td is the duration of the earthquake. In the double sum

method, the absolute value signs in (9.1-18) are included; in the
DSC method, the absolute value signs in (9.1-18) are omitted.
Algebraic sum method:
M SUM = M
i u
i (9.1-19)

CQC (complete quadratic combination) method:

1
2
M SUM = M r M s rs (9.1-20)
r u s u
where
3
8 r s ( r + a s ) a 2
rs =
(1 a )
2 2
(
+ 4 r s a (1 + a 2 ) + 4 2 + 2 a 2
r s
)
s
and a = . You have the option to include only those terms
r
for which rs CUTOFF . CUTOFF = 0.0 implies that all
terms are included, CUTOFF = 1.0 implies that only the
diagonal terms are included.
Note: The results obtained with all methods except the algebraic
sum method and DSC method depend upon the coordinate system
used in modeling the problem. This is because formulas (9.1-15) to
(9.1-18) and (9.1-20) are applied component by component and
because the sign of each component is lost (either the component is
squared or the absolute value of the component is taken).
For example, to evaluate the SRSS displacement at a node, (9.1-
15) is applied to the maximum x displacement, then to the
maximum y displacement and finally to the maximum z
displacement. Figure 9.1-1 shows the results for a two-dimensional
problem. As shown, the expected result occurs only if one
coordinate system direction is aligned with the response direction.
Although the figure illustrates only displacements, the same
phenomenon occurs for all results; accelerations, velocities,
displacements, reactions, stresses and element forces.
This observation also applies to accelerations, velocities and
displacements when skew systems are used at nodes.
Effect of residual (static correction) terms: For the response

spectrum method the residual (static correction) terms are
calculated using the zero period accelerations SZPA. Then the
following residual terms can be calculated:

Residual accelerations:

u r = d k ik i S ZPA (9.1-21)
i u
l (u1)MAX = (u1)SRSS
u
(u1)MAX =
0
u2 u
(u1)SRSS = =
02 0
a) Model coordinate system aligned with the response direction
y (u1)SRSS u sin 309

(u1)MAX =
-u
0 cos 309
x
l
(u1)MAX (u sin 309)2

309 (u1)SRSS =
(-u cos 309)2
u sin 309
=
u cos 309
b) Model coordinate system not aligned with the response direction
Figure 9.1.1: Effects of choice of coordinate system on model

combination responses
Residual displacements:

u r = K 1M d k ik i S ZPA (9.1-22)
i u
Residual stresses, residual element forces and residual reactions

are denoted as r , Fr , R r respectively. These residual

quantities will occur when the displacements of the model are

equal to the residual displacements ur.
Note that when the residual quantities are graphically displayed

separately, then SZPA is set to unity. When the residual quantities
are to be included in a modal combination, however, the
expressions shown in Eqs. (9.1-21) and (9.1-22) are, of course,
used.
Let Mr denote a residual response quantity, then the modal

combination response quantity, including the residual term, can be
calculated as follows:
SRSS addition of residual:
M = ( M SUM ) + ( M r ) 2
2 2
(9.1-23)

Absolute addition method including sign: When the SRSS,

absolute sum, ten percent, double sum or CQC method is used:
M = ( M SUM ) + M r
When the algebraic sum method is used:
M = M SUM + SIGN ( M r , M SUM ) (9.1-24)
where
M r if M SUM 0
SIGN ( M r , M SUM ) =
M r if M SUM < 0
Effect of excitations in multiple directions: In Eqs. (9.1-1) to

(9.1-24) we considered the solution of the response due to
earthquake excitation in one direction only. To consider excitation
in more than one direction, the following spatial combination
methods are available:

SRSS addition of spatial components:

1
2 2
M SPATIAL = ( k M k ) (9.1-25)
k
Absolute addition of spatial components:
M SPATIAL = k M k (9.1-26)
k
where k is a factor for scaling of the component result Mk.
Effect of static load case: Let MST be a response quantity due to a

static load case which has been calculated by ADINA. This static
response quantity would represent the condition of the structure at
the time when the earthquake occurs and can include, for example,
the static response due to gravity load. The final spatial
combination can then be obtained as
M = M ST + SIGN ( M SPATIAL , M ST ) (9.1-27)
where
M SPATIAL if M ST 0
SIGN ( M SPATIAL , M ST ) = (9.1-28)
M SPATIAL if M ST < 0
Initial calculations for response spectrum analysis must be

performed in the ADINA analysis solution; the actual response
spectrum analysis being performed during post-processing with the
AUI.
Note that any number of steps of linear static solution response
calculations can be performed prior to the response spectrum initial
calculations in ADINA.
The initial calculations performed in ADINA are given in the

following. Note that these results are stored in the porthole file.
The modal participation factors ik for each direction k of

ground translational excitation and for each eigenmode, using

equation (9.1-7). Note that the number of eigenmodes used in

the response spectrum calculation must be less than or equal to
the number of eigenmodes available; otherwise, the program
issues an error message and stops the execution.
The modal stress vector si for each specified eigenmode for

all elements in which stress output is requested. See Section
6.2.3 for more information.
These stresses do not have a "physical meaning". They are
used for calculating the spectral response of the structure.
If the nodal force output option is used in an element, the
modal forces corresponding to a displacement vector equal to i
are instead calculated, printed and stored.
In addition, initial calculations can be performed for static

corrections. If requested, ADINA calculates, prints and stores on
the porthole file the following data for each direction k of
excitation.
Residual accelerations u kR , where
N
u kR = d k ik i (9.1-29)
i =1
Residual displacements u kR for the total structural system,

where
u kR = t0 K 1M u kR (9.1-30)
The definitions for the variables on the right side of the equation
t
are given above and 0 K is the stiffness matrix at the start time
of the solution.
Residual forces FRk for the total structural system, where
FRk =M u kR (9.1-31)

Residual stresses s kR for all elements which specify stress

output, where
s kR = t0 C t0 B u kR (9.1-32)
t t
and 0 C, 0 B are the element stress-strain matrix and strain-
displacement matrix, respectively at the start time of the
solution.
If the nodal force output option is used in an element, the
residual forces corresponding to a displacement vector equal to
u kR are instead calculated, printed and stored.
Residual reactions, consistent with the residual

displacements.
During post-processing the AUI performs the response spectrum

analysis. You specify the following information during post-
processing, to obtain the response for a single ground motion
direction:
Loading response spectra (in terms of displacements,

velocities or accelerations) for various damping ratios
Ground motion direction
Damping ratios of the modes
Type of modal combination
Type of combination of residual terms
In the AUI, all of the information needed for the calculation for
one ground motion direction is entered into a response of type
response-spectrum.
To perform spatial combination calculations for more than one
ground motion direction, define a response of type response-
combination and include the response spectrum responses for the
individual ground motion directions in the definition of the
response-combination response.

To include the effects of a static load case, define a response of

type load-step for the static load case, then define a response of
type response-combination and include the response spectrum
responses and the static load case response in the definition of the
response-combination response.
See Section 13.1.4 for information about responses in the AUI.
Additional information about response spectrum analysis can be

found in the following references:
ref. R.W. Clough and J. Penzien, Dynamics of Structures,

McGraw Hill, 1974.
ref. U.S. Nuclear Regulatory Guide 1.92, Combining Modal

Responses and Spatial Components in Seismic Response
Analysis, Revision 1, February 1976.
9.2 Fourier analysis

Given a time history f(t), defined from t0 to t1, the Fourier series
corresponding to f(t) is

f ( t ) = a0 + ( an cos n1t + bn sin n1t )
n =1
where
2
1 = , t p = t1 t0
tp
t1
1
a0 =
tp f ( t ) dt
t0
t1 t1
2 2
an = f ( t ) cos n1t dt , bn = f ( t ) sin n t dt
1
tp t0
tp t0
Since

an cos n1t + bn sin n1t = cn cos ( n1t n )

where
bn
cn = an2 + bn2 , n = tan 1
an
f(t) may be written as

f ( t ) = c0 + cn cos ( n1t n )
n =1
Conceptually, the function f(t) is considered to contain the

frequencies 0, 1 , 21 ,... . . The constants cn and n are the
amplitude and phase angle of that portion of f(t) which oscillates at
frequency n1 .
A simple relationship (Parseval's identity) exists between the

Fourier coefficients and the mean square amplitude of f(t):
t1
1 1 2
f ( t ) dt = c02 + cn
2
tp t0
2 n =1
n1
The power spectral density of f(t) for frequency fn = may
2
be defined as
c02
S f ( f0 ) =
f
cn2
S f ( fn ) = , n = 1, 2,...
2f
1
where f = is the interval between two successive frequencies
2

(in cycles/unit time). The mean square value of f(t) may be

expressed in terms of the power spectral density using
t1
1
f
2
dt = S f ( f n ) f
tp t0 n =1
It is seen that the mean square value of f(t) is related to the area
under the power spectral density curve.
The AUI permits f(t) to be defined as the value of any variable

over any time interval contained within the solution time. The time
step need not be constant in the finite element analysis. The AUI
permits an additional time interval tgap during which f(t) is zero to
be added to the variable response. These concepts are shown in
Fig. 9.2-1.
Time history of variable: Time history used in Fourier analysis:

Variable f(t)
value
tp Periodic
extension of f(t)
l l l l
l l l tgap
l l l
l
l l
l
l
l l l
TSTART TEND Time t0 t1 Time
Figure 9.2-1: Fourier analysis
The Fourier series is truncated by choosing a highest frequency

fmax. This frequency is chosen based on the time step to avoid
aliasing errors:
0.5
f max
t max
( )
Hence, N = int t p f max gives the upper limit of the Fourier series
summation

N
f ( t ) = c0 + cn cos ( n1t n )
n =1
The Fourier coefficient integrals are evaluated using the

trapezoidal rule for numerical integration. These integrals are
typically integrated with less accuracy for higher frequencies.
Because a fast Fourier transform algorithm is not used, the
computer time required to form the Fourier coefficients is
proportional to N 2 .
9.3 Harmonic vibration analysis

It is frequently of interest to determine the steady-state
structural response to given harmonic and random loads. This
structural response is conveniently expressed as the superposition
of modal responses. This section summarizes the theory of
harmonic vibration analysis and discusses those features of the AUI
that perform harmonic vibration analysis.
Consider the finite element system
N
Mu + Cu + Ku = Fk ( t ) (9.3-1)
k =1
where N load cases are being considered. If we perform a mode

superposition analysis, the governing decoupled equations are
N
xi + 2ii xi + i2 xi = ik bk (t ) (9.3-2)
k =1
where ik is the modal participation factor for load case k and

mode i.
These equations apply for the following cases:
Case 1: Structure is grounded and forces Fk ( t ) are applied.

Here k is the load case number:

Fk ( t ) = Fk 0 bk ( t ) (9.3-3)
where bk ( t ) is a dimensionless multiplier for load case k. When

Fk 0 includes mass-proportional loads, the mass-proportional
load calculation does not include the mass coupling between
active and deleted degrees of freedom.
Case 2: A single ground motion is applied in direction k:
Fk ( t ) = Md k bk ( t ) (9.3-4)
where bk ( t ) is the ground acceleration in direction k. Only the

active degrees of freedom are included in this calculation. In
particular, any mass coupling between active and deleted
degrees of freedom is ignored.
Case 2 is seen to be a special case of case 1, except that in case 1

the displacements are interpreted as absolute displacements and in
case 2 the displacements are interpreted as relative displacements.
Each load case is seen to have a spatial variation and a time
variation. The time variation appears within the function bk ( t ) . In
harmonic analysis, we consider functions bk ( t ) of the form
bk ( t ) = bk0 sin (k t k ) (9.3-5)
where bk0 is a function of the forcing frequency k .
We also note here that the transient structure response is

ignored. In most cases, only the mean-square value of a response
quantity can be predicted. However, in harmonic vibration analysis
with all loads acting with the same loading frequency, the
maximum amplitude and phase of a response quantity can be
predicted as well.

Harmonic vibration analysis with all loads acting with the same
loading frequency
Consider the time variation
bk ( t ) = bk0 sin (k t k ) (9.3-6)
for load case k, in which each bk0 is a function of the forcing

frequency , and we write instead of k because the forcing
frequency is assumed to be the same for each load case k. In
this case, the steady-state modal response is
N
xi ( t ) = ( Aki sin (t k ) Bki cos (t k ) ) (9.3-7)
k =1
where
2

1
b i
k
Aki = k0 2 i (9.3-8)
i 2 2 2

1 + 2 i
i i

2
bk0 k
i i
Bki = i
(9.3-9)
i2
2 2 2
1 + 2 i
i i

and the steady-state solution is then
n N
u ( t ) = ( ) xki ( t )
i
(9.3-10)
i =1 k =1
where n is the number of eigenvectors considered in the mode

superposition. Equation (9.3-10) is of course only an

approximation to the steady-state solution response when n is less

than the total number of degrees of freedom in the model.
However the approximation may be adequate when enough
eigenvectors are considered and when harmonic loads of low
enough frequency are considered.
If we consider a single degree of freedom L, then
n N
uL( t ) = L x (t )
(i )
ki
i =1 k =1
n (i) N n (i ) N
= L Cki sin t L Dki cos t
i =1 k =1 i =1 k =1
(9.3-11)
where
Cki = Aki cos k Bki sin k

Dki = Aki sin k + Bki cos k
which can be rewritten as
u L( t ) = uL0 sin (t L ) (9.3-12)

where
2 2
n N
n N

uL0 = L(i ) Cki + L( i ) Dki (9.3-13)
i =1 k =1 i =1 k =1
n (i ) N
L Dki
L = tan i =n1
1 k =1
N
(9.3-14)
(i ) C

i =1
L
k =1
ki

uL0 and L are seen to be the maximum amplitude and phase

angle, respectively.
The quasi-static amplitude of the displacements for forcing

frequency is determined by assuming that the load bk ( t ) is

applied with a very low forcing frequency. One way to determine
the quasi-static amplitude for forcing frequency is to set to
zero within equations (9.3-8) and (9.3-9) (but using bk0
corresponding to forcing frequency ) and then evaluate equation
(9.3-13) as described above. This method is approximate because
the static structural response may not be well modeled in a modal
superposition analysis; in other words, the contribution of the
remaining modes to the static response may not be negligible.
The ratio of the dynamic maximum amplitude to the quasi-static
maximum amplitude is sometimes referred to as the dynamic
magnification factor or the dynamic load factor.
Equations (9.3-11) to (9.3-14) also apply to any quantity that is

proportional to the displacements, such as element forces, stresses
and reaction forces.
Considering the velocity and acceleration, we see from equation

(9.3-12) that

uL( t ) = uL0 sin t L (9.3-15)
2
and
u L( t ) = 2 u L0 sin (t ( L ) ) (9.3-16)
from which the maximum amplitude and phase are easily seen.
For more insight into the above equations, we consider only one
load case (N=1) and draw (9.3-6) and (9.3-12) in the complex
plane (Fig. 9.3-1). Notice that the actual response is the imaginary
component of the vectors.

Im
b
uL b0 sin7t - )
7t - uL0 sin7t - )
7t - Re
Length of b is b0 .
Length of u L is uL0 .
Figure 9.3-1: Force and displacement representations in the

complex plane
To obtain the in-phase or real part of the response, we set

t k to 90o. This orients the force vector to be entirely
imaginary (Fig 9.3-2). The in-phase part of the response is then
obtained from (9.3-12).
Im b
uL
7t - :;
uL0 sin7t - )
7t - Re
Figure 9.3-2: Obtaining the in-phase part of the displacements
To obtain the out-of-phase or imaginary part of the

response, we set t k to 0o. This orients the force vector to be
entirely real (Fig 9.3-3). The in-phase part of the response is then
obtained from (9.3-12).

Im
7t - ;
b Re
#7t - ) uL0 sin7t - )
uL
Figure 9.3-3: Obtaining the out-of-phase part of the displacements
The reason why we use 90o to obtain the real part of the
response is because equations (9.3-6) and (9.3-12) are written in
terms of the sine function, not the cosine function.
Additional information about harmonic vibration analysis can

be found in the following references:
ref. R.W. Clough and J. Penzien, Dynamics of Structures,

McGraw-Hill, 1974.
ref. W.C. Hurty and M.F. Rubinstein, Dynamics of

Structures, Prentice Hall, 1964.
Concerning the usage of the harmonic vibration analysis

feature, ADINA computes the natural frequencies, mode shapes
(including modal stresses and forces) and modal participation
factors. The type of modal participation factor can either be
applied load or ground motion. When the type of modal
participation factor is applied load, ADINA computes the load
vector for each time step in the analysis as if the analysis was a
linear static analysis. However ADINA does not solve for the
displacements corresponding to the load vector, rather ADINA
computes the modal participation factor.
During post-processing the AUI performs the harmonic

vibration analysis. You specify the following information during
post-processing:

Sweep spectrum giving the dependence of bk0 on the forcing

frequency
Load step number (for applied loads) or ground motion

direction (for ground motion loading)
Type of result (maximum amplitude, phase angle, RMS

amplitude or signed amplitude for a given value of t )
Then you can obtain either the entire structural response for a
single forcing frequency or the response at a point for a range of
forcing frequencies.
In order to obtain the in-phase or out-of-phase part of the

response, select signed amplitude for the type of result and set t
to 90 (to obtain the in-phase part of the response) or to 0 (to
obtain the out-of-phase part of the response). (We assume here that
the loading phase angle k is equal to 0.)
The harmonic vibration analysis information is entered into a

response of type harmonic. See Section 13.1.4 for information
about responses in the AUI.
9.4 Random vibration analysis

For general information about the theory of random vibration
analysis, see reference:
ref. W.C. Hurty and M.F. Rubinstein, Dynamics of

Structures, Prentice-Hall, 1964.
In the ADINA system, the only type of result that can be

predicted in random vibrations is the mean-square response,
defined as

T
1
p 2 = lim p
2
(t ) dt (9.4-1)
T 2T
T
for the function p( t ) . The mean-square response can also be

written as an integral in the frequency domain as
1 p ( )
2
T
T 2
p = lim
2
d (9.4-2)
0
T
in which

p ( ) = p ( t ) exp ( it ) dt (9.4-3)

is the Fourier transform of p ( t ) and
p ( ) = p ( ) p*( )
2
(9.4-4)
where p*( ) is the complex conjugate value of p ( ) . The

mean-square response in the frequency domain can also be written
in terms of the power spectral density function
y ( )
2
y ( ) = (9.4-5)
T
as follows:

1
y ( ) d = y ( f ) df
2 0
y (t ) =
2
(9.4-6)
0
where f is the frequency in cycles/(unit time). Using Fourier

transform theory, the Fourier transform of the modal response can
be written in terms of the Fourier transform of the loads:

N
xi ( ) = ik H i ( ) bk ( ) (9.4-7)
k =1
where
1
H i ( ) = (9.4-8)
( ) + i ( 2i i )
i
2 2
Therefore, for component u L ,
n N
uL( ) = L( ) ik H i ( ) bk ( )
i
(9.4-9)
i =1 k =1
and therefore
uL( ) =
2
n n N N
( ) ( )
i =1 j =1 k =1 l =1
L
i
L
j
i
k
lj H i ( ) H *j ( ) bk ( ) bl*( )
(9.4-10)
The following approximations are made to simplify (9.4-10):
Approximation 1)
2
bk ( ) bl*( ) bk ( ) kl (9.4-11)
Equation (9.4-11) is equivalent to neglecting the cross-

spectral densities of the loading combinations.
Approximation 2)
H i ( ) H *j ( ) H i ( ) H j ( ) (9.4-12)
Equation (9.4-12) is equivalent to ignoring the phase

relationships between the response functions.
For more information about approximation 2, see Hurty &

Rubenstein, specifically the discussion after equation (11.75) in

that reference.
Using eqs (9.4-11) to (9.4-12), equation (9.4-10) reduces to
( ) ( ) ) ( )( ) H ( ) H ( ) b ( )
N n n 2
u L( ) = L( ) L(
2 i j k k
i j i j k
k =1 i =1 j =1
(9.4-13)
Using the definition of the power-spectral-density, we write
( )( )
N n n
u L ( ) = (L ) (L ) ( ik )( kj ) H i ( ) H j ( ) bk ( )
i j
k =1 i =1 j =1
(9.4-14)
and therefore
( ) ( ) ( )( ) H ( f ) H
N n n
u = L
2
L
(i ) ( j)
L
k
i
k
j i j ( f ) bk ( f ) df
k =1 i =1 j =1 0
(9.4-15)
In addition to the two approximations (9.4-11) and (9.4-12),

ADINA also allows the user to make an additional approximation
in equations (9.4-14) and (9.4-15):
Approximation 3)
min (i , j )
H i ( ) H j ( ) = 0 for < CUTOFF
max (i , j )
(9.4-16)
When CUTOFF = 0, no approximation is made and when

CUTOFF = 1, eq. (9.4-16) reduces to
2
H i ( ) H j ( ) = H i ( ) ij
The default is CUTOFF=0. Approximation 3 with

CUTOFF=1 is discussed before equation (11.79) in Hurty &
Rubenstein.

In (9.4-15) and (9.4-16), the input power-spectral-density bk ( f )

satisfies

1
bk2 ( t ) = bk ( ) d = bk ( f ) df (9.4-17)
2 0 0
where bk2 ( t ) is the mean-square value of the applied load

(compare (9.4-6) and (9.4-17)). Different authors use different
conventions, for example

b ( t ) = bk ( ) d ,
2
k
0

bk2 ( t ) = b ( f ) df ,
k

bk2 ( t ) = b ( ) d
k

are all possible definitions of the input power-spectral-density.

However, ADINA accepts only the convention given in (9.4-17).
Equations (9.4-14) and (9.4-15) are used when the response
quantity is proportional to the displacements. When the response
quantity is proportional to the velocities or the accelerations, then
the transfer function given in equation (9.4-8) has to be modified:
i
Velocity : H i ( ) =
( ) + i ( 2ii )
i
2 2
2
Acceleration : H i ( ) =
(i2 2 ) + i ( 2ii )
Interpretation of mean-square responses: The mean-square
value of a response quantity can be used to predict the probability
that the actual response will exceed any given value. This is done
by assuming that the response has a Gaussian distribution and has

zero mean value. Then it may be shown that the standard deviation
is well approximated by the root-mean-square (RMS) value of the
response quantity:
= uL2 (9.4-18)
Hence we can write

Pr ( uL > ) = erfc (9.4-19)
2
where the value of the erfc function can be found in standard

tables:
Pr ( uL > ) = 0.3173
Pr ( uL > 2 ) = 0.0455
Pr ( uL > 3 ) = 0.0027
Concerning the usage of the random vibration analysis feature,

ADINA computes the same quantities as in harmonic vibration
analysis (see above).
During post-processing the AUI performs the random vibration

analysis. You specify the following information during post-
processing:
Power-spectral-density bk ( f ) . Note that bk ( f ) must

satisfy (9.4-17). If the input power-spectral-density is obtained
using another convention, you must convert it so that (9.4-17) is
satisfied.
Load step number (for applied loads) or ground motion

direction (for ground motion loading)
Then you can obtain either the RMS response or the power-
spectral-density of the response at a point.

The random vibration analysis information is entered into a

response of type random. See Section 13.1.4 for information about
responses in the AUI.
9.5 SDOF system response

It is frequently of interest to determine the maximum response
of single degree of freedom (SDOF) systems to SDOF support
displacement excitation as a function of the SDOF system natural
frequency.
The support is a node in the finite element mesh (called the

connection point) and the SDOF systems are considered to be
attached to the support. The SDOF systems are assumed to be
much lighter than the structure, so that the SDOF system response
does not affect the structural response.
This type of analysis is called floor response spectrum analysis.

The maximum SDOF system response can be computed when the
structure time history is given, or when the response spectrum
corresponding to structural ground motion is given.
As an example of this type of analysis, consider a machine

within a building that is much lighter than the building. Because
the machine is much lighter than the building, it is assumed that the
machine response does not affect the building response, and
therefore the machine response calculations can be done as a
postprocessing step once the building response (without the
machine) is known.
The response of the building to an earthquake can be computed

either using a time history analysis (when the earthquake ground
motion is given as a function of time) or using a response spectrum
analysis (when the earthquake ground motion is given as a
response spectrum). The response of the machine can also be
computed in each case (Figs. 9.5-1 and 9.5-2) by idealizing the
machine as a SDOF system attached to a point in the building.

m x(t)
c
uc(t) k uc(t)
..Ground acceleration
xg(t) ..
xg(t) input to ADINA as a mass-
proportional load
uc(t) calculated by ADINA
..
x(t)|max calculated by the AUI
Figure 9.5-1: Floor response spectrum analysis from time history

data
Modes 1,..., N calculated by ADINA.

..
x max calculated by the AUI.
..
m x max
.. c
xc max k
Ground acceleration given by response spectrum:

..
xg
2% damping 5% damping
Frequency
Figure 9.5-2: Floor response spectrum analysis from response
spectrum data

Considering many possible machines, each with a different

natural frequency and damping ratio, it is useful to create floor
response spectrum curves (Fig. 9.5-3). Each floor response
spectrum curve shows, for a given machine damping ratio, the
maximum machine response as a function of the machine natural
frequency. One possible use of the floor response spectrum curves
is in the selection of a machine mount that will minimize the
machine's motion in the event of an earthquake, for example, by
selecting the mount stiffness so that the machine has the natural
frequency with the least response.
Maximum
response Damping = 2%
Damping = 5%
Natural frequency of SDOF system

1 k
(f n = 2 4 m )
Figure 9.5-3: Floor response spectrum curves
Structural loading given by one or more time histories
The governing equation of motion for the SDOF system is
mu + cu + ku = kuc + cuc mxg (9.5-1)
where u is the relative displacement of the SDOF system to ground

motions applied as mass-proportional loads, uc is the relative
displacement of the connection point to ground motions applied as
mass-proportional loads and xg are ground accelerations applied as
mass-proportional loads. Equation (9.5-1) is written in terms of
relative displacements because the nodal displacements, velocities
and accelerations output by ADINA are relative to the ground

motion when ground motions are applied as mass-proportional

loads.
Equation (9.5-1) is numerically integrated using the Newmark
method with = 1/ 2, = 1/ 4 from rest initial conditions and the
maximum response is recorded. The maximum response can be the
maximum values of u , u , u , u uc , u uc , u uc , x , x ,
or x . The connection point displacement, velocity and
acceleration are taken directly from the ADINA output, so no
numerical integration is required to evaluate these quantities.
When the absolute displacement or velocity of the SDOF system
are required, the AUI numerically integrates the ground
accelerations to obtain ground velocities and displacements using
the trapezoidal rule.
The damping ratio used for the SDOF systems can be constant
or frequency-dependent.
The AUI can combine the maximum responses from several

time histories using the floor response spectrum from time history
command. In this case, the AUI computes the maximum response
separately for each time history and combines the results using
either the SRSS method or the absolute sum method (see Section
9.1).
This feature is most useful when you have several separate time
history analyses of a structure as separate ADINA runs, each time
history analysis giving the structural response to loading acting in a
different direction, and you want to compute the maximum
response corresponding to the loads acting simultaneously. First,
load the ADINA porthole files into a single database, using the
"append" feature to load all but the first porthole file. Then
compute the response corresponding to combined loads, in which
you define response-ranges for each time history, then specify each
response-range in a separate data input line. (Note: appending the
porthole files shifts the time intervals associated with each time
history so that the time intervals do not overlap.)
The spacing of SDOF system frequencies can be linear,

logarithmic or the recommended USNRC or ASME frequency
spacings, see references:
ref. 1992 ASME Boiler and Pressure Vessel Code, Rules for

Construction of Nuclear Power Plant Components,

Division 1, Appendix N.
ref. U.S. Nuclear Regulatory Guide 1.122, Development of

Floor Design Response Spectra for Seismic Design of
Floor-Supported Equipment or Components, rev. 1, Feb.
1978.
In addition, the SDOF system response is evaluated for each

frequency listed in a frequency table. This feature is useful for
evaluating the response at structural natural frequencies.
After the AUI constructs a response curve, it can broaden peaks

and smooth the curve. There are two peak broadening algorithms
that you can use: peak broadening=yes and peak broadening=all.
These are shown in Figure 9.5-4.
During smoothing, response values are raised (never lowered)
from their unsmoothed values until the curve segments are as close
to straight as possible. By default, it is set to 5 % of the maximum
response value in the curve.
Peak broadening can be used without smoothing, or smoothing
can be used without peak broadening, but these features are most
effective when used together.

Roof for peak with frequency f
Response
Broadened The floor response spectrum
curve commands construct a roof
Slope of roof for each peak on the
is maximum slope unbroadened curve. Set the
of unbroadened delta value in the frequency
curve near peak. table.
The floor response spectrum
Unbroadened commands raise points on
curve unbroadened curve to roof
for the peak.
(1 - delta value)f f (1 + delta value)f Frequency
a) Broadening algorithm used when peak broadening = yes
Roof for point with frequency f The floor response spectrum

Response
commands construct a roof

Broadened for each point on the
curve unbroadened curve. Set the
broadening factor in the floor
response spectrum commands.
Unbroadened
curve The floor response spectrum
commands raise points on
unbroadened curve to
highest possible roof.
(1 - broadening factor)f f (1 + broadening factor)f Frequency
b) Broadening algorithm used when peak broadening = all
Figure 9.5-4: Floor response spectrum broadening algorithms
Structural loading specified by response spectra
Frequently the structural loading is one or more ground motions

specified as response spectra. In this case it is not possible to
compute the structural response using a time history analysis.
Instead the structural response is estimated using the response
spectrum method (see Section 9.1).
The technique, which is presented in the following reference,

ref. K.A. Peters, D. Schmitz and U. Wagner, "Determination

of Floor Response Spectra on the Basis of the Response
Spectrum Method," Nuclear Engineering and Design,
Vol. 44, pp. 255-262, 1977.
can be summarized as follows:

Start with the eigenvalues, eigenvectors, modal participation
factors and modal damping ratios of the structure. These are
termed
i , i , ik , i
for modes i = 1,..., N and ground motion direction k. The

eigenvectors are of length n, where n is the number of degrees of
freedom of the structure. For example, the eigenvalues,
eigenvectors and modal participation ratios are calculated by
ADINA in response spectrum analysis and the modal damping
ratios are specified in the AUI (ADINA-PLOT).
The addition of a SDOF system to the structure contributes an
extra eigenvalue, eigenvector, modal participation factor and modal
damping factor to the mathematical model. These are termed
i* , i* , ( ik ) , i*
*
where the * is used to denote the structure plus SDOF system

mathematical model. The addition of the SDOF system also
modifies the structure eigenvalues, eigenvectors, modal
participation factors and modal damping factors for modes
i = 1,..., N :
i i* , i i* ,
ik ( ik ) , i i*
*
The eigenvectors are now of length n + 1 , with the additional entry

in each eigenvector denoted i0* .
Consider a SDOF system that vibrates in direction . In order
to estimate the maximum absolute acceleration of the SDOF system

due to ground motion in direction k, we use the response spectrum

formula
(x )
k
0 max = modal-combination of ( x0ki )
max
, i = 0,..., N (9.5-2)
where
(x ) = ( ik ) i*0 S a (i* , i* )
k *
0 i max (9.5-3)
is the maximum acceleration due to mode i and S a i* , i* is the ( )

response spectrum acceleration corresponding to natural frequency
i* and damping ratio i* . Any of the modal combination methods
listed in Section 9.1 can be used for the modal combination
formula. For example, if the SRSS method of modal combination
is used,
1
N
2 2
(x ) = ( ik ) i*0 S a (i* , i* )
k *
(9.5-4)
0 max
i =0

( ) ,
*
We see from eq. (9.5-3) that we need ik *
i0 , i = 1,..., N and
( )
k *
0
*
00 to apply the response spectrum method to the structure
plus SDOF system mathematical model. From the last reference
above, if the SDOF system is very light compared with the
structure,

i
* i
(9.5-5a)

*
i 0
i* i , ( ik ) ik , i* i , i = 1,..., N
*
(9.5-5b,c,d)

Also,
1
i*0 2
ia (9.5-6)

1 i
0
where a denotes the degree of freedom corresponding to the node

to which the SDOF system is connected and direction . (We note
that the fraction in this equation is similar to the transfer function
for a SDOF system with natural frequency 0 and zero damping
subjected to harmonic support motion). Finally,
n
( 0k ) 00* = kl ( ik ) i*0
* *
(9.5-7)
i =1
where k =1 for k= and k =0 otherwise. Equation (9.5-7)

follows from equation (9.1-8).
Since we normally have fewer than n eigenvectors, we make the
approximation
N
( )
k *
kl ( ik ) i*0
* *
0 00 (9.5-8)
i =1
by neglecting
n n
1
( ) i
k * *
i0 = i
k
2
ia (9.5-9)
i = N +1 i = N +1
1 i
0
From the right-hand-side of equation (9.5-9), we see that the

neglected term is small provided that the natural frequency of the
SDOF system is much smaller than the natural frequencies of the
neglected modes.
Introducing equations (9.5-5), (9.5-6) and (9.5-7) into equation
(9.5-3) gives the response spectrum approximation to the maximum
absolute acceleration of the SDOF system. We term this

approximation the undamped transfer function, no resonance

correction method because equation (9.5-6) corresponds to an
undamped transfer function and there is no check for the SDOF
system natural frequency being close to a structural natural
frequency.
When the SDOF system natural frequency is close to a
structural natural frequency, the denominator of equation (9.5-6)
tends to zero. In this case, we use the following equations
(x )
k
0 max = modal-combination of ( x0ki )
max
and ( x0k j )
max
,
i = 1,..., N , i j
(9.5-10)
where
( )
k * *
S a ( *j , *j )
(x )
k
=
j ja
(9.5-11)
2 0 ( 0 + j )
0 j max
is an estimate of the maximum acceleration of the SDOF system

due to the motion of that mode j with natural frequency closest to
the natural frequency of the SDOF system. It is recommended that
the results using both equations (9.5-2) and (9.5-10) be calculated
and the smaller result taken; we term this method the undamped
transfer function, resonance correction method.
We have found that equation (9.5-10) can be smaller than
equation (9.5-2) even for SDOF system natural frequencies not
close to any structural natural frequency, for example when the
structural motion of mode j does not affect the SDOF system
motion. When this happens, the response predictions are not
accurate. Therefore we introduce a third method, the damped
transfer function method, which is identical to the undamped
transfer function, no resonance correction method, except that
equation (9.5-6) is replaced by

2

1 + 2 eff i
0
i*0 2
ia (9.5-12)
2
2
1 i + 2 eff i
0 0

where + is used if i 0 and - is used otherwise. The term eff

in this equation is evaluated using one of the following choices:
eff = 0 (9.5-12a)
eff = 0 ( 0 + j ) (9.5-12b)
eff = 02 ( 0 j ) , + if 0 j , if 0 > j (9.5-12c)

2
Equation (9.5-12) corresponds to the transfer function for a damped

SDOF system subjected to support motion. The damped transfer
function method avoids numerical difficulties when the SDOF
system natural frequency equals a structural natural frequency and
gives results close to the undamped transfer function, no resonance
correction method otherwise, provided that the SDOF damping
ratio is small (which is usually the case).
The maximum relative velocity and maximum relative

displacement corresponding to ground motions in direction k can
be estimated using similar techniques by substituting the spectrum
velocity Sv or spectrum displacement Sd for the spectrum
acceleration Sa in the above formulas. Note that, although the
calculated acceleration is an absolute acceleration, the velocities
and displacements are relative to the ground motion.
Once the responses are known for a single load case, the
response for a combination of load cases can be computed as
follows:

( x0 )max = spatial-combination of ( x0k )max , k = 1,..., N (9.5-13)
For example, when the SRSS method is used to combine three load
cases, each load case corresponding to ground motions in one of
the three coordinate directions,
1
3 2 2
( x0 )max = ( x0k ) (9.5-14)
k =1
The options available for selecting SDOF system frequencies,

damping values, broadening peaks and smoothing curves are
exactly the same as when calculating floor response spectrum
curves using time history data, see above.
Note on the use of the above theory with the CQC method of
modal combinations. When the above theory is used, along with
equation (9.5-12), and along with the CQC method, the resonances
tend to be underestimated. This is because, in the CQC method,
the magnitude of the predicted resonance depends on the
cancellation of the quantities M r , where M r is the maximum
modal response component for mode r. To be specific, as two
modes r and s approach each other, the contribution of these modes
to the CQC sum approaches ( M r + M s ) . When equation (9.5-12)
2
is used, then M r and M s remain finite, with nearly equal

magnitudes, and opposite signs, hence the CQC sum tends to zero.
But when equation (9.5-6) is used, then M r and M s grow
unbounded, with nearly equal magnitudes, and opposite signs, so
that the CQC sum remains large.


10.1 Overview
10. Fracture mechanics

10.1 Overview
Linear and nonlinear fracture mechanics analysis can be
performed with the ADINA system.
The following capabilities are included:
Computation of the J-integral in 2-D finite element models,

using the line contour method. See Section 10.9 for more details.

using the virtual crack extension method with nodal virtual shifts.
See Section 10.10 for more details.
The analysis of actual crack propagation in 2-D finite element

models with symmetric geometry and loading. See Section 10.11
for more details.

using the virtual crack extension method with nodal virtual shifts
(NVS method) . See Section 10.14 for more details.
Computation of the J-integral and stress intensity factors

K I , K II , K III in 3-D finite element models, using the virtual crack
extension method with station virtual shifts (SVS method) . See
Section 10.15 for more details.
The fracture mechanics computations can be performed for the

following element types: 2-D solid and 3-D solid. The material
models that can be used are: isotropic-elastic, orthotropic-elastic,
thermo-isotropic, thermo-orthotropic, plastic-orthotropic,
plastic-bilinear, plastic-multilinear, thermo-plastic, creep,
plastic-creep, multilinear-plastic-creep, creep-variable, plastic-
creep-variable, multilinear-plastic-creep-variable, plastic-
cyclic, user-supplied. Any of the kinematic formulations that are
allowed for the chosen materials can be used: small
displacement/small strain, large displacement/small strain, and
large displacement/large strain. The only exception is that the

Chapter 10: Fracture mechanics
ULJ formulation cannot be used for large strain analysis of plastic

materials.
Table 10.1-1 gives some of the basic features for ADINA fracture
mechanics:
Table 10.1-1: Basic features for ADINA fracture mechanics
Line contour NVS method SVS method

2-D model, stationary > >
crack
2-D model, propagating > >
crack
3-D model > >
More than one crack in >
the same analysis
Calculation of the J- > > >
integral
Calculation of stress > (note 1)
intensity factors
Mapped mesh near > (notes 2, 3) > (notes 2, 3) > (note 3)
crack front
Free-form mesh near >
crack front
Blunt crack front > >
(coincident nodes on
the crack front)
Note 1: Only when the elements that are in the SVS domain surrounding the crack are all
linear elastic isotropic without thermal effects, and when the analysis is static.
Note 2: See Section 10.14 for a detailed description of the meshing required for the NVS
method in 3-D analysis.
Note 3: The CRACK-M features for meshing 3-D cracks can be used both with the NVS and
SVS methods.

10.1 Overview
Additional information about the theory used in ADINA can be

found in the following references:
ref. T.L. Anderson, Fracture Mechanics, Fundamentals and

Applications, 3rd edition, CRC Press, 2005.
ref. S.N. Atluri, "Path-independent integrals in finite

elasticity and inelasticity, with body forces, inertia, and
arbitrary crack-face conditions," Engng. Fracture Mech.,
Vol. 16, pp. 341-364, 1982.
ref. H.G. deLorenzi, 'On the energy release rate and the J-
integral for 3-D crack configurations', Int. J. Fract., Vol.
19, pp. 183-193, 1982.
ref. O. Guillermin, "Some Developments in Computational

Methods for Nonlinear Fracture Mechanics,"
Proceedings IKOSS FEM '91, Baden-Baden, Germany,
1991.
ref. T.K. Hellen, "On the Method of Virtual Crack

Extensions," Int. J. Num. Meth. Engng., Vol. 9, pp. 187-
207, 1975.
ref. J.K. Knowles and E. Sternberg, "On a Class of

Conservation Laws in Linearized and Finite
Elastostatics," Archiv. Rat. Mech. Anal., Vol. 44, pp.
187-211, 1972.
ref. D.M. Parks, "The Virtual Crack Extension Method for

Nonlinear Material Behavior," Comp. Meth. Appl. Mech.
Engng., Vol. 12, pp. 353-364, 1977.
ref. J.R. Rice, "A Path Independent Integral and the

Approximate Analysis of Strain Concentration by
Notches and Cracks," J. Appl. Mech., Vol. 35, pp. 379-
386, 1968.
ref. C.F. Shih, B. Moran, T. Nakamura, "Enegy release rate

along a three-dimensional crack front in a thermally
stressed body", Int. J. Fract., Vol. 30, pp. 79-102, 1986.

ref. C.F. Shih, R. J. Asaro, "Elastic-Plastic Analysis of

Cracks on Bimaterial Interfaces: Part I Small Scale
Yielding", J. Appl. Mech., Vol. 55, pp. 299-316, 1988.
ref. M.C. Walters, G. H. Paulino, R.H. Dodds, "Interaction

Integral Procedures for 3-D Curved Cracks Including
Surface Tractions" Engr. Fract. Mech., Vol. 72, pp.
1635-1663, 2005.
10.2 The energy release rate

The energy release rate is defined as
d
G= (10.2-1)
dA
where is the total potential energy and A is the crack area.

is defined by
=U Ru (10.2-2)
where U is the internal energy (strain energy) and Ru is the

sum of the work done by each applied load R moving through its
corresponding displacement u (with the load R held constant as
the displacement increases from 0 to u ). Note that A is the area
of one side of the crack, e.g, if the crack has length a and
thickness b , the crack area is A = ab .
d
The operation is interpreted as follows:
dA
d A
= lim A+A (10.2-3)
dA A0 A
that is, the difference of the total potential energy for two slightly
different crack areas divided by the change in crack area.

10.2 The energy release rate
Although, in theory, the energy release rate could be calculated

using two finite element analyses and (10.2-3), the energy release
rate is more efficiently calculated using the line contour or virtual
crack extension methods, as described below.
10.3 Stress intensity factors

Consider a straight crack within a linear isotropic material under
static conditions without thermal effects or body forces. The
displacement and stress fields close to the crack tip under plane
stress or plane strain conditions can be written in terms of three
fundamental fields, corresponding to the three fundamental modes
of deformation I, II, III. These three fundamental modes of
deformation are shown in Fig. 10.3-1.
a) Mode I b) Mode II c) Mode III
Figure 10.3-1: Fundamental modes of deformation near the crack front
Mode I:
1 r 2
u1I =
2 2 cos 2 1 + 2sin 2 (10.3-1a)


1 r 2
u2I =
2 2 sin 2 + 1 2 cos 2 (10.3-1b)

u3 = 0
I
(10.3-1c)
1 3
11I = cos 2 1 sin 2 sin 2 (10.3-1d)
2 r
1 3
22I = cos 2 1 + sin 2 sin 2 (10.3-1e)
2 r
1 3
12I = cos 2 sin 2 cos 2 (10.3-1f)
2 r
13I = 0 (10.3-1g)
23I = 0 (10.3-1h)
Mode II:
1 r 2
u1II =
2 2 sin 2 + 1 + 2 cos 2 (10.3-2a)

1 r
u2II = cos 1 2sin 2 (10.3-2b)
2 2 2 2
u3II = 0 (10.3-2c)
1 3
11II = sin 2 2 + cos 2 cos 2 (10.3-2d)
2 r
1 3
22II = sin cos cos (10.3-2e)
2 r 2 2 2
1 3
12II = cos 2 1 sin 2 sin 2 (10.3-2f)
2 r
13 = 0
II
(10.3-2g)
23II = 0 (10.3-2h)

Mode III:
u1III = 0 (10.3-3a)
u2III = 0 (10.3-3b)
1 2r
u3III = sin (10.3-3c)
2
11III = 0 (10.3-3d)
22III = 0 (10.3-3e)
12III = 0 (10.3-3f)
1
13III = sin (10.3-3g)
2 r 2
1
23III = cos (10.3-3h)
2 r 2
E
In the above, = is the shear modulus,
2(1 + )
3 4v plane strain

= 3 v (10.3-4)
1 + v plane stress
and, for future reference,
E
plane strain
E = 1 v 2 (10.3-5)
E plane stress
In all cases,
v ( 11 + 22 ) plane strain
33 = (10.3-6)
0 plane stress

These formulas refer to a polar coordinate system with origin at the

crack, as shown in Fig. 10.3-2.
x2
r
q
x1
Figure 10.3-2: Geometry and coordinate system

for stress intensity factor fields
These formulas are well-known and can be found in the book by

Anderson, see the list of references given above.
By superposition, the displacement and stress fields close to the

crack tip can be written in terms of these fundamental fields and in
terms of three multipliers K I , K II , K III :
ui = K I uiI + K II uiII + K III uiIII + uiR (10.3-7)

ij = K + K + K
I
I ij
II
II ij
III
III ij + R
ij (10.3-8)
K I , K II , K III are referred to as the stress intensity factors. Terms

uiR , ijR are referred to as "residual" fields, and are expected to be
much smaller than the other terms, near the crack tip. The residual
terms are added here in anticipation of the developments in Section
10.6.2.
The dimensions of the multipliers K I , K II , K III are

[stress] [length] , so that the units in the SI system are
MPa m .

A positive value of K I is associated with an opening of the crack,

and a negative value of K I is associated with closing of the crack.
Therefore it is expected that K I is positive. However, the other
multipliers K II , K III can either be positive or negative.
10.4 Calculation of the J-integral using the line contour

integration method
The line contour integration method is used in two-dimensional
analysis to calculate a contour independent parameter (J-integral)
characterizing the severity of the displacement, stress and strain
fields at the tip of a crack. The J-integral is given by (see Fig.
10.4-1)
u
J = Wdx2 ij n j i ds (10.4-1)
x1
where = line contour enclosing the crack tip, ij =components of

the stress tensor, ui = components of the displacement vector, n j =
components of the unit vector normal to , ds =length increment
along , W= strain energy density. It is assumed that the crack is
oriented along the x1 direction.
The detailed calculation of the strain energy density is discussed
in Section 11.7.

x2
x1
n
ds
'
Figure 10.4-1: Arbitrary contour around a crack tip

for J-integral calculation
Note that initial strains are not included in the computation of

the J-integral.
For linear elastic materials, the J-integral is related to the stress

intensity factors K I , K II and K III described in Section 10.3 by
K I2 K II2 K III2
J= + + (10.4-2)
E E 2
The J-integral as defined above is equivalent to the energy

release rate under the following conditions:
Isothermal analysis
No body forces and inertia forces; no pressure on the crack

faces
Monotonic and proportional loading; no unloading
Material homogeneous in the x1 direction

10.4 Calculation of the J-integral using the line contour integration method
The J-integral concept can be extended to large displacement

and/or large strain conditions. The J-integral becomes
ui 0
J= W dx2 0T ji 0 n j
0
0 ds (10.4-3)
0

0 x1
in which the left superscript / subscript 0 denotes in the original

configuration, or referred to the original configuration. 0Tij are
the components of the 1st Piola-Kirchhoff stress tensor, defined as
0 ijT = ( det 0t X ) 0t X ik kj (10.4-4)
Under large displacement and/or large strain conditions, J is

path-independent only when the net forces acting on the contour
are zero, i.e.

0
T
0 ji 0 n j 0ds = 0 (10.4-5)

Usually this is the case for contours used in fracture mechanics.
10.5 Calculation of the J-integral using the virtual crack

extension method
In the virtual crack extension method, the J-integral is
calculated using the expression
1 u j X k
G=
Ac V xk
ij W ik
xi
dV (10.5-1)
where V= volume of the cracked body, X k = components of the

virtual crack shift vector, Ac = increase in crack area
corresponding to X k , ij= Kronecker delta, and W = strain
energy density. The calculation of W is discussed in Section 11.7.
In (10.5-1), we have not yet included several corrections described

in more detail below. If there are initial strains, these initial strains
are used in the computation of the stresses and strain energy
density.
We use the symbol G to denote the J-integral when the J-

integral is calculated using a volume integral, as in (10.5-1). Other
authors use the symbol J to denote the J-integral regardless of
how the J-integral is calculated.
The calculation of G using (10.5-1) relies on the definition of a

virtual shift vector field X , which is defined throughout the
volume. The basic concept of the virtual shift vector field is shown
in Figs 10.5-1 and 10.5-2. On the crack front, the virtual shift
vector field specifies the virtual shift of the crack. Within a tube
that surrounds the crack front, the virtual shift vector field specifies
the shift of material particles close to the crack front. Outside the
tube, the virtual shift vector field is zero, and material particles
outside of the tube are not shifted.
The virtual shift vector field X is assumed to vary smoothly
throughout the tube, so that this field can be differentiated as
required in equation (10.5-1).
x2
x1
Crack front
Spacing between cracked

surfaces enlarged for clarity
Tube surrounding crack tip
= Vector in vector field DX
Figure 10.5-1: The virtual crack extension method

in two-dimensional analysis

10.5 Calculation of the J-integral using the virtual crack extension method
Crack front Virtual shift of crack front
Tube surrounding
crack front
Spacing between cracked

surfaces enlarged for clarity
= Vector in vector field DX

in three-dimensional analysis
In three-dimensional analysis, it is convenient to introduce an

orthonormal coordinate system aligned with the crack front, as
shown in Fig. 10.5-3). Direction a is normal to the crack front, in
the plane of self-similar crack propagation, direction b defines the
plane of self-similar crack propagation and direction c is tangent
to the crack front.
b
a
Crack front
Figure 10.5-3: Crack front coordinate system
The virtual shift vector field X causes the crack area to

increase according to the formula

Ac = a X ds (10.5-2)
where ds is the differential length along the crack front (Fig. 10.5-
4) and the integration is performed along the crack front line.
Notice that it is not necessary for X to be perpendicular to the
crack front, however X must lie in the plane defined by direction
b.
For a 2-D crack, this formula simplifies to become
Ac = t X = t X 12 + X 22 (10.5-3)
where t is the thickness at the crack tip ( t = x1 for an axisymmetric

model). X must lie in the line of self-similar crack advance.
X a
Crack front
ds c
Figure 10.5-4: Calculation of virtual crack area increase
Because the integral in (10.5-1) is proportional to the magnitude

of X , and because Ac is also proportional to the magnitude of
X , the value of G is unaffected by the magnitude of X .
The theoretical value of G is not affected by the size or shape

of the tube, or by the values of the virtual shift field X away
from the crack front line. Thus G , like the J-integral, is said to be
path-independent.

Hoop stress correction: In axisymmetric analysis with 2-D

elements, the following correction must be added to G :
1 u x
Ghoop =
Ac
V

xx
x
W X 2 dV

(10.5-4)
where X 2 is the virtual crack extension component in the y

direction.
Thermal correction: For the analysis of structures with temperature

dependent material properties, the effect of thermal loads must be
taken into account in the calculation of the J-integral. The
following correction should be added to G :
1 eijth
Gth =
Ac V
ij
xk
X k dV (10.5-5)
in which eijth are the components of the thermal strain tensor and
is the temperature. Note that, for those materials in which the
material constants are temperature-dependent, the effect of material
constants changing with temperature is not included. (However,
the effect of thermal coefficient changing with temperature is
included.)
Pressure correction: When there are tractions acting on the crack

faces, the following correction should be added to G :
1 ui
G pr =
Ac t S
i
x j
X j dS (10.5-6)
where ti = components of the applied surface traction vector and S=

surface of the cracked body. For example when pressures are
applied, ti = pni , in which p is the pressure and ni is the
outwards surface normal.
Note that tractions from contact conditions and tractions resulting

from fluid loadings in FSI analysis are not included in equation

(10.5-6).
Dynamic correction: In dynamic analysis, the effect of inertia

forces is included in the following correction, which should be
added to G :
1 ui
Gdyn =
Ac
V
ui
x j
X j dV (10.5-7)
in which is the mass density and ui are the components of the

acceleration vector.
Large displacements / large strains: The above expressions are

modified for large displacements / large strains by referring all
quantities to the original configuration, and by using the transpose
of the 1st Piola-Kirchhoff stresses instead of the Cauchy stresses.
Note that the virtual shift vector field X is always defined in the
original configuration.
G plus all of the above corrections is equivalent to the energy

d
release rate under the following conditions:
dA
No body forces.
Monotonic, proportional loading in inelastic analysis
Since the virtual shift vector function is arbitrary, the virtual

shift vector field X can be specified in different ways. There are
two fundamental approaches that can be taken:
Virtual shift vectors specified at the nodes, then these vectors

interpolated within the elements using the element shape
functions. This technique is widely used. In ADINA, this
technique is termed the NVS (nodal virtual shift) algorithm, and
is described in Section 10.10 for 2-D cracks and in Section
10.14 for 3-D cracks.
Because the virtual shift vectors are specified at the nodes, it

is necessary to employ structured meshes at the crack front.
Virtual shift vectors specified directly as a function of

position. In ADINA, this technique is termed the SVS (station
virtual shift) algorithm, and is described in Section 10.15 for
3-D cracks.
A fundamental advantage of specifying the virtual shift
vectors directly as a function of position is that unstructured
(free-form) meshes can be used at the crack front.
10.6 Calculation of stress intensity factors using the virtual

crack extension method
10.6.1 Overview of theory
The theory in this section is based on material given in the

references by Shih, Moran and Nakamura, and Walters, Paulino
and Dodds.
In order to calculate stress intensity factors, it is useful to view G

as an "interaction integral".
We concentrate on the special case of linear isotropic elastic

analysis without thermal or dynamic effects. For this special case,
1
W = ij eij , and ij = Cijkl ekl , where Cijkl is the constitutive
2
tensor.
We now consider an integral closely related to (10.5-1):
1 A u Bj X k
G (u , u ) = V ij xk

A B A, B
W ik dV
Ac x
i
(10.6-1)
1
in which W A, B = ijA eijB . This integral measures the interaction of
2

the stress field corresponding to displacement field u A and the

displacement gradient field corresponding to displacement field
u B . The superscript on the displacement field denotes a field label.
Anticipating the developments below, we will use only the

symmetric form of (10.6-1), namely
1
G (u A , u B ) =
2
(G (u A , u B ) + G (u B , u A ))
1 1 A u Bj u Aj X k
Ac V 2 xk
= + B
W A, B
ik dV
xk xi
ij ij

(10.6-2)
1 A B
in which W A, B =
4
( ij eij + ijB eijA ) . Because ij = Cijkl ekl in
which Cijkl = Cklij , ijA eijB = ijB eijA and therefore
1 1
W A, B = ijA eijB = ijB eijA .
2 2
Equation (10.6-2) (along with some corrections discussed below) is

the basis of the interaction integral method as implemented in
ADINA.
In (10.6-2), G is seen to be a function of two displacement fields

u A , u B . Clearly, if u A = u B = u , (10.5-1) is recovered, and in this
case, we write
G (u) = G (u, u) (10.6-3)
Also,
G (u B , u A ) = G (u A , u B ) (10.6-4)
In addition, G (u A , u B ) is linear in each field u A , u B , so that, for

example,

10.6 Calculation of stress intensity factors using the virtual crack extension method
G ( u A + u B , uC ) = G (u A , uC ) + G (u B , u C ) (10.6-5)
in which u A , u B , uC are three displacement fields and , are

arbitrary scalars.
The advantage of writing G in the form (10.6-2) is that the

fundamental displacement fields used in the definitions of the stress
intensity factors (given in Section 10.3) are orthogonal to each
other, when introduced into G . In fact, analytically, for a straight
crack,
1
G (u I ) = (10.6-6a)
E
1
G (u II ) = (10.6-6b)
E
1
G (u III ) = (10.6-6c)
2
G (u , u ) = 0
I II
(10.6-6d)
G (u I , u III ) = 0 (10.6-6e)
G (u II , u III ) = 0 (10.6-6f)
These results can be obtained by direct computation. In this

computation, we assume a straight enclosed crack under plane
strain conditions, aligned with the global coordinate system, as
shown in Fig. 10.6-1.

y
x
a) Virtual shift domain for enclosed crack
X=(qrqz , 0, 0)
r q
Top crack face
x
Bottom crack face
R
b) Side view of virtual shift domain
Figure 10.6-1: Virtual shift domain used to demonstrate

orthogonality of fundamental fields in an
enclosed crack

We also assume that the virtual shift vector is given by
X = (qr qz , 0, 0) (10.6-7)
in which qr is a function of radius r , with qr = 1 at r = 0 and

with qr = 0 at r = R , and in which qz is a function of axial
coordinate z . Clearly
X qr
=( qz cos , 0, 0) (10.6-8a)
x r
X qr
=( qz sin , 0, 0) (10.6-8b)
y r
X q
= (qr z , 0, 0) (10.6-8c)
z z
The corresponding virtual shift area is
Ac = qz dz (10.6-9)
By directly substituting the fundamental fields from Section 10.3

into equation (10.6-2) and performing the integrations analytically,
the following results are obtained:
G (u I ) G (u I , u II ) G (u I , u III )

G (u II ) G (u II , u III ) =
sym G (u III )

1/ E 0 q
0
qr dr z dz 0 0 0

1/ E 0 + z
0 (1 + 8 ) /16
sym 1/ ( 2 ) qz dz sym 0
(10.6-10)
The second term on the right-hand-side disappears when

qz
z
dz = 0 , in other words when the net change of qz along the

z direction is zero. We will therefore choose qz so that this

condition is satisfied whenever possible. Equations (10.6-6)
directly follow from (10.6-10).
For a boundary crack, the same procedure can be used. There are
two cases: boundary crack with binormal direction pointing into the
material and boundary crack with binormal direction pointing out
of the material. These cases are shown in Figure 10.6-2.
qz
And, under the condition z
dz = 0 , the results
G (u I ) G (u I , u II ) G (u I , u III ) 1/ ( 2 E ) 0 0
III
G (u ) II
G (u , u ) =
II
1/ ( 2 E ) 0
sym
G (u III ) sym 1/ ( 4 )
(10.6-11)
are obtained, regardless of whether the binormal points into or out

of the material.
Considering the hoop stress, thermal, pressure and dynamic

corrections to G , the thermal and dynamic corrections do not
apply because the fundamental fields only apply to linear static
analysis without thermal effects. The hoop stress correction to
(10.6-2) can be written
1 1 A u xB B u x
A

Ghoop (u A , u B ) = V 2 xx x xx x W X 2 dV
+ A, B
Ac
(10.6-12)
The pressure correction can be written
1 1 A uiB B uiA
G pr (u A , u B ) =
Ac S 2 ti x j + ti x j X j dS

(10.6-13)

X=(qrqz , 0, 0)
R r q
Crack face
x
a) Binormal direction points into the material
Crack face x
r -q
R
X=(qrqz , 0, 0)
b) Binormal direction points out of the material
Figure 10.6-2: Virtual shift domains used to demonstrate

orthogonality of fundamental fields in
boundary cracks
There is an issue when calculating (10.6-13) for the fundamental

fields u I , u II , u III because the tractions for the fundamental fields
are unknown. Following the procedure in the reference by Walters,
Paulino and Dodds, we assume that:
The fundamental fields u I , u II , u III apply without

modification when there is a pressure load applied to the crack
The traction for the fundamental fields is zero.

10.6.2 Calculation of stress intensity factors
We use G in the interaction integral form (equation (10.6-2) and

also the corrections from equations (10.6-12) to (10.6-13)) to
calculate the stress intensity factors, as follows.
Let u be the displacement field from the finite element solution.

Then (10.3-7) can be rewritten as
u R = u K I u I K II u II K III u III (10.6-14)
Hence
G (u R ) = G (u) K I G (u, u I ) K II G (u, u II ) K III G (u, u III )

+ K I2G (u I ) + 2 K I K II G (u I , u II ) + 2 K I K III G (u I , u III )
+ K II2 G (u II ) + 2 K II K III G (u II , u III )
+ K III2 G (u III )
(10.6-15)
Now we choose K I , K II , K III to minimize G (u R ) . The result is
G (u I ) G (u I , u II ) G (u I , u III ) K I G (u, u I )

G (u II ) G (u II , u III ) K II = G (u, u II )
sym G (u III ) K III G (u, u III )

(10.6-16)
The same result is obtained using the unsymmetrized G , because

the minimization of G (u R ) gives the matrix equation

2G (u I ) G (u I , u II ) + G (u II , u I ) G (u I , u III ) + G (u III , u I ) K I

2G (u II ) G (u II , u III ) + G (u III , u II ) K II
sym 2G (u III ) K III

G (u, u I ) + G (u I , u)

= G (u, u II ) + G (u II , u)
G (u, u III ) + G (u III , u)

(10.6-17)
Only the symmetric part of G appears in (10.6-17), hence (10.6-

16) and (10.6-17) are the same equation. Therefore it is easier to
simply define G as the symmetric part of G , then employ G
throughout.
10.6.3 Discussion
This procedure is very closely related to the interaction integral

method described in, for example, the reference by Shih and Asaro
and the reference by Walters, Paulino and Dodds. In these
references, the fields u I , u II , u III are referred to as auxiliary fields.
Since the interaction integrals involving only the fundamental

fields (e.g. G (u I ) , G (u I , u II ) , etc.) are analytically known, either
the analytical values of these integrals can be used in (10.6-16), or
the numerical approximations of these integrals can be used.
For example, in the references by Shih, Moran and Nakamura
and Walters, Paulino and Dodds, the analytical values are used.
However, in ADINA, the numerical approximations are used. In
this way, the same numerical integration procedure is used on both
sides of (10.6-16).
The calculation of stress intensity factors using the virtual crack

extension method is implemented in ADINA using the SVS
algorithm, as discussed in Section 10.15.

10.7 Fracture control parameters

Here and below, we describe the input for fracture mechanics in
terms of the command-line commands of the AUI. This input can
also be entered using the AUI user interface. See the Command
Reference Manual for ADINA for detailed information about the
commands.
MASTER command: The FRACTURE parameter of the MASTER

command must be set to YES:
MASTER ... FRACTURE=YES
FRACTURE command: The FRACTURE command specifies the

type of fracture analysis:
FRACTURE TECHNIQUE METHOD DIMENSION TYPE,

PRESSURE TEMPERATURE DYNAMIC
TECHNIQUE: TECHNIQUE is always set to STANDARD.
METHOD: METHOD can be set to the following values:
LINE-CONTOUR: The J-integral is calculated using line

contours. This feature is available only for 2-D cracks.
VIRTUAL-CRACK-EXTENSION: The J-integral is calculated

using the NVS method for virtual crack extension. The virtual
shifts are expressed in terms of nodal point virtual shifts.
BOTH: Same as using LINE-CONTOUR and also VIRTUAL-

CRACK-EXTENSION.
SVS: The J-integral and the stress intensity factors

K I , K II , K III are calculated using the SVS method for virtual
crack extension. The virtual shifts are evaluated at crack
advance stations. This feature is available only for 3-D cracks.
The default is VIRTUAL-CRACK-EXTENSION.
DIMENSION: DIMENSION=2 for 2-D models and =3 for 3-D

10.7 Fracture control parameters
models. The default is 2.
TYPE: TYPE specifies whether the crack is stationary or

propagating (default is stationary).
PRESSURE, TEMPERATURE, DYNAMIC: PRESSURE,

TEMPERATURE, DYNAMIC specify whether or not to compute
the pressure, temperature and dynamic corrections when
calculating virtual shifts. The default is to calculate the corrections.
Other commands
Other commands are discussed in the sections listed below:
2-D cracks
Geometry and meshing: Section 10.8
Method:
Line contour: Section 10.9
NVS virtual shift: Section 10.10
Type:
Propagating: Section 10.11
3-D cracks
Geometry and meshing, general comments: Section 10.12
Meshing using CRACK-M features: Section 10.13
Method:
NVS virtual shift: Section 10.14
SVS virtual shift: Section 10.15
10.8 2-D crack geometry and meshing

10.8.1 Overview
The crack line or surface can be located on the boundary of the

finite element model (symmetric specimen and loading) or inside
the finite element model (asymmetric specimen and/or loading).
However, for crack propagation analysis, the specimen and loading
must be symmetric.
Fig. 10.8-1 shows geometries of 2-D models. The crack can be

located inside the finite element model (Fig. 10.8-1(a)) or on the

boundary of the model (Fig. 10.8-1(b)). When the crack is located
on the boundary of the model, it is recommended that the crack line
be parallel to the Y axis, since some of the solution procedures
described below assume that the crack line is parallel to the Y axis.
Fig. 10.8-1 also shows the line of self-similar crack advance.
As seen below, it can be advantageous to use a mesh in which
nodes are placed along this line. Of course, when the crack is
located on the boundary of the model, then this line also is on the
boundary of the model.
Line of self-similar crack advance

Crack front
a) Both sides of crack modeled
Crack front
Line of self-similar
crack advance
= Crack propagation line
b) One side of crack modeled
Figure 10.8-1: 2-D fracture geometries
The terminology line of self-similar crack advance is used

instead of crack propagation line to emphasize that the crack
might not actually propagate along this line under mixed mode
conditions. However, for symmetric geometries and loadings, the

line of self-similar crack advance is the same as the crack

propagation line.
Illustrative 2-D meshes are shown in Fig 10.8-2. Notice that,

when both sides of the crack are modeled, there might or might not
be nodes in the line of self-similar crack advance. The input
corresponding to these possibilities is described below.
z
y
Crack front
a) Both sides of crack modeled,

no nodes on line of self-similar
crack advance
Crack front
crack advance
b) Both sides of crack modeled,

nodes on line of self-similar
crack advance
Figure 10.8-2: Meshes for 2-D fracture geometries

Crack front
crack advance
= Crack propagation line
c) One side of crack modeled
Figure 10.8-2: Continued
10.8.2 Meshing recommendations
In linear elastic analysis, the variation of the stress and strain

1
fields in the proximity of the crack tip is characterized by a
r
singularity where r is the distance to the crack tip location (see
Section 10.3). Such a variation can be modeled using collapsed
quadrilateral elements with collapsed nodes at the crack tip
location, and midside nodes at the 1/4 points. True triangular
elements with midside nodes at the 1/4 points can also be used. Fig
10.8-3 shows some examples.

1 single node
1/4 L 3/4 L
a) 1/r singularity for linear elastic analysis
3 coincident
untied nodes
1/4 L 3/4 L
b) 1/r singularity for elastic plastic analysis
Figure 10.8-3: Element configurations for the modeling

of stress field singularities
The SINGULAR command can be used to shift the element nodes

near the crack tip to the quarter-points. This command has the
following forms:
SINGULAR POINT Q-POINT

pointi
The nodes at points pointi are considered to be at the crack tip.

The adjacent mid-side nodes are shifted to the quarter-points.
Note that the SINGULAR commands shift the nodes either to or

from the quarter-points, depending upon parameter Q-POINT. The

default is to shift the nodes to the quarter-points.
For the analysis of elastic-plastic structures, collapsed

quadrilateral elements with 3 coincident but untied nodes at the
crack tip location are recommended, in order to reproduce a 1/r
variation of the stress/strain field.
Typical meshes for the evaluation of crack tip blunting, or for

the calculation of local crack opening parameters such as the
CTOD, are shown in Fig. 10.8-4. These meshes can be generated
using the specific meshing features of the AUI. In particular, the
GSURFACE command has several parameters used to generate this
type of mesh.
Several coincident untied Crack tip

nodes at crack tip
Crack tip
Figure 10.8-4: Meshes for the analysis of blunting and local

crack opening criteria
A typical mesh for 2-D crack propagation analysis is shown in

Fig. 10.8-5. Quadrilateral elements with any number of nodes can
be used. To ensure good results, the mesh in the area where the
crack propagates should be made of a regular density of elements
with regular shapes.

Initial crack tip location
Figure 10.8-5: Mesh used for crack propagation
10.8.3 Definition of the crack propagation line, no nodes on the line of

self-similar crack advance
When there are no nodes on the line of self-similar crack

advance, then only the crack front node needs to be specified. This
specification is done with the AUI commands
CRACK-PROPAGATION NODES NCRACK=1

node-1i
or
CRACK-PROPAGATION POINT
pointi
In the above cases, only one data input line of the CRACK-
PROPAGATION command is used.
10.8.4 Definition of the crack propagation line, nodes on the line of self-
similar crack advance
When there are nodes on the line of self-similar crack advance,

then the crack front node and also the line of self-similar crack
advance can be specified. This specification is done with the AUI
commands
CRACK-PROPAGATION NODE NCRACK>1

node-1i node-2i node-NCRACKi
where node-1i is the crack front node, node-2i is the next

corner node on the line of self-similar crack advance, .

or
CRACK-PROPAGATION LINE
linei front-pointi
where linei is the geometry line of self-similar crack advance

and front-pointi is the geometry point at the crack front.
In the above cases, only one data input line of the CRACK-
PROPAGATION command is used.
When only one side of the crack is modeled, ADINA

automatically generates a zero prescribed displacement in the z
direction for the crack front node and for each node on the line of
self-similar crack advance. (If skew systems are used, the zero
prescribed displacement is generated in the skew c direction.)
It will be seen that there are more fracture options available

when there are nodes on the line of self-similar crack advance.
Thus the model should contain nodes on the line of self-similar
crack advance when possible.
10.9 Definition and use of line contours for 2-D cracks

10.9.1 Definition of line contours
In ADINA, the line contour of Eq. (10.4-1) is defined by a

series of adjoining segments passing through the elements located
on the contour (see Fig. 10.9-1). The integration of the J-integral
along each segment is performed numerically using the value of the
variables at the integration points which define the segment.

z
y
a) Physical problem
Line contour '
Integration
point
b) Finite element mesh with line contours
Figure 10.9-1 Definition of line contours in a two-dimensional

finite element model.
As many contours as desired can be requested in ADINA.

Usually, several contours are used simultaneously in order to assess
the J-integral path-independence.
2-D elements with 22 or 33 Gauss integration can be used.

True triangular elements (with triangular integration) cannot be
used. All the elements on one contour must have the same
integration order. If 22 Gauss integration is used, the J-integral is
evaluated simultaneously along two line contours (see Fig. 10.9-2).
If 33 Gauss integration is used, only one J-integral is evaluated,
using a middle path through each element (see Fig. 10.9-2).

4 4 3 4 4
3 3 3
1 2 1 2 1 1
2 2
Path number 1 Path number 2 Path number 1 Path number 2
4 3 4 3 4 3
4 3
1 1
1 2 1 2 2 2
4 4 3 4 3
3 4 3
1 2 1 2 1 2 1 2
4
3 4 3 4 3 4 3
1 2 1 1 2 1 2
2
2x2 Gauss integration 3x3 Gauss integration
Integration point location 1,2,3,4 local node numbers
Figure 10.9-2: Convention for the description of element paths
In order to describe a contour directly in terms of the element

numbers, the element numbers (and the group numbers if
necessary) must be input in a sequence following the line contour
from one end to the other. For the first and last elements on the
contour, at least one face must coincide with the physical boundary
of the model. The AUI command is
J-LINE ELEMENT
elementi groupi
The AUI provides commands for the automatic generation of

line contours.

By geometry point and radius:
J-LINE POINT POINT RADIUS
The line contour is defined by a series of elements intersected by

the circle with origin at geometry point POINT and radius
RADIUS.
By geometry point and ring:
J-LINE RING NRING POINT
The line contour is defined by a series of elements which have ring

number NRING, starting from geometry point POINT. Elements
with ring number 1 are those connected to the crack tip, elements
with ring number 2 are those connected to ring number 1, etc.
10.9.2 Output
Line contour variables
Each line contour outputs the following information to the porthole

file:
LINE_J-INTEGRAL_AVERAGE,
LINE_J-INTEGRAL_DIFFERENCE.
If 22 Gauss integration is used, LINE_J-
INTEGRAL_AVERAGE is the average of the two J-integrals
evaluated from the equations in Section 10.4 and
LINE_J-INTEGRAL_DIFFERENCE is the difference
between these values. If 33 Gauss integration is used,
LINE_J-INTEGRAL_AVERAGE is the value of the J-
integral and LINE_J-INTEGRAL_DIFFERENCE is zero.
10.10 Definition and use of virtual shifts for 2-D cracks

The NVS method of virtual crack extension is always used for
2-D cracks.

A virtual shift requires two items to be specified:
The domain of the virtual shift

The virtual shift vector
Each of these items is described in more detail below.
The virtual shift vector field described in Section 10.5 is

specified by virtual shift vectors at the nodal points. Thus the
virtual shift domain is specified in terms of the node numbers.
Figure 10.10-1 shows a virtual shift domain. Within zone I, all

of the corner nodes are shifted by the same virtual shift vector.
Since zone I contains the crack front, the crack grows slightly, and
the crack area increases by amount Ac . Within zone III, none of
the corner nodes are shifted, thus the zone is unchanged by the
virtual shift.
I: zone rigidly shifted by virtual shift
II: zone distorted by virtual shift x2
III: zone unchanged by virtual shift x1
III III
II II
I I
x1 x1 x1
a) Before virtual shift b) After virtual shift

in two-dimensional analysis
Corner nodes in the virtual shift are shifted by the full value of
the virtual shift vector. Mid-side nodes are shifted by vectors
interpolated from the corresponding corner nodes, so that a mid-

side node between two corner nodes in the virtual shift is shifted by
the full value of the virtual shift vector, and a mid-side node
halfway between a corner node in the virtual shift and an unshifted
corner node is shifted by half the value of the virtual shift vector.
As many virtual shifts as desired can be specified.
It is recommended that several virtual shifts of increasing

domain size be used. Each virtual shift should enclose the next
smaller virtual shift. The virtual shift domain size should increase
equally in all directions from the plane perpendicular to the crack
front.
10.10.1 Domain of the virtual shift
By node numbers: The most basic way to specify a virtual shift is

to specify its corner (vertex) node numbers. In Fig. 10.10-2, corner
nodes in the virtual shift are shown as solid circles and other nodes
are unmarked. All elements with all corner nodes in the virtual
shift are in zone I, and are rigidly translated by the virtual shift. All
elements with only some corner nodes in the virtual shift are in
zone II, and are distorted by the virtual shift. All other elements
are in zone III.
The AUI command is
J-VIRTUAL-SHIFT NODE
nodei
Zone III
Zone II
Zone I
Figure 10.10-2: Description of a virtual shift by nodes
By element numbers: A virtual shift domain can be defined in

terms of element numbers. In this case, each element specified is

in zone I and other elements are in zones II and III. The basic idea
is shown in Fig. 10.10-3.
The AUI command is
J-VIRTUAL-SHIFT ELEMENT
elementi groupi
Figure 10.10-3: Description of virtual shifts by element numbers
By line or surface numbers: The nodes in a virtual shift domain

can be selected using geometry lines or surfaces.
The AUI commands are
J-VIRTUAL-SHIFT LINE
linei
and
J-VIRTUAL-SHIFT SURFACE
surfacei
All nodes lying on the given lines or surfaces are in zone I.
By point and radius: A virtual shift domain can be defined using a

sphere (point at center and radius).
The AUI command is
J-VIRTUAL-SHIFT POINT POINT RADIUS
For 2-D analysis, since the intersection of a sphere with origin in

the y-z plane and the y-z plane is a circle, this command can be

thought of as defining a circle with given center and radius. All

nodes within the circle are in zone I. This is shown in Fig. 10.10-4.
Note that usually the center is at the crack tip, but this is not a
requirement; the center can be at any geometry point, as long as the
circle contains the crack tip.
Point
Radius
Figure 10.10-4: Description of a virtual shift by point and radius
By rings: A virtual shift domain can be defined in terms of rings.

Here the origin of the ring and the number of rings are specified, as
shown in Fig. 10.10-5 for 2-D analysis.
The AUI command is
J-VIRTUAL-SHIFT RING RING-TYPE RING-NUMBER

namei
Elements in rings 1 to 3
Ring origin
Figure 10.10-5: Description of a virtual shift by ring number
The origin of the ring can either be a single node or a set of nodes.
These origin nodes are selected by parameters RING-TYPE and the

data input lines:
RING-TYPE=POINT: The origin nodes are selected by geometry

points, so that each namei is a geometry point.
RING-TYPE=LINE: The origin nodes are selected by geometry

lines, so that each namei is a geometry line.
RING-TYPE=SURFACE: The origin nodes are selected by

geometry surfaces, so that each namei is a geometry surface.
RING-TYPE=NODE: The origin nodes are selected by nodes, so

that each namei is a node.
The number of rings is selected by RING-NUMBER. RING-

NUMBER=0 just selects the origin nodes, RING-NUMBER=1
selects the origin nodes plus nodes in elements connected to the
origin nodes, RING-NUMBER=2 selects nodes in ring 1 plus
nodes in elements connected to nodes in ring 1, etc.
10.10.2 Virtual shift vector
Parameters VECTOR, VX, VY, VZ, N3DSH of the above J-

VIRTUAL-SHIFT commands are used to define the virtual shift
vector, as follows:
VECTOR=INPUT: The shift vector is defined directly using

parameters VX, VY, VZ. Note that both the magnitude and
direction of this vector are used. This vector should lie in the
line or plane of self-similar crack advance.
VECTOR=AUTOMATIC: This option can be used only when

there are nodes on the line of self-similar crack advance and
when these nodes are specified in the CRACK-
PROPAGATION command. The virtual shift vector lies on the
line of self-similar crack advance and has magnitude of 1/100
the distance between the crack tip and the next corner node
along this line.

10.10.3 Output
Virtual shift variables
Each virtual shift prints the following information, and outputs the
accessed in the AUI using the given variable names.
J-PARAMETER_1:
(obsolete)
J-PARAMETER_2:
J-integral without thermal, pressure or dynamic corrections,
equal to G1 + Ghoop .
J-PARAMETER_3:
J-integral with temperature, pressure and dynamic
corrections, equal to G1 + Ghoop + Gth + G pr + Gdyn .
TEMPERATURE_CORRECTION, PRESSURE_CORRECTION,
DYNAMIC_CORRECTION:
The corrections Gth , G pr , Gdyn
10.11 Crack propagation for 2-D cracks

The analysis of crack propagation is available in ADINA for
2-D models. The crack must be planar, on the symmetry line of a
symmetric structure submitted to a symmetric loading.
The "node shift/release" technique used in ADINA combines

the shifting and the releasing of the successive crack tip nodes in
order to model the propagation of the crack tip through the finite
element mesh (see Fig. 10.11-1).
The "node release-only" technique can also be used, whereby

the crack tip node is not shifted but only released. This option
provides better convergence when large deformations occur and
convergence difficulties arise with the node shift/release technique.

Step 1: initial configuration Step 2: step including node shifting
Crack Crack propagation

surface
Step 3: step including node shifting and releasing

Z
Y
Position of crack front

Initial node location
Node shifted
Node released
Nodal conditions along crack propagation surface:

y-direction = free; z-direction = zero prescribed displacement
Figure 10.11-1: The node shift / release procedure

for crack propagation
Definition of crack advance: The crack advance is the distance

that the crack has advanced since the time of solution start (current
crack length minus initial crack length).
Definition of the crack propagation surface: The crack

propagation surface, which typically corresponds to the ligament in
a fracture mechanics test specimen, is defined in ADINA as the set
of nodes which may possibly be released when the crack opens.
The crack propagation surface must be planar. In two-dimensional
models, the crack propagation surface reduces to a line, though still
called a surface, and is the same as the line of self-similar crack
advance.
The commands in Section 10.10.1 should be used to define the
crack propagation surface.

Crack growth control parameter: Different parameters can be

used in ADINA for the control of the crack advance.
Displacement at a node. This displacement is typically

chosen to be at the crack tip (to monitor the crack tip opening
displacement (CTOD)), at the crack mouth (to monitor the crack
mouth opening displacement (CMOD)), or at the point of load
application (see Fig. 10.11-2).
J-integrals with or without thermal, pressure, and dynamic

corrections.
New crack tip F
Initial crack tip

Abscissae of the
initial crack tip V L+
New crack tip
d
F
Initial crack tip
b) V = CMOD.
a) Some definitions of the CTOD. = load point displacement
L = initial distance between
load points
Figure 10.11-2: Definitions of the crack tip opening displacement,

crack mouth opening displacement,
load point displacement
If the J-integral is chosen as the crack growth control parameter,

the associated virtual shift can be either a spatially fixed shift or a
spatially moving shift (see Fig 10.11-3).
A fixed virtual shift can be defined using any of the virtual shift
definition commands given in Section 10.10. The virtual shift
domain of a fixed virtual shift must be chosen large enough to
include the successive crack tip locations at any time/load step of

the analysis.
When a moving virtual shift is requested, ADINA automatically
creates a moving virtual shift as the next higher virtual shift
number (e.g, if 7 virtual shifts are defined using the commands
given above, then the moving virtual shift is shift number 8). The
size of the moving virtual shift is defined by the number of rings of
elements connected to the crack tip node. ADINA automatically
updates the moving virtual shift definition as the crack tip node
changes.
Crack resistance curve: In order to evaluate the amount of crack

advance from the value of the J-integral or the displacement, a
resistance curve must be input. This curve is a material property.
When the crack growth control parameter is a displacement
parameter, then this curve relates the value of the displacement
parameter to the crack advance. When the crack growth control
parameter is the J-integral, then this curve relates the value of the J-
integral to the crack advance.
Fig. 10.11-4(a) shows a resistance curve used when the
materials are temperature-independent. For each value of the crack
growth control parameter, the resistance curve gives the threshold
value for crack advance.
For materials with temperature dependent material properties, a
set of temperature dependent resistance curves can be input. The
temperature at the crack tip node is used with the current value of
the crack growth control parameter to evaluate the threshold value
for crack advance (see Fig. 10.11-4(b)). Notice that the
interpolation is in the horizontal direction.

Crack front
a) Moving virtual shift

(2 rings)
Crack front
b) Fixed virtual shift
Elements included in the virtual shift
Figure 10-11.3: Moving and fixed virtual shifts in crack

propagation analysis

Crack growth
control parameter
Current value
of crack growth
control parameter
Threshold value for crack advance

User-input point
Crack advance
a) Temperature independent
Crack growth User-input resistance

control parameter curve for temperature 1
Interpolated resistance
curve for temperature
Current value
of crack growth
control parameter

User-input resistance
curve for temperature 2
Temperature interpolation is performed
horizontally between the two
Crack advance
b) Temperature dependent
Figure 10.11-4: Resistance curves
Modeling of the crack advance : At the end of each time step,

ADINA computes the crack growth control parameter and
computes the threshold crack advance from the resistance curve. If
the threshold crack advance is less than the current crack advance,

the crack advance remains unchanged. If the threshold crack

advance is greater than the current crack advance, the crack
advance is updated using either the node shift/release technique or
the node release technique. The idea is shown in Fig. 10.11-5.
The crack advance is considered to take place at the beginning
of the next time step. Thus a plot of the crack growth control
parameter vs the crack advance will most likely show the crack
growth control parameter higher than the critical value for at least a
few steps. This mismatch can be reduced by taking smaller time
steps.
When the crack growth control parameter is the J-integral,
ADINA only computes the crack advance for those time/load steps
in which the stresses are printed or saved. Therefore it is
recommended that the stresses be printed or saved for every step.
Crack growth
control parameter
State at end of time step

Crack advance
a) No change of crack advance
Figure 10.11-5: Calculation of crack advance

Crack growth
control parameter

= State at beginning of next time step
Crack advance
b) Change of crack advance -- node shift/release
Crack growth
control parameter
State at beginning of next time step

Crack advance
c) Change of crack advance -- node release
Modeling of the crack advance using the node shift/release

technique: When the crack growth control parameter indicates a
crack increase, the crack tip node is automatically shifted to a new
position (see Fig. 10.11-1 - step 2). When the advance of the crack
tip node becomes larger than the width of the element ahead of the
crack tip node in the propagation direction (and connected to the
crack propagation surface), the crack tip node is released and the
next node on the crack propagation surface starts being shifted
(Fig. 10.11-1 - step 3). Midside nodes are shifted to remain
halfway between the corresponding corner nodes, and are released
when the preceding corner node on the crack propagation surface is
released.
This procedure does not require any specific element size or

mesh refinement, apart from general considerations on the accuracy

of the analysis results. However, when an elastic-plastic material
model is used in the shifted elements, the material state is not
remapped during shifting. This can cause the solution to be
inaccurate.
Quadrilateral elements should be used in the portion of the
model including the crack tip and the crack propagation surface, in
order to allow the nodes on the crack propagation surface to shift as
the crack grows.
Modeling of the crack advance using the node release

technique: When the crack growth control parameter indicates a
crack increase, if the crack increase is less than one-half the
element length, the crack tip node is not released. If the crack
increase is greater than one-half the element length, the crack tip
node (and mid-side node, if any) is released. In general, nodes are
released so that the difference between the desired crack advance
and the actual crack advance (modeled by releasing nodes) is
minimized.
Since the element coordinates are never shifted during this
procedure, the issue of remapping the material state does not arise.
Pressure loading: In case the faces of the crack are subjected to

pressure loads, these loads should be applied to both the initial
crack faces and the crack propagation surface. The pressure loads
are updated as the nodes are shifted and released.
Input specification:
FRACTURE: The following command specifies a crack

propagation analysis:
FRACTURE TYPE=PROPAGATION
CRACK-GROWTH: The following command specifies parameters

associated with crack propagation:
CRACK-GROWTH CONTROL-TYPE J-VERSION,

R-CURVE LOCTYPE DOF SHIFT-RELEASE,
POINT NODE

If CONTROL-TYPE=FIXED or MOVING, the crack growth

control parameter is the J-integral associated with either a fixed or
moving virtual shift; J-VERSION specifies whether the J-integral
includes corrections. If CONTROL-TYPE=NODAL, the crack
growth control parameter is a nodal displacement; parameters
LOCTYPE, DOF, POINT, NODE specify the location of the
displacement and the displacement direction.
R-CURVE specifies the number of the resistance curve to use.

This resistance curve is defined using the R-CURVE command.
SHIFT-RELEASE specifies whether to use the node shift/release

method or the node release method.
CRACK-PROPAGATION: The CRACK-PROPAGATION

command includes a data input line parameter nvshfti, used as
follows:
CONTROL-TYPE=FIXED, nvshfti is the number of the fixed

virtual shift used to control crack growth.
CONTROL-TYPE=MOVING, nvshfti is the number of rings in the

moving virtual shift.
Parameter nvshfti is specified in the data input lines, as shown:
CRACK-PROPAGATION NODES NCRACK>1

node-1i node-2i node-NCRACKi nvshfti
CRACK-PROPAGATION LINE
linei front-pointi nvshfti
R-CURVE: This command defines a set of resistance curves.
R-CURVE NAME MPOINT

thetai x1i y1i x2i y2i xm i ym i
Each data input line specifies the (x,y) points for the resistance
curve for a given temperature. MPOINT is the number of points in
each resistance curve. In this command, x is the crack advance
and y is the crack growth control parameter. If there are no
temperature-dependent materials, then only one curve need be

entered and thetai is not used. When there are temperature-

dependent materials, then at least two curves need to be entered.
The values of x must be the same for each resistance curve.

10.12.1 Overview
Fig. 10.12-1 shows geometries of 3-D models. The crack can

be located inside the finite element model (enclosed crack) or on
the boundary of the model (boundary crack).
Fig. 10.12-1 also shows the plane of self-similar crack
advance. Of course, when the crack is located on the boundary of
the model, then this plane also is on the boundary of the model.
The mesh layout near the crack front strongly influences the
methods that can be used for virtual crack extension. The NVS
method for virtual crack extension can only be used provided that
the mesh has a specific layout near the crack front. The mesh
layout required for the NVS method is discussed in detail in
Section 10.14. The SVS method for virtual crack extension can be
used for any mesh layout, including free-form mesh layouts, as
discussed in detail in Section 10.15.
3-D fracture models can be constructed either with the use of

the CRACK-M meshing features or without the use of the
CRACK-M meshing features. The CRACK-M meshing features
are designed to create mapped meshes of prismatic elements near
the crack front and free-form meshes away from the crack front.
These meshes have the mesh layout required by the NVS method.
However, the CRACK-M features must be used in conjunction
with the ADINA-M solid modeler. The CRACK-M features are
discussed in detail in Section 10.13.

z
y
x
Plane of self-similar
Crack front crack advance
a) Both sides of crack modeled (enclosed crack)
Plane of self-similar crack advance

Crack front = Crack propagation plane
b) One side of crack modeled (boundary crack)
Figure 10.12-1: 3-D fracture geometries
10.12.2 Meshing recommendations
In linear elastic analysis, the variation of the stress and strain

1
fields in the proximity of the crack tip is characterized by a
r
singularity where r is the distance to the crack tip location, see
Section 10.3. Such a variation can be modeled using collapsed

wedge brick and tetrahedral elements with collapsed nodes at the

crack tip location, and midside nodes at the 1/4 points, as shown in
Fig. 10.12-2.
The SINGULAR commands can be used to shift the element nodes

near the crack front to the quarter-points.
SINGULAR LINE Q-POINT

linei
The nodes on lines linei are considered to be at the crack front.

The adjacent mid-side nodes are shifted to the quarter-points.
Note that the SINGULAR commands shift the nodes either to or

from the quarter-points, depending upon parameter Q-POINT. The
default is to shift the nodes to the quarter-points.
The SINGULAR commands should not be used when using the

CRACK-M modeling features. Use the CRACK-M Q-POINT
command instead. The SINGULAR commands also should not be
used when using the SVS method for virtual crack extension, since
the CRACK-SVS command includes the option of shifting nodes to
the quarter-points.
For the analysis of elastic-plastic structures, collapsed

quadrilateral elements with 3 coincident but untied nodes at the
crack tip location are recommended, in order to reproduce a 1/r
variation of the stress/strain field. In 3-D analysis, wedge elements
with coincident but untied nodes on the crack front will reproduce
the 1/r stress/strain singularity (see Fig. 10.12-2).

1 single node
1/4 L 3/4 L
a) 1/r singularity for linear elastic analysis
3 coincident
untied nodes
1/4 L 3/4 L
b) 1/r singularity for elastic plastic analysis
Figure 10.12-2: Element configurations for the modeling

of stress field singularities

10.13 3-D crack geometry and meshing using CRACK-M

Using the CRACK-M features, complex 3-D models with
embedded cracks can be easily constructed. The ADINA-M
geometric modeler must be used.
The CRACK-M meshing features are designed to create mapped

meshes of prismatic elements near the crack front and free-form
meshes away from the crack front.
Meshes created using the CRACK-M features can be used in

conjunction with either the NVS or SVS methods for virtual crack
extension.
The basic ideas are shown in Figs 10.13-1 and 10.13-2. Fig
10.13-1 shows models in which the crack is on the boundary and
Fig 10.13-2 shows models in which the crack is in the interior
(enclosed crack).

Crack front edge

Cracked surface
B3
Crack front
Crack sleeve
B2 Remainder
B1
a) Crack front edge is an open loop
B5
B6
B3
B1
B4
B2
b) Crack front edge is a closed loop
Figure 10.13-1: Schematic division of domain for use of CRACK-M features,

crack on boundary of model

B5
B3
B1
B2
B4
B6
a) Crack front edge is an open loop
Figure 10.13-2: Schematic division of domain for use of

CRACK-M features, crack inside model
(enclosed crack)

B9
B11
B5
B1
B7 B2
B3 B6
B4
B8
B10
B12
b) Crack front edge is a closed loop
In all cases, the model is decomposed into three parts, the crack
front, the crack sleeve and the remainder.
The crack front is that region of the model immediately adjacent
to the crack front edge. This part of the model will be meshed using
mapped meshing, as discussed below.
The crack sleeve is a transition region between the crack front
and the remainder of the model. The crack sleeve encloses the
crack front, and the crack sleeve should include the cracked surface
not represented in the crack front itself. This part of the model will
be meshed using free-form meshing.

The remainder of the model is constructed as usual, and is

meshed using free-form meshing.
Each of these parts is represented with one or more geometry
bodies, as shown. Due to the geometric modeler requirements, two
bodies must be used to represent the closed loop crack front in Fig
10.13-1(b). Also, two bodies must be used to represent the enclosed
crack front in Fig 10.13-2(a), and four bodies must be used to
represent the enclosed closed loop crack front in Fig 10.13-2(b).
For the remainder of the model, up to four bodies are used also, but
these bodies could be merged into a single body, provided that the
faces adjacent to the crack sleeve body remain compatible with the
crack sleeve body faces.
Although a circular crack front is shown in the Figures, the
CRACK-M features can be used whenever the geometry is
topologically equivalent to the Figures.
Crack front body: A typical body used to represent the crack front
is shown in Fig. 10.13-3. The body is prismatic, with one edge
corresponding to the crack front edge, and one face corresponding
to the cracked surface. Any geometric modeler tools can be used to
create the crack front body, provided that the body is topologically
equivalent to the body shown in Fig 10.13-3.
Figure 10.13-3: Body used to represent crack front
Crack sleeve body: Requirements on the crack sleeve body are: the
crack sleeve body should completely surround the crack front
body, the crack sleeve body should have one face on the cracked
surface.
Again, any geometric modeler tools can be used to create the crack

sleeve body.
CRACK-M definition: Once the bodies have been defined, the

CRACK-M DEFINE command is used to define the crack:
CRACK-M DEFINE NAME CFBODY1 CFFACE1 CFEDGE1,

CFBODY2 CFFACE2 CFEDGE2,
CSBODY1 CSFACE1,
CSBODY2 CSFACE2
Fig 10-13.4 shows the parameters of the CRACK-M DEFINE

command superposed on the geometry. Notice that when the crack
is in the interior, the crack front edge is specified twice, once for
each body used to represent the crack front.

CSBODY1
CFBODY1
CSFACE1
CFFACE1 CFEDGE1
a) Crack on boundary (see Fig 10.13-1)
CSBODY1
CFBODY1
CSFACE1
CFFACE1 CFEDGE1
CFFACE2 CFEDGE2
CSFACE2
CFBODY2
CSBODY2
b) Crack in interior (see Fig 10.13-2)
Figure 10.13-4: CRACK-M DEFINE examples
Although the cracked surface is shown at an angle in the figure,

the cracked surface can be in the same plane as the uncracked
ligament, and the two cracked surfaces for the crack in the interior

can be coincident.
When the crack front is a closed loop, two CRACK-M
definitions must be made. Then the CRACK-M COMBINE
command is used to combine the CRACK-Ms:
CRACK-M COMBINE NAME CRACKA CRACKB
where CRACKA, CRACKB are the two CRACK-Ms to be

combined.
CRACK-M subdivision: The CRACK-M SUBDIVIDE command is

used to subdivide the crack front bodies of the CRACK-M.
CRACK-M SUBDIVIDE NAME,

NDIVCF NDIVTAN1 NDIVTAN2 NDIVRAD
NDIVCF is the number of subdivisions along the crack front edge.

NDIVTAN1, NDIVTAN2 and NDIVRAD are the number of
subdivisions in the other directions, see Fig 10.13-5. Notice that
although the CRACK-M contains crack sleeve bodies, the crack
sleeve bodies are not subdivided by CRACK-M SUBDIVIDE. The
crack sleeve bodies are subdivided using the usual subdivision
commands of the geometry modeler.

NDIVTAN1
divisions
NDIVRAD
divisions
a) Crack on boundary (see Fig 10.13-1)
NDIVTAN1
divisions
NDIVTAN2
divisions
a) Crack in interior (see Fig 10.13-2)
Figure 10.13-5: CRACK-M SUBDIVIDE examples

Meshing: The CRACK-M MESH command is used to mesh the

crack front bodies:
CRACK-M MESH NAME,

NODES NCTOLERANCE SUBSTRUCTURE GROUP
MIDNODES
where NAME is the name of a CRACK-M. Mapped meshing is

used to generate a mesh with hexahedra and prisms adjacent to the
crack front edge. The hexahedra can have either 8, 20 or 27-nodes.
Fig. 10.13-6 shows typical meshes. At each station on the crack
front edge, there is one node, and there is no option for generating
multiple nodes at each station on the crack front edge. When the
crack is in the interior of the model (so that the CRACK-M
contains two crack surfaces), there is no coincidence checking on
the crack surfaces, so separate nodes are generated at the matching
positions of the crack surfaces.
One node at each station

on the crack front
CRACK-M Q-POINT used to move nodes to the quarter-points
a) Crack on boundary
Figure 10.13-6: CRACK-M MESH examples

If cracked surfaces are coincident, separate coincident nodes are generated,

for example, here:
b) Crack in interior of model
The CRACK-M Q-POINT command is used to move the mid-

side and mid-face nodes of the (higher-order) elements adjacent to
the crack front edges to the quarter-points.
CRACK-M Q-POINT POSITION
This command must be issued after the crack is meshed. When

using the SVS method of virtual crack extension, it is not necessary
to use the CRACK-M Q-POINT command, because the CRACK-
SVS command can also shift the nodes to the quarter-points.
The CRACK-M GBODY command is used to mesh the crack
sleeve bodies:
CRACK-M GBODY NAME,

NODES NCTOLERANCE SUBSTRUCTURE GROUP
MIDNODES
where NAME is the name of a crack sleeve body. Free-form

meshing is used. When the CRACK-M contains two crack
surfaces, there is no coincidence checking on the crack surfaces, so
separate nodes are generated at the matching positions of the crack
surfaces.
The remaining bodies are meshed using the usual geometry

modeler commands.
It is strongly recommended that the FACELINK 1

OPTION=ALL command be used so that the meshing is
compatible between adjacent bodies.
When the crack is on the boundary of the model, it is necessary

to assign proper boundary conditions to the uncracked ligament.
10.14 NVS method for virtual crack extension for 3-D

cracks
10.14.1 Overview
In the NVS method for virtual crack extension, the virtual shift
vector field is specified by virtual shift vectors at the nodal points.
The fracture mechanics output consists of the values of the J-

integral, calculated using the volume integral G (including
corrections when applicable) for each of the virtual shifts. See
Section 10.5 for a description of G .
The NVS method makes use of a specific mesh layout near the
crack front, as discussed below. If the model does not have this
specific mesh layout near the crack front, the NVS method can still
be used, but with limitations. Therefore, if the model does not have
this specific mesh layout, the SVS method should be used instead.
The following example shows nodal point virtual shift vectors

for a 3-D mesh. Fig. 10.14-1 shows a 3-D mesh near the crack
front. The mesh cross-section is the same for all generator lines.
Element edges lie on the plane of self-similar crack advance. The
plane of self-similar crack advance is also called the crack
propagation surface.
Although the crack front line is straight in the figure, the crack
front line can also be curved.

10.14 NVS method for virtual crack extension for 3-D cracks
Generator line 3 Plane of self-similar

crack advance
Generator line 2
Generator line 1
Crack front line
a) Mesh layout near crack front, showing plane of self-similar crack advance,
crack front line and three generator lines
Generator line 3 Generator line 2 Generator line 1
b) Top view of crack propagation surface and open crack face
c) Cross-section at each generator line
Figure 10.14-1: Example mesh for description of NVS virtual shifts

Fig. 10.14-2 shows a virtual shift for a crack advance at

generator line 1. This virtual shift models a crack advance at the
crack front node at generator line 1, with no crack advance at the
other nodes on the crack front.
Fig. 10.14-3 shows a virtual shift applied at the crack front node
and in one ring of elements at generator line 1. The same crack
advance as in Fig 10.14-2 is modeled, however the region in which
the virtual shift field is non-zero is larger than in Fig 10.14-2.
Figs. 10.14-4 and 10.14-5 show a virtual shift for a crack

advance at generator line 2, both with the virtual shift applied at
only one node, and also on one ring of elements.

a) Isometric view
c) Cross-section at generator line 1
Figure 10.14-2: Virtual shift for crack front node at generator line 1,
virtual shift applied at crack front only

a) Isometric view
one ring of elements

a) Isometric view
virtual shift applied at crack front only

a) Isometric view
one ring of elements

This example shows the general rule for specifying the virtual
shift vectors. The virtual shift vector at the node of the crack front
is specified. This virtual shift vector is applied to the node on the
crack front, and also to a list of neighboring corner nodes on the
cross-section at the given node. As will be seen later on, there are
several ways in which this information is specified.
This example also shows that the ideal mesh layout for use with
the NVS method is as shown in Fig 10.14-6. he mesh
perpendicular to the crack front is topologically swept along the
crack front. This kind of meshing automatically insures that there
is a mapped mesh of nodes both on the plane of self-similar crack
advance and also along the open faces of the crack. Notice that
there is a mesh cross-section corresponding to each generator line.

Mesh section corresponding

to generator line 1
a) 3D mesh b) Sections of 3D mesh
Generator line 1
Open crack face Crack front Plane of self-similar

crack advance
c) Mesh layout on crack plane
Figure 10.14-6: Ideal mesh layout for 3-D fracture model
Corner nodes in the virtual shift are shifted by the full value of
the virtual shift vector. Mid-side nodes are shifted by vectors
interpolated from the corresponding corner nodes, so that a mid-
side node between two corner nodes in the virtual shift is shifted by
the full value of the virtual shift vector, and a mid-side node
halfway between a corner node in the virtual shift and an unshifted
corner node is shifted by half the value of the virtual shift vector.

Fig 10.14-7 gives an illustration.
Crack front
Nodes subjected to a virtual shift
Figure 10.14-7: Virtual shifts at mid-side nodes in 3-D model
As many virtual shifts as desired can be specified. There should

be at least one virtual shift for each node on the crack front. Ideally
there are several virtual shifts for each node on the crack front,
each virtual shift being applied to more and more neighboring
nodes in the element plane perpendicular to the crack front.
If the mesh was created using the CRACK-M features, then the
mesh already contains a mapped mesh of nodes on the plane of
self-similar crack advance.
10.14.2 Definition of the crack propagation surface, CRACK-M

features are not used
The basic procedure for specifying the virtual shifts is

Define a crack propagation surface using the CRACK-

PROPAGATION commands
Define the virtual shift vector and domains using the J-

VIRTUAL-SHIFT commands.
When there is no mapped mesh of nodes on the plane of self-

similar crack advance, then only the crack front line can be
specified. This specification is made using the AUI command
CRACK-PROPAGATION NODES NCRACK=1

node-1i
Here there is one data input line per node on the crack front.
When there is a mapped mesh of nodes on the plane of self-

similar crack advance, in which the mesh is topologically
rectangular, then the crack front line and the plane of self-similar
crack advance can be specified. There are several ways to make
this specification:
By nodes: This specification is made using the AUI command
CRACK-PROPAGATION NODES NCRACK>1

node-1i node-2i node-NCRACKi
Here there is one data input line per generator line, with the first
data input line corresponding to the first generator line, the second
data input line corresponding to the second generator line, etc.
node-1i is the node on the crack front for generator line i,
node-2i is the next corner node along the generator line, etc.
Generator lines for mid-side and mid-face nodes should not be
specified, that is, only corner nodes define generator lines.
By surfaces: This specification is made using the AUI command
CRACK-PROPAGATION SURFACE
surfacei front-linei
Here there is one data input line per geometry line on the crack
front.

When only one side of the crack is modeled, ADINA

automatically generates a zero prescribed displacement in the z
direction for the crack front node and for each node on the crack
propagation surface. If skew systems are used for the nodes on the
crack propagation surface, the zero prescribed displacement is in
the skew c direction.
10.14.3 Definition of the virtual shifts, CRACK-M features are not used
By node numbers: The most basic way to specify a virtual shift is

to specify the virtual shift vector, and to specify the corner node
numbers for the nodes shifted by the virtual shift vector.
The AUI command is
J-VIRTUAL-SHIFT NODE,
VECTOR VX VY VZ N3DSH
nodei
Parameters VECTOR, VX, VY, VZ, N3DSH of the J-VIRTUAL-

SHIFT NODE command are used to define the virtual shift vector,
as follows:
VECTOR=INPUT: The shift vector is defined directly using

parameters VX, VY, VZ. Note that both the magnitude and
direction of this vector are used. This vector should lie in the plane
of self-similar crack advance.
VECTOR=AUTOMATIC: The shift vector is defined automatically,

as follows: The virtual shift vector has direction given by generator
line N3DSH, and has magnitude of 1/100 the distance between the
crack front node and the next corner node along this generator line.
The J-VIRTUAL-SHIFT NODE command defines only a single

virtual shift at a time.
By rings: The virtual shifts can be defined in terms of rings.
The AUI command is
J-VIRTUAL-SHIFT RING,

VECTOR VX VY VZ,
RING-TYPE=AUTOMATIC,
RING-NUMBER,
NORMAL={AUTOMATIC / INPUT} NX NY NZ,
THICKNESS
Although this command has a number of options, we will discuss

only the options RING-TYPE=AUTOMATIC and
NORMAL=AUTOMATIC or INPUT, since the other options are
impractical for 3-D cracks.
When RING-TYPE=AUTOMATIC, the AUI creates several

virtual shift per corner node on the crack front, as follows.
Suppose that there are NWCPS corner nodes on the crack front.
Then
Virtual shift 1: crack front node at generator line 1

Virtual shift 2: crack front node at generator line 2
...
Virtual shift NWCPS: crack front node at generator line
NWCPS
Virtual shift NWCPS+1: crack front node at generator line 1 +
nodes from first ring of elements surrounding the crack front
...
Virtual shift 2*NWCPS: crack front node at generator line
NWCPS + nodes from first ring of elements surrounding the
crack front
...
The total number of virtual shifts is NWCPS * (RING-NUMBER +

1). If RING-NUMBER = 0, then the virtual shifts are defined
using only the crack front nodes. If RING-NUMBER = 1, then the
virtual shifts are defined using the crack front nodes, and also the
first ring of elements surrounding the crack front nodes.
The options for defining the virtual shift vectors are the same as for
J-VIRTUAL-SHIFT NODE. When VECTOR=AUTOMATIC, the
shift vector is determined for each crack front node using the
generator line of that crack front node. When VECTOR=INPUT,
parameters VX, VY, VZ are used to specify the virtual shift vector.
Clearly VECTOR=INPUT is useful only for straight cracks.

Selection of nodes in the virtual shift: We now discuss how nodes

are selected from the rings of elements surrounding the crack front.
Consider RING-NUMBER=1. First, all of the nodes in the first
ring of elements surrounding the entire crack front are selected.
Then these nodes are tested against a disk, and only the nodes that
lie within the given disk are selected.
The disk is defined by a node on the central plane, the
orientation of the central plane, its radius and its thickness, as
shown in Fig. 10.14-8. This information is specified as follows:
Node on the central plane. This is the corner node on the crack
front, corresponding to the virtual shift to be defined. Thus
each corner node on the crack front has its own disk.
Orientation of the central plane. This is determined by a

normal vector as shown in the figure. This normal vector is
specified using one of two options, as follows:
NORMAL=INPUT, normal vector is specified using

parameters NX, NY, NZ.
NORMAL=AUTOMATIC, normal vector is perpendicular

to the generator line of the corner node on the crack front, as
shown in Fig 10.14-9.
Thickness of the disk: This is specified by parameter

THICKNESS.

THICKNESS
Generator line NWCPS
NORMAL
Central plane of disk
Generator line 1
RADIUS
Crack front node
at generator line
N3DSH
Figure 10.14-8: Definition of a disk used to select nodes in 3-D virtual shift
Disk
Generator line N3DSH
THICKNESS
RADIUS
Crack propagation surface is viewed from above.
Figure 10.14-9: Disk defined using NORMAL=AUTOMATIC

It is necessary to specify the crack propagation surface before

using the J-VIRTUAL-SHIFT RING command.
Although the J-VIRTUAL-SHIFT RING command accepts data

input lines, data input lines should not be specified when using
RING-TYPE=AUTOMATIC.
10.14.4 Definitions of the virtual shifts when CRACK-M features are used
When the CRACK-M features are used in the modeling of the

crack, all of the virtual shifts are generated using the command
J-VIRTUAL-SHIFT CRACK-M NAME VSSCALE
where NAME is the number of the CRACK-M and VSSCALE is

used to scale the length of the virtual shift vector.
The virtual shifts are generated in the following order:
Virtual shifts 1 to NDIVCF: Corner nodes at stations 1 to NDIVCF

are shifted. Other corner nodes are not shifted.
Virtual shifts NDIVCF+1 to 2NDIVCF: Corner nodes at stations 1

to NDIVCF are shifted. In addition, the first ring of nodes
surrounding the corner nodes are shifted.
Virtual shifts 2NDIVCF+1 to 3NDIVCF: Corner nodes at

stations 1 to NDIVCF are shifted. In addition, the first and second
rings of nodes surrounding the corner nodes are shifted.
etc., up to virtual shift NDIVRADNDIVCF, in which all rings of

nodes in the mapped mesh of the CRACK-M are shifted.
Fig. 10.14-10 shows typical virtual shifts for station IDIVCF. Fig
10.14-10(a) shows virtual shift IDIVCF and Fig 10.14-10(b) shows
virtual shift NDIVCF+IDIVCF.

Nodes at IDIVCF+1
Nodes at IDIVCF
Nodes at IDIVCF-1
Only corner nodes are shown.
a) Virtual shift IDIVCF: Only node in ring 0 is shifted
b) Virtual shift NDIVCF + IDIVCF: Nodes in rings 0 and 1 are shifted
Figure 10.14-10: Virtual shifts for station IDIVCF

Each virtual shift vector for station IDIVCF is constructed as

follows: The vector is normal to the geometric crack front edge at
station IDIVCF, lies in the plane of the cracked surface, and points
away from the cracked surface. The length of the vector is
VSSCALE(radius of nodal ring 1).
The command
LIST J-VIRTUAL-SHIFT CRACK-M
can be used to create a listing that gives, for each virtual shift, the
crack front node, ring number, number of nodes and virtual shift
vector. (The ring number is the number of nodal rings contained in
the virtual shift.)
10.14.5 Fracture mechanics output
J-PARAMETER_1:
(obsolete)
J-PARAMETER_2:
J-integral without thermal, pressure or dynamic corrections,
equal to G1 ( G1 is the value of G from equation (10.5-1).
J-PARAMETER_3:
J-integral with temperature, pressure and dynamic
corrections, equal to G1 + Gth + G pr + Gdyn .
TEMPERATURE_CORRECTION, PRESSURE_CORRECTION,
DYNAMIC_CORRECTION:
The corrections Gth , G pr , Gdyn

Other variables
VIRTUAL_SHIFT_*_PART_FACTOR
The virtual shift participation factors at the nodes (for
example, VIRTUAL_SHIFT_1_PART_FACTOR can be
used to access the participation factors for virtual shift 1 at
the nodes).
10.15 SVS method for virtual crack extension for 3-D

cracks
10.15.1 Overview
In the SVS method for virtual crack extension, the virtual shifts
are defined using crack advance stations, as described below.
The fracture mechanics output consists of the values of the J-

integral, calculated using the volume integral G (including
corrections when applicable) for each of the virtual shifts. See
Section 10.5 for a description of G . In addition, for a linear
isotropic elastic analysis without thermal or dynamic effects, the
fracture mechanics output also consists of the stress intensity
factors K I , K II , K III calculated for each of the virtual shifts. See
Section 10.3 for a description of the stress intensity factors and
Section 10.6 for a description of how the stress intensity factors are
calculated.
The SVS method does not rely on any specific mesh layout near
the crack front. The mesh layout can be mapped or free-form.
However, the crack front must be sharp, namely the crack front
must be formed from the intersection of the crack top faces and the
crack bottom faces.
It is recommended that the mid-side nodes adjacent to the crack

1
front be shifted to the quarter-points, in order to generate the
r
singularity near the crack front. This shifting is conveniently done
as part of the definition of the SVS crack, see below.

10.15 SVS method for virtual crack extension for 3-D cracks
10.15.2 Definitions of the virtual shifts
The following example shows the virtual shift vector field near
the crack front line. Fig. 10.15-1 shows a crack front and cracked
faces near the crack front. In this example, the crack front line is
straight, see below for an example in which the crack front line is
curved. A coordinate system is aligned with the crack front line, as
shown.
Crack advance stations are placed along the crack front line.
These crack advance stations are not in general coincident with the
node points.
The crack front line is surrounded by a radial domain consisting
of two parts, an inner radial domain and an outer radial domain.
The radii of these domains are constant along the crack front.

a) Crack front, crack advance stations and radial domains
Crack front line

a Crack advance stations
b) Top view of crack front
Outer radius of
radial domain ro
b Inner radius of
radial domain ri
Top crack face a
Bottom crack face
c) Side view of crack front, at crack advance station
Figure 10.15-1: Crack front, crack advance stations, radial domains and
coordinate system for SVS virtual shifts

Fig 10.15-2 shows how the crack advance is modeled for an

advance of the end crack advance station. The virtual shift vector
field X is shown in blue. This vector field is aligned with the
local a direction, and its magnitude varies within the domain both
in the radial direction and in the axial direction. First, consider the
vector field on the plane corresponding to the crack advance station
(Fig 10.15-2(d)). On this plane, the vector field has full strength
within the inner radial domain, and has zero strength outside the
outer radial domain. Between the inner and outer radial domain,
the magnitude of the vector field varies with the radius measured
from the crack front. Next, consider the vector field on a plane
halfway between the end crack advance station and the neighboring
crack advance station. (Fig 10.15-2(c)). On this plane, the strength
of the vector field is given by the strength of the vector field on the
plane corresponding the crack advance station, multiplied by one-
half. The strength of the vector fields on all other planes normal to
the crack front is constructed similarly by interpolation. (The
interpolations used will be discussed in more detail later.)
Fig 10.15-3 shows how the crack advance is modeled for an

advance of an intermediate crack advance station. Again, this
vector field is aligned with the local a direction, and its magnitude
varies within the domain both in the radial direction and in the
axial direction. The same interpolation of the vector field is
employed as already discussed for the end crack advance station.

a) Isometric view Virtual shift vector
b) Top view
c) Side view of crack front, between d) Side view of crack front,

crack advance stations at crack advance station
Figure 10.15-2: Modeling of crack advance at end crack advance

station of open crack front

a) Isometric view
b) Top view
c) Side view of crack front, between d) Side view of crack front,

crack advance stations at crack advance station
Figure 10.15-3: Modeling of crack advance at intermediate crack

advance station

It is important to remember that the crack advance stations and

radial domains are constructed without explicit reference to the
finite element mesh surrounding the crack front. The virtual motion
of the crack front is not represented in terms of nodal virtual shifts.
Notice that the axial crack advance is not modeled using linear
functions. Instead, cubic functions are used. The reasons for this
choice are given below.
When the crack front line is curved, very similar ideas are used.
One difference concerns the construction of the local coordinate
system. Fig. 10.15-4 shows a top view of the crack in the special
case of a curved crack front modeled with straight element edges.
The normal and tangential directions on the element edges are
denoted n and t respectively. At the nodes, the n directions are
averaged to obtain average nodal point normals. The average nodal
points normals are then interpolated to the crack advance station to
obtain the local a direction. Direction c is then constructed to be
orthogonal to direction a and also to the crack plane direction b .
Averaged
nodal point
normal a n
c
t
Crack advance
Node station
Figure 10.15-4: Crack front, crack advance stations and

coordinate system for a curved crack front
In this way, the local coordinate directions a , b , c vary

continuously along the crack front.
The nodal point normal averaging is also done when the
element edges are not straight.
Of course, the difference between the averaged and unaveraged
normals can be made smaller by refining the mesh (assuming that
the crack front to be modeled is continuous, and that the

discontinuities in the crack front are due only to mesh

discretization).
The local coordinates (r , , s ) for global coordinate x are

determined as follows (Fig. 10.15-5). First the plane normal to the
crack front that contains x is found. This plane is associated with
the distance along the crack front s . Then the projections of x
along local directions a and b give local coordinates a and b .
The polar coordinates (r , ) are easily obtained from (a, b) .
Material particle at x
b
(a, b)
s=0 r q
a
s
Figure 10.15-5: Local coordinates for a material particle
If there is more than one plane normal to the crack front that
contains x , the plane for which r is smallest is used. This
situation can occur when the crack front is a closed loop.
The outer radius of the radial domain of integration must be

smaller than the radius of curvature of the crack front. Otherwise
the local coordinates cannot be uniquely determined within the
radial domain of integration, and the radial domain is degenerate
(Fig. 10.15-6).

Extent of outer radial domain
Degenerate domain,
radius of outer radial Crack front
domain is greater than
radius of curvature of
crack front
Top view of crack front
Figure 10.15-6: Degenerate radial domain
In terms of the local coordinates defined above, the virtual shift

field evaluated at position x for a unit crack advance at station
ICAS is given by
X ICAS = qr qs ICAS
a (10.15-1)
In this equation, direction a is considered a function of the crack

front coordinate s . The radial decay of the virtual shift field is
given by the cubic function shown in Fig. 10.15-7(a):
qr = 1, r < ri
= 1 3ur2 + 2ur3 , ri r < ro (10.15-2)
= 0, r ro
r ri
in which ur = . The decay of the virtual shift field along the
ro ri
crack front is given by the cubic functions shown in Fig. 10.15-
7(b):

qs ICAS
= 0, s < sICAS1 or s > sICAS+1
= 3 us2 2 us3 , sICAS1 s < sICAS (10.15-3)
ICAS1 ICAS1
= 1 3 us2 + 2 us3 , sICAS s < sICAS+1

ICAS ICAS
in which sICAS1 , sICAS , sICAS+1 are the distances along the crack
front at the successive stations ICAS 1, ICAS, ICAS + 1 , and in
s sICAS
which us = .
ICAS
sICAS+1 sICAS
qr
1
ri ro r
a) Radial decay
qs ICAS
1
sICAS-1 sICAS sICAS+1 s
b) Axial decay
Figure 10.15-7: Cubic interpolation functions for

radial and axial decay
The gradient of the virtual shift field is obtained in two stages.

First the gradient is obtained in the local coordinate system:

X X X
, , , then the gradient is transformed into the global
r s
X X X X
coordinate system: = , , .
xi x y z
The virtual crack area for a crack advance at station ICAS is

given by equation (10.5-2), with the use of equation (10.15-1):
Ac ICAS
= (a n) qs ICAS
ds (10.15-4)
10.15.3 Calculation of J-integral and stress intensity factors
The following information is calculated for each virtual shift:
The J-integral of the finite element solution (equation (10.5-1)

along with the corrections (10.5-5 to 10.5-7)).
For static analysis in which the elements are all linear elastic
isotropic without thermal effects, the following additional
information is calculated:
The interaction integrals between the finite element solution and

the fundamental fields (e.g. G (u, u I ) ), and also the interaction
integrals of the fundamental fields (e.g. G (u I ) ). Equation (10.6-
2) and the correction equation (10.6-13) are used to calculate G .
The stress intensity factors K I , K II , K III . The matrix equation

(10.6-16) is used, with the numerically calculated values of the
interaction integrals of the fundamental fields used on the left-
hand-side of that equation.
The stress intensity factors K I , K II , K III are calculated for
boundary cracks, even for problems in which K II , K III are
expected to be zero due to symmetry.
Details of the numerical integrations used are given in Section

10.15.4.

One exception concerns the calculations of stress intensity

factors at the ends of an open crack front. Notice that the axial
decay qs has a net change in value along the domain used at the
end of an open crack front (Fig 10.15-2), therefore the integral
qs
s ds is not equal to zero. As a consequence, the second term
in equation (10.6-10) is non-zero, in other words, the fundamental
fields for modes II and III are not orthogonal within the domain.
Therefore ADINA computes only the stress intensity factor K I for
virtual shifts at the ends of an open crack front.
10.15.4 Calculation of G using numerical integration
All integrals involved in the calculation of G are evaluated

numerically, using the integration points and weights of the
underlying finite element mesh. Thus, for example, the integral
u j X k
V
ij
xk
W ik
xi
dV is evaluated using numerical
u j X k
integration of the integrand ij W ik . In this
xk xi
u j
integrand, ij , , W are obtained from the finite element
xk
X k
solution at the integration point, and is obtained from the
xi
formulas given above. We will discuss the accuracy of this
integration below.
For calculation of stress intensity factors, it is also required to

calculate expressions such as G (u I ) and G (u, u I ) . In this
calculation, the fundamental solutions in Section 10.3 are used,
assuming plane strain conditions, and using the local coordinates
(r , ) .
The accuracy of the calculated stress intensity factors depends

on

The accuracy of the numerical integrations used in G
The accuracy of the finite element solution displacement

field u .
These two accuracies are separate from each other. It is possible

for the finite element solution field to be accurate and the
numerical integrations used in G to be inaccurate, and vice versa.
In this section, we discuss the accuracy of the numerical

integrations used in G , without consideration of the accuracy of
the finite element solution field.
The accuracy of the numerical integrations depends upon the

finite element mesh employed. For example, consider the integral
V ICAS
=
V
dV (10.15-5)
ICAS
in which V ICAS
corresponds to the domain in which X ICAS is
non-zero (Fig 10.15-8).
ICAS # 1
ICAS
ro
ICAS + 1
Figure 10.15-8: Domain over which D X ICAS is non-zero
The analytical value of this integral is
V ICAS
= ro2 ( sICAS+1 sICAS1 ) (10.15-6)

If a finite element mesh is used in which this domain is completely

filled by elements, and in which the element boundaries correspond
to the domain boundaries, then this integral is exactly calculated by
the usual numerical integration used in finite element analysis. But
for an arbitrary mesh, the domain contains partial elements (that is,
elements cut by the domain boundary), and the numerical
integration is not exact.
On the other hand, as the mesh is refined, keeping the stations
and radial domains constant, we expect that the size of all of the
partial elements will become smaller, so that the numerical
integration becomes more accurate.
To get an idea of the rate of convergence as the mesh is refined,

consider a cube meshed with cube elements, and in each cube
element, consider a cubic lattice of integration points each with
uniform weight. Thus the entire domain contains a uniformly
spaced lattice of integration points. If there are N 1/ 3 integration
points in each direction, there are N integration points total.
Now consider the calculation of the volume of a domain that is
just within the mesh of cube elements. This volume can be
calculated by numerical integration using the uniformly spaced
lattice of integration points. Any error in the integration will be
confined to the boundary integration points. In this case, there are
approximately 6N 2 / 3 integration points on the boundary. In the
worst case, these boundary integration points, that should have
contributed to the numerical integration, do not contribute. The
6N 2/3 6
relative error is then = 1/ 3 Clearly as N increases, the
N N
error decreases, but the rate at which the error decreases is low.
Another way to express this rate of convergence is in terms of
the element size h . Since the total volume is proportional to Nh3
the relative error is proportional to h .
These results hold when integrating an arbitrary function over
the domain, when the function has non-zero value on the boundary.
These results also hold (with different constants of proportionality)
for arbitrarily shaped domains with smooth boundary.
When the integrand has zero value on the boundary, but is
proportional to h near the boundary, then the relative error in the
integral becomes proportional to h 2 .

Fortunately, for the integrals used in the calculation of G , it is

possible to set the integrand equal to zero at the boundary. The
integrand is set to zero at the boundary by proper choice of the
virtual shift field X ICAS , as follows.
The integrals that need to be evaluated have forms similar to
u j X k
V
ij
xk
W ik
xi
dV . Therefore the integrals can be set
X k
to zero on the boundary by choosing X ICAS so that is
xi
X k
zero on the boundary. Since X ICAS = qr qs ICAS a , can be
xi
qr
set to zero on the boundary by setting both qr , to zero on the
r
qs
boundary r = ro , and by setting both qs ICAS , to zero on
s ICAS
the boundary s = sICAS1 , s = sICAS+1 .
These considerations suggest the use of the cubic interpolation
functions for qr and qs ICAS already presented above in equations
qr
(10.15-2) and (10.15-3). Clearly both qr and are equal to
r
qs
zero on the boundary r = ro , and both qs ICAS and are
s ICAS
equal to zero on the boundaries s = sICAS1 , s = sICAS+1 .
Considering the numerical integration of integrals such as

u j X k
V
ij
xk
W ik
xi
dV , for elements entirely within the
domain, it is recognized that using linear interpolations for

X ICAS (as in the NVS method) corresponds to multiplying
u j
ij W ik by constants. Therefore if the numerical integration
xk

u j
of ij W ik is exact in an element, the numerical integration
xk
u j X k
of ij W ik (in which linear interpolations are used
xk xi
for X ICAS ) is also exact. On the other hand, using cubic
interpolations for X ICAS corresponds to multiplying
u j
ij W ik by quadratic functions. So even if the numerical
xk
u j
integration of ij W ik is exact in an element, the numerical
xk
u j X k
integration of ij W ik (in which cubic
xk xi
interpolations are used for X ICAS ) is not exact. Thus we have
introduced some error into the numerical integration process
through the use of cubic interpolations for X ICAS .
On the other hand, the rate of convergence for numerical
integration, for elements fully contained in the domain, is
proportional to the size of the element h raised to a power of at
least 2. (For example, for rectangular Gauss integration, the rate of
convergence is proportional to h 2m where m is the number of
Gauss points in each direction.) This rate of convergence is faster
than the original rate of convergence (proportional to h ) obtained
when the integrand is non-zero on the domain boundary. This rate
of convergence is also at least as fast as the rate of convergence
anticipated when the integrand is proportional to h near the
boundary.
So we anticipate that the relative error of the numerical
integration of the integrals used for the calculation of G is
proportional to h 2 , where h is the element size.
The above considerations are based on worst-case conditions. In

practice, better rates of convergence are often observed.

There are several ways to assess the actual accuracy of the

numerical integrations performed in the calculation of G . One
way is to numerically calculate an integral in which the actual
value is known, and in which the behavior of that integral is similar
to the behavior of the integrals used in G .
u j
Close to the crack front, ij W ik has magnitude
xk
proportional to 1/ r . Therefore the integral
u j X k ICAS
V xk
ij W ik
xi
dV is similar to the integral
1 ( qr qs ICAS )
V r r
dV .
1 ( qr qs ) dV

ICAS
The integral can be evaluated
V r r
analytically over the given domain as follows:
1 ( qr qs ICAS ) sk +1 2 ro 1 ( qr qs ICAS )
V r r
dV = sk1 0 0 r r
r dr d (a n)ds
sk +1 1 ( qr qs ICAS )
ro
= 2 r dr (a n)ds
sk 1 0 r r
sk +1 ro q
= 2 r
qs ICAS dr (a n) ds
sk 1 0 r
( )
sICAS+1
= 2 qr ro
qr 0
qs ICAS
(a n) ds
sICAS1
= 2 Ac ICAS
So define
1 1 ( qr qs ICAS )
GTEST ICAS
=
2 Ac V r r
dV (10.15-7)
ICAS
Based on the above analysis, we anticipate that this integral, when

evaluated using the same numerical integration as used for the

evaluation of G , should converge to the exact value of 1 with a

rate of convergence of h 2 when cubic interpolation functions are
used for qr and qs( k ) . And the accuracy of GTEST should be
similar to the accuracy of G .
For a boundary crack, substitute for 2 in the above
expressions.
Another way to confirm the actual accuracy of the numerical

integrations is to compare the calculated values of G (u I ) ,
G (u II ) , G (u III ) , G (u I , u II ) , G (u I , u III ) , G (u II , u III ) with their
theoretical values (equations 10.6-6). This confirmation can be
performed only when the crack is straight, however. Also recall
that G (u II , u III ) is theoretically non-zero for virtual shifts
corresponding to the end crack advance stations of open crack
fronts (equation 10.6-10).
10.15.5 Definitions of the cracks
Each crack is defined using the command
CRACK-SVS NAME,
ADV-OPTION ADV-NUMBER ADV-DISTANCE,
RADIUS-INNER RADIUS-MAXIMUM,
RADIUS-NUMBER RADIUS-RATIO,
BINORMAL-DIRECTION,
START-OPTION START-NODE START-POINT,
START-X START-Y START-Z START-SYSTEM,
SHIFT-OPTION SHIFT-VALUE GTYPE
namei bodyi sidei
The parameters of this command are
NAME: number to assign to this crack. There can be more than one
crack in the model.
ADV-OPTION, ADV-NUMBER, ADV-DISTANCE: Parameters

used to place the crack advance stations along the crack front. The
default is to place the crack advance stations at the corner nodes of

the element edges on the crack front.
RADIUS-INNER, RADIUS-MAXIMUM, RADIUS-NUMBER,

RADIUS-RATIO: Parameters used to define the radial domains.
The crack can have more than one radial domain. The default for
RADIUS-INNER is 0, RADIUS-NUMBER is 1 and RADIUS-
RATIO is 1. There is no default for RADIUS-MAXIMUM;
RADIUS-MAXIMUM must be specified.
BINORMAL-DIRECTION: Parameter used to orient the crack

binormal direction (crack plane direction) b . The default is to
orient the binormal direction upwards.
START-OPTION, START-NODE, START-POINT, START-X,

START-Y, START-Z, START-SYSTEM: Parameters used to choose
the starting crack advance station for a closed crack front. The
default is to set the starting crack advance station at the lowest
corner node number.
SHIFT-OPTION, SHIFT-VALUE: Parameters used to shift the mid-

side, mid-face and mid-volume nodes towards the crack front. The
default is not to shift the mid-side, mid-face and mid-volume nodes
of elements adjacent to the crack front to the quarter-points.
GTYPE: Parameter used to specify the type of input on the data

input lines.
namei, bodyi, sidei: Data input lines used to select element faces on
the crack faces.
If GTYPE=THREE-D, then namei is either the number of a

geometry surface (if (bodyi = 0) or the number of a geometry
face (if (bodyi > 0). If GTYPE=ELFACESET, then namei is
the number of an element face-set and bodyi is unused.
sidei specifies whether the specified geometry (or element

face-set) corresponds to the top surface, bottom surface, cracked
surface or uncracked ligament of the crack. For an enclosed
crack, define the crack by specifying the geometry on the crack
top and bottom surfaces. For a boundary crack, define the crack
by specifying the geometry on the cracked surface and on the

uncracked ligament.
For full details, see the Command Reference Manual for ADINA,
CRACK-SVS command.
The CRACK-SVS command determines the crack front edges as

the element edges that are on both the top and bottom surface of the
crack (enclosed crack), or the element edges that are on both the
cracked surface and uncracked ligament (boundary crack).
The crack front edges can form an open curve (open crack front) or
a closed curve (closed crack front). In the case when the crack
front is a closed curve, the crack advance stations at the beginning
and end of the curve are coincident.
Examples
Example 1:
ELFACESET 1 ... // defines an element face-set

for the top surface of a crack
ELFACESET 2 ... // defines an element face-set
for the bottom surface of the crack
CRACK-SVS RADIUS-MAXIMUM=2 GTYPE=ELFACESET
1 ,, TOP
2 ,, BOTTOM
DATAEND
// defines an SVS crack with number 1. The crack
advance stations are at the corner nodes of the
element edges on the crack front. Any mid-side
nodes adjacent to the crack front are shifted to the
quarter-points. There is one radial domain with inner
radius zero and outer radius 2.
Example 2:
// Mesh is defined using geometry bodies. The top of the crack

corresponds to faces 1 and 4 of body 2, and the bottom of the crack
corresponds to faces 3 and 2 of body 4.
CRACK-SVS NAME=2 ADV-OPTION=NUMBER,

ADV-NUMBER=30,

RADIUS-INNER=0.1,
RADIUS-MAXIMUM=2,
RADIUS-NUMBER=5,
SHIFT-OPTION=NO
1 2 TOP
4 2 TOP
3 4 BOTTOM
2 4 BOTTOM
DATAEND
// defines an SVS crack with number 2. 30 crack
advance stations are equally spaced around the crack
front. There are five radial domains with maximum
radius 2. Since the RADIUS-RATIO is 1.0, the
spacing between successive radii is constant with
value (2 - 0.1)/5 = 0.38. So the domains are
Domain 1: inner radius 0.1, outer radius 0.48 (0.1 + 0.38)

Domain 2: inner radius 0.1, outer radius 0.86 (0.48 + 0.38)
Domain 3: inner radius 0.1, outer radius 1.24
Any mid-side nodes adjacent to the crack front are not shifted.
10.15.6 Virtual shift numbers assigned to the virtual shifts
The CRACK-SVS command assigns a virtual shift number to each

virtual shift.
In the following,
ICRACK = the label number of a crack defined with the

CRACK-SVS command.
NCAS(ICRACK) = the number of crack advance stations for

crack ICRACK.
NCRAD(ICRACK) = the number of radial domains for crack

ICRACK.
NCVS(ICRACK) = the number of virtual shifts for crack

ICRACK.
Then the number of virtual shifts for crack ICRACK is given by
NCVS(ICRACK)=NCAS(ICRACK)*NCRAD(ICRACK)
For the first crack (crack with lowest label number), the virtual
shifts are defined in the following order:
virtual shift 1 = virtual shift at crack advance station 1 for radial

domain 1
virtual shift 2 = virtual shift at crack advance station 2 for radial
domain 1
...
virtual shift NCAS = virtual shift at crack advance station
NCAS for radial domain 1
virtual shift NCAS+1 = virtual shift at crack advance station 1
for radial domain 2
...
virtual shift 2*NCAS = virtual shift at crack advance station
NCAS for radial domain 2
...
virtual shift NCRAD*NCAS = virtual shift at crack advance
station NCAS for radial domain NCRAD.
When there is more than one crack defined in the model, then the
virtual shift numbering for each successive crack starts with the
highest virtual shift number for the preceding cracks + 1. For
example, in a model with two cracks, the virtual shifts for the
second crack are defined in the following order:
virtual shift NCVS(1) + 1 = virtual shift at crack advance station

for radial domain 1
...
virtual shift NCVS(1) + NCRAD(2)*NCAS(2) = virtual shift at
crack advance station NCAS(2) for radial domain NCRAD(2).
Example:
CRACK-SVS 1 ... // defines an SVS crack with 7 crack

advance stations and 3 radial domains

CRACK-SVS 5 ... // defines an SVS crack with 4 crack

advance stations and 2 radial domains
The following virtual shift numbers are defined:
Virtual shift Crack-SVS Radial domain Advance

number number number station number
1 1 1 1
2 1 1 2
...
7 1 1 7
8 1 2 1
...
14 1 2 7
15 1 3 1
...
21 1 3 7
22 5 1 1
...
25 5 1 4
26 5 2 1
...
29 5 2 4
10.15.7 Modeling hints
The crack faces must be flat and parallel to each other.
Each radial domain must include at least one element

integration point.
Examples of good and bad modeling are shown in Figure 10.15-

9.

b b
a a
a) Crack faces are parallel and flat within b) Crack faces are not parallel, not OK
the virtual shift domain, OK
In the auxiliary field calculations,

this point appears to lie on the bottom
surface of the crack.
b
a b
c) Radial domain lies outside of d) Crack faces are not flat within the
model domain, not OK radial domain, not OK
Elements adjacent to
crack front
e) Radial domain does not enclose

any element integration points,
not OK
Figure 10.15-9: Examples of good and bad modeling for SVS cracks

It is allowed for the crack front to be curved. However, the

theory used in the calculation of stress intensity factors assumes
that the crack front is straight.
Therefore, in order that accurate results be obtained, the outer

radius of the radial domain should be much smaller than the radius
of curvature of the crack front. This may require a fine mesh near
the crack front.
If the mesh is unstructured, shifting the nodes to the quarter-

points can lead to overdistorted meshes. This is why the default is
not to shift the nodes to the quarter-points.
The number of crack advance stations should not be greater than

the number of corner nodes on the crack front.
10.15.8 Fracture mechanics output
VSINFO command
During post-processing, the VSINFO command can be used to

output a number of useful tables that summarize the definitions of
the virtual shifts, and which also output various error measures.
The error measures are
SVS_VOLUME_PERCENT_ERROR
SVS_GTEST_PERCENT_ERROR
G-I_PERCENT_ERROR
G-II_PERCENT_ERROR
G-III_PERCENT_ERROR
(see descriptions of these error measures below).

VIRTUAL_SHIFT_NUMBER:
The number of the virtual shift
SVS_CRACK_NUMBER:
The crack number
SVS_ADVANCE_STATION_NUMBER:
The crack advance station number
SVS_RADIAL_DOMAIN_NUMBER:
The crack radial domain number
VIRTUAL_SHIFT_DIRECTION-{XYZ}:
The direction of the virtual shift
SVS_CRACK_FRONT_LENGTH:
The total length of the crack front = sNCAS s1
SVS_CRACK_FRONT_DISTANCE:
The distance along the crack front to the crack advance
station = sICAS
SVS_LENGTH:
length of axial domain,
= s2 s1 for a virtual shift at the start of an open crack front,
= sNCAS sNCAS1 for a virtual shift at the end of an open
crack front,
= sICAS+1 sICAS1 otherwise
SVS_INNER_RADIUS:
Inner radius of the radial domain ri
SVS_OUTER_RADIUS:
Outer radius of the radial domain ro
SVS_AREA:
Area of the virtual shift (equation 10.15-4)
SVS_ELEMENT_INT_POINTS:
Number of element integration points used in the domain.
SVS_PRESSURE_INT_POINTS:
Number of pressure load integration points used in the

domain.
SVS_VOLUME_CALCULATED:
Calculated volume of the domain, as computed using
numerical integration
SVS_VOLUME_THEORETICAL:
Theoretical volume of the domain
SVS_VOLUME_PERCENT_ERROR:
Relative error in the volume calculation:
SVS_VOLUME_CALCULATED
SVS_VOLUME_PERCENT_ERROR=100 1
SVS_VOLUME_THEORETICAL
SVS_PRESSURE_AREA_CALCULATED:
Calculated pressure area of the domain, as computed using
numerical integration.
SVS_GTEST_CALCULATED:
Calculated value of the test integral GTEST , as computed
using numerical integration. GTEST is defined in equation
(10.15-7).
SVS_GTEST_THEORETICAL:
Theoretical value of the test integral GTEST .
SVS_GTEST_PERCENT_ERROR:
Error in the calculation of GTEST :
SVS_GTEST_CALCULATED
SVS_GTEST_PERCENT_ERROR=100 1
SVS_GTEST_THEORETICAL
G-FE:
The value of the J-integral computed from the finite element
solution, as described in Section 10.5. For a crack in mode I
loading, G_FE can be interpreted as the energy release rate.
G_FE includes the pressure, temperature and dynamic
corrections, if these corrections are requested in the
FRACTURE command (the default is to request the
corrections).

G-FE_UNCORRECTED:
The value of the J-integral computed from the finite element
solution, without corrections.
G-FE_DYNAMIC_CORRECTION:
The value of the dynamic correction of the J-integral
computed from the finite element solution.
G-FE_PRESSURE_CORRECTION:
The value of the pressure correction of the J-integral
G-FE_TEMPERATURE_CORRECTION:
The value of the temperature correction of the J-integral
G-FE-I, G-FE-II, G-FE-III:

The values of G as computed from the interaction of the
finite element solution and the auxiliary mode I, II, III fields.
G-I, G-II, G-III:

The values of G as computed from the auxiliary mode I, II,
III fields, calculated using numerical integration. For
example G-I = G (u I ) .
G-I-II, G-I-III, G-II-III:

The values of G as computed from the interactions of the
auxiliary mode I, II, III fields, calculated using numerical
integration. For example G-I-II = G (u I , u II ) . In theory,
for a straight crack front, these values should all be zero
(with the exception of G-II-III computed at the end
crack advance stations of an open crack front).
G-I_THEORETICAL, G-II_THEORETICAL,
G-III_THEORETICAL:
The theoretical values of G as computed from the auxiliary
mode I, II, III fields (assuming that the crack front is

1
straight). For example G_I_THEORETICAL = .
E
G-I_PERCENT_ERROR, G-II_PERCENT_ERROR,
G-III_PERCENT_ERROR:
The percent error in G as computed from the auxiliary mode
I, II, III fields (assuming that the crack front is straight). For
example
G-I
G-I_PERCENT_ERROR = 100 1 .
G-I_THEORETICAL
G-RESIDUAL:
G as computed from the residual field, namely the finite
element solution minus the stress intensity factors times the
auxiliary fields.
G-RESIDUAL_PERCENT:
The relative size of G-RESIDUAL, compared with G-FE,
expressed as a percentage:
G-RESIDUAL
G-RESIDUAL_PERCENT= 100
G-FE
K-I, K-II, K-III:

The stress intensity factors. K-II, K-III are not
computed for the end crack advance stations of an open
crack front.
The above list assumes that the analysis is isotropic linear elastic
without thermal or dynamic effects.
If the analysis is not isotropic linear elastic, or the analysis includes

thermal or dynamic effects, the auxiliary field variables, residual
field variables and stress intensity factors are not calculated.

11.1 Substructuring
11. Additional capabilities

11.1 Substructuring
11.1.1 Substructuring in linear analysis
As an alternative to assembling and directly solving the

complete system in linear static analysis, substructuring can be
employed. Through the use of substructuring, in essence, large
new finite elements are established that are then assembled in the
same way as finite elements. Figure 11.1-1 illustrates an
application of substructuring. The computational effort saved
through substructuring can be significant. Also, the time required
for the preparation of the input data for a large problem may be
significantly less when using substructuring.
Typical structure of one story

(consists of beam elements, plate
elements, membrane elements, etc...)
Structure considered
Figure 11.1-1: Substructuring in the analysis of a multistory

building

Chapter 11: Additional capabilities
Each substructure can be reused one or more times; each use of

the substructure is called a "reuse". In Figure 11.1-1, one
substructure is defined, corresponding to one story of the building.
The substructure is reused for each story.
The nodes of a substructure are grouped into two categories,

condensed and retained. A condensed node is one that is not
connected to any nodes in the main structure, a retained node is
connected to at least one node in the main structure.
Substructuring in dynamic analysis can be employed when the

trapezoidal rule of time integration is used in a direct step-by-step
solution (see Section 7.3).
Substructuring can be used in two different ways:
ref. KJB If the mass is lumped to a relatively few degrees of freedom,

Section 8.2.4 the massless degrees of freedom can be statically condensed out
by substructuring. Figure 11.1-2 illustrates schematically the
analysis approach taken. When selecting this approach, you
must have good experience with the physical problem to be
solved in order to be able to assign masses to only a few degrees
of freedom, and yet obtain an accurate solution to the physical
problem considered.
Substructuring in dynamic analysis using ADINA can also

be employed without "lumping masses" to certain degrees of
freedom. In this case all degrees of freedom are treated as mass
degrees of freedom (corresponding to a lumped mass matrix),
and exactly the same solution is obtained as when no
substructuring is employed. The only difference in a solution
without substructuring is the cost of the numerical solution.

pp. 699-707, 1981.
This solution scheme can be very effective when large systems

need to be solved. Note again that in this case no error is
introduced in the solution using the substructuring.

11.1 Substructuring
l l
l l
l l
Main
structure
l l
l l
Substructure without
internal masses
l l
The option of having masses

within the substructures is also
available in ADINA.
Figure 11.1-2: Substructuring with no masses internal to

substructures
11.1.2 Substructuring in nonlinear analysis
The remarks made regarding substructuring in linear analysis

are also applicable when considering the nonlinear analysis of a
structural system.
An important additional consideration (beyond those for linear

analysis) is that it can be particularly effective to perform
substructuring in a nonlinear analysis when there are only local
nonlinearities. In such case, the linear degrees of freedom can be
allocated to the linear substructures and the nonlinear degrees of
freedom to the master structure. This way the equilibrium iteration
is performed in an effective manner.

Some applications are presented in the following reference.

pp. 699-707, 1981.
11.1.3 Substructuring restrictions
The following restrictions apply to substructures:
Cyclic symmetric analysis cannot be performed if substructures

are present.
FSI/TMC analysis cannot be performed if substructures are

present.
Mapping analysis options cannot be performed if substructures

are present.
Nonlinear element groups cannot be employed by the

substructure, i.e. the material model must be linear, the
displacements must be small and the element birth/death option
must not be activated. The main structure can, however, consist of
both linear and nonlinear element groups.
Contact surfaces cannot be formed by nodes belonging to a

substructure.
Any midsurface director vector sets that are manually defined

must be defined separately for each substructure. Skew systems
cannot, however, be defined separately for each substructure since
the skew systems defined for the main structure can also be
referenced by any of the substructures.
When substructures are used, rigid links or constraint equations

cannot be employed by any substructure or between substructures
and the main structure. Hence, constraint equations can only be
employed within the main structure.

11.1 Substructuring
Potential-based fluid element groups cannot be employed by the

substructure.
User-supplied element groups cannot be employed by the

substructure.
Substructure loads can only consist of concentrated loads,

element pressure loads, prescribed displacements and mass
proportional loading. Thus, centrifugal, electromagnetic,
temperature loading and pipe internal pressure loading cannot be
employed by a substructure.
Frequency analysis (and consequently mode superposition

analysis, response spectrum analysis, harmonic or random vibration
analysis) cannot be carried out when substructures are present.
Buckling eigenvalues and mode shapes cannot be calculated

when substructures are present.
Substructures cannot be used in explicit dynamic analysis.
Coupling nodes of a substructure must have the same relative

locations and the nodal degrees of freedom must have the same
directions as for the corresponding main structure nodes. If skew
systems or director vectors are used for the coupling nodes, then
care must be exercised so that the corresponding skew systems or
director vectors for the main structure nodes result in the same
directions for the nodal degrees of freedom.
Substructures must be used with a lumped mass assumption.
The iterative solver, multigrid solver and 3D-iterative solver

cannot be used when substructures are present.
Load penetration and pressure update (i.e., load stiffening effect

of shells) cannot be used when substructures are present.
11.1.4 Additional substructuring modeling techniques
The effectiveness of substructure modeling is significantly

improved when the ADINA substructure/reuse feature is used on

ADINA-M geometry entities. By so doing, the connections

between master structure and substructure parts can be easily
defined from all AUI/ADINA-M geometry entities, along with
meshing options of the commands COPY-TRIANGULATION,
FACE-LINK and EG-SUBSTRUCTURE, etc. It is particularly
effective when only some local nonlinearities (e.g., contact
regions) are considered in a large system analysis: many linear
deformation regions can thereby be condensed out and represented
just by a few degrees of freedom using the substructure/reuse
feature.
In the following examples, we demonstrate the effectiveness of
substructure modeling using the substructure/reuse feature.
Fig. 11.1-3 shows a structure with rotational symmetry (about

the X-axis). We can use 4 (or more) substructures to model the
entire structure. Substructure/reuse parts are meshed with shell
elements and the main structure is represented by isobeam
elements. The line loads are applied to inner isobeam elements and
fixities are then applied to the outer isobeam nodes. The analysis
result will be identical if the model is represented by one master
structure.
master structure with isobeam elements substructure/reuse with shell elements
substructure 2 substructure 1
substructure 3 substructure 4
Figure 11.1-3. Use of substructure/reuse feature: model with

rotational symmetry (1)

11.1 Substructuring
Fig. 11.1-4 shows another structure with rotational symmetry.

Again, we use a 4-substructure representation to model the whole
structure. In this example, we assign a plastic material to the
isobeam elements which represent the master structure. Shell
elements in the substructure/reuse parts are assigned by the linear
elastic material. This analysis thus becomes nonlinear. We then
verify this result with a model represented by one master structure
only. This example shows how local nonlinearities can be handled
by substructure modeling in ADINA, a method that can be very
effective when dealing with a large finite element analysis model
with some local nonlinearities.
substructure 2
substructure 1 main structure

substructure/reuses
substructure 3
substructure 4
Figure 11.1-4. Use of substructure/reuse feature: model with

rotational symmetry (2)
Fig. 11.1-5 shows a substructure/reuse model used to perform a

dynamic analysis on a building. Substructure/reuse parts are
meshed using linear shell elements and the reinforced
columns/beams are modeled by nonlinear isobeam elements. The
impact load is applied at the top left corner node. The analysis is
also performed in another model using a master structure only (left
half of the same figure). The analysis results from these two models
are identical.

reuse
reuse
master structure only master structure substructure/reuse
Figure 11.1-5. Use of substructure/reuse feature: model for

dynamic analysis of a building
Fig. 11.1-6 shows a simple model with wave propagation in a

rod. The rod is represented by five 3-D brick elements in one
substructure. The top and bottom surfaces are represented by shell
elements in the master structure. We define the master structure
element group in order to define the impact load and the initial
condition at the top and to define the boundary condition at the
bottom.
Fig. 11.1-7 shows a 2-D planar contact analysis model.

Substructures can model the regions which have no significant
deformation where coarse meshes can be used. The regions
near contact surfaces should be represented by the master structure,
and fine meshes near contact surfaces need to be used to perform
the nonlinear calculations (e.g., contact search) more accurately.
The connections between the substructure and the master structure
are defined by slave geometry entities and master geometry entities
via REUSE command input. The labels master I and master II
denote the different material properties can be used.

11.1 Substructuring
Z
B
U
B B 3
B - X Y
B
master structure substructure
Figure 11.1-6. Use of substructure/reuse feature: wave propagation in a rod
Z
contact
X Y
master I master II
substructure I
substructure II
Figure 11.1-7. Use of substructure/reuse feature: 2-D contact example

Fig. 11.1-8 shows a 3-D contact analysis model, built from

ADINA-M solid modeling. This example illustrates how the main
structure and the substructure are built for this type of nonlinear
analysis. In the actual analysis, finer meshes should be considered
near the contact surfaces. The connections between the
substructure and the master structure are defined by the REUSE
command as in the 2-D example, after which the mesh
compatibility between the substructure face and the master face
should be enforced by the COPY-TRIANGULATION or FACE-
LINK command.
substructure I
Y
X
master I and contactor

master II and target
substructure II
Figure 11.1-8. Use of substructure/reuse feature: 3-D contact example

11.1 Substructuring
Fig. 11.1-9 shows a long columnar block built from one AUI
geometry volume and 10 reuses of the meshed volume. Due to the
limitations of the current analysis, it is required that dummy shell
elements (Youngs modulus=0.001) are defined on the top/bottom
surfaces in order to define boundary conditions and loadings. The
input steps of Fig. 11.1-9 are as follows:
1. Define Volume 1 (top-surface, label 1 and bottom-surface,

label 6)
2. Define Surfaces 101 to 111 and Transformation translations

1 to 11 with direction vectors along column direction
3. SUBSTRUCTURE 1 / EGROUP THREEDSOLID 1 /

GVOLUME 1
4. REUSE 1 / 1 102 surface,,, matching transform_1

6 101 surface,,, matching transform_1
REUSE 2 / 1 103 surface,,, matching transform_2

REUSE 3 / 1 104 surface,,, matching transform_3

REUSE
REUSE 10 / 1 111 surface,,, matchingtransform_10

6 110 surface,,, matchingtransform_10
5. SUBSTRUCTURE 0 / EGROUP SHELL 1 /

GSURFACE 111 / GSURFACE 101
6. FIXITIES Surface / 111 all
7. LOAD force fx=1.0 / APPLY-LOAD / 1 force 1 surface

101

B B
B B B B
B B
U1 U2 U3
B - - -
substructure
whole structure master structure with 10 reuses substructure - reuse 1
Figure 11.1-9. Use of substructure/reuse feature: column substructure with 10 reuses
Some modeling considerations and limitations:
1. Shell node degrees of freedom are determined as described

in Section 2.7.3 for the master structure and the substructure
shell element nodes with the SHELLMODE =
AUTOMATIC option in the MASTER command, see Fig.
11.1-10.
2. Nodes at the connection locations are always assigned global

six degrees of freedom for the shell nodes, see Fig. 11.1-10.
3. To input boundary conditions and loadings at the master

structure nodes requires the element definitions at these
master nodes, see Fig. 11.1-9.
4. The same material property is assigned in the various

REUSE definitions of the same substructure.
5. The option TRANSFORMATION in the REUSE command

can only be a pure translation type, as shown in Fig. 11.1-5,
Fig. 11.1-9, and Fig. 11.1-10.

11.1 Substructuring
X Y substructure-reuse 2
B
B
B
main structure
6 dofs
5 dofs
6 dofs
5 dofs
main structure nodes
6 dofs
substructure-reuse 1
5 dofs
U U U
1 2 3 1 2 3
B - - - - - -
main structure
Figure 11.1-10. Use of substructure/reuse feature: assignment of degrees of freedom

to nodes at connection locations
11.2 Cyclic symmetry analysis

Cyclic symmetry analysis is useful for components when the
geometry and boundary conditions are rotationally symmetric. One
fundamental cyclic part is modeled and the complete structure is
obtained by rotating the fundamental part N times about an axis of
cyclic symmetry, where N is the number of cyclic parts. Figure
11.2-1 shows a schematic example.
Note that there is no approximation in the analysis when the

cyclic symmetry option is used: the same result is obtained as if the
complete structure is modeled.

Fundamental
Cyclic
cyclic part
boundary 2
Cyclic
boundary 1
o
360
-
N
Cyclic
symmetry axis
Figure 11.2-1: Schematic of a cyclically symmetric component
For a model with N cyclic parts, instead of solving for the

complete model at once, the solution is divided into M harmonics,
where
N +2
2 if N is even

M =
N +1
if N is odd
2
Each harmonic is solved separately, and the results are

superimposed to obtain the solution for the composite model.
For more details on the theory involved in cyclic symmetry, see
ref. W. Zhong and C. Qiu, "Analysis of Symmetric or

Partially Symmetric Structures," Computer Methods in
Applied Mechanics and Engineering, Vol. 38, pp. 1-18,
1983.
Although the geometry, material properties and displacement

boundary conditions must be cyclically symmetric, the external

applied loads can be arbitrary, including pressure loads,
concentrated loads, centrifugal and mass proportional loading, etc.
Prescribed displacement degrees of freedom have to be the same on
all cyclic parts. However, the amplitude of the prescribed
displacement can vary from one part to the next (see Figure 11.2-
2).
Note that solid cyclic structures with nodes along the cyclic
symmetry axis can also be modeled.
If the loads are also cyclically symmetric, the model also

possesses periodic symmetry. This assumption further simplifies
the analysis (see Figure 11.2-2). Note that this assumption
corresponds to having all the loads on the first harmonic (harmonic
0). In this case, only solution of the first harmonic is required.
For general cyclic symmetry (not periodic symmetry), linear

static, linear dynamic, and frequency analysis can be performed. In
linear dynamic analysis, mode superposition or implicit time
integration can be used.
Z p Z
Gravity
p1 p
Side 2
Side 2
F2 F F
F1
Side 1 Side 1
Y p Y
F F
X X F p
F3
p2
(a) General cyclic loading (b) Periodic cyclic loading
Figure 11.2-2: Applied loads in cyclic symmetry analysis

For periodic symmetry, a nonlinear analysis can be performed

as well (including large deformation, large strains, contact, etc.). A
periodic frequency analysis only obtains the periodic natural
frequencies. A dynamic analysis with explicit time integration can
also be performed for periodic symmetry. Thermomechanical
coupling (TMC) problems can also be set up with period
symmetry.
A nonlinear periodic static or dynamic analysis can be followed

by a full cyclic symmetry frequency analysis to obtain all the
natural frequencies and mode shapes of a the deformed structure.
In cyclic frequency analysis, the user can specify either the total
number of frequencies (default) or the number of frequencies per
harmonic. If the total number is specified, ADINA will internally
determine how many frequencies to solve per harmonic. A
conservative (large) value will be used to ensure that no frequency
is skipped.
Also in cyclic frequency analysis, the complete range of

harmonics can be used (default), or alternatively, the lowest and
highest harmonic values can be set by the user.
Another feature in cyclic frequency analysis is to solve for all

frequencies, or just solve for one a every complex frequency pairs.
This is useful because for all except the harmonic 0 (and the last
harmonic is N is even), cyclic frequencies come in pairs.
The cyclic symmetric analysis option has three advantages:
Only a small part of the complete structure needs to be

modeled.
For general cyclic symmetry, the storage required is reduced

up to a factor of N/2, where N is the number of cyclic parts. For
periodic symmetry, the storage required is reduced up to a factor
of N.
The solution time can be significantly less than when the

complete structure is modeled. As an estimate, the cyclic

symmetric analysis will reduce the solution time by a factor of

up to N/2.
Only the finite element model of the fundamental cyclic part

should be created with the AUI. The displacement boundary
conditions are only defined on the fundamental cyclic part. Note
that the displacements on the lines and/or surfaces making up the
cyclic boundaries are unknown (unless they are fixed or
prescribed).
The loads on each of the N cyclic parts must be defined. Loads

on a cyclic boundary can be input as loads onto either of the
adjacent cyclic parts.
Results are saved for all cyclic parts, except in periodic

symmetry when they are only saved for the fundamental cyclic
part. When loading results, the user can select to load only a subset
of cyclic parts. This feature reduces the memory required for post-
processing cyclic symmetry results.
Cyclic symmetry constraints are automatically generated on the

cyclic boundary. Therefore, the degrees of freedom on the cyclic
boundary cannot be constrained by rigid links or constraint
equations.
The nodes of cyclic boundary 1 must coincide with those of

cyclic boundary 2 when rotated counterclockwise 360/N degrees
about the axis of cyclic symmetry. In 2-D analysis, this is achieved
by using the same mesh divisions along both cyclic boundaries.
The same applies to 3-D analysis with mapped meshes. In 3-D
analysis with free-form meshing and 3-D solid elements, using the
same mesh divisions on both cyclic boundaries will not guarantee
that a matching mesh is created throughout the boundary. To
ensure identical meshes on both cyclic boundaries, create a dummy
shell mesh on one cyclic boundary, then copy the resulting mesh
pattern to the other cyclic boundary via the Copy Mesh
Triangulation command, then create the 3-D mesh, and delete the
dummy shell element mesh.
The nodes of cyclic boundary 2 are automatically constrained to

those of cyclic boundary 1 via the constraints based on the theory

of cyclic symmetry. For this reason, nodes on cyclic boundary 2

cannnot be otherwise constrained (by other constraints or rigid
links).
The axis of cyclic symmetry can have any spatial orientation.
Note that if structural elements (truss, beam, iso-beam, plate,

shell or pipe elements) are defined exactly on the cyclic
boundaries, then the complete structural elements must be input in
the definition of the fundamental cyclic part.
Since a cyclic boundary connects two adjacent cyclic parts (for

one cyclic part it will be cyclic boundary 1 and for the other, cyclic
boundary 2), the loads on a cyclic boundary can be input as loads
onto either of the adjacent cyclic parts. However, prescribed
displacements on a cyclic boundary must be applied on the cyclic
part when it is cyclic boundary 1.
In dynamic analysis, the initial conditions for the N cyclic parts

must be defined.
In frequency analysis, the eigenvectors output by the cyclic

symmetry procedure are not M-normalized.
The following analysis options are not possible with cyclic

symmetry:
Potential-based fluid elements
Axisymmetric and generalized plane stress 2-D elements
Piping analysis with ovalization/warping degrees of freedom
Substructure analysis
Response spectrum analysis
Harmonic or random vibration analysis
Fracture mechanics analysis
FSI analysis
Explicit time integration (unless periodic symmetry)
TMC analysis (unless periodic symmetry)

11.3 Reactions calculation

Computations of the reaction forces and moments can be
requested for all degrees of freedom where fixity boundary
conditions are applied and where displacements are prescribed. No
reactions are calculated for the degrees of freedom corresponding
to deleted master boundary conditions.
Reaction calculations are on by default.
Note that applied loads applied to the fixed degrees of freedom

do not contribute to the displacement and stress solutions.
However, these loads are accounted for in the reaction calculations.
Reaction forces and moments at a node are computed using the

consistent force vectors (calculated from the element internal
stresses) of elements attached to the node. Hence, a check on the
balance of the support reactions and the applied loads often
provides a good measure on the accuracy of the solution (in terms
of satisfying equilibrium in a nonlinear analysis).
Reaction calculations in dynamic analysis with consistent mass

matrix and direct time integration take into account the mass
coupling to the deleted degrees of freedom. The reactions exactly
equilibrate the applied forces in all cases.
Reaction calculations in dynamic mode superposition analysis
do not take mass coupling to the deleted degrees of freedom into
account.
Reaction calculations are not available when the Wilson theta

method of time integration is used in dynamic analysis.
11.4 Element birth and death feature

The element birth and death feature is useful for modeling
processes during which material is added to and/or removed from
the structure. Such processes, for example, are encountered in the
construction of a structure (structural members are added in
succession), the repair of a structure (structural components are
removed and new ones are added) or during the excavation of

geological materials (a tunnel is excavated). If the element birth

and death option is used, the corresponding element groups become
automatically nonlinear. Fig. 11.4-1 illustrates two analyses that
require the element birth and death options.
The main features of element birth and death option are as

follows:
If the element birth option is used, the element is added to

the total system of finite elements at the time of birth and all
times thereafter.
If the element death option is used, the element is taken out

of the total system of finite elements at times larger than the
time of death.
If both element birth and death options are used, the element
is added to the total system of finite elements at the time of birth
and remains active until the time of death. The time of death
must be greater than the time of birth. The element is taken out
of the total system of finite elements at all times larger than the
time of death.
Once an element is born, the element mass matrix, stiffness

matrix and force vector are added to the mass matrix, stiffness
matrix and force vector of the total element assemblage (until the
death time, if any). Similarly, once an element dies, the element
mass matrix, stiffness matrix and force vector are removed from the
total assembled mass matrix, stiffness matrix and force vectors for
all solution times larger than the time of death of the element.
Note that an element is born stress free. Hence, even if the

nodal points to which the new element is connected have already
displaced at the time of birth, these displacements do not cause any
stresses in the element, and the stress-free configuration is defined
to occur at the nearest solution time less than or equal to the birth
time.
Note also that the damping matrix is not modified when

elements die or when they are born. Therefore, element groups with
such elements should have no Rayleigh damping. For example, if

11.4 Element birth and death option
you want to simulate a pipe break by setting the death time for
certain elements in an implicit dynamic analysis, then the element
group containing these elements should have damping parameters
set to zero. Otherwise, you might not see a pipe separation.
Installation of Column to be
temporary support replaced
(a) Repair of a bridge
Ground level
l
l l
l l
l l l
l
l
l l
l
Tunnel to be
excavated
(b) Excavation of a tunnel
Figure 11.4-1: Analyses that require the element birth and death
options

When the element birth/death option is used, the tangent

stiffness matrix may at some solution times contain zero rows and
corresponding columns. The equation solver disregards any zero
diagonal element in the tangent stiffness matrix if no elements are
attached to the associated degrees of freedom.
ADINA enables the user to set an element death decay time

parameter which causes the gradual reduction of the element force
vector and stiffness matrix to zero over a finite time rather than
instantly. The reduction starts at the death time and progresses
linearly with time until the decay time has passed. The element
therefore totally vanishes at a time equal to the sum of the death
time and the death decay time. This option is useful for mitigating
the discontinuity that the structure may experience due to the death
of some of its elements.
The element birth/death option applies to any mass effect i.e.,

gravity loading, centrifugal loading and inertia forces. The mass
matrix, therefore, does not remain constant throughout the solution.
To provide the appropriate concentrated and element loading

that takes into account the element birth/death option, you have to
specify time functions on the loading that correspond to the proper
elements (however, see also Section 11.5).
The time at which an element becomes active or inactive is

specified by the parameters TBIRTH and TDEATH respectively
(in the EDATA and EGROUP commands).
If an element is required to be born at time tb, i.e., the

configuration of the element at time tb corresponds to the stress-
free configuration, you should input TBIRTH = tb + where
t
= and t is the time step between time tb and the next
1000
solution time. If an element is required to be inactive at and after
t
time td, you should input TDEATH = td - , where = and
1000
t is the time step between the previous solution time and time td .

Birth option active: Figure 11.4-2(a) shows the activity of an

element for which the birth option is active. Note that if TBIRTH
is input for the range shown (where TBIRTH > tb and TBIRTH is
tb + t), then the stress-free configuration of the element is at time
tb, and the element is first active at time tb + t. Note that the
stress in the element is based on the displacements measured from
the configuration at time tb, irrespective of the input value of
TBIRTH, provided that TBIRTH is input in the range shown in
Fig. 11.4-2(a). See also the example given below.
Death option active: Fig.11.4-2(b) shows the activity of an

element for which the death option is active. Note that if TDEATH
is input for the range shown (where TDEATH td - t and
TDEATH < td), then the element is first inactive at time td.
TBIRTH in this range causes the element

to be included in the stiffness matrix and
the force vector at time t+t
Time
0 t t+t
Stress free First solution time for which
state the element is active
(a) Birth option active
TDEATH in this range causes the element to be

not included in the stiffness matrix and the
force vector at time t
Time
0 t-t t
First solution time for which
the element is inactive
(b) Death option active
Figure 11.4-2: Use of element birth and death option
Birth then death option active: This is a direct combination of the

birth and death options. Initially some elements in an element

group are inactive. At a particular solution time determined by the
time of birth TBIRTH, the elements become active and remain so
until a subsequent solution time determined by the time of death
TDEATH, where TDEATH > TBIRTH.
Example of the element birth option: Consider the materially

linear truss element model shown in Fig. 11.4-3(a) in which the
time of birth for element 2 is slightly larger than t. At time t, the
t t
length L corresponding to the load R is determined as shown in
Fig. 11.4-3(b).
t
Element 2, TBIRTH = t +
1000
u
Both elements are

Element 1 linear 2-node trusses
(always active)
L
a) Model schematic
tR
t F int = t R
1
Element 1:
K1=AE/ 0L
t int
F1 =K1(tL - 0L)
tL
b) Solution at time t
Figure 11.4-3: Example on the use of the element birth option

t
At the time of birth, element 2 with length L is added to the
system which was already in equilibrium, see Fig. 11.4-3(c). Note
t
that L corresponds to the length at which element 2 is stress-free.,

and that the internal force in element 2 is exactly zero immediately

after its addition to the system. Also note that the stiffness of
t
element 2 is based on the length L.
At time t + t, the force in element 2 is determined based on its
deformation with respect to its stress-free state, see Fig. 11.4-3(d).
Hence, the total increment in displacement from time t to time t +
t determines the force in element 2. Identically, the same solution
would be obtained using any value for TBIRTH which satisfies the
relation TBIRTH > t and TBIRTH t + t.
The birth/death feature is available for contact pairs in contact

analysis.

tF =0
Element 2
tL
tR
Element 1
tL
c) Stress-free configuration for element 2
Element 2:
K2=AE/ tL
t+t int
F2 =K2(t+tL - tL)
t+tR
t+t F int + t+t F int = t+t R

1 2
Element 1:
K1=AE/ 0L
t+t int
F1 =K1(t+tL - 0L)
t+tL
d) Solution at time t+t
11.5 Element "death upon rupture"

For the materials and elements listed in Table 11.5-1, element
death is automatically activated when rupture is detected at any
integration point of the element. The element is then considered as
"dead" for the remainder of the analysis, and, in essence, removed
from the model (mass and stiffness contributions).
As elements become dead, contactor segments connected to

11.5 Element death upon rupture
these elements are also removed from the model.
Load penetration can be activated to ensure that pressure loads

are properly redistributed inside voids created by "dead" elements
(see Section 5.3.1).
Table 11.5-1: Elements and material models that include "death upon rupture"
Truss* 2-D 3-D Beam Iso- Shell Pipe
solid solid beam
Plastic-bilinear, plastic-multilinear 9 9 9 - 9 9 9
Mroz-bilinear - 9 9 - - - -
Thermo-plastic, creep, plastic-creep,
multilinear-plastic-creep, creep- 9 9 9 - 9 9 9
variable, plastic-creep-variable,
multilinear-plastic-creep-variable
Plastic-orthotropic - 9 9 - - 9 -
Plastic-cyclic 9 9 9 9 - 9 -
Moment-curvature - - - 9 - - -
User-coded - 9 9 - - - -
*The rupture option is not applicable to a truss element with a gap.
11.6 Initial conditions

11.6.1 Initial displacements, velocities and accelerations
You can specify the initial displacements, velocities and

accelerations of the nodes.
Any initial displacements, velocities and accelerations specified

in a restart run are ignored, except when restarting from a static to a
dynamic analysis; in this case, initial velocities are taken into
account.

Initial rotations should only be applied in small displacement

analysis.
The initial rotations at a node are interepreted differently

depending upon the element attached to the node.
Element without rotational dofs: rotations are ignored
Linear or MNO element with rotational dofs: initial
rotations are used in the calculation of the initial force
vector.
Large displacement element with rotational dofs: initial
rotations are ignored.
Small displacement rigid link: the initial position of the
slave node is determined by the initial rotation of the master
node.
Large displacement rigid link: the initial position of the
slave node is obtained using the initial rotation of the master
node, as if the rigid link is a small displacement rigid link.
This will cause the length of the rigid link to be incorrect if
the initial rotation is large.
11.6.2 Initial temperatures and temperature gradients
The formulas used to determine the initial nodal temperatures

are as follows:
Initial temperature explicitly specified: 0 i = 0 iE

Initial temperature not explicitly specified: 0 i = TINIT
in which 0 i is the initial nodal temperature at node i, TINIT is the

overall initial temperature (also referred to as the default initial
temperature) and 0iE is the explicitly specified initial temperature
at the node.
Initial temperatures can also be read from a temperature file. In
this case, any directly specified initial temperatures are added to
those read from the temperature file.
Initial temperature gradients are handled in an analogous
manner. These are only used at axisymmetric shell nodes and at
shell mid-surface nodes in order to calculate initial temperatures in
these elements. Non-zero initial temperature gradients cannot be
applied to shell nodes.

Initial temperatures and temperature gradients specified in a

restart run are ignored.
When temperatures are directly prescribed (as in Section 5.6),

the directly prescribed temperatures are not used to determine the
initial temperatures. Initial temperatures must be specified as
described above.
The thermal strains are always assumed to be zero initially, see

Section 3.1.4.
11.6.3 Initial pipe internal pressures
The initial pipe internal pressures consist of the sum of the

explicitly defined pipe internal pressure loads (evaluated at time
TSTART) and the pipe internal pressures read from file, if any. If
there are no explicitly defined pipe internal pressure loads and no
pressure loads read from file, the initial pipe internal pressures are
set to zero.
Initial pipe internal pressures specified in a restart run are
ignored.
11.6.4 Initial strains
When the initial strain feature is used, the strains used in the
stress-strain law include the initial strains. Symbolically
= (e + 0 e)
where are the stresses, the function () is the function that

computes stresses from strains, e are the strains computed from the
displacements and 0 e are the initial strains. For example, for a
linear elastic material
= C(e + 0 e)
where C is the stress-strain matrix.
In the above, we assume that a materially-nonlinear-only

formulation is used, we will discuss large displacement

formulations later.
Figure 11.6-1 shows a single 2-D element in which the initial

strain option is used, and in which no other applied loads or body
forces are present. Tensile initial strains are applied. When the
element is unrestrained, = 0 , hence 0 = e + 0 e , hence e is
compressive, corresponding to an element that has shrunk. When
the element is rigidly restrained from motion, e = 0 , hence the
element contains tensile stresses.
Unrestrained element Restrained element

z
tzz
y
tyy
Deformed configuration Deformed configuration

(e = -0e nonzero; t = 0) (e = 0; t = C(e + 0e ) nonzero)
Initial strains: 0eyy = 0.05, 0 ezz = 0.10
Figure 11.6-1: Initial strains applied to 2-D solid element
This example illustrates the following general principles:
Tensile initial strains applied to an unrestrained assemblage

of elements cause the assemblage to shrink.
Tensile initial strains applied to a restrained assemblage of

elements induce tensile stresses in the assemblage.
The initial strain option is available for all solid and structural
elements, but only for nonlinear element groups. If the initial strain
option is applied to an element group that would otherwise be
linear, the element group becomes nonlinear.
All material models except for the Ilyushin material model can
be used.

The strains printed or saved to the porthole file include the

initial strains, that is eoutput = e + 0 e .
Initial strain axes
In general, initial strain components are interpreted in an initial

strain axes coordinate system. These axes are denoted (1-2-3), so
the initial strain components are 0 e11 , 0 e22 , 0 12 , etc. Notice that
the shear strain components use the engineering convention.
The initial strain axes for the truss, beam, iso-beam and pipe
elements have the 1 axes aligned with the axial direction of the
element. In most cases, only the axial initial strain 0 e11 is
considered, however for the axisymmetric shell element, the hoop
strain component 0 e33 is also considered.
The initial strain axes for the 2-D solid, 3-D solid, plate and
shell elements are user-defined. For the 2-D solid, plate and shell
elements, the 1-2 axes must lie in the plane of the element and the 3
axis lies out of the plane of the element. One way to define initial
strain axes is to use the AXES commands of the AUI. In the
following example, axes-system 27 is defined, then the initial strain
axes for the elements on surface 3 are set to axes-system 27:
AXES CONSTANT 27,

AX=0.5 AY=0.3 AZ=0.2 BX=-0.5 BY=0.2 BZ=0.1
SET-AXES-STRAIN SURFACES
ENTRIES SURFACE AXES
3 27
DATAEND
Input of initial strains
The initial strains can be entered as nodal point initial strains,

element initial strains or a combination of nodal point and element
initial strains.
Nodal point initial strains
Nodal point initial strains can be specified as follows:
At the nodes, using the INITIAL STRAINS command.

On geometry, using the INITIAL-CONDITION command to

create an initial condition, then using the SET-
INITCONDITION command to assign the initial condition
to the geometry.
The nodal point initial strains are specified in terms of initial strain
components in the initial strain axes coordinate system. For plate
elements, the membrane and flexural initial strains can be specified,
and for the shell elements, the membrane and gradient initial strains
can be specified.
Within each element, the nodal point initial strain components
are interpolated, component by component, to the integration
points.
Because the initial strain axes can have different orientations in
different elements, the nodal point initial strain components can be
interpreted differently in attached elements. For example, suppose
that node 45 has 0 e11 = 0.02 , and suppose that node 45 is attached
to elements 27 and 65. Within element 27, the value 0 e11 = 0.02 is
interpreted according to the initial strain axes of element 27, and
within element 45, the value 0 e11 = 0.02 is interpreted according to
the initial strain axes of element 45.
Element initial strains truss, beam, iso-beam, shell, pipe

elements
For the truss, beam, iso-beam, shell and pipe elements, the element
initial strain components are specified as element data (and
possibly also assigned to the geometry). For example, to assign
membrane element initial strains to shell elements, commands
similar to
* SHELL ELEMENT
EDATA
ENTRIES EL EPS11 EPS22 EPS12 EPS13 EPS23
1 0.01 0.02 -0.01 -0.02 -0.03
DATAEND
and
SURF-ELEMDATA SHELL

ENTRIES SURFACE EPS11 EPS22 EPS12 EPS13 EPS23

2 0.01 0.02 -0.01 -0.02 -0.03
DATAEND
can be used.
Element initial strains 2-D solid element (except for the 3D

plane stress element)
For the 2-D solid element, the element initial strain components are
specified using strain field coefficients A, B, C, D, E, F and the
formulas
0
ezz = A + Bz
0
eyy = C 0 ezz + D
0
exx = E 0 ezz + F (axisymmetric and generalized plane strain
elements only)
The dependence of the initial strains on the z coordinate is

useful for modeling initial strains in geological materials.
The strain field coefficients are defined using the AUI command
STRAIN-FIELD.
The subscripts x, y, z refer to the 2D coordinate system after
modification by the MASTER 2DPL-AX flag. If MASTER 2DPL-
AX=YZ-Z (the default), the 2-D elements all lie in the global y-z
plane with the global z axis as the axisymmetric axis. In this case
the 2D coordinate system is the same as the global system and the
above formulas give the initial strains in the global system.
However if, for example, MASTER 2DPL-AX=XY-Y, then the 2-D
elements all lie in the global x-y plane with the global y-axis as the
axisymmetric axis. In this case, the y axis of the 2D system is the
global x axis and the z axis of the 2D system is the global y axis.
So then in the above equations z refers to the global y coordinate,
0
ezz = A + Bz refers to the normal initial strain acting in the global
y direction, 0 eyy = C 0 ezz + D refers to the normal initial strain
acting in the global x direction and 0 exx = E 0 ezz + F refers to the
normal initial strain acting in the global z direction.

Element initial strains 2-D solid element, 3D plane stress option
For the 2-D solid element using the 3D plane stress option, the
element initial strain components are specified using strain field
coefficients A, B, C, D and the formulas
0
ezz = A + Bz
0
eyy = C 0 ezz + D
Here variable z is the global z coordinate, however subscripts y, z

refer to the element local y and z directions.
Element initial strains 3-D solid element
For the 3-D solid element, the element initial strain components are
specified using strain field coefficients A, B, C, D, E, F and the
formulas
0
ezz = A + Bz
0
exx = C 0 ezz + D
0
eyy = E 0 ezz + F
See comments for the 2-D solid element concerning the strain field
coefficients. For the 3-D solid element, variable z is always the
global z coordinate, and subscripts x, y, z always refer to the global
coordinate system.
Summary of initial strain input
Table 11.6-1 gives a summary of initial strain input for different

element types.

Table 11.6-1: Initial strains input for nodes and elements
Nodal point Element strain

strain components components
Element e11 e22 e33 12 13 23 e11 e22 e33 12 13 23
Truss 9 - - - - - 9 - - - - -
2-D solid 9 9 92) 9 - - 93) 93) 92,3) - - -
3-D solid 9 9 9 9 9 9 93) 93) 93) - - -
Beam 9 - - - - - 9 - - - - -
Iso-beam 9 - 94) - - - 9 - 94) - - -
Plate:
membrane, 9 9 - 9 - - 9 9 - 9 - -
flexural
Shell:
membrane, 9 9 - 9 9 9 9 9 - 9 9 9
gradient
Pipe 9 - - - - - 9 - - - - -
1) The symbol "9" means that the strain component can be

defined for that element type, either by node or by element
input.
2) Axisymmetric and generalized plane strain elements only.
3) Entered in the global system or 2D local system using strain

field coefficients.
4) Axisymmetric shell elements only.
Initial strains for large displacement / small strain formulation
For the large displacement / small strain formulation, the initial

strains are interpreted as initial Green-Lagrange strains, the initial
strains are added to the Green-Lagrange strains computed from the
displacements, and the result is used to compute the 2nd Piola-
Kirchhoff stresses, that is
0 S = 0 S( 0 + 00 )
This calculation is justified only when the strains (both total and

initial) are small.
Initial strains in large strain formulations, 2D and 3D elements:
The material models discussed in this section are the large strain
plasticity models (in which the ULH formulation is used) and also
the rubber-like material models.
The initial strains are assumed to be small, and the initial spin
tensor is assumed to be equal to zero, so that
X = (I + 0 e)
t0 1
0
t
in which 0 e is the initial strain tensor and 00 X is the initial
deformation gradient. Then a multiplicative decomposition of the
deformation gradient is used at every step, i.e.:
( X)
1
t
t0 X = 0t X t0
0
where 0t X is the total deformation gradient (computed from the

t
element displacements at time t) and t0 X is the mechanical
deformation gradient (used to compute the stresses). Note that with
the above convention, tensile initial strains applied to an element
that is unrestrained cause the element to shrink, as in Fig 11.6-1.
This follows because an unrestrained element has zero stresses,
X = I , 0t X = t00 X = ( I + 0 e ) and det 0t X < 1 (for tensile
t 1
hence t0
initial strains).
t
The components of t0 X are output when deformation gradients
are requested to be output.
Initial strains in inelastic materials
We briefly discuss initial strains in inelastic materials, such as

plasticity, creep and viscoelastic materials. For simplicity, we will
assume that the formulation is materially-nonlinear-only, however
similar considerations arise in large displacement and large strain
analysis.

The rule = (e + 0 e) is used for inelastic materials, where the

function () includes inelastic variables such as the plastic strain
as well as the strain e + 0 e .
One subtle point is that this rule is also applied at the very
beginning of the analysis, namely to compute the stresses at time
TSTART. At the beginning of the analysis, e = 0 , so the stresses
at time TSTART are computed using 0 = ( 0 e) . These stresses
need not be in equilibrium with the applied forces, since
equilibrium is not computed at time TSTART.
In addition, if the material is rate-dependent, the calculation of
0
= ( 0 e) includes the strain rate. The strain rate also includes
effects due to initial strains, and the strain rate depends on the time
step size. The time step size used in the first time step is also used
in the calculation of 0 = ( 0 e) . This can occasionally cause
difficulties, for example in explicit analysis, in which the first time
step is very small, and therefore in which the strain rate as
computed from the initial strain is large.
11.6.5 Initial stresses
An initial stress option can be used with all element types. The
following material models can be used: elastic-isotropic, elastic-
orthotropic, thermo-elastic, thermo-orthotropic, Drucker-
Prager, Cam-clay, Mohr-Coulomb.
The initial stress capability allows initial stresses corresponding to
an unknown loading to be input. These stresses can be provided as
nodal point initial stresses or element initial stresses.
Whenever the constitutive relation is used, the initial stresses are
added to the current stresses caused by the strains. The initial
stresses are included in the printed and saved stresses.
Either initial stresses or initial strains can be accepted by the

program. However the choice of initial stresses or initial strains is
a program-wide choice, so initial stresses and initial strains cannot
be used together in the same analysis.
The following equations are used:

= M + I (11.6-1)
F = BT dV (11.6-2)
V
in which I are the initial stresses (corresponding to the

formulation/material model used, i.e. Cauchy or 2nd Piola-
Kirchhoff), and M are the stresses calculated as usual from the
mechanical strains. Equation (11.6-2) yields at time 0
R I = F I = BT I dV
V
Please note:
The stresses I are in the K NL matrix (stiffening effect)

The stresses I are added to the stresses computed in the
material law.
The stresses I are effects that do not change in time.
There are two options for the initial stress feature:
Initial stresses do not cause deformations
The step-by-step incrementation is performed using
K(
i 1)
U ( ) = t +t R t +t F (
t +t i i 1)
+ RI
in which t +t R is the externally applied load vector. Use the AUI

command MASTER INITIALSTRESS=YES to request this option.
Figure 11.6-2 shows a single 2-D element in which this option
is used, and in which no other applied loads or body forces are
present. Even though the stresses are non-zero, no deformations
occur, because R I is added to the out-of-balance force vector.
This modeling technique can be very useful in the analysis of
geologic materials.
This option can be used in linear and materially-only-nonlinear
analysis.
However, it should be noted that if the birth/death element

option is used (e.g. tunneling), then the initial stresses cause

deformations.

z
tzz tzz
y
tyy tyy

tM = 0 t =tI tM = 0 t =tI
Initial stresses: tyy

I =1000, t I =2000
zz
Figure 11.6-2: Initial stresses applied to 2-D solid element,

initial stresses do not cause deformations
Initial stresses cause deformations
The step-by-step incrementation is performed using
K(
i 1)
U ( ) = t +t R t +t F (
t +t i i 1)
Use the AUI command MASTER INITIALSTRESS=DEFORMATION

to request this option.
Figure 11.6-3 shows a single 2-D element in which this option
is used, and in which no other applied loads or body forces are
present. Tensile initial stresses are applied. When the element is
unrestrained, the stresses are zero, and the element shrinks because
the mechanical stresses are compressive. When the element is
rigidly restrained from motion, M = 0 , hence the element
contains tensile stresses.
This option can be used in linear, materially-only-nonlinear
analysis and large displacement/small strain analysis.


z
tzz
y
tyy

tM =-t I t = 0 t M = 0 t =t I
Initial stresses: tyy

I =1000, t I =2000
zz
Figure 11.6-3: Initial stresses applied to 2-D solid element,

initial stresses cause deformations
Initial stress input
When initial stresses are chosen, then all initial stress input is
entered using the corresponding initial strain input commands.
Initial stress components are interpreted in an initial strain axes
coordinate system.
The initial stress components can be entered either using nodal
point initial stresses or element initial stresses. Simply substitute
the initial stress for initial strain in the commands that define initial
strains. For example, to assign membrane element initial stresses
to shell elements, the commands
* SHELL ELEMENT
EDATA
ENTRIES EL EPS11 EPS22 EPS12 EPS13 EPS23
1 1000.0 2000.0 500.0 200.0 -100.0
DATAEND
could be used. Table 11.6-1 applies to initial stresses, provided

that the stress components are substituted for the strain
components.
Restrictions
The initial stress option can be used for the Drucker-Prager,

Cam-clay and Mohr-Coulomb material models only if the effective

stress corresponding to the initial stresses is smaller than the yield
stress (that is, the initial stress state must lie within the initial yield
surface).
The initial stress option should not be used in large

displacement/large strain analysis.
11.7 Energy calculation
11.7.1 Energy printout
The Energy View utility enables users to monitor energy

variations of the whole model or specific parts of the model during
analysis; it also allows users to load and review the energy data
following analysis. The energy curves can be displayed graphically
in the Energy View dialog box and/or be saved to an energy (.egy)
file.
Currently, ADINA supports Energy View for 2D and 3D solid

elements.
Energy data are calculated and saved either at the level of whole
model or at the element group level, including contact group,
concentrated mass and concentrated damper. The ENERGY
command is used to specify which parts of the model should be
included in the energy data calculation. The ENERGY command
accepts four options: NO, WHOLE, EG-ALL and EG-LIST.
NO: no energy data are saved

WHOLE: energy data of the whole model are saved
EG-ALL: energies of all element groups are saved
EG-LIST: energies of selected element groups are saved
Note that except the option NO, the energy data for the whole
model is always saved. For the last two options, the energy data for
contact groups, concentrated masses, and concentrated dampers are
always saved.

The NODESAVE-STEPS command is used to specify the time

steps for which energy results are to be saved.
Energy calculation for element group
The following energy items are calculated for element groups:

Elastic strain energy EE
Inelastic (viscoplastic) dissipations EP
Kinetic energy EK
The equations used to calculate element strain energies are:
Linear formulation:
1
EE = ij ij dV (11.7-1)
V
2
MNO formulation:
1 t
EE =
2
( ij + t +t ij ) d ije dV (11.7-2)
V t
1 t
EP =
2
( ij + t +t ij ) d ijp dV (11.7-3)
V t
TL formulation:
for linear materials
1
EE = Sij ij dV (11.7-4)
V
2
for plastic materials
1 t
EE =
2
( Sij + t +t Sij ) d ije dV (11.7-5)
V t
1 t
EP =
2
( Sij + t +t Sij ) d ijp dV (11.7-6)
V t

11.7 Strain energy densities
ULH formulation:
t
1
EE =
2
( ij + t +t ij ) dEije dV (11.7-7)
V 0
t
1 t
EP =
2
( ij + t +t ij ) dEijp dV (11.7-8)
V 0
ULJ formulation:
t
1 t
EE =
2
( ij + t +t ij ) dDije dV (11.7-9)
V 0
t
1 t
EP =
2
( ij + t +t ij ) dDijp dV (11.7-10)
V 0
In the above equations:
ij : Cauchy stress
ij : Engineering strain
Sij : 2nd P-K stress
ij : Green strain
ij : Rotated Cauchy stress
Eij : Right Hencky strain
Dij : Jaumann strain
superscript e: elastic part
superscript p: plastic part
The element kinetic energy is

1 T
EK = U MU (11.7-11)
2
where
U : element velocity vector
M: element mass matrix
Energy calculations for contact groups
The following energy items are calculated for contact groups
Contact frictional energy EF

Contact compliance energy stored EC
Contact damping energy dissipated ECD
The equations for these quantities are:
1 t 1
t +t
EF = t EF +
2
( FC + t +t FC ) UTC + ( t FT + t +t FT ) UTT , (11.7-12)
2
1 t 1
t +t
EC = t EC +
2
( FC + t +t FC ) U CN + ( t FT + t +t FT ) UTN , (11.7-13)
2
and
1 t 1
t +t
ECD = t ECD +
2
( FDC + t +t FDC ) U C + ( t FDT + t +t FDT ) UT .
2
(11.7-14)
In the above equations:
FC: Contact force on contactor nodes
FT: Contact force distributed on target nodes
UTC : Incremental displacement of contactor nodes along

tangential direction

UTT : Incremental displacement of target nodes along

tangential direction`
U CN : Incremental displacement of contactor nodes along

normal direction
UTN : Incremental displacement of target nodes along normal

direction
FDC: Contact damping force on contactor nodes
FDT: Contact damping force on target nodes
UC: Incremental displacement of contactor nodes
UT: Incremental displacement of target nodes
Energy calculation for concentrated mass and concentrated

damper
The kinetic energy for concentrated masses is calculated using

equation 11.7-11, but with the velocity vector of the whole model
and mass matrix associated with the concentrated mass.
The Rayleigh damping energy of concentrated dampers is
( CU ) ( U t U)
t +t t +t t +t
ED = t ED + (11.7-15)
where
U : velocity vector of the whole model
U : displacement vector of the whole model
C : Rayleigh damping matrix asscociated with the concentrated

dampers
Energy calculation for whole model

The following energy items are calculated for the whole model:
0
Initial energy E
Work done by external force WEx
Elastic strain energy stored in the model EE
Inelastic dissipation EP
Kenetic energy of the current time EK
Rayleigh damping energy dissipated ED
Contact frictional energy dissipated EF
Contact compliance energy stored EC
Contact damping energy dissipated ECD
Energy balance EB
The initial energy 0E is an internal variable calculated in ADINA,

but it is never saved for Energy View.
The external work is
t +t 1
WEx = tWEx + ( t R + t +t R )( t +t U t U) , (11.7-16)
2
R: global loading vector
U: global displacement vector
Strain energy EE, dissipation EP, and contact energies EF, EC and
ECD for the whole model are calculated by summing the
contributions of each element group or contact group.
The kinetic energy and damping energy of the whole model are
calculated using equations 11.7-11 and 11.7-15, but with the mass
and damping matrices of the whole model.
The energy balance is calculated using the following equation,

reflecting the balance between the energies input to the system
(initial energy and external work) and energies stored in the system
or dissipated.
EB = ( E)
0
+ (W Ex ) ( E K + E E + E C ) ( E P + E D + E F + E CD )
Initial Energy Energy In Energy Stored Energy Dissipated

(11.7-17)
The energy balance should be close to zero. If greater than zero,

energy is being lost, whereas if smaller than zero, energy is gained.
A non-zero energy balance might indicate an inaccurate solution,
for example, due to spurious numerical energy dissipation from a
time step size that is too large. Alternatively, a non-zero energy
balance might be due to a source of energy dissipation not
supported by the ENERGY command (e.g. energy dissipation due to
the Mullins effect, viscoelasticity, creep etc.).
Material models supporting Energy View
Currently, ADINA supports energy calculations for those material

models listed in the Table 11.7-1.
Table 11.7-1 Material models that support the Energy View feature.
Material Models elastic strain energy inelastic dissipation explicit

Linear elastic isotropic 9 N/A 9
Linear elastic orthotropic 9 N/A 9
(1) 9 9 9
Plastic bilinear
(1) 9 9 9
Plastic multilinear
(1) 9 9 9
Plastic orthotropic
Plastic Mroz Bilinear 9 9 N/A
Plastic Gurson 9 9 N/A
Thermo elastic isotropic(2) 9 9 9
(2) 9 9 9
Thermo elastic orthotropic
Thermo plastic(2) 9 9 9
Elastic-plastic Creep 9 9 9
Rubber/Foam 9 X 9
Viscoelastic 9 X X
(3) 9
Nonlinear elastic N/A X
Druck Prager(3) 9 9 X
Mohr Coulomb(3) 9 9 X
(3,4) 9
Plastic cyclic X X
Notes:

1. These material models currently do not support the energy

calculation for mixed-formulation in explicit analysis.
2. For these material models, the energy input by temperature

loading is not accounted for in the energy balance calculation.
3. These material models currently do not support the energy

calculation in explicit analysis.
4. The plastic cyclic material currently only supports the total

strain energy calculation.
11.7.2 Strain energy densities
The computation of strain energy densities at integration points

can be requested for 2-D solid and 3-D solid elements.
The strain energy density can be requested for all material

models, and all nonlinear formulations. The mixed (u/p)
formulation can be used.
The strain energy density cannot be requested in explicit time

integration analysis.
The strain energy density calculations used in this section are

also used for fracture mechanics (Chapter 10).
The strain energy density for rubber-like materials is computed

directly from the definitions given in Section 3.8.
The remainder of this section describes the calculation of the

strain energy density for the other material models.
Small displacements / small strains (linear and MNO formulations)
The basic formula for the strain energy density under small
displacement and small strain conditions is
dW = ij deij (11.7-18)

where ij is the Cauchy stress tensor and eij is the small strain
tensor.
When there are thermal strains, the following formula is used:
dW = ij deijm (11.7-19)
in which eijm = eij eijth are the mechanical strains.
Large displacements / small strains (TL formulation)
The basic formula for the strain energy density under large
displacement and small strain conditions is
d 0tW = 0t Sij d 0t ij (11.7-20)
where the left subscript means referred to the original

configuration, or per unit original volume, 0t Sij are the 2nd Piola-
Kirchhoff stresses and 0t ij are the Green-Lagrange strains. When
there are thermal effects, the thermal strains are subtracted from
0 ij , just as in the small displacement / small strain case.
t
Large displacements / large strains (ULH formulation)
displacement and large strain conditions, using the ULH
formulation, is
d 0tW = t ij d 0t EijR (11.7-21)
where t ij is the rotated Cauchy stress and 0t EijR is the right

Hencky strain. Both of these quantities are discussed in Section
3.1.2. Notice that although the rotated Jaumann stress should
theoretically be used, the rotated Cauchy stress is used instead,
since the plastic strains are incompressible. When there are thermal
effects, the thermal strains are subtracted from 0t EijR , just as in the

small displacement / small strain case.
Large displacements / large strains (ULJ formulation)
displacement and large strain conditions, using the ULJ
formulation, is
d 0tW = t ij Dij (11.7-22)
where Dij is the rate of deformation tensor. As in the ULH case,

the Jaumann stress should theoretically be used, but the Cauchy
stress is used instead. When there are thermal effects, the rate of
change of thermal strains is subtracted from Dij .
Numerical evaluation
The basic formula dW = ij deij is numerically integrated using

the trapezoidal rule, e.g.
1
W = tW + ( t ij + t +t ij )( t +t eij t eij )
t +t
(11.7-23)
2
in which 0W = 0 is assumed. Similar expressions are used for the

large displacement formulations.
The accuracy of the strain energy density depends on the time
step size, when there are nonlinearities in the element group.
11.8 Element group inertial properties

The following element group properties can be requested in
ADINA for printout/saving: total mass, total volume, moments of
inertia, products of inertia, centroid, center of mass.
General elements and spring elements do not contribute to the

element group inertial properties.

11.8 Element group inertial properties
The element group inertial properties are computed at the

beginning of the analysis based on the undeformed model.
In ADINA, the moments and products of inertia are calculated

with respect to the origin of the global coordinate system.
For axisymmetric element groups, the volume, mass and other

properties are calculated using a thickness of 2 radians.
Exception: the masses in the three coordinate directions are
computed using a thickness of 1 radian; this facilitates comparision
with the modal masses.
The AUI can combine the element group quantities over the
whole model or over a selected zone in the model. The AUI also
includes the concentrated masses in the calculations. The moments
and products of inertia are calculated and printed with respect to
the center of mass of the whole model or zone.
When the concentrated masses have different values in different

coordinate directions, for example, when the concentrated mass is
the x direction has a different value than the concentrated mass in
the y direction, then the AUI uses the following procedure: for each
such concentrated mass, the AUI computes an average mass as
the average of the nonzero masses for the different directions. The
AUI then computes the total mass as the sum of the element group
masses plus the sum of the average concentrated masses. In the
listing of mass properties, the AUI lists the total mass, and also the
sum of the mass values in the different directions (without
averaging). For the center of mass and the inertial properties, the
AUI uses the total mass (with averaging) throughout.
When listing the percent modal masses (MPFINFO command),

the AUI uses the total mass values in the different directions
(without averaging the concentrated masses in the different
directions).
It is not possible for the AUI to compute the above-mentioned

properties for a selected part of an element group.

11.9 Restart
Restart is a useful feature in the ADINA System. It can be used
when the user wishes to continue an analysis beyond its previous
end point, or change the analysis type, loads or boundary
conditions or tolerances. A restart analysis is set via the MASTER
command. Recovery of results from a restart file without
continuing the analysis is also possible.
All relevant solution data needed for a restart run are saved in a
file (extension .res) in case they are needed in a restart analysis.
The frequency of data writing to a restart file is set via the
MASTER command.
Note that multiple restart data can be appended to the restart

file. This enables the restart analysis to be based on a solution step
different from the last converged solution. Saving multiple time
step solutions to a restart file can be expensive, however, as it leads
to a large restart file size.
To perform a restart analysis, the .res file from the first model
must be copied to the second model. When running in interactive
mode, the AUI will first look for the restart file, and if it is not
found, the user will be prompted to locate the restart file. When
running in batch mode the .res file must be copied before starting
the second analysis.
If no restart time is provided in the second model (achieved by

setting the restart from 0.0), the program uses the data for the latest
restart time on the .res file.
Note that once the second analysis starts, it will overwrite the
.res file with new data. Therefore, if the user wishes to redo the
second run, then the .res file must be copied again from the first
model. This does not apply to when restart data is appended to the
.res file and when the second run is a frequency analysis (no restart
data in this case).
The geometry, and most element data cannot be changed in a

restart analysis. However, the following changes are allowed:
Type of analysis can change. Static to dynamic and dynamic

11.9 Restart option
to static restarts are allowed.

Solution type can change. Implicit analysis to explicit
analysis restarts are allowed and vice-versa. In this case,
features not available in either solution type cannot be used.
Note that some default settings are different between implicit
and explicit analysis. Some of these have to be manually set by
the user to enable restarts. The most common setting is:
- Default large strain formulation (ULH in implicit and ULJ

in explicit)
Solution control variables can change. The flags, constants
and tolerances for the iteration method, convergence, time
integrations, automatic time stepping and load-displacement-
control can be changed.
Externally applied loads and enforced displacements can be
modified.
The material constants can be changed. However, note that
in a restart run the same material model (with the same number
of stress-strain points and the same number of temperature
points, if applicable) must be used for each element as in the
preceding run.
Boundary conditions can change.
Constraint equations and rigid elements can change.
Contact settings can change. This includes most contact set,
contact pair and contact surface parameters. See Section 4.7.5
for restrictions.
Rayleigh damping coefficients can change.
Time increment and number of solution steps can be
modified.
Time functions describing the load variations can be
changed.
It is not allowed to have no birth-death in the first run, then
birth-death in the restart run. If there is no birth-death in the first
run, give a large value for TDEATH for the element groups that
have birth-death in the restart run.

When restarting from static to dynamic analysis (both implicit

and explicit dynamics), the initial velocities and accelerations are
assumed to be zero. However, if an initial velocity is prescribed in
the restart run, it will be used instead. When restarting from one
dynamic analysis to another, initial velocities and accelerations are
transferred from the first to the second run.
When changing from implicit to explicit analysis (or from

explicit to implicit analysis), be aware that the AUI defaults depend
on the analysis type. For example, by default the AUI selects the
ULH formulation in implicit analysis and selects the ULJ
formulation in explicit analysis.
Result recovery
In result recovery mode, ADINA reads a restart file and generates

results for the last saved restart step. No analysis is performed. If
multiple restarts are saved in the restart file, results can be
recovered at any of the saved times by setting the restart time.
Restart in frequency analysis
For large systems, if the subspace iteration method for

frequency analysis does not converge, it can be efficient to use the
converged eigenvectors as a subset of the starting vectors and redo
the analysis with modified input. The advantage is that the amount
of computational effort needed to compute the remaining
eigenvectors is less than the amount of computational effort needed
to compute all of the requested eigenvectors.
In the following discussion, the first ADINA run is the one in

which not all of the requested eigenvectors are obtained and the
second ADINA run is the one in which the remaining requested
eigenvectors are obtained. Note that the second ADINA run is not
considered to be a restart run.
You supply the converged eigenvectors from the first ADINA
run to the second ADINA run as user-supplied starting vectors.
You also generate a new data file for the second ADINA run, in
which you can change the number of starting vectors and the
maximum number of iterations. It is recommended that the number
of starting vectors be greater than or equal to the number of

11.9 Restart option
converged vectors from the first ADINA run.

Before starting the second ADINA run, you need to copy the
eigenvectors from the first run to the second run. The eigenvectors
are stored in a file with extension .stv.
Restart can also be used to switch from one analysis type to

another. The following table provides details of the possible
analysis type switches.
Table 11.9-1: Switching between analysis options with restart
To Implicit Explicit Frequency Modal

Statics
From Dynamics Dynamics Analysis Analysis
Statics
9 9 9 9 x(1)
Implicit
9 9 9 9 x(1)
Dynamics
Explicit
9 9 9 9 x(1)
Dynamics
Frequency
9 9 9 x 9(2)
Analysis
Modal
9 9 9 x x
Analysis
Notes:
(1) Must restart to frequency analysis first
(2) Can also be in the same run (no restart needed)

11.10 Parallel processing

ADINA supports shared memory parallel processing (SMP) and
distributed memory parallel processing (DMP).
For shared memory parallel processing the following coding is

parallelized: assembly of the global system matrices (loop over
element groups), the in-core sparse solver and the out-of-core
sparse solver. The parallelization of the assembly of the global
system matrices is available for all platforms except for 32-bit
Windows.
To benefit from parallel element assembly, element groups must

be divided into subgroups. The number of subgroups should be
equal or greater to the number of processors (preferably a multiple
of the number of processors). This is set via the NSUBG and
MINEL parameters in the EGCONTROL command.
For distributed memory parallel processing there are currently

two options:
For static and implicit dynamic solutions, the sparse solver is

parallelized using MPI directives, and matrix assembly using
shared memory OpenMP directives. Therefore we have a
combined DMP-SMP algorithm to run ADINA effectively.
For explicit time integration, the program divides models

into subdomains and uses MPI directives to obtain solutions.
Shared memory OpenMP directives are also used in local
subdomains.

11.12 Miscellaneoous
11.11 Usage of memory and disk storage

Depending on the size of the problem and the memory allocated
to ADINA (MTOT), ADINA can perform the solution either in-
core (entirely within real or virtual memory) or out-of-core
(reading and writing disk files). Whenever possible the solution is
performed in-core.
Explicit dynamic analysis can only run in-core. Enough

memory must be available.
Depending on the size of the problem and the available

memory, there are 2 internal in-core/out-of-core settings for the
program without the solver. These 2 memory settings are:
The stiffness (and mass) matrices and the element

information are all stored in-core. (IOPTIM=3)
The stiffness (and mass) matrices are stored in-core, while
the element information is kept out-of-core. (IOPTIM=2)
The memory setting used by the program is provided in the .out

file as parameter variable IOPTIM.
The sparse solver and the 3D-iterative solver may both be set to
in-core or out-of-core.
When using the iterative multi-grid solver (Section 6.5.2), the

solution is always performed in-core. The out-of-core solution
procedure would take an unreasonably long time in most cases.
The disk files that may become large are those with the
following file name extensions:
1: Linear element group data

2: Nonlinear element group data
4: Linear stiffness matrix
7: Implicit dynamic analysis, effective stiffness matrix
8: Restart file
10: Nonlinear stiffness matrix (skyline solver only)
11: Mass/damping matrices

11.12 Miscellaneous
The program can write intermediate files to disk for external
processing, and can read various inputs (such as stiffness and mass
matrices) directly from a user-provided file (such as pipe pressures,
forces, or displacements).
Saving stiffness and mass/damping matrices using disk files
The DISK-STORAGE command can be used to write matrices on

the *.mtx disk file:
DISK-STORAGE GLOBAL-MATRIX=NONE/SAVE/NODE
/FULL STIFFNESS-MATRIX=NONE/SAVE MASS-MATRIX
=NONE/SAVE DAMPING-MATRIX=NONE/SAVE
For a detailed description of the DISK-STORAGE command,
see the ADINA Command Reference Manual.
A non-factorized, linear or nonlinear, global stiffness matrix can be

saved using a disk file (*.mtx file), for a given time step. For
implicit dynamic solutions a mass matrix (lumped or consistent)
and Rayleigh damping matrix can also be stored on the same *.mtx
file if requested. There are three available options:
Option 1. Command input: GLOBAL-MATRIX=FULL

Matrices are given in full upper triangular by column heights form.
The program will then write stiffness, mass and damping matrices
for the last solution time only. The *.mtx file will contain the
following data:
- MAXA array for full by column heights stiffness/mass/damping
matrices
- Full by column heights stiffness matrix followed by mass and
damping matrices (if required)
Option 2. Command input: GLOBAL-MATRIX=SAVE

The matrices are given in a sparse form, i.e. zero matrix elements
in the columns are removed and can be written to the *.mtx file for
any time step. The *.mtx file always contains the following data:
- MAXA array, for now column heights vector with the zero
elements in the columns removed (from the matrix diagonal for
symmetric matrices, from the matrix bottom for non-symmetric

11.12 Miscellaneoous
matrices)
- Equation numbers for the nonzero matrix elements
- Stiffness matrix followed by mass and damping matrices (if
requested) according to the MAXA array used here
Option 3. Command input: GLOBAL-MATRIX=NODE

The matrices (i.e. stiffness, mass and damping) are given in a
sparse form and can be written out to the *.mtx file for any solution
time step. The matrices are written for every node in a direct
matrix input format required by the AUI. The matrices can then be
used in other models as input data for the direct matrix input
options.
Saving heat conductivity and heat capacity matrices
Heat conductivity and heat capacity matrices can also be stored on

the *.mtx file (see command DISK-STORAGE for details).
However, only Option 1 and 2 are available for heat conductivity
and heat capacity matrices.
In certain cases, you can specify data to ADINA using disk

files:
Pipe internal pressures: Pipe internal pressures can be read from

unit 24 in the following format (unformatted sequential Fortran
file):
READ (24) TIME,(PIPEIP(I),I=1,NUMNP)
Here NUMNP is the number of nodes in the main structure and

TIME is the solution time for the pipe internal pressures.
You must specify whether the times in unit 24 correspond
exactly with the ADINA solution times, or whether they are
different than the ADINA solution times. In the latter case,
ADINA interpolates the pipe internal pressures to the ADINA
solution times and writes the interpolated data to unit 24.
Temperatures: Temperatures can be read from unit 56 in a manner

entirely analogous to the reading of pipe internal pressures.
Temperatures read from unit 56 are added to the other specified
temperatures (initial or prescribed), except at the beginning of a
restart run.

Temperature gradients: Temperature gradients can be read from

unit 57 in a manner entirely analogous to the reading of
temperatures. Temperature gradients read from unit 57 are added
to the other specified temperature gradients (initial or prescribed),
except at the beginning of a restart run.
Forces and displacements: Forces can be read from unit 58 in the

following format (unformatted sequential Fortran file):
READ (58) TIME,((F(I,J),I=1,NUMNP),J=1,NOFC)
Here NOFC is the number of force components. You must specify
which force component corresponds to each value of J.
You must specify whether the times in unit 58 correspond
exactly with the ADINA solution times, or whether they are
different than the ADINA solution times. In the latter case,
ADINA interpolates the forces to the ADINA solution times and
writes the interpolated data to unit 58.
The forces read from unit 58 are added to the forces from the
other specified loadings.
At the beginning of the solution analysis (or at the beginning of
the restart run), the total nodal point forces are always defined by
the displacement, velocity and accelerations at the beginning of the
solution analysis (or at the beginning of the restart run), and the
nodal point forces read from unit 58 are not used.
Displacements can be written to unit 58 in the format given
above.
11.13 Remeshing options

In some cases (e.g., rolling, forging), a finite element mesh
might become distorted to such an extent that the solution stops. A
restart analysis cannot be used to continue the solution; usually a
new mesh has to be created in order to solve the problem. This
section describes the remeshing options available in ADINA.
The mapping option allows you to map the last obtained

solution from the current mesh onto a new mesh. The mapping
option is applicable to 2-D solid elements in large strain analysis
only, when the plastic-bilinear or plastic-multilinear material
models are used.
When you anticipate that you will use the mapping option, you

11.13 Remeshing options
must create a mapping file (extension .map) in the first run. In the
AUI user interfaces, choose ControlMapping and select Create
Mapping File or from the AUI command line, use the command
MASTER MAP-OUTPUT=YES. You must also save the AUI
database file.
To create a new mesh on the deformed geometry and continue

the analysis, follow these steps:
Open the AUI database file.
In the AUI user interfaces, choose ControlMapping and

select Remesh and Continue Analysis, or from the AUI
command line, use the command MASTER MAP-OUTPUT =
REMESH.
Read the nodal deformations into the AUI. The nodal

deformations are stored in a file with extension .mds). To read
the nodal deformations in the AUI user interfaces, specify the
nodal deformations file in the ControlMapping dialog box, or
from the AUI command line, use the command MASTER
NODAL-DEFORMATION-FILE = (filename).
When the AUI reads the nodal deformations, it creates new

geometry lines corresponding to the boundaries of the deformed
body. Note, in this AUI version, the AUI does not create 3-D
deformed surfaces.
Now use the dialog boxes and/or commands of the AUI to

create a new mesh. Set the solution start time to the last
obtained solution time. You must use the same element types,
material models and contact conditions as you used in the
previous mesh. However, you can modify the loading, time
stepping and boundary conditions.
Create a new data file and run ADINA. You do not have to
provide the restart file from the previous run to the current
ADINA run, but you must provide the mapping file from the
previous run to the current ADINA run.

11.14 Analysis zooming

The analysis zooming feature is used to analyze details of
models starting from a relatively coarse mesh and subsequently
zooming in on specific regions of interest. The feature is available
in the ADINA program only.
The initial (coarse model) analysis requires the user to specify

that the analysis zooming feature might be used in subsequent runs.
All other options are the same as usual. The ADINA program
creates a mapping file which can be used in subsequent layered
models.
The analysis zooming feature is based on St. Venants principle

that if an actual distribution of forces is replaced by a statically
equivalent system, the distribution of stresses and strains is altered
only near the regions of load applications.
The advantages of this zooming method are
the need for transition regions is eliminated
a user can change the zoomed region and still use the full
model initial solutions for every zooming calculation, thereby
enabling the user to experiment with different designs of the
region of interest
the method can help to demonstrate the adequacy of mesh

refinements.
When using the analysis zooming feature, note that
A user can zoom on any part of the model. Zooming can be

repeated on the already zoomed parts. For every layer, a new
mapping file is created, which can be used in further zooming
on the model.
There can be more than one zoomed part.
Boundary conditions, material models and loading are

derived from the initial model, but they can be modified.

11.15 Stiffness stabilization
The .dat file for the zoomed part must have a different name
from the full model .dat file.
Only 2-D solid and 3-D solid elements can be used in the
zoom modeling.
ADINA performs a step-by-step analysis. The .map file from

the full model solution must be used in the analysis of the
zoomed model. Although this is not a restart analysis, the .map
file should be regarded as the restart file as in a standard restart
run.
For every time step, displacements from the full model

solutions are mapped to the boundary nodes of the zoomed
model. This process should not take a long time since only
boundary nodes are mapped.
The mapped displacements are treated in ADINA as

prescribed displacements in the zoomed model.
The user should be able to verify that the distance between

the cut boundaries and the stress concentration region is
adequate. This can be done by using the AUI (ADINA-PLOT).
The AUI can overlay stress contour plots of the zoomed model
on the same stress contour plots of the whole (coarse) model. In
other words, the AUI can display the zoomed model on the
(coarse) full model for all plotting options, namely, band plots
and vector plots.
The cut boundaries are assumed to be far enough away from

stress concentration regions.

During the solution of equations in static analysis, zero pivots
may arise, for example in the following cases:
Unsupported body: If the forces acting on the body are not in

equilibrium, one or more rigid body motions of the body are

activated and no solution can be expected. Even if the forces

acting on the body are in equilibrium, so that no rigid body motion
is in fact activated, zero pivots are present corresponding to the
rigid body modes.
Contact analysis, in which one or more of the individual parts of

the model (not considering contact) contain rigid body motions.
When the parts are not in contact, then there is nothing to prevent
the rigid body motions. (This includes tied contact.)
Mesh glueing, when one or more of the individual parts of the

model (not considering glueing) contain rigid body motions.
General constraints, when one or more of the individual parts of

the model (not considering the general constraints) contain rigid
body motions.
These zero pivots will stop the solution, unless the zero pivots are
prevented from occuring.
Stiffness stabilization is used to prevent the equation solver

from encountering zero pivots. Stiffness stabilization is available
for static analysis, with or without low-speed dynamics. (In
dynamic analysis, the zero pivots are not present due to the mass
matrix, so stiffness stabilization is not available for dynamic
analysis.)
Parts of the model with rigid body motions can alternatively be

treated by adding weak springs at various locations in the model.
The advantages of using stiffness stabilization, instead of using
weak springs, are:
Determining the number, location and stiffness of the springs

requires a lot of user intervention.
There may be no suitable locations for the springs.
The stiffness of each spring has to be entered as an absolute

value (with dimensions of force/length) while the stiffness
stabilization factor is dimensionless (see below).

The springs generate internal forces which affect the final

solution while stiffness stabilization does not affect the internal
forces. It is sometimes hard to assess how much the springs
affect the final solution.
Stiffness stabilization modifies the diagonal stiffness terms

(except for those belonging to contact equations) as follows:
K ii = (1 + STAB ) K ii
where STAB is a dimensionless stabilization factor. The right-hand
side load vector is not modified. If a diagonal stiffness term is zero
before factorization, the term remains zero.
There are three stiffness stabilization options available, which

are selected using the STABILIZE and STABFACTOR parameters
of the MASTER command:
STABILIZE=NO (no stabilization)

STABILIZE=YES (stabilization, with STAB = STABFACTOR)
STABILIZE=AUTOMATIC (stabilization is activated if
needed)
The defaults are STABILIZE=NO, STABFACTOR=1E-10.
When STABILIZE=AUTOMATIC, the use of stabilization is

determined based on the ratio of the factorized maximum and
minimum diagonals of the stiffness matrix. This determination is
made for every equilibrium iteration in nonlinear analysis. When
stabilization is used, stabilization is applied to all degrees of
freedom using the value STAB (as if STABILIZE=YES).
STABILIZE=AUTOMATIC cannot be used in the DMP
version.
In linear analysis, stabilization should be used with caution,

since the right-hand-side load vector is not modified. The solution
can therefore be affected by stiffness stabilization.
It is recommended to try the analysis first without stabilization.
If the equation solver encounters zero pivots, then try one of the
following methods:

Use stabilization with the smallest possible value of STAB

for which the equation solver gives a solution, or
Use stabilization, and change the analysis to a nonlinear

analysis, for example specify element birth-death in one of the
element groups; or specify a nonlinear material, with material
constants chosen so that the material response is linear (for
example, an elastic-plastic material with a very high yield
stress).
In nonlinear analysis, since the right-hand-side load vector is

not modified, the final converged solution is the same as without
stabilization (assuming that the tolerances are tight enough).
However, the rate of convergence can be worsened due to the
stiffness stabilization, so that more equilibrium iterations are
required.
Stiffness stabilization is only useful for the sparse and 3D-

iterative solvers. The iterative multigrid solver does not fully
factorize the stiffness matrix and hence cannot properly trigger the
automatic stabilization.
Stiffness stabilization is not available in FSI analysis with direct

coupling.
11.16 Geometric imperfections

ADINA supports adding geometric imperfections to a structure
prior to analysis. The initial coordinates of the nodes are modified
based on the imperfection, and no strains (nor stresses) are
generated.
The imperfections can be obtained from a mode shape generated

from a linearized buckling analysis, or a frequency analysis, or
from nodal displacements generated from a static or dynamic
analysis.
Initial geometric imperfections can be useful for transient

dynamic post-buckling type problems where geometric
perturbations help trigger the buckling modes. The initial geometric
imperfection in such cases should ideally be generated from a

linearized buckling analysis.
Initial geometric imperfections can also be useful in obtaining

an approximate stress-free configuration for flexible structures such
as tents. In this case, a static analysis can be performed with gravity
loading, and the displacements of that analysis used as initial
geometric imperfections for subsequent analyses.
11.17 Bolt feature

Overview
The bolt option is a modeling feature which uses element

groups with specific bolt-type loadings to simulate different bolt
handling procedures. The supported bolt-type loadings are:
1. bolt tensioning, i.e., the axial force in the bolt is specified.
2. bolt shrinkage, i.e., the bolt shortening (reduction in bolt

length) is specified.
In both of these cases, the deformations of the rest of the structure

are considered. For example, if a single bolt is loaded with a user-
specified bolt tension, then, after the bolt loading is applied, the
axial force of the bolt has the given value, and the axial force acts
upon the rest of the structure, causing structural deformations.
Different modeling techniques with varying complexity can be

used to model a bolt. Three such techniques are shown in Fig.
11.17-1, in increasing level of complexity.
Any number of bolts can be included in the model. The bolts

can all be loaded simultaneously, or the bolts can be loaded
sequentially.
Bolt type loadings are applied during special 'bolt loading"

iterations. During the bolt loading iterations, time is frozen and the
bolt parameters are automatically iteratively adjusted as described
below. After the bolt loading iterations have converged, then the
time steps proceed as usual.

Constraints Shell elements or rigid surface
One or more beam-bolt elements One or more beam-bolt elements
No contact needed Contact at top and bottom
(a) Bolt modeled with beam-bolt elements (b) Beam-bolt elements used for bolt shank
Solid elements in 3D-bolt element group
Contact all around if needed
(c) Bolt modeled with 3D-bolt element group
Figure 11.17-1: Different bolt modeling techniques
The bolt loading can be performed:

- at the beginning of the analysis, i.e. at time zero.
- at the beginning of a restart analysis
- at any given solution time.

11.17 Bolt feature
A bolt iteration consists of the same process as is normally used

for static analysis, namely the solution of the equilibrium equations
as discussed in Sections 7.1 and 7.2. In each iteration all elements,
including bolt elements, contribute to the global force vector, and
the increment in displacements, including the displacements of the
bolt element nodes, is obtained.
The remainder of the structure (that is all of the rest of the
model, including bolts not being loaded) behaves exactly as usual.
For example, plasticity can occur. Therefore, it is possible that the
bolt iterations might not converge. In this case, the bolt loading can
be divided into sub-steps, so that the change in bolt loads is smaller
during each sub-step.
11.17.1 Beam-bolt element
The beam-bolt element is implemented as an elastic Hermitian

beam element with a modified axial force - length relationship. The
formulation used for the beam-bolt elements can either be small
displacements or large displacements. Any cross-section available
for the beam element can be used. However, the material model
must be linear elastic.
The axial force - length relationship of the beam-bolt element is
F = k ( t L Lb )+ Fb (11.17-1)
where F is the bolt axial force, k is the bolt axial stiffness

EA / Lb , t L is the bolt current length, Lb is the bolt adjusted
length (defined below) and Fb is the bolt force corresponding to
Lb . (Thermal effects are ignored in this discussion.) The bolt
adjusted length is defined as Lb = 0 L b , where 0 L is the bolt
original length and b is the bolt shortening.
Using t L = 0 L + t where 0 L is the bolt initial length and t
is the bolt axial displacement gives
F = k ( t + b )+ Fb (11.17-2)

Initially (before the first bolt-loading is applied), Fb and b are

zero, which means that F = k t and the element behaves exactly
like a usual beam element.
We now discuss the details of the bolt iterations for beam-bolt

elements. For simplicity, in this discussion, we show only a single
bolt element, and we assume that the bolt load is applied at time 0,
that is, before any other deformations have occurred.
The process of bolt tensioning is shown in Figure 11.17-2.

During the bolt iterations, the axial stiffness of the bolt element is
set very small (to f1 k , where f1 << 1 ), and the axial force in the
bolt element is replaced by the user-specified axial force Fb . The
bolt shortening b is determined from the condition b = t in
which t is determined from the displacements of the bolt element
nodes. The bolt iterations continue until equilibrium is satisfied and
also the condition
( ite ) ( ite 1)
b b
< tol
(
max b
( ite 1)
, b
( ite )
)
is satisfied. At this point, the bolt axial stiffness is restored to its
normal value. Since t + b = 0 , the bolt axial force F computed
from (11.17-2) equals the user-specified axial force Fb .

11.17 Bolt feature
Adjusted bolt length = Lb

Bolt element
Specified bolt force Fb
Rest of structure
(a) Before bolt iterations (b) During bolt iterations
Current bolt length = Lb
Axial bolt force

= specified Fb
(c) After bolt iterations
Figure 11.17-2: Bolt force loading
The process of bolt shrinkage is shown in Figure 11.17-3.

During the bolt iterations, the axial stiffness of the bolt element is
set very large (to f 2 k , where f 2 >> 1 ) and the axial length of the
bolt element is set to Lb = 0 L b , where b is user-specified.
Since the bolt still has some flexibility, the bolt contributes axial
force to the nodes using F = f 2 k ( t + b ) , where, since the axial
stiffness is very large, t + b 0 . The bolt iterations continue
until equilibrium is satisfied and also the condition
Fb( ite ) Fb(ite 1)

< tol
(
max Fb(ite 1) , Fb(ite ) )
is satisfied. At this point, the bolt axial stiffness is restored to its
usual value and the bolt axial force Fb is set to F . Again, for

t
+ b = 0 , the bolt axial force F computed from (11.17-1)
equals the axial force Fb .
Specified bolt length = Lb = 0L - b
Bolt element
Rest of structure
(a) Before bolt iterations (b) During bolt iterations
Current bolt length = tL

< user-specified Lb
Axial bolt force

= Fb = F
(c) After bolt iterations
Figure 11.17-3: Bolt shrinkage loading
For all other iterations, in which the bolt is not being adjusted,
the bolt axial force - length relationship is given by (11.17-1),
where Fb and b are in general non-zero and are held constant.
In the beam-bolt element, no additional global equations are

used to include the additional bolt variables b and Fb .
The bolt tolerance in the above equations is set by the BOLT-

OPTIONS TOLERANCES parameter, or by the EGROUP BEAM
BOLT-TOL parameter. The default value is 0.01.

11.17 Bolt feature
The values of f1 and f 2 in the above equations is given by

BOLT-OPTIONS FACTOR1 and BOLT-OPTIONS FACTOR2
respectively. The default values are FACTOR1=1E-6,
FACTOR2=1E6.
11.17.2 3D-bolt element group
An entire 3D element group can be designated as a 3D-bolt

element group. The elements in this group can use any formulation
(small or large displacements, small or large strains), any material
model and any number of nodes per element. The mixed u/p
formulation or the incompatible modes formulation can also be
used.
The geometry of the 3D-bolt element group is also arbitrary.
The element sides and faces can be straight or curved.
Each bolt is modeled as a separate 3D-bolt element group.

Multiple bolts cannot be modeled within a single 3D-bolt group.
The basic ideas used in the 3D-bolt are shown in Fig 11.17-4.
This figure shows a very simple bolt model of four elements, with
the ends of the bolt fixed. For simplicity, it is assumed that the
Poisson's ratio is zero, so that the cross-section of the bolt doesn't
change during the deformations.
Fig. 11.17-4(a) shows the 3D-bolt group along with its bolt plane.
This bolt plane is an additional input for the 3D-bolt group. As
seen, the bolt plane determines the bolt direction and the bolt split
faces. (For more information about the bolt plane, see Section
11.17.5.)
Fig 11.17-4(b) shows an incompatible displacement, called the
bolt displacement, applied to the nodes on the bolt split faces.
Because the bolt model is fixed at its ends, the elements must
increase in length, and, to satisfy equilibrium, the elements must all
be under uniform tension. The nodal point forces acting at the
nodes on the bolt split faces are combined into a single bolt force.
Fig 11.17-4(c) shows the bolt element group after deformations,
as visualized during postprocessing. The elements appear to be
different lengths because the bolt displacement is not included in
the plot. However, the strains are the same in all elements.
In this example, it is clear that the bolt force is determined by

the bolt displacement, and that the bolt force increases as the bolt
displacement increases.
Bolt direction
(normal to
bolt plane) Bolt split faces
Bolt plane
(a) 3D-bolt group with bolt plane
Bolt
displacement
Consistent
nodal point
forces
(b) Bolt displacement and consistent nodal point forces
(c) Bolt element group, as visualized
Figure 11.17-4: 3D-bolt element group example
Fig 11.17-5 shows the same problem, but with one end of the
bolt free. Now when the bolt displacement is applied, the elements

11.17 Bolt feature
on the free end move stress-free, and the resulting plot is shown in
Fig 11.17-5(c). This example justifies the use of the term "bolt
displacement".
When the bolt end is free, it is clear that the bolt force is zero
for any value of the bolt displacement.
(a) Before deformations
Bolt
displacement
(b) After deformations
(c) As visualized
Figure 11.17-5: 3D-bolt element group example, free end
In the general case, a 3D-bolt group has two additional

quantities: the bolt displacement (denoted vb ) and the bolt force
(denoted Fb ). These quantities are used in the global solution of
equations as follows:

1) Bolt force specified:
K KUB u R F
KBU KBB v = R F (11.17-3)
b b b
where K is the usual global stiffness matrix, R is the usual

external loads vector, F is the usual internal loads vector, u is
the usual increment in displacements, KBU, KUB, KBB are
coupling matrices, and Rb is the specified bolt force. During the
equilibrium iterations, both u and vb are obtained, and at
equilibrium, Fb = Rb . Thus the value of vb that results in the
specified Rb is obtained by this procedure.
In addition to the usual convergence criteria, the convergence
criterion
Rb Fb
< tol
(
max Rb , Fb )
is used.
2) Bolt displacement specified:
K KUB u R F
=
KBU KBB v v presc v
(11.17-4)
b ( b b )

where vbpresc is the specified bolt displacement and is very large.

For convergence, vb must be equal to vbpresc . This procedure is
analogous to the procedure in which displacements are prescribed.
3) Usual equilibrium iterations in which the bolt is not being

adjusted:

11.17 Bolt feature
K KUB u R F
KBU KBB v = 0 (11.17-5)
b
The last row of this system of equation results in vb = 0 , so the

global equilibrium equation reduces to the usual case.
In addition to the usual convergence criteria, the convergence
criterion
vbpresc vb
< tol
(
max vbpresc , vb )
is used.
The bolt tolerance in the above equations is set by the BOLT-

OPTIONS TOLERANCES parameter, or by the EGROUP
THREEDSOLID BOLT-TOL parameter. The default value is
0.01.
When there are several 3D-bolts, then each bolt contributes an

additional equation to the global system of equations. There is no
limit to the number of 3D-bolts that can be present in the model.
11.17.3 Usage of bolt loadings
The BOLT-OPTIONS command is used to control general bolt

options, including the use of bolt tables. There are two basic
options:
BOLT-OPTIONS TABLE=NO
Then the type of bolt loading is specified by BOLT-OPTIONS

TYPE (either FORCE or LENGTH) and bolt loading occurs at
solution time BOLT-OPTIONS TIME. The number of bolt loading
steps is specified by BOLT-OPTIONS STEPS. All bolts in the
model are loaded simultaneously.
For example

BOLT-OPTIONS TABLE=NO TYPE=FORCE TIME=2.0

STEPS=2
At time 2.0, all bolts are loaded using force loading. In the first
bolt-loading step, half of the bolt load is applied, and the second
bolt-loading step, the entire bolt loading is applied.
BOLT-OPTIONS TABLE=YES
Then the BOLT-TABLE commands are used to specify the type of

bolt loading, the solution time for the bolt loading, and also the
sequence in which the bolts are loaded. For example
BOLT-OPTIONS TABLE=YES
BOLT-TABLE 1 TYPE=FORCE TIME=2.0
1 2 0.5
1 3 0.2
2 4 0.3
3 2 1.0
3 3 1.0
3 4 1.0
At time 2.0, bolts 2, 3 and 4 are loaded using force loading. In the
first bolt-loading step, bolt 2 is loaded by 0.5 times its specified
bolt force, bolt 3 is loaded by 0.2 times its specified bolt force and
bolt 4 is not adjusted. In the second bolt loading step, bolts 2 and 3
are not adjusted, and bolt 4 is loaded by 0.3 times its specified bolt
force. In the third bolt loading step, all three bolts are loaded by
their specified bolt forces.
The results at any bolt step can be saved to file. If bolt tables are
used, the saved bolt steps all have the same solution time, and are
distinguished by their bolt sequence numbers.
Bolt loading can change from bolt tensioning to bolt shrinkage

at different stages of the analysis. This is achieved using bolt tables
with different bolt loading types.
Bolt loading can be performed in static or implicit dynamic

analysis. During bolt loading in implicit dynamic analysis, the
program temporarily switches to static analysis (unless the low-

11.17 Bolt feature
speed dynamics option is employed).
Explicit dynamic analysis can be performed after a restart from

bolt loading implicit analysis.
Frequency analysis can be performed with the bolt loading

option either in the same run or in a restart run from an analysis
with bolt loading.
If the ATS method is used, it is also applied to the bolt loading

procedure.
For the TLA and TLA-S automatic time step options, the bolt
loading procedure can only be applied at the beginning of the
analysis or at restart. The same holds for the LDC method.
Bolt loading can be used with ADINA heat flow, for both one-
way or fully coupled TMC solutions.
Restarting from a non-converged bolt sequence loading is

possible.
11.17.4 Usage of beam-bolts
A beam element group is identified as a bolt element group

using the parameter OPTION=BOLT in the EGROUP BEAM
command. The bolt element group is assigned a bolt number using
the BOLT-NUMBER parameter. The bolt load (either force or
shortening, depending upon BOLT-OPTIONS and BOLT-TABLE)
is specified using the BOLT-LOAD parameter.
A beam-bolt element group can contain one or more elements.
If the group contains more than one element, it is assumed that the
elements are all connected sequentially and all have the same
length, as shown in Figure 11-17.6.
It is possible to have several bolts with the same bolt label.

Beam-bolt element group
All elements in the group

are equally spaced
Figure 11.17-6: Modeling of a bolt using several

beam-bolt elements.
11.17.5 Usage of 3D-bolts
A 3D element group is identified as a bolt element group using

the parameter OPTION=BOLT in the EGROUP THREEDSOLID
command. The bolt element group is assigned a bolt number using
the BOLT-NUMBER parameter. The bolt load (either force or
shortening, depending upon BOLT-OPTIONS and BOLT-TABLE)
is specified using the BOLT-LOAD parameter.
The 3D bolt element group also uses a bolt plane as part of its
definition. In the general case, the bolt plane is defined using the
BOLT-PLANE command and the bolt plane is specified using the
BOLT-PLANE parameter of the EGROUP THREEDSOLID
command.
The bolt plane gives the bolt direction and the approximate
location of the bolt split faces. The bolt direction is used to
determine the direction of the bolt displacement and the direction
of the bolt force. The bolt plane is used to determine the bolt split
faces. Note that the bolt split faces do not necessarily coincide with
the bolt plane. Rather, the program determines the bolt split faces
to lie "near" the bolt plane.
It is necessary for the bolt plane to intersect the elements, in

such a way that the bolt direction is aligned with the bolt, and the
entire bolt force is transmitted across the bolt plane. Fig 11.17-7
shows some examples of incorrect and correct definitions.

11.17 Bolt feature
Bolt plane
a) Incorrect definition, bolt direction b) Incorrect definition, bolt plane

is not aligned with bolt does not intersect bolt elements
c) Incorrect definition, entire bolt force d) One possible correct definition

is not transmitted across bolt plane
Figure 11.17-7: Examples of bolt plane definition
The following options are available for the BOLT-PLANE

command, see Fig 11.17-8:
OPTION=TWOPOINT: The bolt-plane is specified by two
geometry points and a factor giving the placement of the bolt plane
relative to the two points. Factor=0 corresponds to the bolt-plane
passing through the first point and factor=1 corresponds to the bolt-
plane passing through the second point.
OPTION=TWONODE: Similar to TWOPOINT, except that two
nodes are used.
OPTION=XPLANE: The bolt plane has normal in the x
direction and is positioned at the given X coordinate.
OPTION=YPLANE or ZPLANE: Similar to XPLANE, but
using the y and z coordinate axes instead.
OPTION=POINTAUTO: The bolt plane is positioned at the

given geometry point, and the bolt plane direction is automatically

determined.
OPTION=NODEAUTO: Similar to POINTAUTO, except that a
node is used.
OPTION=POINTNORMAL: The bolt plane is positioned at the
given geometry point, and the bolt plane direction is specified
using the given normal.
OPTION=NODENORMAL: Similar to POINTNORMAL,
except that a node is used.
Option=ZPLANE
x=(0,0,COORDINATE)
x=xP1 + FACTOR(xP2 - xP1) z
P2
Bolt direction
x
P1
Option=XPLANE Option=YPLANE
x=(COORDINATE,0,0) x=(0,COORDINATE,0)
a) Option TWOPOINT b) Options XPLANE,YPLANE,ZPLANE

(TWONODE is similar)
Figure 11.17-8: Options for bolt plane definition
It is also allowed to define a 3D-bolt group without using a bolt

plane. This is done by setting BOLT-PLANE=0 in the EGROUP
THREEDSOLID command.
In this case, the AUI automatically determines the bolt plane as
follows (Fig 11.17-9). First the centroid and moments of inertia of
the bolt group are obtained (assuming unit density). Then the bolt
direction is obtained as the direction of the minimum principal
moment of inertia, and the bolt plane passes through the centroid.
These calculations are all done during data file generation.

11.17 Bolt feature
Bolt direction Direction of minimum

principal moment of inertia
Centroid
Automatically determined
bolt plane
Figure 11.17-9: Automatic determination of bolt-plane
The inertial properties are determined using numerical

integration within the finite element mesh. Because the mesh itself
is used, the inertial properties might not exactly match the
properties corresponding to the underlying geometry. For this
reason, if the direction of minimum principal moment of inertia is
within 1 degree of a global coordinate direction, the global
coordinate direction is used instead.
This algorithm gives a reasonable choice for the bolt plane in
many cases.
The AUI outputs the bolt direction and bolt cross-sectional area
for each bolt during data file generation. This information can be
used to confirm that the bolt plane is defined correctly.
When the bolt force is specified for a system in which one end
of the bolt is free (for example, the system shown in Fig 11.17-5),
the global system of equations (11.17-3) is singular. This is
physically correct as there is no non-zero bolt force that can satisfy
equilibrium.
To prevent numerical difficulties in this case, parameter BOLT-
OPTIONS FACTOR3 can be used. FACTOR3 provides an
additional stiffness relating the bolt force and bolt displacement
displacement. The default for FACTOR3 is zero, and the
dimension of FACTOR3 is stiffness. If FACTOR3 is set nonzero,

its value should be much smaller than the bolt stiffness itself.
Alternatively, stiffness stabilization can be used to prevent
numerical difficulties.
During large displacement analysis, the bolt direction may

change during the solution process. The program computes the
updated bolt direction using three nodes on the bolt split faces; as
these three nodes move, the bolt direction is correspondingly
updated.
One pitfall when importing Nastran files for the model

definition is the following. By default, elements with different
numbers of nodes are placed into different element groups. But the
3D-bolt group feature assumes that all elements modeling the bolt
are placed in the same element group. Therefore, when importing
Nastran files, the option
NASTRAN-ADINA ... SPLIT=NO
should be used, when elements with different numbers of nodes are

used to model the bolt.
It is necessary that the meshing of the 3D-bolt group be

completely compatible, in the sense that the adjacent internal
element faces must match. Free-form meshing can be used
provided that the mesh is compatible.
Fig 11.17-10 shows some simple examples of compatible and
incompatible meshing.

11.17 Bolt feature
Adjacent elements separated

for clarity
(a): Incompatible meshing, internal element faces do not match
(b): Compatible meshing, internal element faces match
Figure 11.17-10: Incompatible and compatible meshing for 3D-bolt groups
Some free-form meshing commands create incompatible

meshes. For example, GBODY NODES=8 PYRAMIDS=NO
creates only tetrahedra and hexahedra, and the meshing is
incompatible since tetrahedra with triangular faces are adjacent to
hexahedra with quadrilateral faces. GBODY NODES=8
PYRAMIDS=YES creates tetrahedral, pyramids and hexahedra,
and the resulting meshing is compatible.
11.17.6 Modeling issues
One modeling pitfall is illustrated in Fig. 11.17-11. The figure

shows a bolt with contact conditions. It is intended that, after the
bolt force is specified, the bolt be in contact with the rigid target.
However, this model will not work as intended, for the following
reason. During the bolt force iterations, the bolt provides no
stiffness between the top and bottom of the bolt (points A and B in

the figure). And since the bolt is not initially in contact with the
target, the contact also does not provide stiffness. Therefore, in the
first equilibrium iteration, point A moves downwards as a rigid
body under the prescribed bolt force, and this motion is very large
since there is no stiffness. The remaining equilibrium iterations will
probably not converge.
To prevent this effect from occurring, make sure that the contact
is established before beginning bolt force iterations. For example,
set up the model so that there is a very small overlap between the
contacting parts. Or specify the bolt shrinkage to establish contact,
then restart and specify the bolt force.
Contact
A conditions A
Bolt-
plane
B B
(a): Beam-bolt (b): 3D-bolt
Figure 11.17-11: Bolts and contact example
11.17.7 Post-processing variables
For each 3D-bolt, the bolt force and bolt displacement is

available for post-processing using the variables BOLT-FORCE,
BOLT-DISPLACEMENT. These variables can be accessed at bolt
points. For example
BOLTPOINT B2 BOLT=2
POINTLIST B2 VAR=BOLT-FORCE BOLT-DISPLACEMENT
lists the bolt force and bolt displacement at bolt 2.
For the beam-bolt elements, the bolt forces are available for
post-processing using the variables BOLT_FORCE,
BOLT_SHEARFORCE-S, BOLT_SHEARFORCE-T,
BOLT_TORSION, BOLT_MOMENT-S, BOLT_MOMENT-T.

11.17 Bolt feature
The bolt moments are calculated at bolt local node 2. Therefore the
bolt moments might not be the maximum moments in the bolt.
The bolt shortening is also available using variable

BOLT_SHORTENING. These variables are stored along with the
elements. For example
ELPOINT G5E2 GROUP=5 ELEMENT=2 OPTION=LABEL,

LABEL=111
POINTLIST G5E2 VAR=BOLT_FORCE BOLT_SHORTENING
11.18 Component mode synthesis (CMS) method
ref. KJB The Craig-Bampton reduction method is used in ADINA for

Section 10.3.7 component mode synthesis (CMS) calculation.
ref. R.R. Craig and M.C.C. Bampton, "Coupling of

Substructures for Dynamic Analysis", AIAA Journal, Vol. 6,
No. 7, July 1968.
A specific capability of ADINA is that the approximate

eigenvectors obtained by the CMS method can be used to
accurately calculate the lowest eigenvalues and eigenvectors of the
finite element system.
Theoretical overview
The degrees of freedom (DOFs) of the analysis model are

divided into internal DOFs and boundary DOFs. The internal
DOFs are represented by the modal displacements of interior nodes
and the boundary DOFs are represented by the physical
displacements of boundary nodes.
Let n = number of DOFs of the original analysis model, q =

number of component modes, k = number of boundary DOFs,
q = q k = number of fixed interface vibration modes.
The following equations are used:
M bb M bi ub K bb K bi ub R b
M + =
ib M ii ui K ib K ii u i R i

u b I 0 u b
u m q
i c
where
ub = Physical displacements of boundary DOFs (vector of

length k )
ui = Physical displacements of interior DOFs (vector of

length n k )
q = Modal displacements of interior DOFs (vector of length

q)
c = K ii1K ib Constraint modes (interior displacements

due to unit boundary displacements, applied one at a
time with the other boundary DOFs fixed), one
column for each constraint mode.
m = Fixed interface vibration modes with boundary DOFs

fixed, one column for each fixed interface vibration
mode.
The first step in this method is the calculation of the static

constraint modes matrix c . One static solution is performed for
each column in c .
The second step is the calculation of the q eigenvalues
1m

m = ... and eigenvectors m of the system
qm

( K ii , ii ). This calculation is done using the subspace iteration
method. The number of starting vectors for the solution of m is
the maximum of (2q, q + 8) .

The third step is the calculation of the reduced matrices K r and

M r using
K r = T K
M r = T M
in which
I 0
=
c m
Performing the indicated multiplications gives
K + K bi c 0
K r = bb
0 m
and
M bb + M bi c + Tc M ib + Tc M ii c M bi m + Tc M ii m
Mr =
symm.
The fourth step is the calculation of the complete eigensystem

of the reduced matrices K r and M r , that is, the solution of
1r

K r Q r = r Q r r where r = ... is a diagonal
qr

matrix containing the eigenvalues and Q r is a matrix with columns
containing the eigenvectors.
The fifth step is the calculation of q approximate eigenvectors

of the original finite element system:
X1 = Q r
ref. KJB Notice that steps 3 to 5 correspond to one step in the Bathe
Section 11.6

subspace iteration method, in which the starting vectors are taken

from the columns of .
Optionally the subspace calculation can be continued in order to

accurately calculate the lowest eigenvalues and eigenvectors of the
original finite element system, that is, for i = 1, 2,...
KXi +1 = MXi
K i +1 = XTi +1KXi +1
M i +1 = XTi +1MXi +1
K i +1Qi +1 = i +1Qi +1 i +1
Xi +1 = Xi +1Qi +1
until convergence is reached ( i +1 , Xi +1 ). The number

of lowest eigenvalues and eigenvectors p is obtained from the
formula p = min(q / 2, q 8) .
From ADINA 9.1, a different approach is taken to obtain the

eigenvalues of the reduced model for the CMS method. The
number of frequencies/modes p to be solved can be much smaller
than the number of static constraint modes r plus the number of
fixed interface dynamic modes s, i.e., p (r+s) = q.
Considering the reduced model matrices as follows,
K r = T K, M r = T M
I 0
where = ,
c n
c represents the static constraint modes, and
n represents the fixed interface dynamic modes.
Now the eigenvalues of

1r

K r Q r = M r Q r r , where r = ... are solved for
r
q

using the full subspace iteration method with p q, e.g., 3p = q.
The reduced system with p = s can be solved as with other finite

element analysis programs.
Different values of p and s with separate inputs for p modes and
s modes can also be used. Thus, the input of NFREQ in the
FREQUENCY command represents the number of frequencies p,
and that of NSFREQ represents the number of fixed interface
frequencies/modes. Further, the usual setting of the paramter
METHOD=CMS can be CMS-SUBSPACE using the subspace
iteration method for solving the fixed interface modes, or CMS-
LANCZOS using the Lanczos method.
In summary, the following two options are available from version

9.1:
K r = T K, M r = T K
in which
I 0
=
c n
Option 1: p = r + s, the old algorithm will be used.

The subspace iteration method is not required for the
reduced matrices but can be activated to obtain the
accuracy.
Option 2: p < (r + s), the new algorithm will be used.

The subspace iteration method is always performed for the
reduced matrices.
Modeling
The user must specify the total number of component modes q

(using the FREQUENCIES command), and also must specify the
boundary DOFs (using the DOF-ACTIVE command). The number
of boundary DOFs gives the value of k .

When FREQUENCIES ... AVL-EXCITE=TEXT, the program

outputs the reduced matrices K r and M r into a file with name
*.out4. The .out4 files can then be imported into other multibody
system software for additional dynamic calculations, e.g., AVL
EXCITE.
By default, the program outputs theq approximate eigenvalues

r and corresponding approximate eigenvectors X1 . However, if
FREQUENCIES ... CMS-SUBSPACE=YES, then the program
outputs the p lowest eigenvalues and corresponding
eigenvectors X instead. In this case, the number of CMS vectors
q must be greater than 8.
User hints
It is allowed for the number of fixed interface vibration modes

to be zero, that is, the number of CMS modes q can be equal to the
number of static constraint modes k .
If the original structure has rigid body modes, then a rigid body
mode shift is required in the solution of the eigensystem of reduced
matrices K r and M r . This rigid body mode shift is specified as
for other frequency analyses, see Section 6.2.
It is assumed that the structure considering boundary degrees of

freedoms fixed ( K ii , ii ) has no rigid body modes.
If the structure is nonlinear and is preloaded, the analysis

requires two runs. In the first run, the preload is applied (gravity
loads, thermal loads, initial strains, etc.) in a nonlinear static
analysis. This run can contain one or more solution steps. The
second run is a restart run, in which the CMS analysis is performed,
based on the stiffness and mass matrix corresponding to the time of
solution start; in this case the time of solution start is the time
corresponding to the restart analysis. Therefore the stiffness and
mass matrices include all geometric and material nonlinearities
corresponding to the end of the nonlinear static analysis.

12.1 Supported analysis options
12. Heat transfer capabilities in the ADINA program

Most of the ADINA-T heat transfer capabilities are also
included in the ADINA program. This chapter documents the
supported analysis options of these capabilities in ADINA, and
describes the features available for modeling.
More details of the theory and modeling of heat transfer and

TMC analysis can be found in the ADINA-T Theory and Modeling
Guide, and the commands available are documented in the AUI
Command Reference Manual, Volume 1.
12.1 Analysis options

Three analysis options are available:
heat transfer analysis only
one-way heat flow coupled to mechanical analysis
fully thermo-mechanical-coupled (TMC) analysis
The first analysis type involves only heat transfer with no

structural equations or coupling. Results from this analysis can be
later used in a structural solution, e.g., via the restart option.
The second analysis type is a one-way coupling, when the

thermal solution affects the structural solution, but the structural
solution does not affect the thermal solution.
In this case, the heat transfer and structural equations are solved
separately, with the temperatures from the heat transfer passed back
to the structural problem for calculation of thermal expansion and
temperature-dependent material properties.
The third analysis type is iterative or direct thermo-mechanical

coupling which is a two-way coupling where both the thermal and
structural solutions are interdependent.
In this case, the thermal solution can affect the structural
solution and the structural solution can affect the thermal solution.
In the case of iterative TMC, at the beginning of each time step,

Chapter 12: Heat transfer capabilities in the ADINA program
the structural model is solved for the displacements using the

current temperatures. Then the heat transfer model is solved for the
temperatures using the current displacements. This cycle
constitutes one TMC equilibrium iteration. TMC convergence is
then assessed, and if it is not reached, then the structural and heat
transfer models are solved again using the new current
displacements and new current temperatures. This process is
repeated until TMC convergence is reached. Note that within each
TMC equilibrium iteration, the heat transfer and structural models
each have their own internal iteration procedure and convergence
criteria.
The same TMC convergence parameter is used in the

displacement and temperature convergence checks.
The temperature convergence is checked as follows:
t +t
T (i ) t +t T ( i 1)
t +t (i )
2
< TOLL (12.1)
T
2
The displacement convergence is checked as follows:
t +t
U (i ) t +t U ( i 1)
t +t (i )
2
< TOLL (12.2)
U
2
where i denotes the TMC iteration.
For the direct TMC option, the program solves one system of
equations which includes displacement and temperature increments
and there is no TMC iteration process; only the standard Newton-
Raphson iteration procedure is applied for which convergence or
displacements/forces and temperatures are checked separately
using equations (12.1) and (12.2) respectively.
In strongly coupled problems, a temperature relaxation factor

can be used to help reach convergence. This parameter defaults to
1.0, which corresponds to no relaxation.
The temperatures used in the structural analysis in the case of

12.1 Supported analysis options
temperature relaxation at a TMC iteration k are based on the

temperatures in the last heat transfer TMC iteration k-1 as well as
the prior heat transfer TMC iteration k-2.
structure
(k )
= (1 1 ) heat
( k 2)
+ 1 heat
( k 1)
where 1 is the temperature relaxation factor.
Note that decreasing the relaxation factor usually reduces the

chances of an oscillating solution, but if decreased too much will
also slow down convergence.
The following analysis options, available in ADINA-T, are not

available in the ADINA program for heat transfer solutions:
- diffusive radiosity
- thermal eigenvalue calculations
- seepage
- analogous field problems
- Joule heating
12.2 Modeling
One finite element model is used for mechanical and heat flow
solutions with the following features:
Element groups have two different sets of material models,

one mechanical and the other one, thermal (called in the AUI
input TMC-MATERIAL).
The same element integration order is used for mechanical

and thermal solutions.
Element birth/death times are the same for mechanical and

thermal solutions.
For higher order elements used in the displacement and stress

solutions, the ADINA program has an option to calculate
temperatures using corner nodes only (TMC-CONTROL TEMP-
INTERP=CORNER).

Different time steps can be used for temperature and

displacement solutions in case of the one way coupling option. For
the fully coupled solution, the same time step is used.
Steady and transient heat transfer solution can be used with

either statics or implicit dynamics structural solution.
No temperature tape is used during heat flow analysis in the

ADINA program.
Different time spans can be used for thermal and mechanical

solutions for one way coupling. If the thermal time span is shorter
than the mechanical time span, then the last calculated temperature
field is used.
Different solution starting times can be used for thermal and

mechanical solutions for one way coupling. For example, first
thermal solution can be completed and then mechanical solution
can follow with the solution starting time larger than the thermal
solution ending time. The last obtained temperature field will then
be used in the mechanical solution.
Explicit dynamics cannot be coupled to heat transfer analysis.
For shell elements, two degrees of freedom are present at each

node, one corresponding to the top surface and one to the bottom
surface.
For shell elements, heat flow loads can be applied either on top,
bottom or mid-surface. Note that different temperature loads can be
applied to the top and bottom surface. When temperatures are
applied to the midsurface they are assigned to both top and bottom
shell surfaces.
Single and layered shells can be used in TMC solutions.
Automatic time stepping (ATS) can be used to vary the time

step size in order to help obtain a converged solution. If no
convergence is obtained in the original step, the program
automatically subdivides the time steps, and attempts to solve

12.2 Modeling
again. Further subdivision can be done until convergence is

reached or the time step size becomes smaller than a minimum
value. This minimum value is set as the original time step divided
by a user-specified scaling factor.
This automatic time stepping procedure is used in the solution

of heat transfer analyses and one-way coupled analyses. For
iteratively coupled TMC analyses the structural ATS procedure of
Section 7.2.4 is used instead. Note that the structural ATS
procedure has many more features, and is better suited for
nonlinear problems involving contact, geometric and material
structural nonlinearities. For direct coupled TMC, the structural
ATS procedure is applied.
Line search can also be used in the heat transfer solution in both
heat transfer only and coupled heat transfer analyses.
Mesh glueing can be used for temperature DOFs for 2-D and 3-
D solid elements. If the GLUEMESH command is specified in the
input and a TMC analysis is requested, then, in addition to
displacement DOFs, temperature DOFs are also constrained
between glued domains. If heat flow only option is used, then only
temperature DOFs are constrained.
12.3 Soil consolidation analysis

Coupled diffusion-stress analysis problems can be solved with
the ADINA system. Such is the phenomenon of soil consolidation,
in which a soil under load settles due to the dissipation of excess
internal fluid pressure.
This chapter discusses the solution of soil consolidation

problems using ADINA-TMC. However, it is recommended that
soil consolidation problems be solved instead using the porous
media formulation of ADINA, see Section 3.12.
In ADINA-TMC, the soil consolidation modeling capability is

based on the linear consolidation theory, as described below.
The soil is regarded here as a porous medium, consisting of a

solid skeleton (soil particles in contact) and of interconnected voids

partially or totally filled with fluid. The soil porosity, n, is defined

as the proportion of voids per unit total volume. The void ratio, e,
is defined as the proportion of voids to solids in any given volume.
The fluid content, , at a given fluid pressure , is defined as the
change in fluid volume, per unit total volume, between the strained
state corresponding to and a reference unstrained state.
The linear consolidation theory provides a macroscopic

description of the soil response, based on the following
assumptions:
< The soil skeleton behavior is linear elastic isotropic.

< The fluid is incompressible.
< The fluid flows through the porous soil according to Darcy's
law:
v = k (12.1)
where
v = fluid velocity vector

k = soil permeability matrix
= fluid pressure
Considering only small (macroscopic) strains in the soil and

small velocities in the fluid, a linear stress-strain relation can be
derived:
= Ce 1 (12.2)
where
= macroscopic stress tensor

e = macroscopic strain tensor
C = macroscopic stress-strain law matrix of the soil
skeleton
= first soil consolidation parameter
1 = Kronecker delta vector
It is also assumed that the fluid content varies linearly with

the fluid pressure and the soil volumetric strain, i.e.

= ev + (12.3)
where
ev = soil skeleton volumetric strain
= second soil consolidation parameter
The general equations governing transient soil consolidation can

then be established. First, the macroscopic stresses defined in
equation (12.2) must satisfy the equilibrium condition:
+ f b = 0 (12.4)
where f b are body forces. Second, the continuity condition for

incompressible fluid flows:

q b = v (12.5)
t
where q b is the internal fluid flow generation, used together with

Darcy's equation (12.1), and with equation (12.3), yields the
diffusion equation controlling the variation of the fluid pressure:
ev
( k ) = + qb (12.6)
t t
We can now see that the stress equation (12.4) and the diffusion
equation (12.6) constitute a coupled diffusion-stress equation
system.
Note on the soil consolidation parameters: From equation

(12.6), it can be seen that the first soil consolidation parameter
is a ratio between the outward flux of fluid out of the soil and the
variation in the soil volume, whereas the second soil consolidation
parameter is a ratio between the outward flux of fluid and the
variation in the fluid pressure.
yield the following equation:

u
0 0 t K u K u u f
K T + = (12.7)
u M 0 K q
t

where
K u = BTu C Bu dV
V
K u = BTu H dV
V
K u = HT Bu dV
T
K = BT k B dV
V
M = HT H dV
V
f,q = load vectors (forces, fluid flows)

H = interpolation matrix
Bu , B = gradient matrices for the displacements u and
pressure
Equation (12.7) can be written as follows:
K u u = f K u (12.8)
u
M + K = q K T u (12.9)
t t
Equation (12.8) is solved using the ADINA model, for a given fluid
pressure field, and equation (12.9) is solved, for a given
displacement field. An iterative procedure can therefore be used to
solve this equation system.
UCMAT3C, TUSR2C and/or TUSR3C, should be used in soil

consolidation analysis. Note that, in TUSR2C and TUSR3C,

KEY=5 provides the first soil consolidation parameter .
12.4 Piezoelectric analysis

The solution of piezoelectric problems (direct and converse
effects) can be performed with ADINA-TMC.
Similar coupled mechanical/electrical problems including
magnetostrictive, electrostrictive materials, and shape memory
alloys, can also be solved with ADINA-TMC.
In a piezoelectric material, the stresses depend on the electric

potential, and the electric potential depends on the stresses. Hence,
the following system must be solved:
K UU K U U FUB + FUS
K = (12.10)
U K FS
where
K UU = mechanical stiffness
K = electric permittivity matrix
K U =piezoelectric matrix
K U = KU
FUB = body force loading vector
FUS = surface force loading vector
FS = surface charge density vector
U = vector of structural displacements
= vector of electric potentials
For solution with ADINA-TMC, the equation system (12.10)

above is written as follows:
K UU U = FUB + FUS K U (12.11)
K = FS KU U (12.12)

Equation (12.11) can be solved using the ADINA model,

assuming that the electric potential field is known, and equation
(12.12) can be solved using the same model, assuming that the
displacements in the structure are known.
The ADINA user-supplied subroutines UCMAT2P and/or

UCMAT3P, and the user-supplied subroutines TUSR2P and/or
TUSR3P, should be used n piezoelectric analysis.
The piezoelectric solution option can also be used in FSI

analysis.
More information about the theory and applications of

piezoelectricity with the finite element method can be found in the
ref. Gaudenzi P., Bathe K. J., "Recent Applications of an

Iterative Finite Element Procedure for the Analysis of
Electrostatic Materials," Proceedings of the 4th
International Conference on Adaptive Structures, Nov.
2-4, Kln, Germany, 1993.
ref. Tzou H. S., Tseng C. I., "Distributed Piezoelectric

Sensor/Actuator Design for Dynamic
Measurement/Control of Distributed Parameter Systems:
A Piezoelectric Finite Element Approach," Journal of
Sound and Vibration, Vol. 138 (1), pp. 17-34, 1990.
12.5 Thermal coupling between solids and fluids

ADINA provides two solution options for thermo-fluid-structure
interaction problems:
One-way thermal coupling: The temperature field is solved

entirely in the CFD model, where solid elements are modeled as
fluid elements with fluid velocity and pressure set to zero i.e., only
temperature DOFs are active in the solid part. The solid
temperature field is then mapped to the solid model for every FSI
iteration. This method does not account for any heat generation in
the solid.

12.5 Temperature coupling between solids and flulids
Two-way (and one-way) thermal coupling between solids and

fluids: The energy equation is solved separately in the fluid model
and the solid model, and the domains are coupled together by
enforcing the same temperature and heat flux at the fluid-structure
interface. This method is known as thermo-FSI (or TFSI) analysis.
The benefits are:
heat generation in the solid model (e.g. due to frictional
heating, plasticity, viscous losses, etc.) affects the
temperature in the fluid model
no need for temperature mapping
no need to create a solid mesh in the fluid model
Note that the BTFSI option in version 9.0 has been replaced by
the above standard TFSI option. If a user specifies in the input the
BTFSI option (formerly known as TQ and QT), the program uses
the standard TFSI solution procedure.
Figures 12.5-1 and 12.5-2 below illustrate the difference between

the one- and two-way thermal coupling methods.
In TFSI:
Two-way thermal coupling (TFSI) is only supported for
FCBI-C elements with iterative FSI coupling. It is not
supported for direct FSI coupling.
All thermo-mechanical coupling (TMC) options for heat
flow solution in the solid model are available. That is, one-
way TMC coupling, iterative TMC coupling, and direct
TMC coupling methods can all be used in the solid model.
Iso-beam and shell elements account for the through-
thickness temperature gradient due to different temperatures
on the top and bottom surfaces of the element. However, the
beam element assumes uniform through-thickness
temperature equal to the average temperature between the
top and bottom surfaces.

Fluid model Solid model
Fluid Solid Solid

element element element
group group group
in in
= k, cp ADINA- ADINA
CFD
= k, cp E,
FSI
boundary
condition
Figure 12.5-1: One-way thermal coupling (conjugate heat transfer)
Fluid model Solid model
Fluid Solid
element element
group group
= k, cp
E,
= k, cp
FSI
boundary
condition
Figure 12.5-2: Two-way thermal coupling (TFSI)

13.1 Calculation of results within the AUI
13. Postprocessing considerations

In this chapter we discuss those aspects of the AUI (ADINA
User Interface) which require an in-depth discussion.
This chapter discusses the postprocessing of results from all five

solution programs ADINA, ADINA-T, ADINA-F, ADINA-FSI,
ADINA-TMC.

The ADINA system programs generate a wide variety of results,
for example, stresses, displacements, reactions, heat fluxes, contact
segment tractions. You access all of these results using several
quite general AUI commands. To understand the inputs to these
commands, the following concept is important to remember.
The AUI must know the answers to the following four questions
before it can obtain a result.
What result is to be obtained?
Where (in the model) is the result to be evaluated?
When (for which load step, mode shape, etc) is the result to
be evaluated?
How is the result to be evaluated?
You can supply answers to each of these questions within the AUI.
"What" is controlled by variables. "Where" is controlled by model
points, model combination points, model lines, zones and result
grids. "When" is controlled by response depictions and
response-range depictions. "How" is controlled by smoothing
depictions and result control depictions.
It is also important to understand that the AUI can transform

results computed by the solution programs (the available results) to
the results that you request (the requested results). The AUI can
make the following transformations:

Chapter 13: Postprocessing considerations
Spatial transformations:
The AUI can interpolate results available at the element

integration points to other points within the elements, such as
the element local nodes.
The AUI can combine the results from all of the elements
attached to a node into a single result (smoothing).
The AUI can interpolate the results from the nodes to points
within elements.
Time transformations:
The AUI can interpolate results available at two solution

times to solution times between these two times.
Resultant transformations:
The AUI can combine results at a single point into a single

result. For example, the AUI can compute the effective stress at
a point from the stress components at the point.
Combination transformations:
The AUI can combine results according to response

spectrum formulas, harmonic and random vibration formulas or
user-supplied combinations (known as response-combinations).
The AUI can also scan for the most extreme result for given
solution times and return the most extreme value (envelope
combinations).
These transformations are organized in a hierarchy such that

each transformation can request information from the
transformation below it. The hierarchy is described in the
following examples.
Example 1: As an example involving only the spatial
transformation, consider the evaluation of a smoothed stress
component at a point within an element (not necessarily a point at
which the results were computed by ADINA) for an available time

(a time for which results were computed by ADINA). This stress

component is calculated by the spatial transformation as follows.
The spatial transformation interpolates the available stress
component data from the integration points to the nodes, performs
smoothing, then interpolates the stress component to the requested
point.
Example 2: As an example involving only the time and spatial

transformations, consider the evaluation of a smoothed stress
component at a point within an element for a requested time (not
necessarily a time for which results were computed by ADINA).
This stress component is calculated by the time transformation.
The time transformation requests the stress component for the
available solution times nearest to the requested time, then
interpolates in time to the requested solution time. The stress
components at the available solution times are calculated by the
spatial transformation as in Example 1.
In the above example, the spatial transformation acts as a
subroutine that is called by the time transformation.
Example 3: As an example involving the resultant, time and spatial

transformations, consider the evaluation of smoothed effective
stress at a point within an element for a requested time. The
effective stress is computed by the resultant transformation. The
resultant transformation requests the stress components from the
time transformation, then computes the effective stress. The time
transformation computes the stress components as in Example 2.
In the above example, the time transformation acts as a
subroutine that is called by the resultant transformation.
Example 4: As an example involving the combination, time and

spatial transformations, consider the evaluation of stress component
difference at a point within an element between two requested
times. (E.g., xx ( time 2.0 ) xx ( time 1.0 ) .) The stress component
difference is computed by the combination transformation. The
combination transformation requests the stress components from
the time transformation, then computes the difference. The time
transformation computes the stress components as in Example 2.
In the above example, the time transformation acts as a
subroutine that is called by the combination transformation.

In the above examples, the AUI automatically determines the

hierarchy order. However, when considering examples involving
both the resultant and the combination transformations, you can
control the order of these transformations. There are two cases:
Resultants before combinations: The combination transformation

can request results from the resultant transformation, the resultant
transformation can request results from the time transformation, the
time transformation can request results from the spatial
transformation.
In other words, the combination transformation acts on
resultants.
Resultants after combinations: The resultant transformation can

request results from the combination transformation, the
combination transformation can request results from the time
transformation, the time transformation can request results from the
spatial transformation.
In other words, the combination transformation acts on the
variables contained within the resultant.
The default is resultants before combinations.
The following examples show cases in which the hierarchy order is

important.
Example 5: Consider the calculation of the radial stress in response

spectrum analysis. The response spectrum calculation is a
combination transformation and the radial stress calculation is a
resultant transformation.
If resultants before combinations is used, the combination
transformation requests the radial stresses for each mode shape
from the resultant transformation, then the combination
transformation combines the modal radial stresses using the
response spectrum formulas.
If resultants after combinations is used, the resultant

transformation requests the stress components from the
combination transformation, then the resultant transformation
computes the radial stress. The combination transformation
calculates the stress components using the modal stress components

and the response spectrum formulas.

In this example, resultants before combinations is correct, and
it would be incorrect to use resultants after combinations.
Example 6: Consider the calculation of the effective stress for a

response combination involving the addition of two load cases. The
response combination calculation is a combination transformation
and the effective stress calculation is a resultant transformation.
transformation requests the effective stresses for each load case
transformation adds the effective stresses.
computes the effective stress. The combination transformation
adds the stress components from the two load cases.
In this example, resultants after combinations is correct, and it
would be incorrect to use resultants before combinations.
Example 7: Consider the calculation of the effective stress in

harmonic vibration analysis, in which you enter the value of t .
The harmonic vibration analysis calculation is a combination
transformation and the effective stress calculation is a resultant
transformation.
transformation requests the effective stresses for each mode shape
transformation combines the modal effective stresses using the
harmonic vibration formulas.
and the harmonic vibration formulas.
Example 8: Consider the calculation of the resultant 1 + ui , where

ui is a displacement component, in harmonic vibration analysis.

transformation requests the quantity 1 + i for each mode shape
from the resultant transformation (where i is the corresponding
eigenvector component for each mode shape), then the combination
transformation combines the quantities 1 + i using the harmonic
vibration formulas.
transformation requests ui from the combination transformation,
then the resultant transformation computes 1 + ui . The
combination transformation calculates ui using i for each mode
shape and the harmonic vibration formulas.
Example 9: Consider the calculation of effective stress in random

vibration analysis.
transformation requests the effective stresses for each mode shape
transformation combines the modal effective stresses using the
random vibration formulas.
and the random vibration formulas.
In this example, neither resultants before combinations nor
resultants after combinations is correct. It would be better if the
AUI could estimate the effective stress by changing the signs of the
stress components output from the combination transformation, so
as to maximize the effective stress.
The general rule of thumb is, if both the combination

transformation and the resultant transformation are linear, it doesnt
matter which rule to use; both will give the correct results. If the
combination transformation is nonlinear and the resultant

transformation is linear, use resultants before combinations. If

the combination transformation is linear and the resultant
transformation is nonlinear, use resultants after combinations. If
both transformations are nonlinear, then neither rule will give the
correct results.
There is an additional consideration when the model point is a

combination point, for example a node combination point (in which
the results at the nodes in the combination point are summed).
When resultants before combinations is used, the combination
transformation requests the summed resultant (the summing being
performed by the combination point).
When resultants after combinations is used, the summing (by
the combination point) is performed on the variables referenced by
the resultant, then the resultant is taken.
Example 10: Consider the calculation of the effective stress for a

response combination involving the subtraction of two load cases.
Here the effective stress is evaluated at a node combination point.
transformation requests the summed effective stresses for each load
case, then the combination transformation subtracts the summed
effective stresses.
If resultants after combinations, the combination point
requests the effective stresses from the resultant transformation and
then sums the effective stresses. The resultant transformation
computes the effective stresses from the stress components. The
combination transformation subtracts the stress components from
the two load cases.
In the following, we discuss what results are available, where

they can be evaluated and how they are evaluated.
13.1.1 What results are available
The results computed by the ADINA system programs are made

available to the AUI when the corresponding porthole file is
loaded. In many cases you control the results that are available
when you define the model during preprocessing. For example,
reactions are computed if you check the "Calculate Reaction

Forces/Moments" box within the ControlMiscellaneous Options

dialog box.
Regarding results output along with the elements, in general you
can either output stresses or forces, but not both. This choice is
made within the element group definitions dialog box. For
example, for a 2-D solid element group, the choice is made with the
"Calculated Element Results" buttons.
When you choose to output stresses, the actual results output
depend upon the material model and upon the stress reference
system employed. These results are described in Chapter 2 of this
manual for each element group.
Regarding results output along with the contact surfaces, you
can output either segment tractions or forces. This choice is made
within the contact group definitions dialog box. However, ADINA
calculates segment tractions only for contactor contact surfaces.
The actual results output are described in Chapter 4 of this manual.
Each type of result output by ADINA is referred to by a variable

name.
The variables can be classified according to where (in the
model) they are output by the solution program and how they are
interpolated within the model. The variable location types are:
Location-independent: Variables such as time or frequency that

are not associated with specific points in the model.
Node field: Variables such as displacement, velocity or

acceleration that are output at node points and that can be
interpolated within elements (using the nodal shape functions).
Element/layer field: Variables such as stresses that are output

within elements or element layers and that can be interpolated
within elements. These variables are output by the solution
program at the integration points.
Contact: Variables such as normal traction that are output at

contact segments (ADINA only).
Draw bead: Variables such as restraining traction and uplifting

traction that are output at draw bead segments (ADINA only).

Section field: Variables such as moments that are output within

the sections of elements (midsurface of shells or neutral line of
Hermitian beams) and that can be interpolated within element
sections (ADINA only).
Radiation field: Variables that are output at radiosity segments

(ADINA-T only).
Line contour: Variables that are output for each fracture

mechanics line contour (ADINA only).
Virtual shift: Variables that are output for each fracture

mechanics virtual shift (ADINA only).
Node discrete: Variables such as reaction that are output at node

points and that cannot be interpolated within elements.
Element local node: Variables such as consistent nodal forces

that are output at the element local nodes and that cannot be
interpolated within elements.
Element local node field: Variables such as shell thicknesses

that are output at the element local nodes and that can be
interpolated within elements.
Coordinate: Variables such as coordinate.
Element face: Variables that are output at the element edges (2D
analysis) or element faces (3D analysis) of ADINA-F elements.
Mesh surface: Variables, such as the surface normal, that can

only be evaluated at mesh integration or mesh extreme points.
The variables can also be classified according to function. The

classifications used in the AUI are: displacement, velocity,
acceleration, temperature, fluid, latent heat, stress, strain, force,
traction, reaction, flux, electric field, fracture, failure criterion,
eigenvector, prescribed load, coordinate, thickness, time,
frequency/mode, miscellaneous, user-defined.
Tables 13.1-1 to 13.1-23 show all of the variables. Each table

shows all of the variables of the given function category. In the

tables, we compress the variable names using characters enclosed
in parentheses. The compression rules are as follows:
Numerical lists: Entries enclosed in parentheses are to be

expanded with one variable per character in the parentheses.
For example, OVALIZATION-(123456) represents the six
variables OVALIZATION-1, OVALIZATION-2,
OVALIZATION-3, OVALIZATION-4, OVALIZATION-5,
OVALIZATION-6.
Vectors: Entries enclosed in parentheses are to be expanded

with one variable per character in the parentheses. For example,
(ABC)-DISPLACEMENT represents the three variables A-
DISPLACEMENT, B-DISPLACEMENT, C-DISPLACEMENT.
Symmetric tensors: Entries enclosed in parentheses are to be

expanded to give components as in this example:
STRESS-(XYZ) represents the six variables STRESS-XX,
STRESS-YY, STRESS-ZZ, STRESS-XY, STRESS-XZ,
STRESS-YZ.
Other cases: Entries enclosed in parentheses and separated by

commas are to be expanded as in this example:
CURVATURE-(RB,SB,RBSB,T) represents the four
variables CURVATURE-RB, CURVATURE-SB,
CURVATURE-RBSB, CURVATURE-T.
We also use the * to denote any number starting from 1, as in this

example: TIME_FUNCTION_* represents the variables
TIME_FUNCTION_1, TIME_FUNCTION_2, etc.
The tables give notes for those variables that are not self-
explanatory.

Table 13.1-1: Displacement variables

Variable name Variable location type Notes
(123)-DISPLACEMENT Node field 1
(ABC)-DISPLACEMENT Node field 2
(XYZ)-DISPLACEMENT Node field
DISPLACEMENT_MAGNITUDE Node field 3
FLUID_P0 Node field
FLUID_POTENTIAL Node field
OVALIZATION-(123456) Node field
(123)-ROTATION Node field 1
(ABC)-ROTATION Node field
(V1,V2)-ROTATION Node field
(XYZ)-ROTATION Node field
WARPING-(123456) Node field
1) These are the components of the displacements and rotations transformed by the
AUI into a user-specified coordinate system. More information is given at the end of
this section.
2) The A, B and C components refer to the skew coordinate system (if one is used).
The V1, V2 components refer to the shell rotation directions (for shell midsurface
nodes with five degrees of freedom). The X, Y and Z components always refer to the
global coordinate system. The AUI can transform results that were calculated by
ADINA in one coordinate system to another coordinate system, see Table 13.1-24.
This comment holds for velocities, accelerations and eigenvectors as well as
displacements.
3) DISPLACEMENT_MAGNITUDE is the square root of the sum of the squares of the

displacement components, as computed by the AUI.

Table 13.1-2: Velocity variables

(123)-ANGULAR_VELOCITY Node field 1
(ABC)-ANGULAR_VELOCITY Node field
(V1,V2)-ANGULAR_VELOCITY Node field
(XYZ)-ANGULAR_VELOCITY Node field
ELEMENT_(123)-VELOCITY Element field 1
ELEMENT_(XYZ)-VELOCITY Element field
FLUID_P0_VELOCITY Node field 2
FLUID_POTENTIAL_VELOCITY Node field 3
NODAL_SLIP_VELOCITY Node field 4
NODAL_SLIP_VELOCITY-(XYZ) Node field 5
OVALIZATION_VELOCITY-(123456) Node field 6
PORE_PRESSURE_VELOCITY Node field 7
(123)-VELOCITY Node field 1
(ABC)-VELOCITY Node field
(XYZ)-VELOCITY Node field
VELOCITY_MAGNITUDE Node field 8
WARPING_VELOCITY-(123456) Node field 9
1) These are the components of the velocities, angular velocities, etc., transformed by
the AUI into a user-specified coordinate system. More information is given at the
end of this section.
2) This is the 1st time derivative of the fluid P0 degree of freedom.
3) This is the 1st time derivative of the fluid potential degree of freedom.
4) NODAL_SLIP_VELOCITY is the magnitude of the contact slip velocity. It is

output when the new contact segments are used in frictional contact.
5) NODAL_SLIP_VELOCITY-(XYZ) are the components of the contact slip

velocity. These are output when the new contact segments are used in frictional
contact.
6) These are the 1st time derivatives of the ovalization degrees of freedom.

7) This is the 1st time derivative of the pore pressure (see Table 13.1-7 for the pore
pressure).
8) VELOCITY_MAGNITUDE is the square root of the sum of the squares of the

velocity components, as computed by the AUI.
9) These are the 1st time derivatives of the warping degrees of freedom.
Table 13.1-3: Acceleration variables

(123)-ACCELERATION Node field 1
(ABC)-ACCELERATION Node field
(XYZ)-ACCELERATION Node field
ACCELERATION_MAGNITUDE Node field 2
(123)-ANGULAR_ACCELERATION Node field 1
(ABC)-ANGULAR_ACCELERATION Node field
(V1,V2)-ANGULAR_ACCELERATION Node field
(XYZ)-ANGULAR_ACCELERATION Node field
FLUID_P0_ACCELERATION Node field 3
FLUID_POTENTIAL_ACCELERATION Node field 4
OVALIZATION_ACCELERATION-(123456)Node field 5
PORE_PRESSURE_ACCELERATION Node field 6
WARPING_ACCELERATION-(123456) Node field 7
1) These are the components of the accelerations and angular accelerations

transformed by the AUI into a user-specified coordinate system. More information is
given at the end of this section.
2) ACCELERATION_MAGNITUDE is the square root of the sum of the squares of the

acceleration components, as computed by the AUI.
3) This is the 2nd time derivative of the fluid P0 degree of freedom.
4) This is the 2nd time derivative of the fluid potential degree of freedom.
5) These are the 2nd time derivatives of the ovalization degrees of freedom.
6) This is the 2nd time derivative of the pore pressure (see Table 13.1-7 for the pore
pressure).

7) These are the 2nd time derivatives of the warping degrees of freedom.
Table 13.1-4: Temperature variables

Variable name Variable type Notes
ELEMENT_TEMPERATURE Element field 1
TEMPERATURE Node field
1) This is the temperature as output at the element integration points.
Table 13.1-5: Fluid variables

CELL_PECLET_NUMBER Element field
CELL_REYNOLDS_NUMBER Element field
EFFECTIVE_VISCOSITY Element field
MASS_RATIO_* Node field
NODAL_VISCOSITY Node field
OMEGA-(XY,XZ,YZ) Element field
STREAM_FUNCTION Node field
TOTAL_HEAD Node field
TURBULENCE_EPSILON Node field
TURBULENCE_K Node field
TURBULENCE_W Node field
Table 13.1-6: Latent heat variables

LATENT_HEAT_INTERFACE_* Node field

Table 13.1-7: Stress variables

AXIAL_STRESS Element field 18
DISTORTIONAL_STRESS Element field
EFFECTIVE_STRESS Element field 1,3
EQUIV_INTERNAL_AXIAL_PRESSURE Element field
EQUIV_INTERNAL_HOOP_PRESSURE Element field
FE_EFFECTIVE_STRESS Element field 2,3
FE_MAX_SHEAR_STRESS Element field 2
FE_PRESSURE Element field 2,4
FE_SIGMA-P1 Element field 2,5
FE_SIGMA-P1_ANGLE Element field 2,5
FE_SIGMA-P1_DIRECTION-(XYZ) Element field 2,5
FE_SIGMA-P2_ANGLE Element field 2,5
FLUID_REFERENCE_PRESSURE Element field
GASKET_PRESSURE Element field
GASKET_STRESS-(ABC) Element field 17
GASKET_YIELD_STRESS Element field
GENERAL_ELEMENT_STRESS Element field
GRAVITY_IN-SITU_PRESSURE Element field
HILL_EFFECTIVE_STRESS Element field
MAX_SHEAR_STRESS Element field 1,7
MEAN_STRESS Element field
NODAL_PRESSURE Node field 4
PORE_PRESSURE Node field 15
PRESSURE Element field 1,4
PRESSURE_GRADIENT_MAGNITUDE Element field 16
SIGMA-NORM2 Element field 1,8
SIGMA-(P1,P2,P3) Element field 1,6
SIGMA-(P1,P2,P3)_DIRECTION-(XYZ) Element field 1,14
STRESS-(123) Element field 9
STRESS-(ABC) Element field 10
STRESS-(IJK) Element field 11
STRESS-(RST) Element field 12

STRESS-(XYZ) Element field 13

STRESS_THRU_THICKNESS Element field
YIELD_STRESS Element field
1) These variables are computed by the AUI from data output by the solution
program.
2) Variables that begin with FE_ denote quantities that are output directly from the
solution program. For example, FE_EFFECTIVE_STRESS is the effective stress as
output from ADINA for certain elements/material models.
These variables should be distinguished from variables defined in the AUI as
predefined resultants (see Note 1).
3) Variable FE_EFFECTIVE_STRESS is the effective stress as output by ADINA.

Usually FE_EFFECTIVE_STRESS and EFFECTIVE_STRESS are numerically
identical to each other, when they are evaluated at the element integration points.
One case in which FE_EFFECTIVE_STRESS and EFFECTIVE_STRESS are not
numerically identical is when ADINA uses a large displacement/small strain
formulation. In this case, ADINA uses 2nd Piola-Kirchhoff stresses internally and
variable FE_EFFECTIVE_STRESS is a 2nd Piola-Kirchhoff stress; but since
variables STRESS-XX, etc are Cauchy stresses, EFFECTIVE_STRESS is also a
Cauchy stress.
The AUI interpolates (and possibly smoothes) EFFECTIVE_STRESS from the
integration points differently than it interpolates FE_EFFECTIVE_STRESS. The
AUI interpolates EFFECTIVE_STRESS by interpolating the stress components
(possibly smoothing them as well) individually; then, after interpolation, the AUI
computes EFFECTIVE_STRESS using the formula
(
e = 1/ 2 ( xx yy ) + ( xx zz ) + ( yy zz ) + 6 ( xy2 + xz2 + yz2 )
2 2 2
)
However, the AUI interpolates and smoothes FE_EFFECTIVE_STRESS using the
same algorithms that it uses for any other stress component.
4) The comments given in Note 3 apply to the variables PRESSURE (computed as

xx + yy + zz
p= by the AUI) and FE_PRESSURE. The variable
3
NODAL_PRESSURE is output by ADINA-F (since pressure is calculated at the nodes

in ADINA-F), and, when postprocessing ADINA-F results, you can use

NODAL_PRESSURE to plot or list the nodal pressures.
5) These are the principal stresses and therir directions, as output by the ADINA
program. Note that the concrete and curve description models, these are the true
principal stresses only before cracking has occurred. After cracking, the directions
are fixed corresponding to the crack directions and these variables are no longer
principal stresses.
For the data fitted concrete material model, FE_SIGMA-P1_ANGLE and
FE_SIGMA-P2_ANGLE for 2-D solid elements, and FE_SIGMA-
P1_DIRECTION-(XYZ), FE_SIGMA-P2_DIRECTION-(XYZ) and
FE_SIGMA-P3_DIRECTION-(XYZ) for 3-D solid elements define the normal
directions of the crack planes; FE_SIGMA-P1, FE_SIGMA-P2 and FE_SIGMA-
P3 are the stress values at the time of crack formation along the normal directions of
the crack planes.
For 2-D solid elements, FE_SIGMA-P3 is always out of the plane of the element
and FE_SIGMA-P1, FE_SIGMA-P2 are always in the plane of the element
(FE_SIGMA-P1 FE_SIGMA-P2) only for the concrete and curve description
material models.
6) These are the principal stresses as computed by the AUI from the stress
components (SIGMA-P1 SIGMA-P2 SIGMA-P3).
7) This is the maximum shear stress, defined as one-half the difference between the
maximum and minimum principal stress, as computed by the AUI from the stress
components.
8) This is the 2-norm of the stress tensor (the larger of the absolute values of the
maximum and minimum principal stress) as computed by the AUI from the stress
components.
9) These are the stresses transformed by the AUI into a user-specified coordinate
system. More information is given at the end of this section.
Note that these variables have nothing to do with the initial stresses entered in the
initial strain directions.
10) These are the stress components in the axes of the material coordinate system, for
orthotropic materials. Note that the AUI cannot transform stresses output in other
coordinate systems to the material coordinate system.
11) These are equal to the global stress components for those elements in which the

stresses are output in the global coordinate system, equal to the isoparametric
coordinate system stress components for those elements in which the stresses are
output in the isoparametric coordinate system, and equal to the material coordinate
system stress components for those elements in which the stresses are output in the
material coordinate system. These variables are mainly used for constructing
resultants that are invariants (that is, do not depend on the coordinate system
direction). For example, resultant variable EFFECTIVE_STRESS is defined in
terms of STRESS-II, STRESS-JJ, etc.
12) These are the stress components in the axes of the isoparametric coordinate
system. For shell elements, component directions R and S refer to directions r and
s (see Section 2.7.1 and Figure 2.7-6). Note that the AUI cannot transform stresses
output in other coordinate systems to the isoparametric coordinate system.
13) These are the stress components in the global coordinate system. Note that the
AUI cannot transform stresses output by ADINA in other coordinate systems to the
global coordinate system.
14) These are the directions of the principal stresses SIGMA-P1,SIGMA-P2,

SIGMA-P3 (see note 6).
15) This is the value of the pore pressure, used in the porous media formulation.
16) PRESSURE_GRADIENT_MAGNITUDE is implemented for the ADINA-F 3- and

4-node 2D elements, and for the 4- and 8-node 3D elements. The purpose of
PRESSURE_GRADIENT_MAGNITUDE is to provide an ADINA-F variable that
should be smooth in the mathematical solution but that may not be smooth in the
finite element solution. A band plot of PRESSURE_GRADIENT_MAGNITUDE will
show jumps between elements.
17) In 2-D analysis, the BB, AB and CC components are defined; in 3-D analysis, the
BB, CC, AB, AC, BC components are defined.
18) In band plots, AXIAL_STRESS can be used instead of STRESS-RR.
Table 13.1-8: Strain variables

ACCUM_EFF_CREEP_STRAIN Element field 11
ACCUM_EFF_PLASTIC_STRAIN Element field 5

ACCUM_EFF_TRANSF_STRAIN Element field 12

ACCUM_EFF_VISCOPLASTIC_STRAIN Element field
ACCUM_PLASTIC_AXIAL_STRAIN Element field
ACCUM_PLASTIC_CURVATURE-(ST) Section field
ACCUM_PLASTIC_TWIST Section field
AUSTENITE_FRACTION Element field 12
AXIAL_STRAIN Section field
CREEP_STRAIN-(ABC) Element field 1
CREEP_STRAIN-(RST) Element field 1
CREEP_STRAIN-(XYZ) Element field 1
CREEP_STRAIN_THRU_THICKNESS Element field
CURVATURE-(RB,SB,RBSB) Section field
CURVATURE-(ST) Section field
CURVATURE-(XL,YL,XLYL) Element field
DEFORMATION_GRADIENT-(XYZ) Element field 6
DETWINNED_MARTENSITE_FRACTION Element field 12
EFF_VISCOPLASTIC_STRAIN_RATE Element field
EFFECTIVE_CREEP_STRAIN Element field
FE_STRAIN-P1 Element field 14
GASKET_CLOSURE_STRAIN Element field
GASKET_PLASTIC_CLOSURE_STRAIN Element field
GASKET_STRAIN-(ABC) Element field 9
GASKET_THERMAL_STRAIN Element field
GASKET_THERMAL_STRAIN-(ABC) Element field 9
IRRADIATION_STRAIN Element field
LOGSTRAIN_P1 Element field 10
LSTRETCH-(XYZ) Element field 7
MEMBRANE_STRAIN-(RB,SB,RBSB) Section field
MEMBRANE_STRAIN-(XL,YL,XLYL) Element field
NEUTRAL_AXIS_POSITION-(RB,SB) Section field
PLASTIC_AXIAL_STRAIN Section field
PLASTIC_CURVATURE-(ST) Section field
PLASTIC_STRAIN-(ABC) Element field 1
PLASTIC_STRAIN-(RST) Element field 1
PLASTIC_STRAIN-(XYZ) Element field 1

PLASTIC_STRAIN_THRU_THICKNESS Element field

PLASTIC_TWIST Section field
RSTRETCH-(XYZ) Element field 8
SHEAR_STRAIN-(RB,SB) Section field
STRAIN-(123) Element field 3
STRAIN-(ABC) Element field 1,4
STRAIN-(IJK) Element field 1,4
STRAIN-(RST) Element field 1,4
STRAIN-(XYZ) Element field 1,4
STRAIN-(P1,P2,P3) Element field 1,13
STRAIN_THRU_THICKNESS Element field
STRETCH-(XYZ) Element field 2
THERMAL_STRAIN Element field
THERMAL_STRAIN-(ABC) Element field 1
THERMAL_STRAIN-(RST) Element field 1
THERMAL_STRAIN-(XYZ) Element field 1
THERMAL_STRAIN_THRU_THICKNESS Element field
TRANSFORMATION_STRAIN-(RST) Element field 12
TRANSFORMATION_STRAIN-(XYZ) Element field 12
TWINNED_MARTENSITE_FRACTION Element field 12
TWIST Section field
VISCOPLASTIC_STRAIN-(XYZ) Element field 1
VOLUMETRIC_STRAIN Element field
1) The off-diagonal strain terms are engineering strains; i.e. twice the tensorial
strains.
2) These are the Cartesian components of the left stretch tensor V, as output by
ADINA. See Section 3.1, equation 3.1-5 for the definition and meaning of V.
The principal stretches can be computed in the AUI by defining a resultant using
functions SIGP1, SIGP2 and SIGP3. For example, you can define, for 3-D problems,
MAX_STRETCH = SIGP1(<STRETCH-XX>, <STRETCH-YY>,

<STRETCH-ZZ>, <STRETCH-XY>,
<STRETCH-XZ>, <STRETCH-YZ>)
The left stretch tensor is also used in the AUI for plotting the stretches as element
vectors.
3) These are the strain components transformed by the AUI into a user-specified

coordinate system. See the end of this section for more information.
Note that these strains have nothing to do with the initial strains entered in the
initial strain directions.
4) The same comments as are given for the stress variables (notes 10, 12, 13 of Table
13.1-7) apply to the strain variables.
5) The definition of the accumulated effective plastic strain is given in Section 3.4.1.
6) These are the components of the deformation gradient tensor 0t X , as output from
ADINA. Note that there are 9 components, since the deformation gradient tensor is
in general not symmetric.
7) These are the Cartesian components of the left stretch tensor V, as computed by
the AUI from the deformation gradients output by ADINA. See Section 3.1,
equation 3.1-5 and Figure 3.1-2 for the definition and meaning of V.
MAX_STRETCH = SIGP1(<LSTRETCH-XX>, <LSTRETCH-YY>,

<LSTRETCH-ZZ>, <LSTRETCH-XY>,
<LSTRETCH-XZ>, <LSTRETCH-YZ>)
The left stretch tensor is also used in the AUI for plotting the stretches as element
vectors.
8) These are the Cartesian components of the right stretch tensor U, as computed by
the AUI from the deformation gradients output by ADINA. See Section 3.1,
equation 3.1-1 for the definition and meaning of U.
MAX_STRETCH = SIGP1(<RSTRETCH-XX>, <RSTRETCH-YY>,

<RSTRETCH-ZZ>, <RSTRETCH-XY>,
<RSTRETCH-XZ>, <RSTRETCH-YZ>)
9) In 2-D analysis, the BB, AB and CC components are defined; in 3-D analysis, the
BB, CC, AB, AC, BC components are defined.
10) These are the maximum, intermediate and minimum principal logarithmic
strains. They are available in large strain analysis.

11) ACCUM_EFF_CREEP_STRAIN is defined in Section 3.6.3.
12) These variables are used with the SMA material defined in Section 3.14.
13) These are the maximum intermediate and minimum principal strains, as
computed from variables STRAIN-II, STRAIN-IJ, etc.
14) These variables apply to the data fitted concrete material model. They are the
strain values at the time of crack formation along the normal directions of the crack
planes.
Table 13.1-9: Force variables

AVERAGE_BENDING_MOMENT Section field 1,6
AVERAGE_MEMBRANE_FORCE Section field 2,6
AVERAGE_SHEAR_FORCE Section field 6
AXIAL_FORCE Section field
BENDING_MOMENT-(RB,SB,RBSB) Section field 3
BENDING_MOMENT-(ST) Section field
BENDING_MOMENT-(XL,YL,XLYL) Element field 4
CONTACT_FORCE-(123) Node discrete 15
CONTACT_FORCE-(ABC) Node discrete
CONTACT_FORCE-(XYZ) Node discrete
DAMPING_FORCE Element field
EFFECTIVE_BENDING_MOMENT Section field
EFFECTIVE_MEMBRANE_FORCE Section field
ELASTIC_FORCE Element field
FORCE-(ABC) Element field
FORCE-(RST) Element field
FORCE-(XYZ) Element field
FORCE_RST_MAGNITUDE Element field 11
MAX_PRINCIPAL_BENDING_MOMENT Section field 6, 7
MAX_PRINCIPAL_MEMBRANE_FORCE Section field 6, 8
MEMBRANE_FORCE-(RB,SB,RBSB) Section field 3
MEMBRANE_FORCE-(XL,YL,XLYL) Element field 4
MIN_PRINCIPAL_BENDING_MOMENT Section field 6, 9
MIN_PRINCIPAL_MEMBRANE_FORCE Section field 6, 10
MOMENT-(ABC) Element field

MOMENT-(RST) Element field

MOMENT-(XYZ) Element field
MOMENT_RST_MAGNITUDE Element field 11
NODAL_FORCE-(RST) El local node 5
NODAL_FORCE-(XYZ) El local node 5, 14
NODAL_FORCE_RST_MAGNITUDE El local node 11
NODAL_MOMENT-(RST) El local node 5
NODAL_MOMENT-(XYZ) El local node 5, 14
NODAL_MOMENT_RST_MAGNITUDE El local node 11
RESTRAINING_FORCE-(XYZ) Node discrete 12
SHEAR_FORCE-(RB,SB) Section field 3
SOLITARY_CONTACT_FORCE-(ABC) Node discrete
SOLITARY_CONTACT_FORCE-(XYZ) Node discrete
TORSIONAL_MOMENT Section field
UPLIFTING_FORCE-(XYZ) Node discrete 13
YIELD_AXIAL_FORCE Section field
YIELD_BENDING_MOMENT-(ST) Section field
YIELD_TORSIONAL_MOMENT Section field
( )
1) This is evaluated by the AUI as 0.5 M r r M s s , which is an invariant (does not
depend on the directions of r and s .
( )
2) This is evaluated by the AUI as 0.5 Rr r Rs s , which is an invariant (does not
depend on the directions of r and s .
3) RB denotes the r direction, SB denotes the s direction.
4) XL denotes the element local x direction, YL denotes the element local y

direction.
5) These are the consistent nodal point forces corresponding to the stresses within the
element. The AUI does not transform these from the coordinate system in which
they are output by ADINA.
6) These are computed by the AUI from the data output by ADINA.
M s s Mr s
7) This is evaluated by the AUI as the maximum principal value of .
Mr s M r r

Rr r Rr s
8) This is evaluated by the AUI as the maximum principal value of .
Rr s Rs s
M s s Mr s
9) This is evaluated by the AUI as the minimum principal value of .
Mr s M r r
Rr r Rr s
10) This is evaluated by the AUI as the minimum principal value of .
Rr s Rs s
11) Each of these variables is the square root of the sum of the squares of the
components, as computed by the AUI.
12) RESTRAINING_FORCE-(XYZ) is the restraining forces on the nodes of the

drawbead segments.
13) UPLIFTING_FORCE-(XYZ) is the uplifting forces on the nodes of the

drawbead segments.
14) For 2D, 3D, plate, shell and spring elements, these variables are output in the
coordinate system of the node to which the element is attached (global system or
skew system). For example, if a 3D element is attached to nodes with skew systems,
variable NODAL_FORCE-X is the nodal force in the skew A direction.
15) These are the components of the contact forces, transformed by the AUI into a
user-specified coordinate system. More information is given at the end of this
section.
Table 13.1-10: Traction variables

NODAL_NORMAL_TRACTION Node field 6, 7
NODAL_NORMAL_TRACTION-(123) Node field 6,10
NODAL_NORMAL_TRACTION-(XYZ) Node field 6
NODAL_TANGENTIAL_TRACTION Node field 8, 9
NODAL_TANGENTIAL_TRACTION-(123) Node field 8,10
NODAL_TANGENTIAL_TRACTION-(XYZ) Node field 8
NORMAL_TRACTION Contact 1

NORMAL_TRACTION-(XYZ) Contact
RESTRAINING_TRACTION-(XYZ) Drawbead segment 4
SURFACE_TRACTION-(XYZ) Mesh surface 2
TANGENTIAL_TRACTION Contact 3
TANGENTIAL_TRACTION-(XYZ) Contact
1) This is the magnitude of the normal traction vector.
2) These are computed by the AUI as predefined resultants using the formulas
SURFACE_TRACTION-X = <STRESS-XX>*<SURFACE_NORMAL-X> +
<STRESS-XY>*<SURFACE_NORMAL-Y> +
<STRESS-XZ>*<SURFACE_NORMAL-Z>
SURFACE_TRACTION-Y = <STRESS-XY>*<SURFACE_NORMAL-X> +
<STRESS-YY>*<SURFACE_NORMAL-Y> +
<STRESS-YZ>*<SURFACE_NORMAL-Z>
SURFACE_TRACTION-Z = <STRESS-XZ>*<SURFACE_NORMAL-X> +
<STRESS-YZ>*<SURFACE_NORMAL-Y> +
<STRESS-ZZ>*<SURFACE_NORMAL-Z>
See Table 13.1-17 for the definitions of the surface normal variables.
3) This is the magnitude of the tangential traction vector.
4) RESTRAINING_TRACTION-(XYZ) is the restraining tractions on the drawbead

segments.
5) UPLIFTING_TRACTION-(XYZ) is the uplifting tractions on the drawbead

segments.
6) These variables are output when the new contact segments are used.
7) NODAL_NORMAL_TRACTION is the magnitude of the normal traction.
8) These variables are output when the new contact segments are used in frictional
contact.
9) NODAL_TANGENTIAL_TRACTION is the magnitude of the tangential traction.
10) These are the components of the nodal normal and tangential tractions,
transformed by the AUI into a user-specified coordinate system. More information is
given at the end of this section.

Table 13.1-11: Reaction variables

(123)-MOMENT_REACTION Node discrete 1
(ABC)-MOMENT_REACTION Node discrete 2
(V1,V2)-MOMENT_REACTION Node discrete 2
(123)-MOMENT_REACTION Node discrete 1
(XYZ)-MOMENT_REACTION Node discrete 2
(123)-MOMENTUM_FLUX_ELEDGE Element face 1
(XYZ)-MOMENTUM_FLUX_ELEDGE Element face 4
(123)-MOMENTUM_FLUX_ELFACE Element face 1
(XYZ)-MOMENTUM_FLUX_ELFACE Element face 5
(ABC)-REACTION Node discrete 2
(XYZ)-REACTION Node discrete 2
(123)-REACTION_ELEDGE Element face 1
(XYZ)-REACTION_ELEDGE Element face 4
(123)-REACTION_ELFACE Element face 1
(XYZ)-REACTION_ELFACE Element face 5
REACTION_MAGNITUDE Node discrete 3
1) These are the components of the reactions, moment reactions, etc., transformed by
the AUI into a user-specified coordinate system. More information is given at the
end of this section.
2) Reactions are only output for those nodes with fixities. Reactions that are
numerically zero might not be output by ADINA.
3) REACTION_MAGNITUDE is the square root of the sum of the squares of the

reaction components, as computed by the AUI.
4) These are the components of momentum flux output on ADINA-F element edges
(2D analysis). On boundaries with no flow, momentum flux is equal to reaction.
5) These are the components of momentum flux output on ADINA-F element faces
(3D analysis). On boundaries with no flow, momentum flux is equal to reaction.
Table 13.1-12: Flux variables


AXIAL_HEAT_FLUX Element field

HEAT_FLUX_SURFACE Mesh surface 1
HEAT_FLUX-(123) Element field 7
HEAT_FLUX-(XYZ) Element field
HEAT_FLUX_MAGNITUDE Element field 6
KINETIC_ENERGY_FLUX_SURFACE Mesh surface 2
MASS_FLUX_ELEDGE Element face 8
MASS_FLUX_ELFACE Element face 9
MASS_FLUX_SURFACE Mesh surface 3
MOMENTUM_FLUX-(XYZ)_SURFACE Mesh surface 4
RADIATION_HEAT_FLUX-(XYZ) Radiation field
SEEPAGE_FLUX-(123) Element field 7
SEEPAGE_FLUX-(XYZ) Element field
SEEPAGE_FLUX_MAGNITUDE Element field 6
VOLUME_FLUX_SURFACE Mesh surface 5
1) This is computed by the AUI as a predefined resultant using the formula
HEAT_FLUX_SURFACE = <HEAT_FLUX-X>*<SURFACE_NORMAL-X> +
<HEAT_FLUX-Y>*<SURFACE_NORMAL-Y> +
<HEAT_FLUX-Z>*<SURFACE_NORMAL-Z>
KINETIC_ENERGY_FLUX_SURFACE =
KINETIC_ENERGY_DENSITY*VOLUME_FLUX_SURFACE
MASS_FLUX_SURFACE = ELEMENT_DENSITY*
(<X-VELOCITY>*<SURFACE_NORMAL-X> +
<Y-VELOCITY>*<SURFACE_NORMAL-Y> +
<Z-VELOCITY>*<SURFACE_NORMAL-Z>)
See Table 13.1-17 for the definitions of the surface normal variables. You must
define variable ELEMENT_DENSITY before using variable
MASS_FLUX_SURFACE.

MOMENTUM_FLUX-X_SURFACE =
<X-VELOCITY>*<MASS_FLUX_SURFACE>
MOMENTUM_FLUX-Y_SURFACE =
<Y-VELOCITY>*<MASS_FLUX_SURFACE>
MOMENTUM_FLUX-Z_SURFACE =
<Z-VELOCITY>*<MASS_FLUX_SURFACE>
You must define variable ELEMENT_DENSITY before using these variables. See
Table 13.1-17 for the definitions of the surface normal variables.
VOLUME_FLUX_SURFACE = <X-VELOCITY>*<SURFACE_NORMAL-X> +
<Y-VELOCITY>*<SURFACE_NORMAL-Y> +
<Z-VELOCITY>*<SURFACE_NORMAL-Z>
6) Each of these variables is the square root of the sum of the squares of their
components, as computed by the AUI.
7) These are the components of the fluxes, transformed by the AUI into a user-
specified coordinate system. More information is given at the end of this section.
8) This is the mass flux output at ADINA-F element edges (2D analysis). The mass
flux is positive when the mass flux is in the same direction as the element edge
inwards normal.
9) This is the mass flux output at ADINA-F element faces (3D analysis). The mass
flux is positive when the mass flux is in the same direction as the element face
inwards normal.
Table 13.1-13: Electric field variables

ELECTRIC_POTENTIAL Node field
CURRENT_DENSITY-(123) Element field 2
CURRENT_DENSITY-(XYZ) Element field
CURRENT_DENSITY_MAGNITUDE Element field 1

HEAT_DISSIPATION Element field

HEAT_DISSIPATION_RATE Element field
TOTAL_HEAT_DISSIPATION Loc-independent
TOTAL_HEAT_DISSIPATION_RATE Loc-independent
1) CURRENT_DENSITY_MAGNITUDE is the square root of the sum of the squares

of the current density components, as computed by the AUI.
2) These are the components of the current density, transformed by the AUI into a
user-specified coordinate system. More information is given at the end of this
section.
Table 13.1-14: Fracture variables

CRACKED_SURFACE Virtual shift 1
CRACK_BOUNDARY_FLAG Node discrete 2
CRACK_LENGTH Loc-independent
DYNAMIC_CORRECTION Virtual shift 3
J-PARAMETER_1 Virtual shift 3,4
LINE_CONTOUR_NUMBER Line contour 7
LINE_J-INTEGRAL_AVERAGE Line contour 7,8
LINE_J-INTEGRAL_DIFFERENCE Line contour 7
PRESSURE_CORRECTION Virtual shift 3
TEMPERATURE_CORRECTION Virtual shift 3
VIRTUAL_CRACK_SURFACE Virtual shift 3,9
VIRTUAL_SHIFT_NUMBER Virtual shift 3
VIRTUAL_SHIFT_*_PART_FACTOR Node field 3,10
(XYZ)-VIRTUAL_SHIFT Virtual shift 3,11
1) This variable is used to plot the cracked surface in 3-D crack propagation analysis
(ADINA fracture mechanics analysis). It is equal to 1.0 if the node is on the crack
front or has been released; otherwise it is not defined. This variable can be plotted in
a band plot to shade the area that has cracked during the analysis.
2) This variable indicates the current boundary condition applied to a node on a crack
propagation surface (used in ADINA fracture mechanics analysis). It has the
following values: 0=node is released, 1=node is on the crack front and is fixed,

2=node is on the crack front and is constrained, 11=node is unreleased and is fixed,
12=node is unreleased and is constrained.
The crack boundary flag is not defined on nodes that are not on the crack
propagation surface.
3) These variables are used in ADINA fracture mechanics analysis, virtual crack
extension method, see Chapter 10.
4) This value of the J-integral is obsolete.
5) This value of the J-integral does not include temperature, pressure or dynamic
corrections.
6) This value of the J-integral includes temperature, pressure and dynamic

corrections.
7) These variables are used in ADINA fracture mechanics analysis, line contour
method.
8) This is the average of the J-integrals for the outer and inner paths within elements
that use 22 Gauss integration.
9) This is the crack surface area.
10) This variable gives, for each node, the amount of participation in the virtual shift,
from "undefined" (node is not included in virtual shift) to 0.0 (node is included, but
does not shift) to 1.0 (node is included and shifts completely). It can be used in a
band plot to show the virtual shift.
11) These are the components of the virtual shift vector.
Table 13.1-15: Failure criterion variables

CAM-CLAY_FLAG Element field 14
CAP_LOCATION Element field
CRACK_FLAG Element field 1
DF-CONCRETE_FLAG Element field 17
DRUCKER-PRAGER_FLAG Element field 16
EFFECTIVE_STRESS_RATIO Element field

FAILURE_CRITERION Element field 2,3

FAILURE_CRITERION_COMP-FIBER Element field 2,3
FAILURE_CRITERION_COMP-MATRIX Element field 2,3
FAILURE_CRITERION_SURFACE-(1234) Element field 2,3
FAILURE_CRITERION_TENS-FIBER Element field 2,3
FAILURE_CRITERION_TENS-MATRIX Element field 2,3
FAILURE_FLAG Element field 2,4
FAILURE_FLAG_COMP-FIBER Element field 2,5
FAILURE_FLAG_COMP-MATRIX Element field 2,5
FAILURE_FLAG_COMPRESSION-(ABC) Element field 2,5
FAILURE_FLAG_SHEAR-(AB,AC,BC) Element field 2,5
FAILURE_FLAG_SURFACE-(1234) Element field 2,5
FAILURE_FLAG_TENS-FIBER Element field 2,5
FAILURE_FLAG_TENS-MATRIX Element field 2,5
FAILURE_FLAG_TENSION-(ABC) Element field 2,5
FAILURE_SURFACE_FUNCTION Element field 10
GASKET_DEFORMATION_MODE Element field 11
GASKET_STATUS Element field 12
MOHR-COULOMB_FLAG Element field 15
NUMBER_OF_CRACKS Element field 6
PLASTIC_FLAG Element field 7
PLASTIC_FLAG_2 Element field 8
PLASTIC_FLAG_AXIAL Section field
PLASTIC_FLAG_BENDING-(ST) Section field
PLASTIC_FLAG_TORSION Section field
SMA_FLAG Element field 13
SPECIFIC_VOLUME Element field
VOID_RATIO Element field
VOID_VOLUME_FRACTION Element field
WRINKLE_FLAG Element field 9
YIELD_FUNCTION Element field 10
YIELD_SURFACE_DIAMETER_P Element field
1) This is a three-digit integer (P1)(P2)(P3) interpreted as follows. Each digit

represents a principal stress coordinate direction. If the digit equals 0 (zero), there is
no crack; if the digit equals 1, there is an open crack; if the digit equals 2, there is a
closed crack. If the crack flag equals 333, the material is crushed.
The related variable NUMBER_OF_CRACKS returns the number of cracks at the
point, as computed from this variable.

2) The failure variables are used in analyses employing the anisotropic failure criteria
within shell element groups, see Section 2.7.8.
3) Each failure criterion variable stores the value of the failure criterion used in the
analysis.
4) The failure flag has the following values: 0=material at the point considered has
not failed, 1=material at the point considered has failed.
5) All of the failure flags with a qualifier have the following values: 0=material at the
point considered has not failed in that mode, 1=material at the point considered has
failed in that mode.
6) This variable returns the number of cracks at the point, as computed by the AUI
from the variable CRACK_FLAG. If the material is crushed, the value of
CRACK_FLAG is 4.
7) The plastic flag has the following values: 1=elastic conditions, 2=plastic
conditions.
For the Mohr-Coulomb material, the plastic flag can have the following values:
1=elastic conditions; 2=Mohr-Coulomb plastic deformation; 3=plastic deformation at
the yield surface vertices (Drucker-Prager yielding); 4=tension cutoff. Note that, for
the Mohr-Coulomb material, variable MOHR-COULOMB_FLAG should be used
instead of PLASTIC_FLAG. MOHR-COULOMB_FLAG has the same numerical
values as PLASTIC_FLAG, and the AUI knows the allowed values of MOHR-
COULOMB_FLAG.
8) This variable is used for the Drucker-Prager material model. It has the following
values: 1=elastic conditions, 2=Drucker-Prager yielding, 3=cap yielding, 4=vertex
yielding, 5=tension cutoff yielding. Note that, for the Drucker-Prager material,
variable DRUCKER-PRAGER_FLAG should be used instead of PLASTIC_FLAG_2.
DRUCKER-PRAGER_FLAG has the same numerical values as PLASTIC_FLAG_2,
and the AUI knows the allowed values of DRUCKER-PRAGER_FLAG.
9) This variable has the following values: 0=no wrinkling, 1=wrinkling in one
direction, 2=wrinkling in two directions.
10) This variable is the value of the yield function of failure surface function.
11) GASKET_DEFORMATION_MODE has the following values: 1=elastic, 2=plastic,

3=rupture.

12) GASKET_STATUS has the following values: 1=open, 2=closed, 3=sealed,

4=leaked, 5=crushed.
13) SMA_FLAG is used with the SMA material. It has the following values: 1=elastic
state; 2=martensite transformation; 3=austenite transformation.
14) CAM-CLAY_FLAG is used with the Cam-clay material. It has the following
values: 1=elastic state, 2=hardening, 3=softening, 4=critical.
15) MOHR-COULOMB_FLAG is used with the Mohr-Coulomb material. It has the

following values: 1=elastic state, 2=Mohr-Coulomb yielding, 3=vertex yielding
(Drucker-Prager yielding), 4=tension cutoff yielding.
16) DRUCKER-PRAGER_FLAG is used with the Drucker-Prager material. It has the

following values: 1=elastic state, 2=Drucker-Prager yielding, 3=cap yielding,
4=vertex yielding, 5=tension cutoff yielding.
17) DF-CONCRETE_FLAG is used with the data fitted concrete material model. It
has the following values: 1 = elastic unloading; 2 = elastic reloading; 3 = nonlinear
loading; 4 = one-crack; 5 = two-crack; 6 = fully cracked; and 7 = crushed.
Table 13.1-16: Eigenvector variables

(123)-EIGENVECTOR Node field 1
(ABC)-EIGENVECTOR Node field
(XYZ)-EIGENVECTOR Node field
(123)-EIGENVECTOR_ROTATION Node field 1
(ABC)-EIGENVECTOR_ROTATION Node field
(V1,V2)-EIGENVECTOR_ROTATION Node field
(XYZ)-EIGENVECTOR_ROTATION Node field
EIGENVECTOR_MAGNITUDE Node field 2
FLUID_P0_EIGENVECTOR Node field
FLUID_POTENTIAL_EIGENVECTOR Node field
OVALIZATION_EIGENVECTOR-(123456) Node field
PORE_PRESSURE_EIGENVECTOR Node field
THERMAL_EIGENVECTOR Node field
1) These are the components of the eigenvectors transformed by the AUI into a user-

specified coordinate system. More information is given at the end of this section.
2) EIGENVECTOR_MAGNITUDE is the square root of the sum of the squares of the

eigenvector components, as computed by the AUI.
Table 13.1-17: Prescribed load variables

(ABC)-PRESCRIBED_ACCELERATION Node field
(XYZ)-PRESCRIBED_ACCELERATION Node field
(ABC)-PRESCRIBED_ANG_ACCELERATION Node field
(V1,V2)-PRESCRIBED_ANG_ACCELERATION Node field
(XYZ)-PRESCRIBED_ANG_ACCELERATION Node field
PRESCRIBED_CONC_HEAT_FLOW Node field
PRESCRIBED_CONC_SEEPAGE_FLOW Node field
PRESCRIBED_CONV_TEMPERATURE Node field
(ABC)-PRESCRIBED_DISPLACEMENT Node field
(XYZ)-PRESCRIBED_DISPLACEMENT Node field
(ABC)-PRESCRIBED_FLUID_TRACTION Node field
(XYZ)-PRESCRIBED_FLUID_TRACTION Node field
PRESCRIBED_FLUID_P0 Node field
PRESCRIBED_FLUID_POTENTIAL Node field
(ABC)-PRESCRIBED_FORCE Node field
(XYZ)-PRESCRIBED_FORCE Node field
PRESCRIBED_MASS_RATIO_* Node field
(ABC)-PRESCRIBED_MOMENT Node field
(V1,V2)-PRESCRIBED_MOMENT Node field
(XYZ)-PRESCRIBED_MOMENT Node field
PRESCRIBED_OVALIZATION-(123456) Node field
PRESCRIBED_PRESSURE Node field
PRESCRIBED_RAD_TEMPERATURE Node field
(ABC)-PRESCRIBED_ROTATION Node field
(V1,V2)-PRESCRIBED_ROTATION Node field
(XYZ)-PRESCRIBED_ROTATION Node field
PRESCRIBED_TEMPERATURE Node field
PRESCRIBED_TOTAL_HEAD Node field
(ABC)-PRESCRIBED_VELOCITY Node field
(XYZ)-PRESCRIBED_VELOCITY Node field
PRESCRIBED_WARPING-(123456) Node field

Table 13.1-18: Coordinate variables

AREA_ELEDGE Element face 11
AREA_ELFACE Element face 12
AREA_ORIGINAL_ELEDGE Element face 11
AREA_ORIGINAL_ELFACE Element face 12
(XYZ)-COORDINATE Coordinate 1
(XYZ)-COORDINATE_INCREMENT Coordinate 2
(XYZ)-ORIGINAL Coordinate 7
DIRECTOR_VECTOR-(XYZ)_CURRENT Element local node field 9
DIRECTOR_VECTOR-(XYZ)_ORIGINAL Element local node field 10
DISTANCE Coordinate 3
DISTANCE_POSITION Coordinate 3
(XYZ)-LEVER Coordinate 4
(XYZ)-POSITION Coordinate 5
SURFACE_NORMAL-(XYZ) Coordinate 6
SURFACE_NORMAL-(XYZ)_ORIGINAL Coordinate 8
1) These are the initial coordinates (coordinates of the original configuration of the
model), when used with responses of type load-step. These are the reference
coordinates (coordinates of the reference configuration), when used with responses of
type mode-shape, response-spectrum, harmonic and random (recall that these
responses include a reference time).
2) These give the changes in the X, Y and Z coordinates between the original
configuration and the current configuration. They are used when the coordinates of
the model change during the analysis, for example, in fracture mechanics crack
propagation analysis.
3) These variables, when used in a line list or line graph, are defined as follows:
DISTANCE = last computed DISTANCE + (distance between previous point

and current point),
DISTANCE of first point = 0.0
with a similar definition for DISTANCE-POSITION. (The difference between

DISTANCE and DISTANCE-POSITION is that DISTANCE is computed based on

the original configuration of the model and DISTANCE-POSITION is computed

based on the deformed configuration of the model.) These variables are path
dependent and can only be used in a line list or line graph.
4) The lever variables are defined as follows: xL = ( x - xREF )N with similar definitions
for the y and z lever variables. ( x, y ,z ) are either the original or current coordinates,
( xREF, yREF , zREF ) are the coordinates of a reference point and N is an integer greater
than or equal to 1. The choice of original or current coordinate, ( xREF, yREF , zREF )
and N are entered as part of a result control depiction.
The purpose of the lever variables is to make it easier to compute moments about
a reference point.
5) These are the current coordinates (coordinates of the current configuration of the
model), when used with responses of type load-step. These are the reference
coordinates (coordinates of the reference configuration), when used with responses of
type mode-shape, response-spectrum, harmonic and random (recall that these
responses include a reference time).
6) These variables are used to access the face outwards normal of mesh integration or
mesh extreme points.
These variables can also be used when plotting bands (for example, a resultant can
contain these variables when plotting bands). When plotting bands, these variables
refer to the normal at the current time (when the response is of type load-step), or at
the reference time (when the response is of type mode-shape, response-spectrum,
harmonic or random).
7) These are the initial coordinates (coordinates of the original configuration of the
model), when used with any response type.
8) These variables can be used when plotting bands (for example, a resultant can
contain these variables), except when plotting bands on cutting surfaces. When
plotting bands, these variables refer to the normal of the original configuration of the
model.
9) DIRECTOR_VECTOR-(XYZ)_CURRENT are the current components of the shell

nodes director vectors. These variables are available from ADINA at the local nodes
of shell elements.
10) DIRECTOR_VECTOR-(XYZ)_ORIGINAL are the original components of the

shell nodes director vectors. These variables are available from ADINA at the local
nodes of shell elements.

11) These are the current area and original area of an ADINA-F element edge (2D
analysis). For planar analysis, unit thickness is assumed. For axisymmetric analysis,
one radian is assumed.
12) These are the current area and original area of an ADINA-F element face (3D
analysis).
Table 13.1-19: Thickness variables

THICKNESS Element local node field 1, 2
THICKNESS_CHANGE Element local node field 1, 3
THICKNESS_CURRENT Element local node field 1, 4
THICKNESS_ORIGINAL Element local node field 1, 2
THICKNESS_STRAIN Element local node field 1
THINNING Element local node field 1
1) These variables are available from ADINA at the local nodes of ADINA shell
elements.
2) THICKNESS and THICKNESS_ORIGINAL are the shell thicknesses in the

original configuration of the model.
3) THICKNESS_CHANGE = THICKNESS_CURRENT - THICKNESS_ORIGINAL.
4) THICKNESS_CURRENT is the shell thickness in the current configuration.

THICKNESS_CURRENT is different than THICKNESS and
THICKNESS_ORIGINAL only for the top-bottom nodes of transition shell elements
and for large strain shell elements.
5) THICKNESS_STRAIN is the natural logarithm of THICKNESS_CURRENT

divided by THICKNESS_ORIGINAL.
6) THINNING=-(THICKNESS_CHANGE/THICKNESS_ORIGINAL).
Table 13.1-20: Time variables

LAMBDA Loc-independent 1
LOAD_STEP Loc-independent

TIME Loc-independent
TIME_FUNCTION_* Loc-independent
1) LAMBDA is the load vector multiplier in the LDC method.
Table 13.1-21: Frequency/mode variables

BUCKLING_LOAD_FACTOR Loc-independent 1
EIGENVALUE Loc-independent 2
FREQUENCY Loc-independent 3
(XYZ)-MODAL_PARTICIPATION_FACTOR Loc-independent 4
MODE_NUMBER Loc-independent
NATURAL_FREQUENCY Loc-independent 5
EIGENSOLUTION_ERROR_MEASURE Loc-independent 6
1) This is the buckling load factor computed in linearized buckling analysis.
2) This is the eigenvalue as computed by ADINA. The natural frequency is the

square root of the eigenvalue.
3) The dimensions of frequency are cycles/(unit time)
4) These variables refer to modal participation factors for ground motions acting in
the X, Y and Z directions.
5) The dimensions of natural frequency are radians/(unit time).
6) The eigensolution error measure is used to measure the accuracy of the computed
eigenvalues, see Section 6.2.
Table 13.1-22: Miscellaneous variables

CENTER_OF_MASS-(XYZ) Loc-independent 1,2
CENTROID-(XYZ) Loc-independent 1,3
CONTACT_STATE_FLAG Node discrete 4
COORDINATE_SYSTEM_NUMBER Loc-independent 5
FLUID_DOF_TYPE Node discrete 6
INERTIA-P1 Loc-independent 1,7

INERTIA-P1_DIRECTION-(XYZ) Loc-independent 1,8

KINETIC_ENERGY_DENSITY Node field 13
LINE_INDEX_NUMBER Loc-independent 14
MASS Loc-independent 1,15
MOMENT_OF_INERTIA-(XX,YY,ZZ) Loc-independent 1,16
NODAL_CONTACT_GAP Node field 26
NODAL_CONTACT_STATUS Node field 27
NORMAL-(XYZ) Contact
NUMBER_OF_SUBINCREMENTS Element field
PIPE_CROSS-SECTION_AREA Node field 17
PIPE_SKIN_AREA Node field 18
PRODUCT_OF_INERTIA-(XY,XZ,YZ) Loc-independent 1,19
RADIOSITY Node field
REACTION_MOMENT-(XYZ) Node discrete 25
ROTATIONAL_SYSTEM Node discrete 20
SEGMENT_AREA Contact
SEGMENT_LENGTH Contact
STRAIN_ENERGY_DENSITY Element field 23
SURFACE_MOMENT-(XYZ) Mesh surface 24
TRANSLATIONAL_SYSTEM Node discrete 21
VOLUME Loc-independent 1,22
1) These variables are available for ADINA models when mass properties
calculations are requested in ADINA-IN for all element groups in the model. They
represent the given properties for the entire structure (or for that part of the structure
selected in the AUI) and are therefore location-dependent.
2) This is the center of mass in the global coordinate system.
3) This is the centroid (center of volume) in the global coordinate system.
4) The contact state flag has the following values: 0=unknown, 1=free, 2=sticking,
3=sliding. This variable is defined only for node to node contact.
5) This variable stores the number of the coordinate system into which
stresses/strains are transformed, see Section 13.1.3.

6) The fluid dof type has the following values: 0=free velocity potential, 1=fixed,
2=constrained, 3=free velocity potential, 4=free P0 degree of freedom.
7) This is the largest principal inertia, given in a coordinate system with origin at the
center of mass and axes parallel to the global coordinate system.
8) This is the principal axis associated with the largest principal inertia.
9) This is the intermediate principal inertia, given in a coordinate system with origin
at the center of mass and axes parallel to the global coordinate system.
10) This is the principal axis associated with the intermediate principal inertia.
11) This is the smallest principal inertia, given in a coordinate system with origin at
the center of mass and axes parallel to the global coordinate system.
12) This is the principal axis associated with the smallest principal inertia.
KINETIC_ENERGY_DENSITY = 0.5*ELEMENT_DENSITY*
(<X-VELOCITY>**2 + <Y-VELOCITY>**2 + <Z-VELOCITY>**2)
You must define variable ELEMENT_DENSITY before using variable

KINETIC_ENERGY_DENSITY.
14) This is used when listing or graphing results along a line.
15) This is the total mass.
16) These are the moments of inertia, given in a coordinate system with origin at the
center of mass and axes parallel to the global coordinate system. The XX component
of the inertia tensor is equal to the XX moment of inertia.
17) This is the sum (over all pipe elements attached to the node) of the pipe internal
cross-sectional area, based upon the current geometry.
18) This is the sum (over all pipe elements attached to the node) of the internal pipe
skin area, based upon the current geometry.

19) These are the products of inertia, given in a coordinate system with origin at the
center of mass and axes parallel to the global coordinate system. The XY component
of the inertia tensor is the negative of the XY product of inertia.
20) Rotational system has the following values: 0=rotational results are stored in the
global coordinate system, 1=rotational results are stored in a skew system,
2=rotational results are stored in a shell midsurface coordinate system (the shell node
has five degrees of freedom).
21) Translational system has the following values: 0=translational results are stored
in the global coordinate system, 1=translational results are stored in a skew system.
22) This is the total volume.
23) This variable is defined only in ADINA fracture mechanics analysis.
SURFACE_MOMENT-X =
<Y-LEVER>*<SURFACE_TRACTION-Z> -
<Z-LEVER>*<SURFACE_TRACTION-Y>
SURFACE_MOMENT-Y =
<Z-LEVER>*<SURFACE_TRACTION-X> -
<X-LEVER>*<SURFACE_TRACTION-Z>
SURFACE_MOMENT-Z =
<X-LEVER>*<SURFACE_TRACTION-Y> -
<Y-LEVER>*<SURFACE_TRACTION-X>
See Table 13.1-18 for the definitions of the lever variables and Table 13.1-10 for the
definitions of the traction variables.
25) NODAL_CONTACT_GAP is output when the new contact segments are used.
26) NODAL_CONTACT_STATUS is output when the new contact segments are

used. It has the following values: =1, node is dead (on a dead contact segment), =2,
node is open, = 3, node is closed, frictionless contact, = 4, node is closed, slipping
contact, = 5, node is closed, sticking contact.
REACTION_MOMENT-X =

<Y-LEVER>*<Z-REACTION> -<Z-LEVER>*<Y-REACTION>
REACTION_MOMENT-Y =
<Z-LEVER>*<X-REACTION> -<X-LEVER>*<Z-REACTION>
REACTION_MOMENT-Z =
<X-LEVER>*<Y-REACTION> -<Y-LEVER>*<X-REACTION>
See Table 13.1-18 for the definitions of the lever variables and Table 13.1-11 for the
definitions of the reaction variables.
Table 13.1-23: User-defined variables

INT_USER_VARIABLE_* Element field 1
ONE Loc-independent 2
USER_VARIABLE_* Element field 1
1) These variables are used in conjunction with the user-supplied material model.
2) This variable has the constant value 1. It is included so that you can determine
the total area of a domain that is defined as a mesh integration point.
3) Resultants that you define in the AUI are added to the list of user-defined
variables. The variable location type of a resultant is determined by the variables that
define the resultant.

The AUI can transform node field and node discrete results
calculated by ADINA in one coordinate system to another
coordinate system as shown in Table 13.1-24.
Table 13.1-24: Coordinate transformations for node field and

node discrete variables performed by the AUI
Variable prefix
Coordinate system of X,Y,Z A,B,C V1,V2
ADINA/ADINA-F
result
Global 1 1 3
Skew 2 1 3
V1,V2 3 1 1
1) The result is computed without any coordinate transformation.

2) The result is computed using a transformation from the skew
system to the global system.
3) No result is computed.
Displacement and stress transformations: The AUI can

transform displacements, stresses, strains and similar quantities to
user-defined coordinate systems, including cylindrical and
spherical systems.
In order to activate this feature, you set the coordinate system of
the result control dialog box to the number of the coordinate system
that you define. By default, the coordinate system is 0 (the global
Cartesian system).
You access the transformed displacements using variables 1-
DISPLACEMENT, 2-DISPLACEMENT, 3-DISPLACEMENT, and
the transformed stresses using variables STRESS-11, STRESS-
22, STRESS-33, STRESS-12, STRESS-13, STRESS-23.
Other transformed quantities are accessed using similar names.
These variables can be used in any plotting or listing command.
The AUI evaluates the transformed stress and strain variables
from the stresses and strains in the global system, as computed by
the solution program. Therefore the stresses and strains must have
been calculated in the global system by the solution program.

The direction numbers in the variable names 1-

DISPLACEMENT, STRESS-11, STRESS-22, etc. correspond to
coordinate directions as shown in the following table:
System type Direction 1 Direction 2 Direction 3

Cartesian XL YL ZL
Cylindrical R THETA XL
Spherical R THETA PHI
Thus, by default, STRESS-11 = STRESS-XX, STRESS-22 =

STRESS-YY, etc.
You can plot the directions used for displacement and stress
transformations in a mesh plot. This is done from within the
Element Depiction dialog box reached using the Modify Meshplot
icon. This feature is useful for verifying that the displacement and
stress transformation directions are correct.
Please note:
You can smooth the transformed stresses. The smoothing is

done in the global coordinate system and the smoothed stresses
are transformed into the user-specified system.
The transformed off-diagonal strain components (STRAIN-

12, STRAIN-13, STRAIN-23) are engineering quantities, not
tensorial quantities.
You can use the variables STRESS-11, STRESS-22, etc.

within resultants.
13.1.2 Where results can be obtained
Results can be evaluated at various types of locations within the

model.
Location-independent: Used for variables that are location-

independent, see above.
Node: locations at nodes.
Element/layer: locations within elements or element layers. These

can be the integration points, the element local nodes or other
locations within the element.

To specify the element, you need to enter the element group

number and element number. If the element is an ADINA
multilayer shell element, you also need to enter the layer number.
Otherwise you can enter a layer number of 1.
To specify the point within the element (or element layer), there
are several conventions, as follows. In ADINA system elements,
there are two subtypes of element points: integration points and
local nodes. In any given analysis, for any given element, the
results are computed by the solution program at only one of these
subtypes.
In the AUI, you can request that the results be evaluated at the
element points where they were computed by the solution program,
or you can request that the results be extrapolated to other element
points. This latter option is available for some elements as shown
in Table 13.1-25.
Table 13.1-25 Extrapolation of element results

ADINA
Element type Result Dimension Can extrapolate
location for results to
Truss Int. pts. 1-D -1.0 r 1.0
2-D solid or fluid Int. pts. 2-D -1.0 r,s 1.0*
3-D solid or fluid Int. pts. 3-D -1.0 r,s,t 1.0*
Iso-beam (3-D) Int. pts. 3-D -1.0 r,s,t 1.0
Iso-beam (2-D) Int. pts. 2-D -1.0 r,s 1.0
Plate Int. pts. 2-D 0.0 r,s 1.0
Shell Int. pts. 3-D -1.0 r,s,t 1.0
ADINA-T
1-D Int. pts. 1-D -1.0 r 1.0
2-D conduction Int. pts. 2-D -1.0 r,s 1.0*
3-D conduction Int. pts. 3-D -1.0 r,s,t 1.0*
Shell conduction Int. pts. 3-D -1.0 r,s,t 1.0
ADINA-F:
2-D fluid Int. pts. 2-D -1.0 r,s 1.0*
3-D fluid Int. pts. 3-D -1.0 r,s,t 1.0*

*) If the element is triangular or tetrahedral, the element uses

triangular/tetrahedral area coordinates as the isoparametric
coordinates and the coordinates range from 0.0 to 1.0.
(end of table)
You use the Defined By parameter and its related parameters in

the element result point dialog box to select the element point as
follows:
Grid point: This option is available only if results were

calculated at the element integration points by the solution
program.
The element conceptually contains an evenly spaced grid in
the elements' isoparametric system. The grid can have one, two
or three dimensions depending on the element type.
To specify a grid point using a 1-D grid, give a two digit
number AB: digit A gives the total number of grid divisions and
digit B gives the desired grid point.
The isoparametric coordinate corresponding to the grid point
is computed using
B 1
C = Cmin + ( Cmax Cmin )
A 1
where Cmin and Cmax are the minimum and maximum

isoparametric coordinates that the elements can contain. For
example, AB=52 corresponds to r = - for elements in which r
can be between -1 and 1.
To specify a grid point using a 2-D grid, give a four digit
number ABCD: digits A and B give the desired grid point in the
r direction and digits C and D give the desired grid point in the s
direction. For example, ABCD=2121 corresponds to r = -1, s =
-1 and ABCD=3243 corresponds to r = 0, s = 1 3 .
To specify a grid point using a 3-D grid, give a six digit
number ABCDEF: digits A and B give the desired grid point in
the r direction, digits C and D give the desired grid point in the s
direction and digits E and F give the desired grid point in the t
direction.
You can also specify the element centroidal point by giving

a grid value of 0.
Label point: You specify the label point in the element at which
results were computed by the solution program.
If the results were calculated at the integration points, then
the label is a 1, 2, 3 or 4 digit number giving the integration
point as shown in Table 13.1-26.
If the results were calculated at the element local nodes, then
the label is the local node number.
No extrapolation is performed by the AUI when using the
label option.
Table 13.1-26: Label point numbering

ADINA:
Element type Result locations Label number
Truss Int. pts. (INR)
2-D solid or fluid Int. pts. 10(INR) + INS*
3-D solid or fluid Int. pts. 100(INR) + 10(INS) + INT*
Hermitian beam Int. pts. 100(INR) + 10(INS) + INT
Iso-beam (3-D) Int. pts. 100(INR) + 10(INS) + INT
Iso-beam (2-D) Int. pts. 10(INR) + INS
Plate Int. pts. (ITRI)
Shell (rectangular) Int. pts. 100(INR) + 10(INS) + INT
Shell (triangular) Int. pts. 10(ITRI) + (INT)
Pipe Int. pts. 1000(INA) + 10(INB) + INC
Spring ---- 1
General Stress transformation (J)
matrix
ADINA-T:
1-D Int. pts. (INR)
2-D conduction Int. pts. 10(INR) + INS*
3-D conduction Int. pts. 100(INR) + 10(INS) + INT*
Shell (conduction) Int. pts. 100(INR) + 10(INS) + INT
ADINA-F:
2-D fluid Int. pts. 10(INR) + INS*
3-D fluid Int. pts. 100(INR) + 10(INS) + INT*
Notes:

Each quantity in parentheses represents one integer:
(ITRI) integration point number in triangular integration
(ITET) - integration point number in tetrahedral integration
(INR), (INS), (INT) integration point numbers for r, s and t coordinates
(INA), (INB), (INC) integration point numbers for a, b and c coordinates
(J) stress component number
*) If the element is triangular, the label number is (ITRI); if the element is

tetrahedral, the label number is (ITET).
(end of table)
Node: This option is available only if results were calculated at

the element integration points by the solution program.
You enter the global node number of a node connected to the
element.
The results are extrapolated to the point within the element
corresponding to the position of the node in the element. For
example, if the first node of a 2-D element is number 65, then
selecting node 65 is equivalent to selecting the point (r,s) = (1,1)
in the element.
It is necessary to specify the element number when this
option is used because, when the results are unsmoothed, the
results at a node point are different from the different elements
attached to the node.
The use of this option does not, by itself, smooth results
from neighbor-ring elements to the specified node. But if
smoothing is turned on, then any element that contains the
specified node will give the same results (because the smoothed
results are continuous between the elements).
Isoparametric coordinates: This option is available only if

results were calculated at the element integration points by the
solution program.
You directly enter the r, s and t coordinates of the desired

point. If the element is 1-D or 2-D, the unused coordinates are

ignored.
Special considerations for the ADINA shell element: The element

is considered to be a 3-D element for the purpose of specifying
element points. You can specify that the element point lies on the
shell midsurface using the following options:
Node number: Specify the node number of a midsurface node.
Grid location: Specify a grid location of 0, or a grid location in

which the 5th and 6th digits are equal to 1.
When the element point lies on the shell midsurface, when the
results extrapolated from the element integration points are
evaluated at the point, the t coordinate and layer used for the
evaluation are selected by the parameters of the result control
depiction.
Element section: locations within element sections (ADINA only,

beam and shell elements). These can be the section integration
points, the element local nodes or other locations within the
element section.
Two types of elements can have results output at element
sections. The Hermitian beam element can have results output on
the neutral line of the element; the neutral line is considered to be
the element section. The ADINA shell element can have results
output on the neutral surface of the element; the neutral surface is
considered to be the element section.
Element section results are very similar to element result points.
You can request that results be evaluated at the locations at which
they were computed by ADINA or that the results be extrapolated
to other locations using the grid, node and isoparametric coordinate
parameters. It is important to remember the correct dimension of
the element section: the section is a line for beam elements and a
surface for shell elements. Hence, for example, a grid location is of
the form AB for beam element section points and of the form
ABCD for shell element section points.
Note that it is possible to list shell element stresses at shell
element section result points if shell midsurface calculations are
activated in the result control depiction.

Contact segments: (ADINA only). Locations in contact segments.

A contact surface contains one or contact segments, each of which
contains a single location where ADINA computes results.
When you request the coordinates of a contact segment, the
coordinates of the center of the segment are used.
Within a contact surface, contact segments are numbered
sequentially starting from 1. Hence, to specify a contact segment
result point location, you must specify three numbers: the contact
group number, the contact surface number and the contact segment
number.
Note that no results are available for contact segments on target
contact surfaces. Those contact results that are output at nodal
points (for example, concentrated contact forces) are considered by
the AUI to be associated with node points.
Draw bead segments: (ADINA only). Locations in draw bead

segments. A contact surface optionally contains one or more
drawbead segments, each of which contains a single location where
ADINA computes results.
Within a contact surface, draw bead segments are numbered
sequentially starting from 1. Hence, to specify a draw bead
segment result point location, you must specify three numbers: the
contact group number, the contact surface number and the draw
bead segment number.
Radiosity segments: (ADINA-T only). Locations in radiosity

segments. A radiosity surface contains one or more radiosity
segments, each of which contains integration points at which
ADINA-T computes results.
The options for entering locations within radiosity segments are
very similar to those used for entering locations within element
layers. You can request that results be evaluated at the locations
where they were computed by ADINA-T or that the results be
extrapolated to other locations within the segment using the grid,
node and isoparametric coordinate parameters. It is important to
remember the correct dimensions of the radiosity segment. Hence,
for example, a grid location is of the form AB for segments within
radiosity lines and of the form ABCD for segments within radiosity
surfaces.

Virtual shifts: (ADINA only). In order for you to access the

results associated with virtual shifts in ADINA fracture mechanics
analysis, the AUI allows you to define a result point at a virtual
shift. (In this case, the virtual shift does not occupy a point in
space.)
Line contours: (ADINA only). In order for you to access the

results associated with J-integral line contours in ADINA fracture
mechanics analysis, the AUI allows you to define a result point at a
J-integral line contour. (In this case, the line contour does not
occupy a point in space.)
Element faces: (ADINA-F only). Results output at individual

element edges (2D analysis) and element faces (3D analysis) by
ADINA-F are generally most useful when they are summed on
model boundaries, or other selected areas of the model. The AUI
uses the element edge-set and element face-set concepts in order to
group the element edges and faces. For example, in order to
calculate the mass flux over a selected area of the model,
1) Create an element edge-set (2D analysis) or element face-set

(3D analysis) corresponding to the selected area (from the
command-line, use the commands ELEDGESET or
ELFACESET). In many cases ADINA-F automatically creates
useful element edge-sets and element face-sets.
2) Create a model result point of type element edge-set or element

face-set (from the command-line, use the commands
ELESETPOINT or ELFSETPOINT).
3) List or plot the mass flux at the model result point. The output
mass flux is the sum of the mass fluxes at all of the element
edges or faces of the element face-set or edge-set.
You can also list the individual mass fluxes at the element edges or
faces of an element edge-set or element face-set. For example,
given an element edge-set or face-set:
1) Create a model result line of type element edge-set or element

face-set (from the command-line, use the commands
ELESETLINE or ELFSETLINE).

2) List the mass flux on the model result line.
Mesh integration points: If you want to integrate a quantity over

the faces of a mesh plot, define a mesh integration point.
The mesh integration point consists of: 1) a domain over which
the integration is to take place and 2) the rules for performing the
integration.
Domain: The domain is described by the name of a mesh plot

and the name of a zone as follows. If the mesh plot does not
have a cutting surface, the domain is the external faces of the
meshplot (the skin of the meshplot). If the mesh plot has a
cutting surface, the domain is the cutting surface faces.
Faces can be, for example, the boundaries of 3-D elements,
the midsurfaces of shell or plate elements, or the interiors of 2-
D elements. They cannot be the boundaries of shell or 2-D
elements.
Another way to think of the domain is in terms of a band plot
on the mesh plot. The domain is that area of the mesh plot that
is covered by the bands.
Use the zone name to restrict the domain. A face is in the
domain only if its associated element is in the specified zone.
Rules: The rules are the integration type, a multiplying factor,

an integration order and the configuration over which the
integration is performed. Let z be the integrand and S be the
domain. Then the integration types are
Integral: zdS
S
zdS
Averaged: zav = S
z
2
dS
Mean-square: zms = S

z
2
dS
S
Root-mean-square:
S
( z z ) dS
2
av
S
Variance:
S
( z z ) dS
2
av
S
Standard deviation:
S
2
zz
S zav av dS
Relative variance:
S
2
zz
S zav av dS
Relative standard deviation: (also
S
referred to as the coefficient of variation)
Use the multiplying factor to multiply the result after the

indicated integration is performed. Use the integration order to
indicate the numerical integration order over each element face.
Use the configuration to indicate whether to perform the integration
in the original configuration of the model or in the current
deformed configuration.
Please note that you do not specify the integrand within the
mesh integration point. You specify the integrand when you
reference the mesh integration point within another dialog box.
Once you have defined a mesh integration point, you can use it
within any command that accepts a model result point. For
example, you can plot the time history of an integrated quantity
using GraphResponse Curve (Model Point).
There are certain variables and predefined resultants that can
only be used in conjunction with mesh integration (or mesh

extreme) points. These are all based upon the face outwards
normal. The face outwards normal can be accessed through
variables SURFACE_NORMAL-X, SURFACE_NORMAL-Y,
SURFACE_NORMAL-Z. Useful quantities for which predefined
resultants are defined are VOLUME_FLUX_SURFACE,
MASS_FLUX_SURFACE, SURFACE_TRACTION-X,
SURFACE_TRACTION-Y, SURFACE_TRACTION-Z,
MOMENTUM_FLUX-X_SURFACE, MOMENTUM_FLUX-
Y_SURFACE, MOMENTUM_FLUX-Z_SURFACE,
KINETIC_ENERGY_DENSITY,
KINETIC_ENERGY_FLUX_SURFACE,
HEAT_FLUX_SURFACE. Any of these variables can be used
along with a mesh integration point. For example, you can
determine the volume flux of flow through a pipe by defining a
mesh integration point corresponding to a cutting surface, then
graphing or listing the variable VOLUME_FLUX at the mesh
integration point.
You can also obtain the surface area of the mesh integration
point by evaluating variable ONE at the point (variable ONE is
defined to have the value 1.0).
You can also obtain the approximate volume of the mesh by use
of the divergence theorem (also known as Gauss theorem or
Greens theorem), which is:
iA dV = Ain dS
V S
Simply define resultant ADOTDS to be <X-COORDINATE>*

<SURFACE_NORMAL-X> and evaluate it at a mesh integration
point. This works because i(< X-COORDINATE >, 0, 0) = 1 .
Mesh extreme points: If you want to determine the extreme value

of a quantity over the faces of a mesh plot, define a mesh extreme
point.
The mesh extreme point consists of: 1) a domain over which the
search for the extreme value is to take place and 2) the rules for
determining the extreme value.
Domain: The domain is defined in exactly the same way as for a

mesh integration point, see above.

Rules: The rules are the extreme type, the multiplying factor,
the search grid order and the domain configuration. The
extreme types can be absolute maximum, maximum or
minimum. Use the multiplying factor to multiply the result after
the extreme value is found. Use the search grid order to indicate
the number of sampling points within each element face. Use
the domain configuration to indicate whether to search for the
extreme value using the original or current deformed
configuration of the model.
Please note that you do not specify the quantity within the
mesh extreme point. You specify the quantity when you
reference the mesh extreme point within another dialog box.
Once you have defined a mesh extreme point, you can use it
within any command that accepts a model result point. For
example, you can plot the time history of an extreme quantity
using GraphResponse Curve (Model Point).
The variables that can be used for mesh integration points
can also be used for mesh extreme points.
Not all results can be obtained at all result locations. Table

13.1-27 shows the result location types for each type of available
variable. Notice that certain combinations are allowed only when
smoothing is requested (see below).
13.1.3 How results are evaluated (including smoothing)
If a node field variable is requested within an element, the nodal

shape functions are employed to interpolate the variable within the
element.
If an element/layer variable is requested within an element at a

point, the variable is interpolated or extrapolated as follows:
RST interpolation: If the variable represents a floating-point

value, the variable is interpolated or extrapolated using the
nearest integration points and bilinear interpolation. If the
variable represents an integer value, the value of the variable at
the nearest integration point is used.

Table 13.1-27: Locations at which available variables can be

evaluated
Variable location type Result location types
Location-independent All
Node field Node, element/layer, contact
segment, radiation segment
Element/layer field Node (smoothing only),
element/layer, section
Contact segment Contact segment
Drawbead segment Drawbead segment
Section field Node (smoothing only),
section
Radiation field Node (smoothing only),
radiosity segment
Line contour Line contour
Virtual shift Virtual shift
Node discrete Node
Element local node Element/layer (at local nodes
only)
Element local node field Element/layer, section
Coordinate Node, element/layer, section,
contact segment, radiation
segment
Element face Element edge-set or element
face-set
Face interpolation: For 2D elements, this is the same as

centroid interpolation, see below. For 3D elements, the closest
element face is determined for the point. Then the results are
extrapolated using RST interpolation to the center of the closest
element face. Therefore the extrapolated results for all points in
the element that are closest to a given element face are the same
as the extrapolated element face center value.
Centroid interpolation: The results are extrapolated using RST

interpolation to the element centroid. Therefore the
extrapolated results at all points in the element are the same as
the extrapolated centroid value.

Integration point interpolation: The results at the nearest

integration point are used.
The choice of interpolation is made within the result control dialog

box. RST interpolation is the default.
Smoothing
The element/layer field or section results can be smoothed (that

is, made continuous between adjacent elements or element
sections). The smoothing process is now described in detail for
element results; the smoothing process is similar for element
section results.
Smoothing consists of two steps:
1) Smoothing to the node points. For each node,
a) Within each element attached to the node, the element

integration point results are extrapolated to the node point.
The extrapolation process is exactly the same as used when
element integration point results are requested at other points
within the element.
b) The contributions of each element are combined into a

single result, as specified by the type of smoothing:
Averaged: averages
Minimum: minimum result is taken
Maximum: maximum result is taken
Difference: difference between the maximum and

minimum result is taken
Extreme: extreme result (result furthest from 0.0) is

taken
Error: Same as difference, except that only the element

corner nodes are considered and the result is divided by
a reference value that you enter (the default reference
value is 1.0).

2) Extrapolation to the requested point. The extrapolation is

exactly the same as used when nodal point results are requested
within the elements. If the requested point is a nodal point, no
action is taken in this step.
Normally, all elements attached to a node contribute results to

the node. However, you can choose to smooth results using only
selected elements. This is done by specifying a zone name in the
smoothing depiction. In this case, only elements within the zone
contribute results to the node.
This option can be important when, for example, the model has
several materials and you do not want to smooth stresses across the
material interfaces.
The following restrictions for smoothing are important to

remember:
Results from 2-D, 3-D (solid or fluid) and plate elements can
be smoothed. Results from shell elements can be smoothed on
the midsurface of the shell elements if shell element midsurface
calculations are turned on in the result control depiction.
You must have requested that results be saved at the element

integration points during preprocessing.
Results that evaluate as integer values, such as the crack flag

from the concrete model, can be smoothed into non-integer
values, even when the non-integer values have no meaning.
Smoothing results across element groups may not be

physically appropriate, for example, when the elements groups
have different material properties.
Results are smoothed component-by-component without

taking into account possible differences in coordinate systems
between adjacent elements. This is especially important for
results from shell elements.
Control of smoothing is done by choosing a smoothing

depiction, or by defining and using a smoothing depiction. By

default, smoothing is turned off in the AUI.
Plotting results onto midsurfaces of shells: The AUI allows you

to specify which shell layer and t coordinate value to use when
plotting results onto shell midsurfaces. This allows you to plot, for
example, the results from the top of the shells onto the shell
midsurfaces. You can also choose to plot results evaluated at the
shell midsurface onto the shell midsurface. These choices are made
using a result control depiction. By default, results from the top of
the shell are presented on the shell midsurface.
13.1.4 When (for which load step, etc.) are results available
Results available from solution programs of the ADINA system

are
Load step results: Results associated with load steps (time

steps). Each load step has a unique load step number and
solution time.
Mode shape results: Results associated with mode shapes.

Each mode shape has two numbers that define it uniquely, the
mode shape number and the solution time corresponding to the
solution from which the mode shapes were calculated.
Residual results: Results associated with residual (static

correction) responses in response spectrum analysis (see Section
9.1 for theoretical background).
Corresponding to these three types of results, the AUI has three

response types:
Load-step: Allows you to reference results from a specific load

step. A load step is referenced by its solution time.
If you enter a solution time between two solution times for
which results are output by the solution program, the AUI
automatically interpolates the results to the entered solution
time. This is an example of a time transformation discussed in
Section 13.1.
Mode-shape: Allows you to reference results from a specific

mode shape. A mode shape is referenced by its mode shape

number and solution time.
Residual: Allows you to reference residual results. Residual

results are referenced by the ground motion direction and
solution time corresponding to the solution for which the
residual results were obtained.
The AUI includes additional response types used in various

analyses.
Response spectrum: A calculation method used for computing

results in response spectrum analysis. You specify the modal
combination method, loading spectrum, modal damping table
and other information. See Section 9.1 for theoretical
background.
Harmonic: A calculation method used for computing results in

harmonic vibration analysis. You specify the desired harmonic
response (maximum, phase angle, rms amplitude or signed
amplitude), the loading frequency, the loading spectra, the
damping table and other information. See Section 9.3 for
theoretical background.
Random: A calculation method used for computing results in

random vibration analysis. You specify the desired random
response (rms amplitude or power-spectral-density), the loading
frequency (for power-spectral-density only), the loading spectra,
the damping table and other information. See Section 9.4 for
theoretical background.
Response-combination: A calculation method for combining

previously defined responses. These responses can be of any
type except for response-combination and any mixture of the
allowed response types can be used. For each previously
defined response, you can specify a weighting factor and a
combination method.
The AUI uses the following algorithm to evaluate the response-
combination:
accum=0.0

For (each response in the response-combination) {

value=(result for the response) * multiplying factor
if (method = algebraic) {
accum=accum + value
}
else if (method = abs) {
accum=abs(accum) + abs(value)
}
else if (method = srss) {
accum=sqrt(accum**2 + value**2)
}
else if (method = signed 1) {
if (value >= 0.0) {
accum=value + abs(accum)
}
else {
accum=value - abs(accum)
}
else if (method = signed 2) {
if (value >= 0.0) {
accum=accum + abs(value)
}
else {
accum=accum - abs(value)
}
}
}
You might use methods signed 1 and signed 2, for example,

when including a static load case along with the combined
responses for different spatial directions in response spectrum
analysis.
Envelope: At a point in the model, a given result (such as the

value of a stress component) has a time history, and the AUI can
perform a calculation on the time history, for example, the AUI
can obtain the maximum value and the corresponding solution
time. Of course, the magnitude of the maximum value and the
corresponding solution time is different from point to point.
The envelope of the result is a listing of all of the calculated
results; the listing gives the model points, the calculated results

and the corresponding solution times.

To obtain an envelope listing in the AUI, define a response
of type envelope. This response associates a response name
with a calculation method in which several responses are
evaluated, a calculation is made on the responses and one value
is returned. The type of calculated value can be an extreme
value, a time-integral value or a sampling-based value:
Extreme value: The minimum, maximum, absolute value of

the maximum or difference between maximum and
minimum.
t2
Time integral: v(t ) dt

t1
t2
v(t ) dt
t1
Time averaged: vtav =
t2 t1
t2
v(t )
2
dt
t1
Time mean-square: vtms =
t2 t1
Time root-mean-square: vtms
t2
( v(t ) v )
2
tav dt
t1
Time variance: vtvar = = vtms vtav
2
t2 t1
Time standard deviation: vtsd = vtvar
t2 2
v(t ) v
t vtav tav dt v
Time relative variance: vtrvar = 1 = tvar
t2 t1 2
vtav

Time relative standard deviation: vtrsd = vtrvar
N
Sample sum: v
i =1
i
v i
Sample averaged: vsav = i =1
v 2
i
Sample mean-square: vsms = i =1
Sample root-mean-square: vsms
(v vsav )
2
i
Sample variance: vsvar = i =1
= vsms vsav
2
Sample standard deviation: vssd = vsvar
2
vi vsav
N

v

v
= sav = svar
i =1
Sample relative variance: vsrvar 2
N vsav
Sample relative standard deviation: vsrsd = vsrvar
In these formulas, v(t ) is the time history of a solution quantity

and vi is the value of a solution quantity.
There are two ways to select the responses included in the

envelope:
Range: The AUI evaluates the value for a range of load-

steps and returns the extreme value. You can choose the start

and end times, the intermediate steps to sample, and whether

to interpolate results to steps for which results are not saved.
Selected: The AUI evaluates the value at each response

specified in the data input lines of this command and returns
the extreme value. Time-integral values cannot be evaluated
in this case.
The AUI performs the calculation independently at each

requested point in the model.
The AUI includes dialog boxes for each of the response types.
Each dialog box defines a response of the desired response type.
You then use the response in another command when you want to
plot or list the results using that response.
For response types response spectrum, harmonic, random,
response combination, random or envelope, the AUI does not
actually compute the response until the response is referenced by
another command.

13.2 Evaluation of meshes used

Once the solution with a given finite element mesh has been
obtained, you may want to identify whether the mesh used has been
fine enough. This can be achieved in the AUI by plotting bands of
quantities that are the derivatives of the primary element variables.
For example, a band plot of pressures (or effective stress or any

stress variable) can be produced). The unsmoothed or smoothed
pressures can be displayed. Consider that unsmoothed pressures
are plotted. If the stresses are to be continuous in the exact solution
(which may not be the case when there are discontinuities in
material properties), the pressure bands will be reasonably
continuous if the mesh has been fine enough. The breaks in bands
indicate the severity of the stress discontinuities.
ref. T. Sussman and K.J. Bathe, "Studies of finite element

procedures stress band plots and the evaluation of
finite element meshes," Engineering Computations, Vol.
3, No. 3, pp. 178-191, September 1986.
The breaks in the bands are easily seen if the bands are drawn
using two colors that alternate (so that the resulting plot looks like
a photoelastic image). This option is available in the AUI when a
band table of type repeating is requested.
Another error indicator based on jumps in results between

elements when too coarse a mesh has been used is the maximum
difference between unsmoothed results at nodes. This error
indicator is obtained by using a smoothing depiction of type error;
see Section 13.1.3 for a description of the smoothing depictions.
This error measure can be divided by a reference value.
For convenience, the AUI has an Error Plots icon that allows
you to easily plot an error indicator. For example, suppose that a
band plot of unsmoothed pressure is displayed. When the Error
Plots icon is clicked,
1) The maximum pressure in the unsmoothed band plot is taken as

the error reference value.
2) At each node, the maximum difference between unsmoothed

pressures is computed. This maximum difference divided by the
error reference value is the nodal error indicator.
3) The nodal error indicator is then drawn as a band plot.
See the AUI Primer, Problem 13, for an example showing the
evaluation of meshes using band plots.

Index
Index
2 A
2-D solid elements, 15 Acceleration variables, 1257
3-D plane stress, 15 Accumulated effective plastic strain,
axisymmetric, 15 426, 461
element output, 27 Active column solver, 910
formulations, 22 ADINA heat transfer capabilities,
generalized plane strain, 15 1233
mass matrices, 26 ADINA System documentation, 4
material models, 22 ADINA-TMC
numerical integration, 24 Piezoelectric analysis, 1241
plane strain, 15 soil consolidation analysis, 1237
plane stress, 15 Alignment elements, 336
recommendations for use, 33 element output, 378
triangular, 21 Analysis zooming, 1202
nd
2 Piola-Kirchhoff stresses, 391 Analytical rigid targets, 765
Anand material model, 661
3 Arc length method, 929
3-D solid elements, 37 Arrival time, 829
element output, 50 Arruda-Boyce material model, 547
formulation, 46 3-D analysis, 548
mass matrices, 50 axisymmetric analysis, 548
material models, 45 plane strain analysis, 548
numerical integration, 47 plane stress analysis, 548
prisms, 44 selection of material constants, 548
pyramids, 44 ATS method, 827, 923, 941
recommendations for use, 58 dynamic analysis, 926
tetrahedra, 44 low speed dynamics, 922
3D-iterative solver, 914 Automatic step incrementation, 917
3D-shell elements, 225 ATS method, 923
LDC method, 929
5 Axisymmetric shell elements, 162,
164, 172
5 degrees of freedom node, 188 element output, 170, 173
formulations, 171
6 linear, 168
6 degrees of freedom node, 190 mass matrices, 173
material models, 171
nonlinear, 171

Index
numerical integration, 166, 172 stresses, 72

torsional response, 72
B U cross-section, 129
Beam element Wagner effect, 76
cross-section, 65 warping beam, 91
Beam elements, 60 warping degrees of freedom, 159
2-D action, 102 warping function, 82
3-D action, 102 Bergstrom-Boyce viscoelasticity, 588
auxiliary node, 63 BFGS method, 917, 919, 939
box cross-section, 119 Bimoment, 73
coefficient of thermal expansion, Bishear, 73
100 Bolt option, 1207
cross-section geometric properties, Boundary conditions
67 displacement, 825
degrees of freedom, 63 essential, 825
displacements, 69 force, 825
end-release option, 152 moment, 825
fixed-end force correction, 105 natural, 825
forces/moments, 63
C
general cross-section, 111
geometry, 62 Cable elements, 7
I cross-section, 124 Cam-clay material model, 621
kinetic energy, 81 initial stiffness, 624
L cross-section, 134 initial stresses, 623
large displacement formulation, 94 Cauchy stresses, 387, 391
mass matrices, 98 Centers of mass
modeling hints, 157 inertial properties, 1190
moment curvature, 141 Central difference method, 959
nodal point force output, 103 Centrifugal loads, 851
nonlinear elasto-plastic, 100 Centroids
numerical integration, 101 inertial properties, 1190
orientation vector, 63 Cohesive elements, 381
pipe cross-section, 116 element output, 385
rectangular cross-section, 111 Component mode synthesis (CMS)
rigid-end option, 156 method, 1227
section force output, 103 Composite shell elements, 200
shear center, 66 specification of layer thicknesses,
standard beam, 90 201
stiffness matrix, 99 transverse shear stresses, 200
strains, 71 Composite time integration, 949
stress output, 103 Computers supported by ADINA, 2

Index
Concentrated dampers, 953 discretization, 773

Concentrated loads, 830 forces, 759
Concrete material model, 715 initial penetration, 746
compressive failure envelopes, 508 output, 756
data fitted, 513 rigid target method, 739
material behavior after failure, 509 segment method, 739
material failure envelopes, 500 three-dimensional, 727
Poissons ratio in the compressive tractions, 756
region, 512 two-dimensional, 726
post tensile cracking behavior, 509 Contact surfaces extension, 749, 890
postcompression crushing behavior, Contact variables, 1252
511 Convergence criteria, 933
stress-strain relations, 496 contact, 935
temperature effects, 512 energy, 934, 937, 941, 961
tensile failure envelopes, 505 force and moment, 934, 937, 941,
Concrete material models, 493 961
Condensed node, 1142 translation and rotation, 934, 941
Consistent contact surface stiffness, Coordinate variables, 1253, 1279
754 Crack propagation, 1027, 1069
Consolidation analysis, 638 crack growth control, 1071
Constraint equations, 881 definition of the crack propagation
Constraint-function method, 737, 740 surface, 1070
Contact algorithm, 743 modeling of the crack advance,
selection of, 739 1074, 1076, 1077
Contact analysis features, 766 node release-only technique, 1069
Contact birth/death, 754 pressure loading, 1077
Contact body, 765 Crack resistance curves, 1072
Contact compliance, 767 Crack tip blunting, 1058
Contact damping, 770, 783 CRACK-M meshing features, 1083
Contact definition features, 765 Creep coefficients
Contact detection, 768 user-supplied, 485
Contact group properties, 743 Creep laws, 483
Contact oscillations, suppressing, 769 eight-parameter, 483
Contact pairs, 727 eight-parameter with variable
Contact slip, 755 coefficients, 485
Contact surface depth, 745 exponential, 483
Contact surface offsets, 744 LUBBY2, 484
Contact surfaces, 725 LUBBY2 with variable coefficients,
constraint-function method, 737, 486
740 power, 483
convergence, 780 Creep material model, 9, 474, 483

Index
Creep strains, 483 Direct time integration, 334, 948

O.R.N.L. rules for cyclic loading Director vectors, 178
conditions, 487 Disk files, 1197
strain hardening, 487 displacements, 1200
time offset, 490 forces, 1200
Creep-variable material model, 474, pipe internal pressures, 1199
485 temperature gradients, 1200
Curve description material model, 610 temperatures, 1199
cracking, 614 Displacement variables, 1255
cyclic parts, 1157 Displacement-based finite elements,
tension cut-off, 614 17, 39
Curve fitting for rubber and foam Displacement-based fluid elements,
material models, 569 277
Arruda-Boyce model, 573 3-D, 278
eight-chain model, 575 axisymmetric, 278
hyper-foam model, 573 element output, 279
Mooney-Rivlin model, 571 irrotational conditions, 277
Ogden model, 572 planar, 278
Cyclic symmetry analysis, 1153 Distributed loads, 834
periodic symmetry, 1155 beam, 839
restrictions, 1158 deformation dependent, 835
iso-beam, 840
D pipe, 840
Damping, 953, 955 shell, 841
consistent, 953 Draw bead variables, 1252
lumped, 953 Drucker-Prager material model, 616
modal, 952 Dynamic analysis, 948, 959
Rayleigh, 953
E
Strain Energy Proportional, 955
Darcy=s law, 638 Earthquake loading, 952
Data fitted concrete material model, Effective plastic strain, 425
513 Effective-stress-function algorithm,
Deformation gradient tensor 443
elastic, 396 Eigenvalue problems, 893
inelastic, 396 Eigenvector variables, 1277
total, 393 eight-chain material model
Deformation-dependent distributed 3-D analysis, 568
loads, 835 axisymmetric analysis, 568
Deformation-dependent pressure plane strain analysis, 568
loads, 835 plane stress analysis, 568
Determinant search method, 902 selection of material constants, 568

Index
Eight-chain material model, 559 3D-shell, 225

Eight-parameter creep law alignment, 336
variable coefficients, 485 axisymmetric shell, 162
Eight-parameter creep law, 483 cable, 7
Elastic-isotropic material model, 402, cohesive, 381
403 general, 257
Elastic-orthotropic material model, iso-beam, 162
402, 404 pipe, 237
2-D solid elements, 404 plate, 176
3-D solid elements, 409 rebar, 12
fabric, 633 ring, 7
plate elements, 409 shell, 177
shell elements, 410 spring/damper/mass, 257
Electric field variables, 1272 End-release option, 152
Electromagnetic loads, 871 Energy release rate, 1030
Element death upon rupture, 1166 Engineering strains, 387, 390
Element birth/death, 917, 1159 Engineering stresses, 387, 391
Element local node field variables, Envelope response type, 1305
1253 Equilibrium iterations
Element local node variables, 1253 BFGS method, 919
Element locking, 223 full Newton method, 919
Element output for modified Newton method, 918
2-D solid elements, 27 Error indicators, 1309
3-D solid elements, 50 Explicit dynamic analysis, 971
alignment elements, 378 Explicit integration
axisymmetric shell elements, 172 central difference method, 959
cohesive elements, 385 Explicit time integration, 948, 959
displacement-based fluid elements, Exponential creep law, 483
277
general elements, 259 F
iso-beam elements, 173 Fabric material model, 633
nonlinear spring/damper/mass Failure criteria, 206
elements, 266 Hashin failure criterion, 212
pipe elements, 254 maximum strain failure criterion,
potential-based fluid elements, 335 208
shell elements, 214 maximum stress failure criterion,
truss elements, 11 206
Element/layer field variables, 1252 tensor polynomial failure criterion,
Elements 210
2-D solid, 15 Tsai-Hill failure criterion, 209
3-D solid, 37 user-supplied failure criterion, 213

Index
, 1274
Failure criterion variables Lanczos iteration method, 904
Fiber-matrix composites, 201 of preloaded structures, 1232
Five degrees of freedom node, 188 potential-based fluid elements, 299,
Flanges, 241, 244 334
Floor response spectrum analysis, restart, 1194
1014 rigid body modes, 901
curve smoothing, 1018 subspace iteration method, 903
damped transfer function method, Frequency domain analysis, 987
1024 Fourier analysis, 987
peak broadening, 1018 harmonic vibration analysis, 987
response spectra loading, 1019 random vibration analysis, 987
spatial combinations, 1024 response spectrum analysis, 987
time history loading, 1016 SDOF system response, 987
undamped transfer function, no , 1282
Frequency/mode variables
resonance correction method, Friction

1023 basic models, 759
undamped transfer function, user-supplied models, 761
resonance correction method, Friction delay, 755
1023 Full Newton iterations, 917, 919, 939
Fluid variables, 1258 line searches, 920
Follower loads, 833
Force variables, 1266 G
Formulations for Gap override, 748
2-D solid elements, 22 Gasket material model, 643
3-D solid elements, 46 General elements, 257, 953
axisymmetric shell elements, 171 element output, 259
iso-beam elements, 171 Geometric imperfections, 898, 1206
pipe elements, 247 Geotechnical material models, 610
shell elements, 185 Global mass matrix, 963
truss elements, 9 Green-Lagrange strains, 388, 390
Fourier analysis, 998 Gurson material model, 420
power spectral density, 999
Fracture mechanics, 1027 H
crack propagation, 1027, 1069
line contour method, 1027 Harmonic response types, 1304
meshing recommendations, 1080 Harmonic vibration analysis
virtual crack extension method, applied forces, 1001
1027 ground motion direction, 1008
Free surface modeling, 281 ground motions, 1002
Frequency analysis, 898 initial calculations, 1007
error bound, 901 modal damping ratios, 1008

Index
phase angle, 1008 initial displacements, 1167

potential-based fluid elements, 302, initial pipe internal pressures, 1169
334 initial strains, 1169
quasi-static amplitude, 1004 initial stresses, 1177
real part of the response, 1008 initial temperatures, 1168
RMS amplitude, 1008 temperature gradients, 1168
signed amplitude, 1008 Initial displacements, 1167
sweep spectra, 1008 Initial loads, 874
Hencky strain, 390 Initial pipe internal pressures, 1169
Hermitian beam elements, 60 Initial strains, 1169
Hill yield condition, 440 element, 1171
Holzapfel model for finite strain large strain formulations, 1176
viscoelasticity, 579 nodal point, 1171
Hyper-foam material model, 549 strain fields, 1173, 1174
3-D analysis, 550 Initial stresses, 1177
axisymmetric analysis, 550 restrictions, 1180
plane strain analysis, 550 Initial temperature gradients, 1168
plane stress analysis, 550 Initial temperatures, 1168
selection of material constants, 550 Initial thermal strains, 1169
Initial velocities, 1167
I Intial conditions
Implicit time integration, 948, 959 initial velocities, 1167
Newmark method, 949, 959 Irradiation creep material model, 475,
trapezoidal rule, 950, 960 486
Wilson theta method, 949 Iso-beam elements, 162
Improperly supported bodies, 782 element output, 171, 173
Incompatible modes finite elements, formulations, 171
20, 43 general 3-D, 162
Inelastic deformations, 396 linear, 168
Inertial properties, 1190 mass matrices, 173
center of mass, 1190 material models, 171
centroid, 1190 nonlinear, 171
element group, 1190 numerical integration, 166, 172
moments of inertia, 1190 plane strain 2-D, 162
products of inertia, 1190 plane stress 2-D, 162
total mass, 1190 shear deformations, 166
total volume, 1190 warping effects, 165
Infinite elements, 281 Isothermal plasticity material models,
Initial accelerations, 1167 420
Initial conditions, 1167 Isotropic hardening, 421, 448
initial accelerations, 1167 Isotropic hyperelastic effects, 540

Index
Iterative solvers, 912 Linear static analysis, 909, 981, 982

Linearized buckling analysis
J classical formulation, 894
Jaumann strain, 390 error bound, 897
J-integral, 1027 restart analysis, 895, 896
Load penetration, 837
K Load step response type, 1303
Load step results, 1303
Kinematic hardening, 421, 448 Load/displacement incrementation,
Kirchhoff stresses, 391 917
Loading
L centrifugal, 851
Lanczos iteration method, 904 concentrated, 830
Large displacement formulation, 9, distributed, 834
171, 247, 402, 472, 494, 610 Electromagnetic, 871
Large displacement/large strain initial, 874
formulation, 23, 46, 388, 389, 422, mass-proportional, 857
440, 447, 475, 539, 616, 635, 638 pipe internal pressures, 870
Large displacement/small strain prescribed temperature gradients,
formulation, 23, 46, 388, 422, 440, 867
447, 449, 475, 616, 633, 635, 638 prescribed temperatures, 867
Large strain analysis pressure, 833
ULJ formulation, 398 user-supplied, 875
Latent heat variables, 1258 Location-independent variables, 1252
LDC method, 827, 929, 941 Logarithmic strains, 390
constant increment of external work Low speed dynamics, 922
method, 931 LUBBY2 creep law, 484
constant spherical arc length variable coefficients, 486
constraint method, 931
load vector multiplier, 931 M
Limiting maximum incremental Mapping files, 867, 1200
displacement, 784 Mass
Line contour method, 1027, 1035 inertial properties, 1190
output, 1063 Mass matrices for
Line contour variables, 1253 2-D elements, 26
Line search, 920 iso-beam elements, 173
Linear buckling analysis pipe elements, 253
secant formulation, 895 shell elements, 205, 206
Linear dynamic analysis, 948 truss elements, 10
Linear formulation, 9, 23, 46, 188, Mass matrix, 978
247, 403 Mass scaling, 984

Index
Mass-proportional loads, 857 plastic-multilinear material, 421

potential-based fluid elements, 315 plastic-orthotropic, 440
Material models, 387 Ramberg-Osgood, 720
Anand, 661 rubber and foam, 539
Arruda-Boyce, 547 Shape Memory Alloy, 652
Cam-clay, 621 SMA, 652
concrete, 715 Sussman-Bathe, 551
concrete, 493 temperature dependence, 602
creep, 474 thermal strain effect, 600
creep variable, 485 thermo-elastic, 472
creep-variable, 474 thermoelasticity and creep, 714
curve description, 610 thermo-elasto-plasticity and creep,
curve fitting, 569 474
Drucker-Prager, 616 thermo-isotropic, 472
eight-chain, 559 thermo-orthotropic, 472
elastic-isotropic, 402, 403 thermo-plastic, 474
elastic-orthotropic, 402, 404 user-coded, 688
fabric, 633 viscoelastic, 634, 719
gasket, 643 viscoelastic effects, 578
geotechnical, 610 viscoplastic, 706
Gurson, 446 Material models for
hyper-foam, 549 2-D solid elements, 22
irradiation creep, 475, 486 3-D solid elements, 45
isothemal plasticity, 420 axisymmetric shell elements, 171
isotropic hyperelastic effects, 540 iso-beam elements, 171
Mohr-Coulomb, 625 pipe elements, 246
Mooney-Rivlin, 542 plate elements, 176
Mroz-bilinear, 432 shell elements, 185
Mullins effect, 592 truss elements, 9
multilinear-plastic-creep, 474, 483 Materially-nonlinear-only
multilinear-plastic-creep-variable, formulation, 9, 23, 46, 171, 188,
474, 485 247, 423, 440, 447, 466, 472, 475,
nonlinear elastic, 413, 417 494, 610, 616, 621, 635, 638
Ogden, 545 Matrices for
orthotropic effect, 597 3-D solid elements, 50
piezoelectric, 664 axisymmetric shell elements, 173
plastic, 713 Mean-square response, 1008
plastic creep, 474, 483 interpretation of, 1012
plastic-bilinear, 421 Memory allocation, 911
plastic-creep-variable, 474, 485 in-core solution, 1197
plastic-cyclic material, 448 out-of-core solution, 1197

Index
Mesh evaluation, 1309 plane strain analysis, 543

Mesh glueing, 887 plane stress analysis, 543
Mesh surface variables, 1253 selection of material constants, 544
Miscellaneous variables, 1282 Mroz-bilinear material model, 432
Mixed-interpolated finite elements, Mullins effect, 592
19, 41 Multigrid solver, 914
Mixed-interpolation formulation, 23, Multilayer shell elements, 200
46, 424, 467, 476, 543, 639, 689 specification of layer thicknesses,
Modal damping, 952 201
Modal mass, 989 transverse shear stresses, 200
Modal participation factors Multilinear-plastic-creep material
properties of, 995 model, 474, 483
Modal stresses, 906 Multilinear-plastic-creep-variable
Mode shape response type, 1304 material model, 474, 485
Mode shape results, 1303
Mode superposition, 948, 951, 959, N
962 Newmark method, 949, 952, 959
potential-based fluid elements, 300 Node discrete variables, 1253
Modeling of gaps, 418, 432 Node field variables, 1252
Modified Newton iterations, 917, 918, Node release-only technique, 1069
939 Node shift/release technique, 1069
line searches, 919 Node-node contact, 736
Mohr-Coulomb material model, 625 Noh-Bathe method, 976, 979
Moment-curvature beam elements, Nominal strains, 391
141 Nonconvergence, 935, 937
axial behavior, 142 Nonlinear dynamic analysis, 959
cyclic behavior, 149 Nonlinear elastic material model, 413,
elastic-plastic material model, 144 417
element output, 150 Nonlinear static analysis, 916
flexural behavior, 142 selection of incremental solution
mass matrices, 150 method, 938
nonlinear elastic model, 143 Non-positive definite stiffness matrix,
numerical integration, 143 911
section results, 150 Numerical integration for
stress resultants, 150 2-D solid elements, 24
torsional behavior, 142 3-D solid elements, 47
Moments of inertia axisymmetric shell elements, 166,
inertial properties, 1190 172
Mooney-Rivlin material model, 542 iso-beam elements, 166, 172
3-D analysis, 543 pipe elements, 250
axisymmetric analysis, 543 shell elements, 197

Index
truss elements, 9 474, 485

NVS virtual shifts, 1027, 1094 Plastic-cyclic material model, 448
Plastic-multilinear material model,
O 421
O.R.N.L. rules for cyclic loading Plastic-orthotropic material model,
conditions, 487 440
Ogden material model, 545 Plate elements, 176
3-D analysis, 546 material models, 176
axisymmetric analysis, 546 Pore pressures, 638
plane strain analysis, 546 Porous media formulation, 638
plane stress analysis, 545 Porous structures, 638
selection of material constants, 546 Positive definite stiffness matrix, 909
Operating systems supported by Post-collapse response, 929, 941
ADINA, 2 Postprocessing, 1245
Ovalization degrees of freedom, 238, Potential-based fluid elements, 280
244 3-D, 280
axisymmetric, 280
P choice of formulation, 304
direct time integration, 334
Parallel processing element output, 335
sparse solver, 1196 free surfaces, 281
Parallel-Network Framework, 674 ground motions, 295
Periodic symmetry, 1155 infinite elements, 281
Piezoelectric analysis, 1241 initial pressure in, 332
Piezoelectric material model, 664 mass-proportional loads, 315
Pipe elements, 237 planar, 280
element output, 254 pressure loads, 315
formulations, 247 static analysis, 333
mass matrices, 253 Power creep law, 483
material models, 246 Power spectral density, 999, 1009
numerical integration, 250 Prescribed accelerations, 860
ovalization degrees of freedom, 238 Prescribed displacements, 860
pipe-beam, 237 Prescribed load variables, 1278
pipe-shell, 237 Prescribed temperature gradients, 867
recommendations for use, 256 Prescribed temperatures, 867
warping degrees of freedom, 238 Prescribed velocities, 860
Pipe internal pressures, 241, 247, 870 Pressure loads, 833
Plastic strains, 479 deformation-dependent, 835
Plastic-bilinear material model, 421 load penetration, 837
Plastic-creep material model, 474, 483 plate/shell, 841
Plastic-creep-variable material model, potential-based fluid elements, 315

Index
shell, 841 initial calculations, 995

three-dimensional, 836 loading response spectra, 997
two-dimensional, 836 modal combinations, 990
Products of inertia modal damping ratios, 997
inertial properties, 1190 modal-combinations, 997
potential-based fluid elements, 302,
Q 334
Quarter-point midside nodes, 1056, residual responses, 988
1081 residual terms, 992, 997
Quasi-static amplitude, 1005 spatial combinations, 997
SRSS method, 990
R static correction terms, 992
static load case, 995
Radiation field variables, 1253 ten percent method, 990
Ramberg-Osgood material model, 720 Response spectrum response type,
Random response type, 1014, 1304 1304
Random response types, 1008 Response sprectrum analysis
Random vibration analysis, 1008 static load case, 998
ground motion direction, 1013 Response type
initial calculations, 1013 response-combination, 997
mean-square response, 1008 Response types
modal damping ratios, 1013 envelope, 1305
potential-based fluid elements, 302, harmonic, 1304
334 load step, 1303
power spectral density, 1009, 1013 mode shape, 1304
Rayleigh damping, 953, 959, 979 random, 1008, 1014, 1304
Reaction variables, 1270 residual, 1304
Reactions, 1159 response spectrum, 1304
Rebar elements, 12 response-spectrum, 997
Recommendations for use of Response-combination response type,
shell elements, 221 997, 1304
Remeshing, 1200 Response-spectrum response type,
Residual response type, 1304 997
Residual results, 1303 Restart, 1192
Response spectrum analysis, 987 Restart option
absolute sum method, 990 frequency analysis, 1194
algebraic sum method, 991 Result calculation within the AUI,
CQC method, 992 1245
double sum method, 991 combination transformations, 1246
DSC method, 991 examples, 1246
ground motion direction, 997 resultant transformations, 1246

Index
resultants after combinations, 1248 652

resultants before combinations, Shape memory effect, 652
1248 Shell elements, 177
spatial transformations, 1246 16-node, 223
time transformations, 1246 3D, 225
Result locations 4-node, 221
contact segments, 1294 9-node, 223
draw bead segments, 1294 basic assumptions in, 177
element section, 1293 composite, 200
element/layer, 1288 director vectors, 178
line contours, 1295 displacement-dependent pressure
location- independent, 1288 loading, 188
mesh integration points, 1296 element output, 214
node, 1288 failure criteria, 206
radiosity segments, 1294 locking, 223
virtual shifts, 1295 mass matrices, 205, 206
Result recovery, 1192, 1194 material models, 185
Results MITC16, 223
load step, 1303 MITC4, 221
mode shape, 1303 MITC9, 223
residual, 1303 multilayer, 200
Resuses (of substructures), 1142 nodal forces, 220
Retained nodes, 1142 nodal point degrees of freedom, 188
Rigid body modes, 901 numerical integration, 197
Rigid links, 883 recommendations for use, 221
Rigid target contact algorithm, 784 section results, 217
Rigid target method, 739 shear deformations, 183
Ring element, 7 stress resultants, 217
Rubber and foam material models, thick, 221
539 thin, 221
Rubber stability indicators, 604 transition, 195
Rupture conditions, 430, 446, 467, Shell midsurfaces
477, 480, 488 plotting results onto, 1303
Shell stiffeners, 166
S Shell-shell intersections, 196
SDOF system response, 1014 Shell-solid intersections, 196
Section field variables, 1253 Shifting procedure, 905
Segment method, 739 Six degrees of freedom node, 190
Selection of contact algorithm, 739 Skew systems, 825
Shape Memory Alloy, 652 SMA, 652
Shape Memory Alloy material model, SMA material model, 652

Index
Small displacement contact feature, Hencky strain, 390

725, 735 Jaumann strain, 390
Small displacement formulation, 9, logarithmic strains, 390
171, 247, 403, 472, 494, 610, 621 stretches, 390
Small displacement/small strain Strain variables, 1262
formulation, 23, 46, 387, 422, 440, Strain-rate effects
447, 449, 475, 616, 635, 638 2-D elements, 431
Smoothing, 1301 3-D elements, 431
restrictions, 1302 shell elements, 431
Soil consolidation analysis, 1237 Stress intensity factors, 1027, 1031
Solution of KU=R, 909 Stress measures, 387, 391
active column solver, 910 2nd Piola-Kirchhoff stress, 391
iterative solvers, 912 Cauchy stress, 391
multigrid solver, 914 Cauchy stresses, 387
sparse solver, 910 engineering stress, 391
Solvers, 966, 984 engineering stresses, 387
3D-iterative solver, 914 Kirchhoff stress, 391
Sparse matrix solver, 910 Stress variables, 1259
Sparse solver, 1196 Stretch tensor
in-core, 911 left, 395
memory allocation, 911 Stretches, 390
out-of-core, 911 Structural vibration, 951
parallel processing, 1196 Sturm sequence check, 904, 905
Spring/damper/mass elements, 257, Subspace iteration method, 903
953 Sturm sequence check, 904, 905
linear single-degree-of-freedom, Substructures, 1142
260 Substructuring, 1141
linear two-degree-of-freedom, 261 condensed nodes, 1142
nonlinear, 262 linear analysis, 1141
Stabilized TLA method, 927 nonlinear analysis, 1143
Static analysis, 909, 981, 982 restrictions, 1144
potential-based fluid elements, 296, retained nodes, 1142
333 Superelastic effect, 652
Stiffness stabilization, 783, 917 Supported computers and operating
Strain energy density, 1188 systems, 2
Strain Energy Proportional Damping, Suppressing contact oscillations, 769
955 Sussman-Bathe material model, 551
Strain hardening, 487 SVS virtual shifts, 1027, 1112
Strain measures, 387, 390 Swelling analysis, 638
engineering strains, 387, 390 Symmetry condition (for pipe
Green-Lagrange strains, 390 elements), 245

Index
T material models, 9
numerical integration, 9
Temperature files, 867
rebar, 12
Temperature variables, 1258
recommendations for use, 12
Temperature-dependent material, 602
Thermal coupling between solids and
U
fluids, 1242
Thermal strains, 400, 478 UL formulation, 403, 423, 472, 476
Thermo-elastic material models, 472 ULH formulation, 46, 388, 389, 390,
Thermoelasticity and creep material 466, 476, 635, 639, 689
model, 714 ULJ formulation, 388, 389, 390, 423,
Thermo-elasto-plasticity and creep 440, 466
material models, 474 Undrained analysis, 638
Thermo-isotropic material model, 472 User-coded material model, 688
Thermo-orthotropic material model, User-defined variables, 1286
472 User-supplied creep coefficients, 485
Thermo-plastic material model, 474 User-supplied elements, 270
Thermorheologically simple material, User-supplied friction models, 761
539, 602 User-supplied loads, 875
Thickness variables, 1281
Tied contact, 733 V
Time functions, 826, 828 Variables, 1252
constant, 828 acceleration, 1257
short circuit, 828 contact, 1252
sinusoidal, 828 coordinate, 1253, 1279
Time offset, 490 displacement, 1255
Time step management, 983 draw bead, 1252
Time variables, 1281 eigenvector, 1277
TL formulation, 24, 46, 403, 423, 440, electric field, 1272
466, 472, 476, 494, 539, 610, 616, element local node, 1253
621, 635, 638, 689 element local node field, 1253
TLA method, 927 element/layer field, 1252
TLA-S method, 927 , 1274
Total Load Application method, 927
failure criterion
fluid, 1258
Traction variables, 1268 force, 1266
Trapezoidal rule, 950, 960 , 1282
TRS material, 539, 602
frequency/mode
interpolation within elements, 1299

True strains, 391 latent heat, 1258
Truss elements, 7 line contour, 1253
element output, 11 location-independent, 1252
formulations, 9 mesh surface, 1253
mass matrices, 10

Index
miscellaneous, 1282 pressure correction, 1041

node discrete, 1253 thermal correction, 1041
node field, 1252 Virtual shift variables, 1253
prescribed load, 1277, 1278 Virtual shifts, 1064
radiation field, 1253 spatially fixed, 1071
reaction, 1270 spatially moving, 1071
section field, 1253 Viscoelastic effects, 578
smoothing, 1301 Viscoelastic material model, 634, 719
strain, 1262 Prony series, 637
stress, 1259 Viscoplastic material model, 706
temperature, 1258 Volume
thickness, 1281 inertial properties, 1190
time, 1281 von Karman ovalization modes, 242
traction, 1268
User-defined, 1286 W
velocity, 1256 Wagner effect, 76
virtual shift, 1253 Warping degrees of freedom, 238, 244
Velocity variables, 1256 Wave propagation, 950
Virtual crack extension method, 1027, Wilson theta method, 949
1037
description of virtual shifts, 1064 Z
dynamic correction, 1042
output, 1069, 1111, 1136 Zero-slope-of-pipe-skin, 244

Tmg-A 92

Uploaded by

Copyright:

Available Formats

Tmg-A 92

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Tmg-A 92

Uploaded by

Copyright:

Available Formats

UTOMATIC

Theory and Modeling Guide

ADINA 9.2 October 2016

ADINA R & D, Inc.

Volume I: ADINA Solids & Structures

ADINA R & D, Inc.

tel. (617) 926-5199

The information contained in this document is subject to change without notice.

ADINA is a registered trademark of K.J. Bathe / ADINA R & D, Inc.

ADINA R & D, Inc. 1987 - 2016

ADINA R & D, Inc. iii

2.4.3.6 Strain energy for a linear elastic material........................................ 74

iv ADINA Structures Theory and Modeling Guide

2.7.4 Transition elements .............................................................................. 195

ADINA R & D, Inc. v

2.11.18 Modeling: Considerations for frequency analysis and modal

vi ADINA Structures Theory and Modeling Guide

3.4.4 Gurson material model ......................................................................... 446

ADINA R & D, Inc. vii

3.9.1 Curve description material model ........................................................ 610

viii ADINA Structures Theory and Modeling Guide

3.18.8.3 Thermo-elasticity and creep material model ............................... 714

ADINA R & D, Inc. ix

4.9 Rigid target contact algorithm..................................................................... 784

x ADINA Structures Theory and Modeling Guide

5.14.2 Usage of the user-supplied loads........................................................ 875

ADINA R & D, Inc. xi

7.4.3 Global mass matrix............................................................................... 963

xii ADINA Structures Theory and Modeling Guide

ADINA R & D, Inc. xiii

11.6 Initial conditions...................................................................................... 1167

xiv ADINA Structures Theory and Modeling Guide

13.2 Evaluation of meshes used ...................................................................... 1309

ADINA R & D, Inc. xv

To provide a document that summarizes the methods and

To provide specific references that describe the finite

Hence, this manual has been compiled to provide a bridge

ref. K. J. Bathe, Finite Element Procedures, 2nd ed.,

Following this introductory chapter, Chapter 2 describes the

ADINA R & D, Inc. 1

the material models available in ADINA. Chapter 4 describes the

1.2 Supported computers and operating systems

2 ADINA Structures Theory and Modeling Guide

Table 1.2-1: Supported computers/operating systems

Platform Operating system Version ADINA- Parallelized Fortran

ADINA R & D, Inc. 3

Table 1.3-1: Units

SI SI (mm) USCS (inch) USCS (kip)

length meter (m) millimeter inch (in) inch (in)

pressure, Pascal (Pa) = N/mm2 psi = lb/in2 ksi = kip/in2

1) A body that weighs 1 lb has a mass of 1/386.1 = 0.002590 lb-sec2/in.

1.4 ADINA System documentation

4 ADINA Structures Theory and Modeling Guide

ADINA User Interface Command Reference Manual

Volume I: ADINA Solids & Structures Model Definition

Volume II: ADINA Thermal Model Definition

Volume III: ADINA CFD & FSI Model Definition

Volume IV: ADINA EM Model Definition

Volume V: Display Processing