Predictive Engineering PDF

LS-DYNA® Analysis for Structural Mechanics
An overview of the core analysis features used by LS-DYNA® to simulate highly nonlinear
transient behavior in engineered structures and systems.
LS-DYNA Analysis for Structural Mechanics 2014
Acknowledgements
These notes were constructed from numerous sources but special thanks should be given to the
following people:
Technical Support Team at Livermore Software Technology Corporation (LSTC)

With special mention to:
Jim Day, LSTC
Philip Ho, LSTC
External to LSTC, we have:

Paul Du Bois, Hermes Engineering, NV
And
DYNAmore, Gmbh, Germany
Trademarks:
LS-DYNA® and LS-PrePost® are registered and protected trademarks of LSTC.
Femap® is a registered and protected trademark of Siemens PLM Software.
Disclaimer:
The material presented in this text is intended for illustrative and educational purposes only. It is not intended to be
exhaustive or to apply to any particular engineering design or problem. Predictive Engineering nor the organizations
mentioned above and their employees assumes no liability or responsibility whatsoever to any person or company for
any direct or indirect damages resulting from the use of any information contained herein.
Proprietary Information to Predictive Engineering, Please Do Not Copy or Distribute Page 2 of 105
COURSE OUTLINE
1. INTRODUCTION ................................................................................................................................................................. 9
1.1 WHAT THE STUDENT CAN EXPECT ........................................................................................................................................................................9
1.2 WHAT WE COVER ..............................................................................................................................................................................................9
1.3 HOW WE DO IT .................................................................................................................................................................................................9
1.4 GENERAL APPLICATIONS ...................................................................................................................................................................................10
1.5 SPECIFIC APPLICATIONS (COURTESY OF PREDICTIVE ENGINEERING)............................................................................................................................11
2. IMPLICIT VERSUS EXPLICIT ANALYSIS.............................................................................................................................. 17
2.1 WHAT WE ARE SOLVING ..................................................................................................................................................................................17
2.2 EXPLICIT (DYNAMIC).........................................................................................................................................................................................18
2.3 IMPLICIT (DYNAMIC).........................................................................................................................................................................................18
3. FUNDAMENTAL MECHANICS OF EXPLICIT ANALYSIS ..................................................................................................... 19
3.1 TIME STEP SIGNIFICANCE ..................................................................................................................................................................................19
3.1.1 Explicit Time Integration ...................................................................................................................................................................20
3.2 TIME STEP SIGNIFICANCE (COURANT-FRIEDRICHS-LEWY (CFL) CHARACTERISTIC LENGTH) .............................................................................................21
3.3 MASS SCALING: (EVERYBODY DOES IT BUT NOBODY REALLY LIKES IT) .......................................................................................................................22
3.3.1 Workshop: Femap to LS-DYNA with Mass Scaling Basics .................................................................................................................23
3.3.2 Instructor Led Workshop I: Mass Scaling Advanced.........................................................................................................................28
3.4 IMPLICIT MESH VERSUS EXPLICIT MESH CHARACTERISTICS .......................................................................................................................................29
3.4.1 Instructor Led Workshop II: Implicit versus Explicit Mesh Differences ............................................................................................29
4. FIRST LS-DYNA MODEL: GETTING STARTED .................................................................................................................... 30
4.1 LS-DYNA KEYWORD MANUAL ..........................................................................................................................................................................30
4.2 KEYWORD SYNTAX ...........................................................................................................................................................................................30
4.3 UNITS ...........................................................................................................................................................................................................31
4.4 REFERENCE MATERIALS AND PROGRAM DOWNLOAD..............................................................................................................................................32
4.5 SUBMITTING LS-DYNA ANALYSIS JOBS AND SENSE SWITCHES .................................................................................................................................32
4.6 WORKSHOP I: LS-DYNA GETTING STARTED EXAMPLE ..........................................................................................................................................33

5. EXPLICIT ELEMENT TECHNOLOGY ................................................................................................................................... 34
5.1 ELEMENT TYPES IN LS-DYNA ............................................................................................................................................................................34
5.2 ONE GUASSIAN POINT ISOPARAMETRIC SHELL ELEMENTS AND HOURGLASSING ...........................................................................................................35
5.2.1 Instructor Led Workshop III: Explicit Element Technology | A: Side Bending..................................................................................35
5.2.2 Instructor Led Workshop III: Explicit Element Technology | B: Out-of-Plane Bending with Plasticity ............................................36
5.2.3 Workshop: Building the Better Beam ...............................................................................................................................................37
5.2.4 Workshop II: Hourglass Control/Hourglass ......................................................................................................................................38
5.2.5 Workshop III: Meshing for Explicit Success | Meshing for Explicit Success Start.modfem ..............................................................39
5.3 SCALAR ELEMENTS (NASTRAN CBUSH EQUIVALENTS) ...........................................................................................................................................40
5.3.1 Workshop Discrete Elements (Non-Linear Buckling Analysis)..........................................................................................................42
6. LS-PREPOST ..................................................................................................................................................................... 43
6.1 WORKSHOP IV: LS-PREPOST | WORKSHOP 7 & 8 ...............................................................................................................................................43
7. MATERIAL MODELING ..................................................................................................................................................... 44
7.1 PART 1. METALS .............................................................................................................................................................................................44
7.1.1 Engineering Stress-Strain vs True Stress-Strain ................................................................................................................................44
7.1.2 Review of Material Models Available in LS-DYNA ............................................................................................................................45
7.1.3 Material Failure and Experimental Correlation ................................................................................................................................46
7.2 WORKSHOP V: ELASTIC-PLASTIC MATERIAL FAILURE .............................................................................................................................................47
7.3 STRAIN RATE SENSITIVITY OF METALS..................................................................................................................................................................48
7.4 PART 2: ELASTOMERS AND FOAMS .....................................................................................................................................................................49
7.4.1 Modeling Elastomers vs Foams (Viscoplasticity) ..............................................................................................................................49
7.5 MATERIAL MODELS FOR MODELING FOAMS .........................................................................................................................................................50
7.6 WORKSHOP VI: MODELING AN ELASTOMER BALL WITH HEX AND TET ELEMENTS .........................................................................................................51
7.7 EQUATION OF STATE (EOS) MATERIAL MODELING ................................................................................................................................................52
7.7.1 Modeling Water with *EOS_GRUNEISEN and *MAT_NULL .............................................................................................................54
7.8 MATERIAL FAILURE SIMULATION ........................................................................................................................................................................55

7.8.1 Basic Methods of Modeling Failure: Material versus Bond Failure..................................................................................................55
7.9 WORKSHOP VII: MODELING GENERAL FAILURE ....................................................................................................................................................56
7.10 MODELING RIGID BODIES .............................................................................................................................................................................57
7.10.1 Rigid Materials (*MAT_020 or *MAT_RIGID) ...................................................................................................................................57
7.10.2 Instructor Led Workshop V: RBE2 to CNRB ......................................................................................................................................58
7.10.3 Workshop VIII: Using Rigid Bodies ...................................................................................................................................................59
8. CONTACT ......................................................................................................................................................................... 60
8.1 DEFINITION OF CONTACT TYPES .........................................................................................................................................................................60
8.2 GENERAL CONTACT TYPES .................................................................................................................................................................................61
8.2.1 Additional Options: SOFT=2 “The Default” ......................................................................................................................................61
8.2.2 Contact when things ERODE .............................................................................................................................................................62
8.3 WORKSHOP IX: UNDERSTANDING BASIC CONTACT MECHANICS ..............................................................................................................................63
8.3.1 Student Bonus Option: Contouring Interface Pressures .................................................................................................................63
8.3.2 Workshop IX: Comparison of Results ...............................................................................................................................................64
8.4 INSTRUCTOR LED WORKSHOP VI CONTACT: NEGATIVE SLIDING INTERFACE ENERGY ....................................................................................................66
8.5 INTERMEDIATE CONTACT ANALYSIS .....................................................................................................................................................................67
8.5.1 Beam and Edge-to-Edge Contact Modeling......................................................................................................................................67
8.5.2 Special Contact Options ....................................................................................................................................................................67
8.6 WORKSHOP X: EDGE-TO-EDGE CONTACT ............................................................................................................................................................68
8.7 TIED CONTACT FOR MESH TRANSITIONS, WELDING AND GLUING .............................................................................................................................69
8.7.1 Tied Contact or Gluing ......................................................................................................................................................................69
8.8 SUMMARY TABLE FOR TIED CONTACT ..................................................................................................................................................................70
8.9 WORKSHOP XI-A: TIED CONTACT FOR HEX-TO-TET MESH TRANSITIONS (TIED_SURFACE_TO_SURFACE) .................................................................71
8.9.1 Student Bonus...................................................................................................................................................................................73
8.10 WORKSHOP XI-B: TIED CONTACT FOR GLUING THINGS TOGETHER (BEAM_OFFSET) ...........................................................................................74
8.10.1 Instructor Led Bonus Workshop Bad Energy with Tied Contact ......................................................................................................75
9. DAMPING ........................................................................................................................................................................ 76
9.1 GENERAL, MASS AND STIFFNESS DAMPING ..........................................................................................................................................................76
9.1.1 *DAMPING_option............................................................................................................................................................................76
9.1.2 *DAMPING_FREQUENCY_RANGE .....................................................................................................................................................76
9.1.3 Material Damping (e.g., elastomers and foams) ..............................................................................................................................77
9.1.4 Student Example on Material Damping............................................................................................................................................77
9.2 INSTRUCTOR / STUDENT LED WORKSHOP VII: DAMPING........................................................................................................................................78
9.3 BULK VISCOSITY ..............................................................................................................................................................................................79
10. LOADS, CONSTRAINTS AND RIGID WALLS .................................................................................................................. 80
10.1 LOADS ......................................................................................................................................................................................................80
10.1.1 Initialization Loads ............................................................................................................................................................................80
10.1.2 Point and Pressure Loads..................................................................................................................................................................80
10.1.3 Body Loads ........................................................................................................................................................................................80
10.2 WORKSHOP: APPLYING MOVING BOUNDARY CONDITIONS ................................................................................................................................81
10.3 WORKSHOP XII: DROP TEST OF PRESSURE VESSEL ............................................................................................................................................82
10.3.1 Student Bonus...................................................................................................................................................................................84
11. CONNECTIONS VIA JOINTS, BOLTS AND SPOTWELDS ................................................................................................ 85
11.1 JOINTS OR *CONSTRAINED_JOINT_..........................................................................................................................................................85
12. LOAD INITIALIZATION BY DYNAMIC RELAXATION AND IMPLICIT ANALYSIS .............................................................. 86
12.1 INITIALIZATION OF GRAVITY, BOLT PRELOAD AND OTHER INITIAL STATE CONDITIONS ...............................................................................................86
12.2 WORKSHOP XIII: DYNAMIC RELAXATION.........................................................................................................................................................87
13. IMPLICIT-EXPLICIT SWITCHING FOR BURST CONTAINMENT ...................................................................................... 88
13.1 HIGH-SPEED ROTATING EQUIPMENT – *CONTROL_ACCURACY .....................................................................................................................88
13.1.1 Workshop XIV: Implicit-Explicit Turbine Spin Up .............................................................................................................................89
14. SMOOTHED PARTICLE HYDRODYNAMICS (SPH) ......................................................................................................... 90
14.1 INTRODUCTION ..........................................................................................................................................................................................90
14.1.1 A Little Bit of Theory (skip this if you don’t like math…) ..................................................................................................................90
14.1.2 Lagrangian vs Eulerian ......................................................................................................................................................................92
14.1.3 Types of Simulations with SPH..........................................................................................................................................................93
14.1.4 Common Keywords for SPH ..............................................................................................................................................................93
14.2 SPH WORKSHOP I: GETTING STARTED............................................................................................................................................................94
14.3 SPH WORKSHOP II: FLUID MODELING ...........................................................................................................................................................95
14.4 SPH WORKSHOP III: VERIFICATION................................................................................................................................................................96
14.4.1 SPH Verification III: Workshop Bonus ..............................................................................................................................................97
14.5 SPH WORKSHOP IV: BIRD STRIKE .................................................................................................................................................................98
14.5.1 Bird Strike Models.............................................................................................................................................................................99
14.6 REFERENCES ........................................................................................................................................................................................... 100
15. EXPLICIT MODEL CHECK-OUT AND RECOMMENDATIONS ....................................................................................... 101
15.1 UNITS.................................................................................................................................................................................................... 101
15.2 MESH.................................................................................................................................................................................................... 101
15.3 D3HSP FILE (LS-DYNA EQUIVALENT TO THE NASTRAN F06 FILE)..................................................................................................................... 101
15.4 HISTORY PLOTS ....................................................................................................................................................................................... 102
15.5 MATERIAL MODELING ERRORS ................................................................................................................................................................... 102
15.6 CONTACT OPTIONS WITH RECOMMENDATIONS ............................................................................................................................................. 103
15.7 CONTROL CARDS WITH RECOMMENDATIONS ................................................................................................................................................ 104
15.8 DATABASE CARDS WITH RECOMMENDATIONS ............................................................................................................................................... 105
15.9 ETC ....................................................................................................................................................................................................... 106
16. IMPLICIT CHECK-OUT AND RECOMMENDATIONS .................................................................................................... 107
16.1 IMPLICIT KEYWORD CARDS AND RECOMMENDATIONS ..................................................................................................................................... 108
17. DISCRETE ELEMENT METHOD ................................................................................................................................... 109
18. FLUID STRUCTURE INTERACTION AND MULTI-PHYSICS IN LS-DYNA........................................................................ 110
19. LS-DYNA RECENT DEVELOPMENTS ........................................................................................................................... 111
END ........................................................................................................................................................................................ 112
1. INTRODUCTION
1.1 WHAT THE STUDENT CAN EXPECT
This class is directed toward the engineering professional simulating highly nonlinear, transient dynamic problems
involving large deformations and contact between multiple bodies. What this means in more layman terms is that we
will provide a realistic foundation toward the practical usage of LS-DYNA.
1.2 WHAT WE COVER

• Explicit FEA Mechanics
• The technology of creating accurate nonlinear, transient FEA models
• How to do your own research to create more advanced simulations
• Our condensed experience and that of our colleague’s toward inoculating you against repeating our mistakes
1.3 HOW WE DO IT
• The class covers the basics in a hands-on manner as taught by an engineer that has had to live by what they have
killed.
• Each day will have six to eight Workshops. Each Workshop is part theory, part demonstration and part hands-on
practice. Videos are provided for each Workshop allowing the student to relax and follow along at their own pace.
These videos cover the basics and also provide insight into the many tips and tricks that make LS-DYNA the world’s
most complete and accurate simulation code.
• Breaks are provided every two hours where students can pause, relax and ask the instructor more detailed
questions.
• Students are encouraged to turn off their email, text messaging and other forms of digital/social media during class
time (8:00 am to 5:00 pm).
1.4 GENERAL APPLICATIONS
Crashworthiness Driver Impact Train Collisions
Earthquake Engineering Metal Forming Military
1.5 SPECIFIC APPLICATIONS (COURTESY OF PREDICTIVE ENGINEERING)
Drop Test Consumer

Crash Analysis of Seats Sporting Goods Equipment
Products
Large Deformation of
Drop Test of Composites / Electronics Human Biometrics
Plastics
Drop Test of Nuclear Waste

Crash Analysis of Cargo Net Impact Analysis of Foams
Container
Plastic Thread Design PSD / Modal Analysis Digger Tooth Failure
Electron Beam Welding Pyro-Shock Analysis Medical Equipment
Ballistic Shock Loading of Optical Hyperelastic Medical Seal

Fracture Mechanics of Glass
Equipment Analysis
Discrete Element Method for the Drop-Test of Hand Held

Blade-Out Analysis
Mining Industry Electronics
Ballistic Penetration of Al/Foam High-Speed Spinning Disk Locomotive Fuel Tank

Panel Containment
Airplane 16g Crash Analysis Impact Analysis of Safety Block Device
Snap-Fit Analysis
2. IMPLICIT VERSUS EXPLICIT ANALYSIS

LS-DYNA is a non-linear transient dynamic finite element code with both explicit and implicit solvers.
2.1 WHAT WE ARE SOLVING

Explicit only works when there is acceleration (dynamic) whereas an implicit approach can solve the dynamic and the
static problem. For dynamic problems, this means that we are solving the following equation:
where n=time step. A common terminology is to call the part the internal force in the structure. The basic problem
is to determine the displacement at some future time or , at time .
In conceptual terms, the difference between Explicit and Implicit dynamic solutions can be written as:
( )
All these terms are known at time state “n” and thus can be solved directly.
For implicit, the solution depends on nodal velocities and accelerations at state n+1, quantities which are unknown:
( )
Given these unknowns, an iterative solution at each time step is required.
2.2 EXPLICIT (DYNAMIC)

Internal and external forces are summed at each node point, and a nodal acceleration is computed by
dividing by nodal mass. The solution is advanced by integrating this acceleration in time. The maximum
time step size is limited by the Courant condition, producing an algorithm which typically requires many
relatively inexpensive time steps. Using this criterion, the solution is unconditionally stable. Since the
solution is solving for displacements at nodal points, the time step must allow the calculation to
progress across the element without “skipping” nodes. Hence, the explicit solution is limited in time
step by the element size and the speed sound in the material under study. Even worse, the smallest
element in the mesh can dictate the time step for the whole solution and likewise combined with the
stiffest material (fastest speed of sound).
2.3 IMPLICIT (DYNAMIC)

A global stiffness matrix is computed, decomposed and applied to the nodal out-of-balance force to
obtain a displacement increment. Equilibrium iterations are then required to arrive at an acceptable
“force balance”. The advantage of this approach is that time step size may be selected by the user. The
disadvantage is the large numerical effort required to form, store, and factorize the stiffness matrix.
Implicit simulations therefore typically involve a relatively small number of expensive time steps. The key
point of this discussion is that the stiffness matrix (i.e., internal forces) has to be decomposed or inverted
each time step whereas in the explicit method, it is a running analysis where the stiffness terms are re-
computed each time step but no inversion is required.
3. FUNDAMENTAL MECHANICS OF EXPLICIT ANALYSIS

3.1 TIME STEP SIGNIFICANCE
3.1.1 EXPLICIT TIME INTEGRATION
• Very efficient for large nonlinear problems (CPU time increases only linearly with DOF)
• No need to assemble stiffness matrix or solve system of equations
• Cost per time step is very low
• Stable time step size is limited by Courant condition
• Time for stress wave to traverse an element
• Problem duration typically ranges from microseconds to tenths of seconds
• Particularly well-suited to nonlinear, high-rate dynamic problems
• Nonlinear contact/impact
• Nonlinear materials
• Finite strains/large deformations
Figure 1: How Solution Time and Result Outputs Are Defined in Explicit
3.2 TIME STEP SIGNIFICANCE (COURANT-FRIEDRICHS-LEWY (CFL) CHARACTERISTIC LENGTH)
• In the simplest case (small, deformation theory), the timestep is

controlled by the acoustic wave propagation through the material.
• In the explicit integration, the numerical stress wave must always
propagate less than one element width per timestep. √
• The timestep of an explicit analysis is determined as the minimum
stable timestep in any one (1) deformable finite element in the mesh.
(Note: As the mesh deforms, the timestep can similarly change)
• The above relationship is called the Courant-Friedrichs-Lewy (CFL)
condition and determines the stable timestep in an element. The CFL
condition requires that the explicit timestep be smaller than the time
needed by the physical wave to cross the element. Hence, the
numerical timestep is a fraction (0.9 or lower) of the actual theoretical
timestep. Note: the CFL stability proof is only possible for linear
problems. ( )
• In LS-DYNA, one can control the time step scale factor (TSSFAC). The
default setting is 0.9. It is typically only necessary to change this factor
for shock loading or for increased contact stability with soft materials.
Based on this conditions, the time step can be increased to provide faster solution times by artificially
increasing the density of the material (e.g., mass scaling, lowering the modulus or by increasing the
element size of the mesh.
3.3 MASS SCALING: (EVERYBODY DOES IT BUT NOBODY REALLY LIKES IT)
Explicit Time Step Mass Scaling (*Control_Timestep)*

• Mass scaling is very useful and directly increases the
√
timestep. The concept is simple, Larger Timestep =
Lower Solution Time
• One can also just simply increase the global density of √ ( )
=5,384 mm/ms
the material for non-dynamic simulations (i.e., where
inertia effects can be considered small).
• *CONTROL_TIMESTEP: Conventional mass scaling (CMS)
(negative value of DT2MS): The mass of small or stiff  LS-DYNA time step is different between Femap and
elements is increased to prevent a very small timestep. LS-DYNA due to TSSFAC=0.9 (default)
Thus, artificial inertia forces are added which influence  Mesh quality affects Time Step – just tweak it
all eigenfrequencies including rigid body modes. This
means, this additional mass must be used very carefully Clean Mesh Mesh Toolbox Tweaked
so that the resulting non-physical inertia effects do not
dominate the global solution. This is the standard
default method that is widely used.
• With CMS, a recommended target is not to exceed 5%
of the mass of the system or 10% of the mass of any
one part. Added mass can be tracked with *DATABASE
options of GLSTAT for entire model and MATSUM for
individual parts. (Note: All general recommendations
and tips are given in Explicit Model Check-Out and
Recommendations.)
3.3.1 WORKSHOP: FEMAP TO LS-DYNA WITH MASS SCALING BASICS
What You Will Learn Units: kN-mm-ms-kg

How to build your first LS-DYNA model from the ground up Linear, elastic material model of aluminum:
and how to improve the run times using mass scaling. E v 
Tasks 70 0.33 2.71e-6
• Open Femap model (Femap to LS-DYNA with Mass
Scaling Basics Start.modfem) and create LS-DYNA
Isotropic Material and then plate property with thickness
= 1.0 (investigate Formulation for plates and just note
that there is a lot going on with LS-DYNA).
• Mesh surface and check Explicit Time Step (see Model
Data Contour / Show Model Data Contour). Note that
the time step is in milliseconds given the unit system and
that the element time step is defined by its shape.
• Change elastic modulus from 70 to 35 and re-contour
time step. Note change and then un-do (CTRL-Z) within
Femap.
• Setup Analysis Manager for LS-DYNA Analysis. The
analysis will run for 8 milliseconds with results sets at
intervals of 0.01. Make sure to select a Load for the
analysis. Export to the same directory as the Femap
model file. While doing this verify the load application
and the constraint set (there ain’t any….).
Femap to LS-DYNA with Mass Scaling Basics

(continued)
Tasks
• Run LS-DYNA model file using LS-DYNA.
• Take a look at the Analysis Output and find the time
step listing. One will note that it is reduced by 0.9
from what Femap calculates. The reason for this is to
ensure that the Courant condition is met.
• Open up LS-PrePost (LSPP) and load the d3plot file
(File / Open / LS-DYNA Binary Plot). This “mother
file” will then load its subsequent children files, i.e.,
d3plot01, d3plot02, etc. automatically. To make it
easier the first time around, go to View / Toolbar and
set the toolbars to Text and Icon.
• Then we will contour the von Mises Stress (Toolbar
Post, FrinComp, von Mises Stress. Then hit the play
button on the Animate dialog box. Sit back and enjoy
the stress wave colors.
Femap to LS-DYNA with Mass Scaling Basics (continued)

With the model working, let’s harvest some data. We are going to make several runs of this model to investigate the
relationship between mesh, explicit time step and mass scaling. As part of this process, you’ll get comfortable working
with Femap, LS-DYNA Program Manager and LSPP. Our test metric is going to be the maximum displacement from a
node at the end of the bar.
Tasks
• Within existing LSPP model, open History, select Node, Y-
Displacement and then pick a node at the very top of the bar
near the center (any’ol node near the center) and then
likewise at the bottom, near
the center. When done you
should have two nodes
selected and then hit Plot
within the History dialog box.
When finished something like
this should appear as shown
on the right.
• Note that the maximum displacement at the top is 0.00781
mm with an explicit timestep = 0.0334 ms.

Tasks
• Go back to the Femap model file and open up the Meshing
Toolbox Pane. Decrease the mesh sizing on the right-hand
side curve in the middle by four increments. Although we
are making the mesh size larger (i.e., the timestep should
increase), one will note that the explicit time step will
decrease due to non-uniform shaped elements used to
transition the mesh. This is more reflective of FEA reality
where the mesh is rarely uniform.
• Export Model to LS-DYNA, analyze and create a plot as in the
steps above.
• Apply mass scaling to arrive at an explicit time step of
0.0334 ms. See LS-DYNA Keyword *CONTROL_TIMESTEP
and insert this keyword into the Femap Analysis Set
Manager (see graphic) and rerun
the model. Note: keywords can
be entered in a simple comma
separated format as shown in the
graphic on the right and when the
deck is exported, they are
inserted into the LS-DYNA analysis
file.

The following task is to get comfortable reading the LS-DYNA
output. Although this information can be found in the d3hsp and
messag files, it is a bit easier to grab it directly off of the screen.
The information we are looking for is the “percentage increase” of
mass that the –DTMS option within the *CONTROL_TIMESTEP has
added to our model. In this example, we have added mass of
16.9%.
Objective
Apply mass scaling to obtain similar Max. Displacements. Keep in
mind that the higher the time step, the faster your analysis will
run and in practice, it can take a 120 minute run and allow it to
run in 80 minutes with no adverse effects. It is a huge
productivity boost.
Model Time Step % Mass Added by Mass Scaling Max. Displacement

Starting Point 0.0334 ms 0.00% 0.00781 mm
Skewed Mesh (-4x) 0.0184 ms 0.00% 0.00781 mm
Skewed Mesh with Mass Scaling 0.0334 ms 16.8% 0.00684 mm
Class Bonus: What happens when the CFL Criterion is not met? (e.g., set TSSFAC=2.0)
3.3.2 INSTRUCTOR LED WORKSHOP I: MASS SCALING ADVANCED
Explicit Time Step Mass Scaling (*CONTROL_TIMESTEP)* Time step ranges from 2.89 to 4.74e-4
• Mass scaling is no free lunch. For dynamic systems, added mass can
affect the response of the system.
• It is just something to monitor and make an engineering judgment
about its effectiveness; time savings versus potential detrimental
effects. Mass scaling is my universal modeling condiment and the
aim is typically no more than 5% additional mass.
Analyst Note: Would this make your dynamic (F=ma) analysis more
conservative or less?
• Conventional mass scaling (CMS) has morphed to using the negative
(-)DT2MS option as the recommended default.
• Selective mass scaling (SMS): Using selective mass scaling, only the
high frequencies are affected, whereas the low frequencies (rigid
body bodies) are not influenced; thereby, a lot of artificial mass can
be added to the system without adulterating the global solution.
• This method is very effective, if it is applied to limited regions with
very small critical timesteps. SMS is invoked with the IMSCL
command over a single part or multiple parts.
Solution time is 10 seconds for no mass scaling and 5 and 6 seconds for
SMS and CMS respectively. SMS is more computationally expensive but
has large benefits for some models.
Example Courtesy of www.DynaSupport.com
3.4 IMPLICIT MESH VERSUS EXPLICIT MESH CHARACTERISTICS

3.4.1 INSTRUCTOR LED WORKSHOP II: IMPLICIT VERSUS EXPLICIT MESH DIFFERENCES
Meshing for Accuracy

• Solution time (number of nodes + time step) is often
one of the most important considerations in setting up
an explicit analysis; care should be exercised in setting
up the mesh density.
• A good implicit mesh does not typically work well for an
explicit analysis.
• In an explicit analysis, linear, elastic stresses are not
often the most important analysis result. Typically,
plastic strain, energy, crushing depth, etc. are more
important. These parameters are not as mesh sensitive
as linear, elastic stresses and permit a much larger
element size to be used.
Since the time step is controlled by wave propagation, the
mesh should be graded gradually to likewise allow a
smooth wave propagation through the structure whenever
possible.
Analyst’s Note: Mass scaling is great but it needs to be
combined with a reasonable mesh gradient.
4. FIRST LS-DYNA MODEL: GETTING STARTED

4.1 LS-DYNA KEYWORD MANUAL
LS-DYNA has perhaps one of the most basic learning methods. It is

organic. One simply has to dig in and learn the basics and there is no
substitute for doing it yourself. The Keyword Manual also provides
recommended usage guidelines and examples on how to use the
commands. It is your first and best resource.
4.2 KEYWORD SYNTAX
• Commands are strings of words separated by an underscore, e.g.,

Required Commands:
*BOUNDARY_PRESCRIBED_MOTION_RIGID. *KEYWORD
• Text can be uppercase or lowercase *CONTROL_TERMINATION
• Commands are arranged alphabetically in User's Manual *NODE
• Order of commands in input deck is unimportant (except *ELEMENT
*KEYWORD and *DEFINE_TABLE)
*SECTION
• Keyword command must be left justified, starting with an
asterisk *MAT
• A "$" in the first column indicates a comment *PART
• Input values can be in fixed fields or comma-delimited *DATABASE- BINARY- D3PLOT
• A blank or zero parameter indicates that the default value of *END
parameter will be used (or taken from *CONTROL_option)
4.3 UNITS
Many a fine analysis model has been brought down by bad units. Although one may wonder why in this modern age one
still has to twiddle with units and not have it addressed by the interface is philosophical-like engineering debate between
the ability to hand-edit the “deck” or be hand-cuffed to a gui (pronounced “gooey”) interface. Moving past this
discussion, to use LS-DYNA effectively, one should have a rock-solid and un-shakable conviction in your chosen system of
units.
Since the majority of LS-DYNA work is dynamic, the analyst will often be looking at the energies of the system or
velocities, in addition to displacements and stresses. Hence, a consistent set of units that are easy to follow can provide
significant relief in the debugging of an errant analysis. A general guide to units can be viewed within the Class Reference
Notes / Units (see Consistent units — LS-DYNA Support.pdf). Saying all that, here are the four unit systems that I have
standardized on for analysis work. It doesn’t mean they are the best but at least they are generally accepted.
Consistent Unit Sets for LS-DYNA Analysis
Mass Length Time Force Stress Energy Density Steel Young’s Gravity
kg m s N Pa J 7,800 2.07e+9 9.806
g mm ms N MPa N-mm 7.83e-03 2.07e+05 9.806e-03
Ton (1,000 kg) mm s N MPa N-mm 7.83e-09 2.07e+05 9.806e+03
Lbf-s2/in (snail) in s lbf psi lbf-in 7.33e-04 3.00e+07 386
4.4 REFERENCE MATERIALS AND PROGRAM DOWNLOAD
The first site to visit: www.lsdynasupport.com LSTC Program Download Site

Another great site: www.dynasupport.com ftp://user:computer@ftp.lstc.com
LS-DYNA Examples: www.DYNAExamples.com SMP Version: ls-dyna
LS-DYNA Conference Papers: www.dynalook.com MPP Version: mpp-dyna
Newsletter: www.FEAInformation.com SMP/Windows: pc-dyna
Yahoo Discussion Group: LS-DYNA@yahoogroups.com
Aerospace Working Group: awg.lstc.com
4.5 SUBMITTING LS-DYNA ANALYSIS JOBS AND SENSE SWITCHES
Analysis jobs can be submitted directly with command line syntax or using the
Windows manger (shown on the right).
While LS-DYNA is running, the user can interrupt the analysis and request
mid-analysis information. This interrupt is initiated by typing ctrl-c on
keyboard and then a "sense switch“ can be activated by typing the following:
• sw1 A restart file is written and LS-DYNA terminates
• sw2 LS-DYNA responds with current job statistics
• sw3 A restart file is written and LS-DYNA continues
• sw4 A plot state is written and LS-DYNA continues
• swa Dump contents of ASCII output buffers
• stop Write a plot state and terminate
4.6 WORKSHOP I: LS-DYNA GETTING STARTED EXAMPLE
This workshop uses the LSTC Getting Started Example

material and a LS-DYNA model has been prepared. This
material can be found in the Students’ “Class Reference
Notes” folder.
Goals
• Open Windows Notepad and build LS-DYNA
Keyword deck by hand-entry. Use comma
separated format.
• For each command, consult the Keyword Manual.
• Analyze your model.
• Post process the results within LS-PrePost
• If time exists proceed to other examples.
5. EXPLICIT ELEMENT TECHNOLOGY

5.1 ELEMENT TYPES IN LS-DYNA
There are many different element types in LS-DYNA:

• Point elements (mass, inertia)
• Discrete elements (springs, dampers)
• Beams
• Solids (20 and 3D, Lagrangian, Eulerian, ALE)
• Shells
• Thick Shells (8 noded)
• Seatbelts (and related components)
• EFG and SPH (meshless)
Extremely Brief Recommendations

• Hughes-Liu Integrated Beam, ELFORM=1, is default.
Stresses are calculated at the mid-span of the beam.
Special requirements for stress output.
• For solid elements, the default is ELFORM=1 and uses Detailed Element Recommendations (see Student’s Class
one-point Guassian Integration (constant) stress. Reference Notes)
This element is excellent for very large deformations.
Review of Solid Element Formulations Erhart.pdf
It is the standard recommend for explicit simulations.
Aerospace Working Group - Aerospace_MGD_v12-1.pdf
• Shell elements are covered in detail.
5.2 ONE GUASSIAN POINT ISOPARAMETRIC SHELL ELEMENTS AND HOURGLASSING

5.2.1 INSTRUCTOR LED WORKSHOP III: EXPLICIT ELEMENT TECHNOLOGY | A: SIDE BENDING
Isoparametric Shell Elements

Default element is one Guass point in-plane (ELFORM=2)
• This default formulation is efficient and generally the most
robust formulation for large deformations.
• The example shows that under-integrated elements have severe
problems in bending. The recommended number of through
thickness elements is three (3). However, fully integrated
(ELFORM=16) does an adequate job with one or two.
Computationally 3x more expensive than the default
formulation (ELFORM=2).
• Importantly, it is not always possible to use only ELFORM=16
due to computational expense and care must be taken with
using the default formulation in situations where only one
element through thickness is possible.
• Increasing the number of elements can be problematic due the
CFL timestep condition since three elements over a narrow
width of strip will always cause a severe reduction in timestep.
• Recommended size is 5 mm for steel and aluminum and thus
yields a time step of approximately 1 s.
This section courtesy of LSTC and Paul Du Bois, Hermes Engineering NV
5.2.2 INSTRUCTOR LED WORKSHOP III: EXPLICIT ELEMENT TECHNOLOGY | B: OUT-OF-PLANE BENDING WITH PLASTICITY
Isoparametric Shell Elements*

• Only two element formulations are recommended:
ELFORM=2 and ELFORM=16 (keep it simple).
• Number of through-thickness integration points (NIP)
controlled by user:
NIP 1: Membrane Behavior
NIP 2: Linearly Elastic Behavior (default)
NIP 3+: Recommended for Nonlinear Materials
Optimum NIP for Nonlinear Plasticity = 5
• ELFORM=2 not well-suited to warped geometries
unless BWC warping stiffness with full projection is
invoked or use ELFORM=16. Warping is something
that is troublesome whether the analysis is implicit
or explicit.
5.2.3 WORKSHOP: BUILDING THE BETTER BEAM

Objective
The importance of mesh density and plate out-of-plane
integration will be demonstrated through the use of this
simply-support I-beam model (half-symmetry). The material
model is steel with a yield stress of 100,000 psi and a
tangent modulus of 200,000 psi. The workshop will start
with a Femap model shown on the right.
Tasks
• Setup LS-DYNA Analysis (Femap Analysis Set Manager)
with a termination time = 0.05 and Output Time Interval
=0.0001.
• Measure maximum displacement at the end of the beam
and record this value on the table below.
• Then, increase mesh density in web, record maximum
displacement. Lastly increase Plate Integration Points = 5
and record maximum displacement.
Extra Task: Open up LS-DYNA deck in LSPP (File / Open /

Model Mesh Density Integration Max. Disp. LS-DYNA Keyword File) and look at the Keyword *SECTION
Start 1 0 60 (Model / Keywrd / *SECTION and note where the plate
integration points are defined, read the manual if
Refine 4 0 3 necessary for clarification. Go back to one element thru
5 the web and change the element type to full-integration,
fast.
5.2.4 WORKSHOP II: HOURGLASS CONTROL/HOURGLASS
Isoparametric Shell/Solid Elements and Hourglassing*

• All under-integrated isoparametric elements (one Guass
point) have hourglassing present. It is a non-physical “zero-
energy” mode of deformation.
• Fully Integrated formulations do not hourglass. Additionally,
tetrahedron and triangular elements do not hourglass but are
overly stiff in many applications.
• *Control_Hourglass or *Hourglass to set hourglass control.
• Use default unless additional documentation is consulted
(e.g., see Review of Solid Element Formulations Erhart.pdf
(Class Reference Notes / Solid Elements).
• Hourglass energy should be less than 10% of the internal
energy at any stage of the analysis (use *CONTROL_ENERGY
(HGEN=2) to calculate hourglass energy).
• In LSPP, check glstat for total hourglass energy and then
matsum for individual part energy.
How to Limit Hourglassing Workshop II: Hourglass Control

• Apply pressures instead of point loads. • Evaluate current model for hourglassing.
• Refine mesh Plot internal energy and hourglass energy.
• Selectively use ELFORM=16 (3x computational cost) • Read Hourglass Material.
• Attempt fix with different hourglass type.
• Switch ELFORM to 16.
5.2.5 WORKSHOP III: MESHING FOR EXPLICIT SUCCESS | MESHING FOR EXPLICIT SUCCESS START.MODFEM
Element Quality for Explicit Analysis

Orthogonal Meshes
In explicit analysis, regular meshes are important to represent
the wave propagation problem with minimal dispersion. When
the mesh is irregular, the wave front will become dispersed and
degrade the quality of the results.
Ideally, mesh lines should be orthogonal or parallel to the stress
wave front (and thus orthogonal ).
Note: The stress wave front is that induced by the dominant load.
Workshop III: Mesh for Explicit Success
The student meshes the geometry to obtain the best quality
results possible with the minimum element count and CPU cost.
Tasks
• Mesh geometry using a mesh sizing of 5 mm.
• Setup Analysis Manager with a termination time of 10
milliseconds with data output at 50 s (0.050 ms) intervals.
• Review results and plot resultant displacement at the top
corner node where the load was applied (+Y).
• Investigate the following by recording the effects of: i) Plate
NIP, ii) Mesh Density, iii) ELFORM & iv) Mass Scaling on the
displacement at the end of the beam.
5.3 SCALAR ELEMENTS (NASTRAN CBUSH EQUIVALENTS)

The CBUSH element is of such general utility that it merits its own special section on how to obtain equivalent behavior
within LS-DYNA. One of the concepts that merit attention is that an explicit analysis always requires mass whereas a
static implicit analysis only requires stiffness. Hence, in a static analysis, one can have elements with zero length (i.e.,
zero mass) whereas in an explicit analysis mass must be present and likewise a finite length.
Although LS-DYNA has several methodologies to arrive at simulation the behavior of Nastran CBUSH element (e.g.,
*ELEMENT_DISCRETE), we will present the most basic method and the one recommended by LSTC. In Nastran, the
CBUSH property card specifies orientation and stiffness. In LS-DYNA, these capabilities are handled by two cards: (i)
*ELEMENT_SECTION, ELFORM=6 (orientation) and (ii) *MAT_LINEAR_ELASTIC_BEAM (stiffness values).
Nastran CBUSH LS-DYNA Equivalent *SECTION_BEAM & *MAT_66
If the connection has zero length, then the *SECTION_BEAM (ELFORM=6) field SCOOR should be set to a value
corresponding to the requirements of the analysis. In other words, take a look at the manual and www.dynasupport.com
under discrete beams since the choice of SCOOR is not obvious. Our general default is SCOOR=3. Importantly, a CID
must be defined or if CID=0, then the beam follows the global coordinate system (just like Nastran).
The mass of the element is calculated the field VOL and INER (if torque is present) and the mass density of the material.
A basic element is given below for a zero length “CBUSH” element in a unique coordinate system:
Nastran CBUSH Property Card Equivalent LS-DYNA Keywords

*SECTION_BEAM_TITLE
CBUSH Equivalent Element
$# secid elform shrf qr/irid cst scoor nsm
1 6 1.000000 2 2 3.000000 0.000
$# vol iner cid ca offset rrcon srcon trcon
1.000000 1.000000 1 0.000 0.000 0.000 0.000 0.000
*ELEMENT_BEAM
$# eid pid n1 n2 n3 rt1 rr1 rt2 rr2 local
1 1 400 401 0 0 0 0 0 2
*DEFINE_COORDINATE_SYSTEM
4, 488.999, -28.99994, 414.4192, -105.6686, -180.7164, 227.1846
670.4222, 17.28623, -199.297
*MAT_LINEAR_ELASTIC_DISCRETE_BEAM_TITLE
Spring Stiffnesses
$# mid ro tkr tks tkt rkr rks rkt
1 0.001000 0.000 1.0000E+6 1.0000E+6 0.000 0.000 0.000
$# tdr tds tdt rdr rds rdt
0.000 0.000 0.000 0.000 0.000 0.000
$# for fos fot mor mos mot
0.000 0.000 0.000 0.000 0.000 0.000
Some things to note is that a third node is not defined within the *ELEMENT_BEAM card since the orientation of the
element is handled by the CID definition. One will note that VOL and INER are both given values of 1 for simplicity since
the mass of the element is then controlled by just the mass density on the material card. And, don’t forget that a *PART
card is also required to tie together the *SECTION and *MAT cards. In this example the coordinate system is fixed and
does not rotate with the system.
5.3.1 WORKSHOP DISCRETE ELEMENTS (NON-LINEAR BUCKLING ANALYSIS)

Femap Model LS-DYNA Non-Linear Buckling Results
6. LS-PREPOST
6.1 WORKSHOP IV: LS-PREPOST | WORKSHOP 7 & 8
Introduction to LS-PrePost
Reference and tutorial materials are provided at the www.LSTC.com
site.
Model manipulation is by Shift Key and the left, right and middle mouse
button.
Class Referenced Note’s Section
The detailed usage of LS-PrePost in its own rights is a two day class (see
Class Reference Notes / LS-PrePost Introduction Manual / LS-PrePost
Introduction.pdf).
Workshop Goal: To Get Comfortable with LSPP Post-Processing

• Open PDF file and review contents. Take 10 minutes.
• Walk through PDF until arriving at Workshop 7.
• Perform exercises as listed as Bookmarks.
7. MATERIAL MODELING
7.1 PART 1. METALS
7.1.1 ENGINEERING STRESS-STRAIN VS TRUE STRESS-STRAIN
The constitutive large-strain modeling of all materials is based on

the true stress-strain response of the material. Starting with a
simple tensile test, the engineering stress strain is just the force
over the original cross-sectional area of the coupon. The true
stress-strain response accounts for the necking of the cross-section
and can be elastically stated as shown on the graph on the right.
This method has its limitations and once the coupon starts to neck
locally, this approach is no longer valid and an iterative approach
must be used to calculate the true stress-strain response.
In many simulations, it is not exactly critical to have the necking
response accurately characterized since once the material starts to
neck we only have the end point data available and hence we draw
a straight line from the initiation of necking to failure.
7.1.2 REVIEW OF MATERIAL MODELS AVAILABLE IN LS-DYNA

The broad array of material models in LS-DYNA can be Major Categories
over-whelming. In the material manual they are listed in (LS-DYNA_manual_Vol_II_R7.0.pdf)
numerical order based on their insertion into the code.
Hence, the elastic material model (*MAT_001) was the
first material model developed. These earlier models are • Elastic • Heart/Lung/Tissue
well validated since they have been used extensively • Elastic-Plastic • Acoustic material
over the years. Later material models, e.g., *MAT_181 • Rigid • Fabric
Simplified Rubber, was developed in early 2000, and
• Orthotropic/Anisotropic • Concrete/Soil
although a somewhat recent development, it has seen
• Hyperelastic (Rubber) • High Explosives
wide-spread usage due to its advanced formulation and
robustness. For elastomers, it is my default choice. • Foams • Laminated Glass
• Composites • User-defined
Although many material models exist for metals, one of
the most robust models is *MAT_024 or • Viscoelastic
*MAT_PIECEWISE_LINEAR_PLASTICITY. This material
model is the standard workhorse and is the
recommended starting point for elastic-plastic simulation
of metals and general plastics since it can also handle
viscoelastic behavior (i.e., strain-rate dependency).
Note: It is recommended practice to build a pilot model

using the material law of choice and replicate the
material’s experimental data.
7.1.3 MATERIAL FAILURE AND EXPERIMENTAL CORRELATION
Material Failure
Simulation of material failure is a broad research avenue.
Prediction of brittle material failure, as shown by the glass
shatter patterns on the right, can be extremely mesh
sensitive. For ductile materials, the failure prediction is
more robust since ferrous and non-ferrous materials often
have high-energy absorption characteristics prior to failure.
That is to say, metallic materials generally tend to tear
while brittle materials tend to snap.
Image from article: A finite element model for impact simulation with laminated
glass, Timmel, Kolling, Osterrieder, DuBois, IJIE, 2006.
Experimental Correlation
The development of a material model often starts with
stress-strain data from a standard mechanical test. This
data can then be converted into true stress-strain and a
very simplified approach used to extend this curve to
failure.
In the high-strength steel shown on the right (see Class
Reference Notes; MMPDS, AerMet 100 Steel Bar), the
stress-strain curve drops as necking occurs. The presented
data is all that is needed to generate a useful true stress-
strain curve for the material.
7.2 WORKSHOP V: ELASTIC-PLASTIC MATERIAL FAILURE
Objective: To approximately model the material

deformation and failure of the AerMet 100 steel.
A simplified approach is presented using the piecewise
linear plasticity approach available in *MAT_024. Data is
estimated from the presented curve and corrected to true
stress-strain up to the point of necking. After necking, a
straight line is drawn to the ultimate failure point. The final
processed, true stress-strain curve is shown on the right.
The data spreadsheet can be found within the Workshop
file folder.
Goal: Get Comfortable With Constitutive Modeling

Run initial model (Elastic-Plastic Failure Start.dyn) and
investigate the current material law. Plot von Mises and
Plastic Strain (History Plots).
Change material law to Piecewise Linear Plasticity using
Excel Spreadsheet or PDF file in Workshop Directory and
rerun model. Note correct format for entering plastic
strain.
Compare results, check hourglass energy balance and
validate material law model.
7.3 STRAIN RATE SENSITIVITY OF METALS
• This mechanical metallurgical behavior is due to the

movement of dislocations within the crystalline
structure. Dislocations move up to the speed of sound
within the metal.
• The graph on the right provides a rough-order-of-
magnitude idea of how strain rate affects the true
stress in steels and aluminum. The strain rate is in
seconds.
• An example of strain rate effects might be that for
rotor burst. The rotor
spins at 55,000 rpm. The
tip velocity of the turbine
blade at burst is 575 m/s.
As the blade impacts the
containment ring, one
could expect to see
significant strain rate
effects.
• For example, given a 10 mm tall, carbon steel cylinder
that is compressed 10% at 575 m/s. This would give a
strain rate of 575 s-1 or log ε’ = 2.8. From the chart on
the right, one could expect an increase in the yield
stress of >50%.
• Strain-rate effects can be evaluated by numerical
testing (i.e., exploratory work looking at maximums).
7.4 PART 2: ELASTOMERS AND FOAMS

7.4.1 MODELING ELASTOMERS VS FOAMS (VISCOPLASTICITY)
The Material section of the LS-DYNA keyword manual provides a wealth of

practical information on the modeling of elastomers (soft plastics and rubbers)
and foams. Plastic materials have very unusual engineering stress-strain curve
due to necking. However, once corrected for true stress-strain, the curve looks
very reasonable.
The curve on the right is for a uniaxial test performed on rubber. Experimental
uniaxial data can be directly entered into the material card. For the modeling of
rubbers, this is a very common approach.
Viscoplasticity
This concept is fundamental in the
modeling of plastic, rubber or foam
materials. These materials deform via
the stretching of their long-chain hydro-
carbon network. As such, they are very
sensitive to strain rate effects. As the
strain rate increases, their complete
stress/strain curve will shift upward.
Analyst’s Note: Whenever a new material model is simulated, a virtual test

coupon analysis should be done and the results compared to the mechanical test
data.
7.5 MATERIAL MODELS FOR MODELING FOAMS
The following table presents the currently available foam material models in use within LS-DYNA. One recommended
foam model is that of *MAT_083 or *MAT_FU_CHANG_FOAM. Foams are perhaps the most challenging material to
model due to their extreme nonlinear behavior upon loading and unloading plus their tendency to become crushed or
damaged during loading, and then upon unloading, present a completely different stress/strain response.
Foam Material Models Available in LS-DYNA

5 * MAT_SOIL_AND_FOAM 142 *MAT_TRANSVERSELY_ANISOTROPIC_CRUSHABLE_FOAM
14 * MAT_SOIL_AND_FOAM_FAILURE 144 *MAT_PITZER_CRUSHABLE_FOAM
26 *MAT_HONEYCOMB 154 *MAT_DESHPANDE_FLECK_FOAM
126 *MAT_MODIFIED_HONEYCOMB 163 *MAT_MODIFIED_CRUSHABLE_FOAM
53 *MAT_CLOSED_CELL_FOAM 177 *MAT_HILL_FOAM
57 *MAT_LOW_DENSITY_FOAM 178 *MAT_VISCOELASTIC_HILL_FOAM
62 * MAT_VISCOUS_FOAM 179 *MAT_LOW_DENSIlY_SYNTHETIC_FOAM
63 *MAT_CRUSHABLE_FOAM 180 *MAT_LOW_DENSIlY_SYNTHETIC_FOAM_ORTHO
73 *MAT_LOW_DENSIlY_VISCOUS_FOAM
75 *MAT_BILKHU/DUBOIS_FOAM
83 *MAT_FU_CHANG_FOAM
Analyst’s Note: Since foams are modeled using solid elements, it is not uncommon to
have numerical problems as the foam becomes highly compressed and crushed since
the elements used to idealized this behavior, likewise become highly distorted or
crushed. Typically, the workarounds are to use highly structured meshes with large
element sizes. Another technique that is gaining utility is to use SPH (Smooth Particle
Hydrodynamics) to model the foam material. This “mesh-free” technique will be
covered at the end of these course notes.
7.6 WORKSHOP VI: MODELING AN ELASTOMER BALL WITH HEX AND TET ELEMENTS
Solid Element Meshing for Soft Materials:

Hexahedral versus Tetrahedral
Whenever possible, a hex mesh should be used for
the modeling of soft materials. The recommended
element formulation is ELFORM=1 (hex) or =13
(tet). Both are one-point Guassian integration
formulations and can handle large-deformations
without element aspect failure problems (not
negative volume). Another recommendation is to
use ELFORM=-1 for non-perfect hex’s.
Please take a read within: Review of Solid Element
Formulations Erhart.pdf (Class Reference Notes /
Solid Elements) and it has an excellent section on *CONTROL_SOLID |ESORT=1 (?)
Hourglass Control.
Element Quality and Negative Volumes
Large-deformation behavior in soft materials is highly
sensitive to mesh characteristics. In an ideal situation, brick
elements are always preferred due to their regularity of
formation (shape and distribution). Even with a high-quality
mesh, under high compression loading, brick elements can
generate negative volumes. The best solution is to refine the
load application and/or improve the mesh quality. Prior to
embracing any one path, it is recommended to consult
references (i.e., www.DYNASupport.com and Keyword
Manual).
7.7 EQUATION OF STATE (EOS) MATERIAL MODELING

An equation of state is required for materials that undergo significant deformation (can be very large plastic deformation
or a compressible fluid). The Cauchy stress tensor can be separated into a hydrostatic stress tensor ( ) and a deviatoric
stress tensor ( ):
The deviatoric stress is calculated by the material model constitutive law. The pressure term, p, must then come from an
equation of state (EOS). The EOS provides a relationship between pressure and the volume (can also be a relation of
temperature and/or energy). Depending on the compressibility of the material, different types of EOS’s are possible. A
very popular EOS is the Gruneisen equation of state. The full version of this EOS (compression) is:
[ ( ) ]
( )
[ ( ) ( ) ( )]
( )
The equation appears rather complicated at first glance, if we focus on a weakly compressible fluid (many engineering
fluids can be considered this way), the equation of state can be reduced to:
Where is the initial reference density for the material, is the speed of sound in the material and = . All the
other parameters are curve found by curve fitting to a set of compression experiments. These parameters are typically
only needed when the pressure loading is very high as in shock waves. In the figure shown on the next page, we can see
a comparison between an EOS specified with only the speed of sound (linear dependency of pressure on volume) and an
EOS to give a cubic dependency of pressure on volume.
In the subsequent workshops, the units are [ ], [ ] (specified as RO on *MAT_NULL) ,and [ ] (specified
as C0 on *EOS_GRUNEISEN) and is a dimensionless parameter.
Comparison between weakly compressible and compressible
Fluids in LS-DYNA Explicit need to be described by a constitutive material law (such as *MAT_NULL for example) and an
appropriate EOS. The reason is that solving the set of Euler equations (or full Navier-Stokes in the presence of viscocity)
with a strictly explicit time integration scheme requires an equation of state to directly determine the pressure at each
node point. A truly incompressible algorithm requires solving a Poisson equation (elliptical partial differential equation)
to ensure that the flow is divergence free. The Poisson equation can only be solved iteratively or using Implicit time
integration.
All this truly means is that in LS-DYNA SPH Explicit, a fluid that is commonly considered incompressible can be treated as
weakly compressible with a simple EOS by defining only two parameters; initial density and the speed of sound in the
material. Throughout the workshops, we use *MAT_NULL, but other material models such as *MAT_JOHNSON_COOK,
*Elastic_Plastic_Hydrodynamic, etc. can be used with an EOS to describe various engineering materials.
7.7.1 MODELING WATER WITH *EOS_GRUNEISEN AND *MAT_NULL
As basic as this may sound, it is not that obvious. The

technique for modeling water is well described in Class
Reference Notes / Aerospace Working Group /
Aerospace_MGD_v12-1.pdf at page 66, Section 3.9.2
Water. The entry is as simple as the graphic shown on
the right.
The only real difficulty is ensure that one gets the units
handled correctly.
Analyst’s Note: Make sure you remember to set the

*PART card to use the *EOS law.
7.8 MATERIAL FAILURE SIMULATION

7.8.1 BASIC METHODS OF MODELING FAILURE: MATERIAL VERSUS BOND FAILURE
Standard failure in the modeling of materials is

by specifying some sort of material based failure
criterion. My favorite approach is to use the
*MAT_ADD_EROSION card to specify the exact
failure criteria that is needed. For metals, one
approach is to set the EFFEPS (maximum
effective strain at failure) to 3x MXEPS
(maximum principal strain at failure). This
ensures that the material does not prematurely
fails under compressive plastic deformation but
still remains true to the mechanical test data.
Exercise: Open LS-
DYNA_manual_Vol_II_R7.0.pdf (see Class
Reference Notes / Keyword Manuals) and read
the *MAT_ADD_EROSION card section.
Another way of modeling failure is by *CONSTRAINED_TIED_NODES_FAILURE. With this formulation, bond failure can be
modeled in a direct and simple manner by setting the plastic strain required to pull apart the nodes. Of course, this
plastic strain is taken from that elements integration point. The setup for this failure mechanism is to take a clean mesh
and let LSPP create the tied connections. This is done by breaking apart the elements and then tying together the
adjacent nodes. Upon failure, the elements fly apart but are not deleted. An example of this concept is can be found at
www.dynaexamples.com / Intro by J. Reid / Sphere Plate. Given all that, I prefer the simplicity of *MAT_ADD_EROSION.
7.9 WORKSHOP VII: MODELING GENERAL FAILURE
Objective: Modeling of material failure is not as complex as one might think if a reasonable expectation is taken from the
outset. It should be noted that: “All models are wrong, but some models are useful” and hence when trying to replicate
failure in a structure one should strive for upmost simplicity prior to adding complexity. This concept segues into another
saying: “elegant simplicity is deceptively difficult to achieve”. In the following workshop, a basic failure mechanism
within the *MAT_24 card is improved upon by using the *MAT_ADD_EROSION approach.
Model Introduction: A flat bar is held at its end (near the hole) and the rail is given an initial velocity to impact against
the bar and move it upward.9
Workshop Tasks
• Open Flat Bar Bending Start.dyn and inspect material law used for the default simulation.
• Understand unit system since it is not completely uncommon in the world of LS-DYNA.
• Run Model and inspect failure mode. Note: Rigid material modeling technique.
• Delete failure criterion from material card.
• Open LS-DYNA Material Manual and read-up on the
*MAT_ADD_EROSION card. Then, within LSPP add a tensile strain
failure criterion (MXEPS) = 0.25 to the existing material law.
• Rerun model
Student Bonus: Open up LSPP with your final model file and increase
the initial velocity of the bar and note the results.
7.10 MODELING RIGID BODIES

7.10.1 RIGID MATERIALS (*MAT_020 OR *MAT_RIGID)
This is one of the most powerful modeling techniques within LS-DYNA. By

setting bodies (i.e., parts) to use *MAT_RIGID, where deformation and
stresses are not of interest, significant CPU savings can be realized. In the
background, LS-DYNA retains the surface mesh of the part for contact
behavior and calculates an inertia matrix to simulate the dynamic
behavior of the body. What is useful with this approach is that the body
still retains its inertia and physical characteristics as it interacts with other
bodies within the simulation but at a fraction of the numerical cost of
dragging around a fully deformable body. For example, the model on the
right is of a deep drawing operation and only the plate is deformable.
Note: Rigid bodies cannot have constraints applied to them. To constrain
a rigid body, the CMO flag is set within the *MAT_RIGID card.
For a very nicely done reference, please see Class Reference Notes / Rigid
Bodies / LS-DYNA Intro Class Chapter 9 Rigid Bodies.pdf.
Other Common Usage: Rigid Links and Joints
Nastran multi-point-constraints (MPC) equations, of which their two most common flavors are the RBE2 and RBE3
elements, are translated into two different LS-DYNA formulations. Our discussion will just focus on the RBE2 translation.
The RBE2 is translated into a rigid body where the nodes are placed into a group and then constrained rigidly per the
number of dependent DOF’s specified in the RBE2 element. The inertia or mass properties for this nodal rigid body are
obtained from the elements attached to the nodes of the rigid body. Although this may sound a bit odd coming from the
implicit world, in explicit mechanics everything needs a bit of mass to enable its calculation. Hence, for the rigid link to
behave correctly, it borrows mass from its attached elements and two node CNRB’s should be avoided.
The LS-DYNA card that is used is *CONSTRAINED_NODAL_RIGID_BODY (CNRB) with the CMO card specifying what DOF’s
are to be released.
7.10.2 INSTRUCTOR LED WORKSHOP V: RBE2 TO CNRB
A Y-direction load is applied to the

independent node of a RBE2 element around
a hole in a plate model. The corners of the
plate model are pinned. This example will be
used to demonstrate how a RBE2 element is
translated into LS-DYNA and its behavior.
The process is self-discovery and no walk-
through video is provided. The instructor will
walk through the steps and then the student
is left to explore.
Goals: Open Femap model and inspect Rigid
Link. Export to LS-DYNA model and review
Stress results from the default run.
card within LSPP. Run model. Then release Given the model setup, what stress
CNRB in Y-Direction and re-run model. units are we seeing?
Check out the DRFLAG and RRFLAGS.
7.10.3 WORKSHOP VIII: USING RIGID BODIES
The implementation of rigid bodies open doors to making your models run
faster and allows you to focus on what is important and avoid getting
distracted. For complex system level models, the use of rigid bodies is
invaluable and if desired can always be switched later on to deformable.
What You Will Learn
• How Rigid Bodies work within a simulation and how easy it is to switch to
deformable to gather deformation and stress information.
• Defining a fixed movement (i.e., punch movement) as a load case.
• Getting more familiar with the LSPP Interface for making model updates.
Student Led Workshop I: Switching Rigid Bodies to Deformable Bodies

The pendulum model starts out rigid and then is switched to deformable at the
last moment prior to impact. See Class Reference Notes / LSTC General
Examples Manual and Examples / EXAMPLES / DEFORMABLE_TO_RIGID.
This example provides insight in how to do
Restarts (See Class Reference Notes / Keyword
Manuals / LS-DYNA_manual_Vol_I_R7.0.pdf /
Section RESTART INPUT DATA). Two files are
provided that covers the complete run in the
directory noted above. Run the regular model
and then perform a restart. A small movie file
is provided to show how the process works (see
Workshops / Student Led Workshops.
8. CONTACT
8.1 DEFINITION OF CONTACT TYPES
LS-DYNA was developed specifically to solve contact

problems (see Class Reference Notes / History of LS-
DYNA). Contact behavior is enforced by three methods:
(i) Penalty-Based using finite springs (see graphic); (ii)
Constraint-Based where no penetration is allowed and
cannot be used with rigid bodies; and (iii) Tied Contact
(discussed within its own section). The most commonly
used approach is that of the penalty method.
Contact can be effortlessly implemented or it can be
bewitching in complexity. A reasonable treatment of
contact is a multi-day course in itself. To start the
learning process see Class Reference Notes / Contact
User’s Guide / Contact User's Guide.pdf.
Efficient Contact Modeling

Whenever possible, interferences between parts should
be avoided. It is standard contact practice that any initial
interference is removed (nodes are shifted) and as such,
sharp stress spikes can occur where parts/plates overlap.
Setting up contact surfaces appropriately that account for
plate thickness can be time consuming. If necessary,
contact thickness can be overridden within the *CONTACT
Keyword card or tracked via IGNORE=1
8.2 GENERAL CONTACT TYPES
Although there are numerous contact types given in the Keyword Manual, these Classical Contact
workhorse formulations are recommended: “Keeping Nodes on the Right Side”
1. *CONTACT_AUTOMATIC_GENERAL
2. *CONTACT_AUTOMATIC_SINGLE_SURFACE
3. *CONTACT_AUTOMATIC_SURFACE_TO_SURFACE.
The first formulation is the “kitchen sink” and is computationally expensive but is
very robust and will enforce contact between beam elements and other
components. It is a single-surface formulation and only the slave side is defined
(it assumes that everything contacts everything else).
The second formulation is a less CPU expensive form of GENERAL (e.g., edge-to-
edge contact is not checked). While the last formulation is general purpose and
numerically efficient.
8.2.1 ADDITIONAL OPTIONS: SOFT=2 “THE DEFAULT”

The standard contact search routine is based on nodes looking for faces (i.e.,
segments). Sometimes, contact might be missed for a few steps, and when
finally engaged a large restoring force is required to separate the interfaces or
contact might just not occur. Additionally, the standard penalty approach
calculates the spring stiffness based on global material stiffnesses. This can lead
to contact instability in soft materials.
To counter these problems, the contact option SOFT=2 is recommended. This
switches the contact search routine to segment-to-segment and locally calculates
the stiffness for the penalty approach. It should be considered the “default”.
8.2.2 CONTACT WHEN THINGS ERODE

For numerical efficiency, the contact algorithm only looks at free edges and faces. If element erosion occurs (i.e.,
element failure), the standard contact algorithm is not prepared to look for contact on these newly generated faces. If
one knows ahead of time, then the contact can be switched to _ERODING with EROSOP=1. The EROSOP option is
required to allocate memory storage for the newly created element surfaces.
This option should not be evoked for most element erosion situations since it is not very numerically efficient. For
example, plate models are typically just fine without using element erosion, but as shown below, it is critical.
_AUTOMATIC_SINGLE_SURFACE _ERODING_SINGLE_SURFACE
8.3 WORKSHOP IX: UNDERSTANDING BASIC CONTACT MECHANICS
Often times, the challenge to modeling contact is not setting up the contact Keywords Discussed
model but checking the results. In this workshop, the goal is to verify the contact *DATABASE
behavior and plot the contact force for each tube. The contact behavior should
_ASCII (GLSTAT, RCFORCE, SLEOUT,
make good engineering sense.
RWFORC)
Workshop Tasks _BINARY_INTFOR
• Run model (Pipe on Pipe Contact Start.dyn) and look at contact behavior.
• Investigate logical contact behavior and change to Automatic and try using
*RIGIDWALL_GEOMETRIC_FLAT
SOFT=2 (FINISH I)
• Measure contact forces between the parts (*CONTACT_FORCE_TRANSDUCER)
and set ASCII file parameters RCFORCE & RWFORCE = 1e-6 (FINISH II)
• Plot rigid wall impact force (RWFORCE) via PADD on XY Plot Dialog
• Lastly, if one has time, setup model to show contact pressure (see Student
Bonus Option).
8.3.1 STUDENT BONUS OPTION: CONTOURING INTERFACE PRESSURES
LS-DYNA doesn’t automatically generate interface pressures developed during

contact. To obtain this information, three items are required: (i)
*DATABASE_BINARY_INTFOR DT={time interval} must be set; (ii) Print flag(s) on
card 1 of *CONTACT_ must be set to SPR=1 and/or MPR=1; and (iii) and upon
analysis (LS-DYNA Program Manager), one must use the Advanced tab and
include the Interface Force s=(provide your own filename) option. This creates a
separate binary file can then be read by LS-PrePost as a separate post-processed
item (just load the filename specified above).
8.3.2 WORKSHOP IX: COMPARISON OF RESULTS
START FINISH I FINISH II
*CONTACT_FORCE_
*CONTACT_SINGLE_SURFACE *CONTACT_AUTOMATIC_
TRANSDUCER w/ ASCII outputs
SINGLE_SURFACE (SOFT=2)
RCFORC & RWFORC 1E-6
Interpenetration between surfaces is With _AUTOMATIC and SOFT=2, the Although _SINGLE_SURFACE is used,
noted which should never happen. correct contact behavior is noted. one can still obtain surface-to-surface
contact forces. The rigid wall forces
are also reported with the ASCII option
RWFORC. Note that the rigid wall
force is much greater than the tube-to-
tube force since it represents the
combined force of both tubes hitting
the “ground floor”.
FINISH III FINISH IV FINISH V
*SECTION_SHELL (Thickness=50) *CONTACT_AUTOMATIC_ *CONTACT_AUTOMATIC_

SURFACE_TO_SURFACE SURFACE_TO_SURFACE (SOFT=2)
With _SINGLE_SURFACE contact, With _SURFACE_TO_SURFACE contact, With the Soft=2 option
initial interpenetrations are not Interpenetration is enforced and the interpenetrations are gracefully
instantly removed but automatically slave surface nodes are immediately handled (i.e., tracked). It is what
tracked. This is equivalent of setting moved to accommodate the profile of makes the Soft=2 option quite useful in
the IGNORE=1 option. the master surface. Note: One should many contact applications.
add back in SINGLE_SURFACE contact
for the top tube with SOFT=2.
Analyst Note: The reason that so many contact options exist is partly due to legacy requirements and also for numerical
efficiency. When working with large assemblies, the contact algorithm can often represent the most numerically costly
part of the simulation (e.g., >50%). To pare down run time, it can be very handy to remove the SOFT option and to
explore the many other more economical contact algorithms; such as NODES_TO_SURFACE or to remove the
_AUTOMATIC option.
8.4 INSTRUCTOR LED WORKSHOP VI CONTACT: NEGATIVE SLIDING INTERFACE ENERGY
Contact sliding energy has three contributions: (i) interfacial forces; (ii) energy dissipated due to friction (FS ≠ 0.0); and
(iii) energy dissipated by contact damping (on the Contact Card VDC ≠ 0.0). In a well-defined model, the sliding interface
energy (SLEOUT) should be positive and perhaps no more than 10% of the total internal energy of the model. However, a
common numerical pathology for contact is to generate negative sliding energy and is a “red-flag” for that particular
contact. Numerically, negative sliding energy is typically generated by: (i) parts sliding past each other (not friction) and
the penalty method has difficulty maintaining the surfaces apart (interpenetrations) and (ii) rough mesh, large time step,
etc. where the contact behavior has trouble pushing the surfaces apart cleanly and can get numerically lost in which
direction to push the surfaces apart.
Analysts Note: It is common practice to report Sliding Energy in comparison to the analysis internal energy and also to
provide a plot of individual Sliding Energy for each contact (SLEOUT).
How to Fix (when it is more than a couple of percent of internal energy but even then it depends)
Eliminate initial penetrations, check for redundant contacts, reduce the time step scale factor (*CONTROL_TIMESTEP
TSSFAC <0.9), refine the mesh, set contact card options back to default except SOFT=1 and IGNORE=1, explore the use of
the DEPTH option.
8.5 INTERMEDIATE CONTACT ANALYSIS
8.5.1 BEAM AND EDGE-TO-EDGE CONTACT MODELING Where Regular _AUTOMATIC Contact Fails
*CONTACT_AUTOMATIC_GENERAL
Beam and direct edge-to-edge contact creates special challenges since
contact is based on a line between two nodes. For beam elements, the
contact surface is enforced as a cylinder regardless of the beam’s cross-
section. The algorithm checks contact along the length of the beam and at
its end. With this capability, all of the more complex interactions shown
on the right are found and prevented. These are other features to this
contact such as
_AUTOMATIC_GENERAL is a “single-surface” contact and the user only
defines the slave set for contact.
8.5.2 SPECIAL CONTACT OPTIONS
*CONTACT_AUTOMATIC_ End Contact Occurs for all “Automatic”
To handle contact at the end of plates, a semi-circular projection is made Contacts via Projection
as shown on the graphic on the right for all _AUTOMATIC contacts. The
radius of the projection is ½ the plate thickness and the option SHLEDG
lets the user chose between round or square edge.
ISYM
This option controls how the contact algorithm handles symmetry edges.
If your model has a symmetry plane enforced by SPC constraints, you’ll
want to set ISYM=1.
*CONTROL_CONTACT This end contact option is the “gotcha” when
constructing the “perfect mesh”
Allows “one-stop-shopping” to set your contact options.
8.6 WORKSHOP X: EDGE-TO-EDGE CONTACT
Problem Statement
A thin (0.002 m) corrugated plate is slammed into two other plates using a recommended default contact algorithm of
_AUTOMATIC_SINGLE_SURFACE. Upon inspection, it doesn’t work. How
would you fix it?
Script
• Run model and look at contact behavior.
• Change contact to _AUTOMATIC_GENERAL
• Re-analyze model and check contact
Analyst’s Note: One may note that LS-DYNA is warning you that the time step
should not exceed 1.4e-5 while the model’s running time step is 2.1e-4.
• Reduce timestep via TSSFAC to 0.5 (add *CONTROL_TIMESTEP card)
• Re-analyze and then lastly remove the interpenetration by “translation”
8.7 TIED CONTACT FOR MESH TRANSITIONS, WELDING AND GLUING

8.7.1 TIED CONTACT OR GLUING
Given the idealization difficulty of system modeling, the ability

to tie together different mesh densities (e.g., hex-to-hex or tet-
to-hex), snap together parts along a weld-line or just glue Analyst’s Note: Tied contacts are not really
sections together (e.g., plate edge to a solid mesh) is an “contacts” but a constraint relationship that uses the
amazingly useful ability and LS-DYNA provides a very complete *CONTACT card entry format. For an explicit
Tied Contact tool box to work with. analysis, the constraint option ties the adjacent nodal
The emphasis of this course to provide an overview of the basics velocities together while for implicit, the
to get started efficiently with LS-DYNA, a short list of displacements are tied. This explains why the nodes
recommended *KEYWORDS for Tied Contact are presented that must be on the same plane or adjacent and also why
work for both implicit and explicit solution sequences. this formulation can’t be used with rigid bodies or
have SPCs attached to any node that is tied.
When the Mesh is Co-Planar (Translational DOF Tied) Additionally, it is only generally applicable for just
• *CONTACT_TIED_SURFACE_TO_SURFACE translational DOF (TX, TY & TZ).
• *CONTACT_TIED_NODES_TO_SURFACE With the OFFSET formulation, the penalty method is
used. If the BEAM option is employed all six DOF’s
are tied together by essentially using little springs
When the Mesh is Co-Planar (All Six DOF Tied)
between the nodes. This is the most computationally
• *CONTACT_TIED_SHELL_EDGE_TO_SURFACE expensive tied contact but the most robust. Since the
tied contact is enforced by springs, the nodes can be
When the Mesh is Offset (All Six DOF Tied) offset, rigid bodies can be tied together and even
• *CONTACT_TIED_SURFACE_TO_SURFACE_BEAM_OFFSET SPCs can be applied to the tied nodes.
• *CONTACT_TIED_SHELL_EDGE_TO_SURFACE_BEAM_OFFSET The reason that we like constraint Tied Contact is
that it is stable whereas penalty Tied Contact _Offset
The utility of using a very brief subset is that one can build up is penalty based and has all the standard pathologies
experience and confidence without the expense of trying out a of regular contact such as the possibility of creating
rather daunting list of Tied Contact Options. negative sliding interface energy.
8.8 SUMMARY TABLE FOR TIED CONTACT
*CONTACT_TIED_ Recommended Usage Type Pros/Cons

Gluing solid mesh transitions Pros: Provides smooth displacement and
together where the two stress interpolation across dissimilar meshes
_SURFACE_TO_SURFACE Constraint
meshes are co- between hex-to-hex or tet-to-hex. Cons: If
planar/adjacent. the meshes are co-planar, there are no cons.
Useful for creating weld edge Pros: Allows the logical modeling of edge
lines between two solid contact between solid parts. If the meshes
_NODES_T0_SURFACE elements parts that are co- Constraint are co-planar, there are no cons. Cons: Not to
planar; simulates a “fillet be used with plate or beam elements since
weld”. rotational DOF are not correctly handled.
Welding plate or beam nodes
Pros: Handles all six DOF’s using a constrain
together when the mesh is co-
_SHELL_EDGE_TO_SURFACE Constraint method. Cons: Not designed for solid
planar and captures all six
elements.
DOF.
Pros: Allows one to glue together dissimilar
_SURFACE_TO_SURFACE_ Gluing solid mesh transitions
meshes that are offset. Also one may glue
when the meshes are not co- Penalty
BEAM_OFFSET deformable bodies onto rigid bodies. Cons:
planar.
Not suitable for plate and beam connections.
Pros: The grand slam of tied contact. Handles
Ideal for welding together
all six DOF and offsets. Typically very
_SHELL_EDGE_SURFACE_ plates/beams or for
numerically stable with little negative sliding
plates/beams onto solids Penalty
BEAM_OFFSET interface energy. Also may glue deformable
whether deformable and/or
bodies onto rigid bodies. Cons: None except
rigid with offsets.
that it is penalty based.
8.9 WORKSHOP XI-A: TIED CONTACT FOR HEX-TO-TET MESH TRANSITIONS (TIED_SURFACE_TO_SURFACE)
Background for TIED Contact Analysis

The advantages of using a hex mesh for explicit work centers on better shape control during large deformation and the
ability to maintain a larger time step. The last items is often pivotal in keeping your solution running fast without having
the program add excessive mass if automatic mass scaling is invoked (*CONTROL_TIMESTEP (DTMS = negative timestep
value)). In the implicit world, the use of hex elements are desired for the ability to provide equivalent stress mapping
using far less nodes (eight node brick versus the use of five 10-node tetrahedrals to fill the same space or 8/26 nodes)
and often times, cleaner stress contours. This workshop shows how to setup the mesh transition and run the analysis
using the implicit solution.
Workshop Script
A small endoscopic, surgical stapling anvil is given a 40 lbf
QC proof load. The yield stress of the powder metallurgy SS
is 110 ksi. Is this device safe for surgical use?
• One starts with the provided Femap model. Inspect how
contact was defined. Export model using the LS-DYNA
translator.
• Open model in LSPP, inspect how contact was translated
and verify *CONTROL cards for analysis.
• Run Analysis and contour stresses
• Use the XY plot capability to verify reaction forces
• Would you recommend this design?
Export to LSPP and Verify Loads, Constraints and Contact Run Implicit Analysis and Verify Load
This image created by Application / Model Checking / General After contour stresses, make an XY plot SPC force plot by
Checking / Contact Check. Highlight the “Contact” and hit the summing the SPC’s (see ASCII button).
Check button.
Keyword Commands Used

*CONTROL_IMPLICIT_GENERAL (IMFLAG=1), *CONTACT_TIED_SURFACE_TO_SURFACE, *DATABASE_ASCII (SPCFORC=1)
8.9.1 STUDENT BONUS

If this is working for you, change the DTO field to 0.1 within the *CONTROL_IMPLICIT_GENERAL card and watch what
happens. Then, change the TIED contact to that of *CONTACT_TIED_SHELL_EDGE_TO_SURFACE_BEAM_OFFSET. If you
are wondering what is what, read the manual.
Another avenue is to investigate how plasticity affects the analysis. Implement material plasticity using a yield stress of
100,000 psi with a tangent modulus of 100,000 psi. If this is too boring, change the yield stress to 30,000 psi with a
tangent modulus of 10,000 psi.
8.10 WORKSHOP XI-B: TIED CONTACT FOR GLUING THINGS TOGETHER (BEAM_OFFSET)
This workshop takes a typical example of where to use Tied Contact and walks through the setup and debugging process.
A comparison is also provided between the Nastran and LS-DYNA model.
Script: Review contact setup within Femap; Run NX Nastran Analysis; Set Analysis Manager to LS-DYNA model and
export; Run LS-DYNA model and interrogate results; Change contact to _BEAM_OFFSET; Verify Tied Status and correct
accordingly; Re-analyze model and interrogate results.
Analyst’s Note: The Tied option considers a node “tied” if it is within 5% of the element’s thickness. This applies to all
_TIED formulations. As mentioned, the constraint option moves the slave node to be adjacent to the master surface while
the offset option accounts for the gap; but whether or not it is tied, depends upon the separation of the nodes. To
override the default setting, one can set the SST and MST to a negative number that reflects the absolute distance to
search for a tie relationship between the slave and master nodes.
Required Reading: Class Reference Notes / Keyword Manuals / LS-DYNA_manual_Vol_I_R7.0.pdf – General Remarks
on *CONTACT on page 549 to 555.
8.10.1 INSTRUCTOR LED BONUS WORKSHOP BAD ENERGY WITH TIED CONTACT
If anything this little dialog is to remind myself to be careful with Tied Contact’s with “OFFSET”. As mentioned, the Offset
option indicates that the algorithm is using the penalty method to enforce the locked motion between parts. When
there is “penalty” one has the opportunity to create negative sliding interface energy. This behavior killed a rather
simple analysis. It was a bit amazing how it completely changed the behavior of the structure.
The fix is just to change the contact to _BEAM_OFFSET.
The two models are provided.
9. DAMPING
9.1 GENERAL, MASS AND STIFFNESS DAMPING
In dynamics, there often can be some oscillations that the analyst would prefer to have damped out or to account for
viscous behavior of some materials (see Material Damping). By default, LS-DYNA is undamped.
9.1.1 *DAMPING_OPTION
Introduces Rayleigh proportional damping based on: [ ] [ ] [ ]
• The mass damping constant is specified by *DAMPING_GLOBAL, *DAMPING_PART_MASS and
*DAMPING_RELATIVE.
• The stiffness damping constant is activated by *DAMPING_PART_STIFFNESS.
Mass damping is for low frequency response (rigid body modes), while stiffness damping is more effective at higher
frequencies. Since they are dissipative, their energy loss should be tracked. This can be done with the
*CONTROL_ENERGY option of RYLEN=2. Energy loss is then reported in the glstat and matsum files.
9.1.2 *DAMPING_FREQUENCY_RANGE
This is a more elegant approach to damping and allows the user to specify the critical damping coefficient and the
frequency range to damp. It is effective when used with low amounts of damping (e.g., 1 or 2%) and when the frequency
range is no more than a factor of 30x (e.g., 100 to 3,000).
Analyst’s Notes: I know of no shortcut to producing good agreement with the observed loss in a test I can only suggest
good judgment and a trial-and-error approach in order to tune the numerical damping.
9.1.3 MATERIAL DAMPING (E.G., ELASTOMERS AND FOAMS)

Elastomers (e.g., *MAT_181 Simplified Rubber) and foams (e.g., *MAT_053 Fu Chang foam) have the ability to add
damping directly within the material card. Recommended values are between 0.05 and 0.5. However, there is no true
recommended value since each material application is a bit different and requires some observation by the analyst to
determine the appropriate value.
9.1.4 Student Example on Material Damping

An example of material damping is provided in the Class Reference Notes / Damping titled: Baseball Ball Material
Damping Example.dyn along with a movie file of what one can expect to witness.
9.2 INSTRUCTOR / STUDENT LED WORKSHOP VII: DAMPING
An example is given of a vibration problem run by NX Nastran normal modes analysis and then by undamped and
damped transient analysis. If the student desires, the model is easy to switch to LS-DYNA implicit for an Eigenvalue run.
Script: The model is inspected within Femap and and Nastran normal modes analysis is run. The first natural frequency is
noted. The model is then exported out to LS-DYNA and directly analyzed. A History plot is made of node 1 in the Z-
direction. Using LSPP X-Y plot tools, under Oper, a fft is applied and note the frequency at the first big spike on the FFT
graph. The model will then be given 10% critical damping between 250 and 350 Hz. Save and plot results.
Student Bonus: One can switch it to an Eigenvalue analysis by adding these two *CONTROL cards:
_IMPLICIT_EIGENVALUE (NEIG=5) and IMPLICIT_GENERAL (IMFLG=1). Once analyzed you’ll want to view the d3eigv.
9.3 BULK VISCOSITY

Although the use of bulk viscosity as a form of damping is not exactly classical since it was developed for the treatment of
shock waves through the FEA mesh, it has grown into a recommended procedure for general dynamic structural FEA
work even when shock waves are not present. The *CONTROL_BULK_VISCOSITY card has several options and the default
are recommended. If shell elements are present in the analysis, one should set TYPE=-2 to activate this calculation for
shells. The reason it is not default is that shock waves are more of a solid element behavior. In the case of shells, we are
more interested in the linear bulk viscosity mechanism (see LS-DYNA Theory Manual for more details).
Example of the effects of bulk viscosity on an impact analysis is shown below courtesy of Suri Bali and his excellent
website Blog2.d3view.com.
Standard Response Slightly Smoothed w/ Bulk Viscosity
Images courtesy of Suri Bali via his Blog2.d3view.com – full article can be found in Class Reference Notes / Bulk Viscosity
Analyst Note: Although not necessary, it is recommended to add *CONTROL_BULK_VISCOSITY to any transient structural
FEA analysis since it tends to slightly damper out crazy oscillations at little numerical cost and with little energy cost.
10. LOADS, CONSTRAINTS AND RIGID WALLS

10.1 LOADS
10.1.1 INITIALIZATION LOADS
The most common initialization load is *INITIAL_VELOCITY_option. For example, for any type of drop test, the structure
is given a uniform initial velocity and then allowed to instantaneously hit the target. The other common initialization
command is *INITIAL_TEMPERATURE for thermal analysis work.
10.1.2 POINT AND PRESSURE LOADS

There is nothing complex to these loads. Point loads are those loads applied at nodes while pressure loads are applied
over element faces. In LS-PrePost, pressure loads are applied onto segments (i.e., faces).
10.1.3 BODY LOADS

Body loads are most commonly defined as constant
acceleration to capture the effect of gravity. Keep in mind
that LS-DYNA treats body acceleration loads differently and
that to obtain the same direction of gravity in Nastran one
must switch the sign of the acceleration load.
10.2 WORKSHOP: APPLYING MOVING BOUNDARY CONDITIONS

This workshop is quite simple and just demonstrates how one can create a variety of boundary conditions using some of
the LS-DYNA commands.
Objective: Interrogate the Femap model and notice

how the load condition is defined with a 5 Hz
function with gravity. We are going to bounce the
cube down this shaker but the model needs some
work.
Tasks
• Setup contact between the cube and the shaker
pan within Femap with a static friction of 0.3
• Define the two Regions using Properties and then
setup the Connector.
• Following this procedure, define glued contact
(TIED) between the shaker table springs (the lats)
and the bed supports. Keep it simple and use
OFFSET. To bridge the gap between the
components, set the slave/master thickness to 2.
For ease, create Regions using Properties.
• Set up Analysis Manager for a run time of 5.0
seconds with a results frequency of 0.05. Run
model. Please note estimated run time and time
step.
• Assume we don’t need to have the shaker table
bed, supports as “deformable”, switch the
material model to “rigid” and note the speed-up.
10.3 WORKSHOP XII: DROP TEST OF PRESSURE VESSEL
This exercise is geared toward increasing your confidence in using LS-DYNA. We are going to do a standard drop test
where the body is given an initial velocity and a rigid wall is placed directly beneath the vessel. At the end of this exercise
we’ll fill the vessel with a fluid. The objective is to get familiar with manipulating a model within LS-PrePost (LSPP). The
workshop starts by exporting the model from Femap where one has defined the materials (aluminum and rubber) and the
mesh with a few predefined LS-DYNA options within the Femap Analysis Set Manager.
Workshop Script
• A thin-walled (0.05”) aluminum vessel that is 24” in diameter and 36” tall is impacted at a speed of 100 MPH (1,760
in/s) against a rigid wall. The first task is to set the initial velocity for the vessel followed
by creating the rigid wall.
• Use *INITIAL_VELOCITY_GENERATION to get the vessel going. One will need to create a
part set to define what you want to initialize (see *SET_PART). Set initial velocity to Vy=-
1,760 and all other options default.
• Define rigid wall with *RIGIDWALL_PLANAR. We are placing this wall perpendicular to
the bottom of the vessel. The bottom of the vessel is at 0, 0, 0. To avoid initial contact
with the rigid wall, it is placed below the outer skin of the vessel at 0, -0.025, 0 and the
head of the vector at 0, 1, 0 to define its position.
• Run model and notice that the pressure vessel skin folds in upon itself.
• Add *CONTACT_AUTOMATIC_SINGLE_SURFACE (use prior part set for your “single surface”) and set SOFT=2 for
improved contact behavior. Run model and it should look better. This model has Finish I appended.
• The next step is to apply a pressure load. Pressure loads are typically applied to “segments” which is LS-DYNA
terminology for element faces. This operation could be done in Femap but we’ll proceed with LSPP. One will need to
create a segment set, define a load curve and then create the pressure load.
• To create your segment set, use the Create Entity option within LSPP. This option is located within the Model and Part
toolbar below the Keyword Manager button. Within the Create Entity screen, expand the Set Data option and select
*SET_SEGM option. Another pane will appear. Within this pane you’ll want to Create (Cre) the new segment set. I
would recommend picking the segments using the ByPart option. After picking, hit Apply and you’re done.
• Create your pressure load curve using the *DEFINE_CURVE command. Set the curve to unity over time, e.g., 0,1 and
1,1. When done, this should be curve number 2.
• The pressure load is finally created using the *LOAD_SEGMENT_SET command. We are setting the pressure to 5 psi by
setting SF=5. Run the model and interrogate. This model has Finish II appended to its name.
• To simulate the incompressible fluid behavior within the

vessel, add *AIRBAG_LINEAR_FLUID to the model with
BULK=5e5 and RO=1e-4 with all other options default. The
segment set created for the pressure load is re-used to
define the enclosed fluid volume. The completed card is
shown on the right.
• Rerun model. You should see the rubber portals bulge. This
model is noted as FINISH III.
10.3.1 STUDENT BONUS

Take the prior model and reorient the *RIGIDWALL_PLANAR to make it vertical or in the XY plane. Then change the
*INTIAL_VELOCITY_GENERATION to have the vessel hit normal to the XY plane or in the Z-direction. We are simply re-
orienting the impact from smashing down on the ZX plane to hitting “sideways” against a XY plane (so-to-speak).
FINISH I FINISH II FINISH III

(Change Colors) (Create Section Cut) (Contour Stresses)
Clean and neat drop test against rigid The pressure is inward (function of the As the vessel is collasped, the fluid
wall. plate element normals or sign of the volume pushes the rubber portals
pressure load. outward.
11. CONNECTIONS VIA JOINTS, BOLTS AND SPOTWELDS

11.1 JOINTS OR *CONSTRAINED_JOINT_
To model the motion and likewise, large deformation in engineering systems, one needs joints. LS-DYNA has a very
sophisticated set of commands that will allow one to model many types of common joints (e.g., hinges, spherical
bearings, etc. If time permits, we’ll do a simple model.
12. LOAD INITIALIZATION BY DYNAMIC RELAXATION AND IMPLICIT ANALYSIS

12.1 INITIALIZATION OF GRAVITY, BOLT PRELOAD AND OTHER INITIAL STATE CONDITIONS
Stress Initialization
Given that explicit analysis work often involves timesteps in the range of microseconds, one can image the challenges of
using an explicit approach to obtain quasi-static or static stress states in structures subjected to uniform constant
loading. There are many applications where the start of the explicit analysis requires the initialization of steady-state
loads within the structure. Here’s a short list: rotating equipment (e.g., fans, turbine blades or rotating flywheels),
pressurized vessels or tires, bolt preloads, shrink-fit parts, or spring mounted structures under constant gravity.
These static stress-state conditions can be simulated in LS-DYNA using two techniques: Explicit Dynamic Relaxation or
Implicit Static Analysis.
Dynamic Relaxation (DR) *CONTROL_DYNAMIC_RELAXATION

DR is a heavily damped explicit analysis that is initiated prior to the main transient analysis. It has all the characteristics
of a standard explicit run but it is assumed that stresses are relatively elastic and that displacements are small. The
solution is heavily damped and unexpected results may be generated. Nevertheless, with some models, it does a great
job with bolt preload, tire inflation or application of a shrink-fit. In the DR process, the load is applied (e.g., bolt preload)
as a transient load with a sharp ramp up and then a steady-state application. The model dynamically responds to this
load application with all the characteristics of a standard explicit transient analysis. As the model is solved, the nodal
velocities are reduced at each timestep by the dynamic relaxation factor (default = 0.995). The kinetic energy (KE) is
calculated at prescribed steps and when this energy has decreased sufficiently against the initial KE, the solution is
considered converged and the DR process shuts down and the solution is handed over to the regular, normal explicit
transient analysis sequence. There are lots of options to manage this process and the student is referred to the Class
Reference Notes for more information.
Analyst’s Note: I have often struggled to get DR to work correctly. It seems “finicky” since the process is uses a heavily
damped response and it just ain’t natural. As such, whenever possible I strive to use the implicit approach.
12.2 WORKSHOP XIII: DYNAMIC RELAXATION
As modal frequency response analysis is known as “the poor man’s

transient dynamics”, Dynamic Relaxation might also be known as “the
poor man’s implicit”. Nevertheless, in many cases it gets the job done
a lot faster than straight-up implicit and is easier to setup.
Script
• Using the “Start” model, create bolt loading curve (*DEFINE_CURVE)
with the following three points: 0, 0 | 1e-4, 5000 | 1e-2, 5000. Set
the SIDR=1 (please see Manual about details on this option).
• Apply bolt preload load to the four beam elements. This is done by
*INITIAL_AXIAL_FORCE_BEAM (please see Manual about details on
this Keyword). However, one will first need to create a beam set.
This has already been done for you – thus proceed gently forward
using this predefined beam set. Please note unique Material Law
required for the use of this Keyword.
• Apply load of 1,000 units at the far end of the structure’s tip. This
done by creating another curve with the following four points: 0, 0 |
5e-4, 0 | 6e-4, 1000 | 1e-2, 1000 and SIDR=0. Then, under
*LOAD_NODE_POINT, pick the node at the far end of the beam (it’ll
be 6352) with DOF=2 and LCID=2 (with any luck) and hit Accept.
• Run analysis.
Student Extra: Take the prior model and add mass scaling to model. Remember to set the DT2MS value to a negative
number. Forgot why? Please read the manual since “structured organic learning sticks the best.”
Analyst’s Note: After completing Workshop XIV, come back to this example and switch to implicit.
13. IMPLICIT-EXPLICIT SWITCHING FOR BURST CONTAINMENT

An application for implicit startup is the initialization of the steady-state stress field for rotating equipment. For systems
with high-speed rotating components, the model can be initialized to its steady-state spinning condition using an implicit
analysis. To model any downstream event that may be highly nonlinear and dynamic, the model can then be switched
over to an explicit analysis. The trick is performing this type of analysis is to remember that LS-DYNA lets you apply
boundary conditions and manage contact behavior in a “birth / death” manner. For example, one can apply constraints
to all secondary structures that are not relevant, or in the case of a turbine analysis, not spinning to ensure that the
implicit analysis runs smoothly and then remove them (i.e., death) once the explicit analysis starts up. Likewise, this can
be done with contact behavior within the model to avoid numerical difficulties during the implicit solve. Both of these
tricks can dramatically speed-up the implicit solution without affecting the accuracy of the simulation.
Analyst Note: The implicit-to-explicit switching can be a bit tricky. What is recommended is that one switches the
simulation to explicit prior to desired end-point of the implicit simulation. For example, if the implicit end point is 1.0 and
the implicit time step is 0.25, then the curve used to control the switch from implicit-to-explicit could be set to any time
point after 0.75 and before 1.0(e.g., 0.9) and the explicit “on” switch set to the end time of the implicit simulation (e.g.,
1.000001).
13.1 HIGH-SPEED ROTATING EQUIPMENT – *CONTROL_ACCURACY
For structures that rotate, it is recommended that the *CONTROL_ACCURACY option OSU = 1 and INN = 2 be set. The
OSU option adds additional terms to the stress update and improves the accuracy of the simulation while the INN option
sets invariant node numbering to ensure the accurate calculation of element forces within elements that are become
highly twisted and/or rotate through space. Interesting enough, the INN = 2 option is default for implicit calculations.
Both of these options will slow the simulation down by as much as 10%.
13.1.1 WORKSHOP XIV: IMPLICIT-EXPLICIT TURBINE SPIN UP
This workshop will show the student how to setup a turbine simulation from implicit ramp-up to steady-state rotating
movement. The model has been mostly prepared and the student is required to finish the setup using the following
Keywords:
*CONTROL_ACCURACY (OSU=1, INN=2), *CONTROL_IMPLICIT_GENERAL (IMFLAG=-1, DTO= 0.25),
*CONTROL_IMPLICIT_SOLVER (LSOLVR=6), *INITIAL_VELOCITY_GENERATION (PHASE=1),
*INITIAL_VELOCITY_GENERATION_START_TIME (STIME=1.00001), *INITIAL_VELOCITY_RIGID_BODY (VZR=225),
*LOAD_BODY_PARTS (PSID=1), *LOAD_BODY_RZ (LCID=2, SF=225), *BOUNDARY_SPC_SET_BIRTH_DEATH (NSID=1,
DOFX=1, DEATH=1.0).
For any implicit analysis, it is recommended to use the double-precision solver.
This is activated within the LS-D YNA Program Manager under Solver, Select
LSDYNA Solver. When completed, one should see the graphic on the right a small
“d” within the solver name.
14. SMOOTHED PARTICLE HYDRODYNAMICS (SPH)

14.1 INTRODUCTION
A short introduction video for this section can be found in Workshops / SPH / Introduction to SPH Modeling
14.1.1 A LITTLE BIT OF THEORY (SKIP THIS IF YOU DON’T LIKE MATH…)
Kirk Fraser, Sr. Staff Engineer, Predictive Engineering
Smoothed particle hydrodynamics (SPH) was developed in the 1970’s Continuity Equation
by Monaghan, Gingold and Lucy for astrophysics problems. Monaghan
has published an enormous number of papers on the SPH method. ∑ ( )
Libersky et al. [1] were the first to apply the method to solid mechanics
problems. Lacome [2] was one of the first to implement SPH in LS-
DYNA.
Mesh-based methods do a great job for all kinds of engineering Conservation of Momentum
calculations. When the deformation gets really large, mesh-based
methods start to fail due to negative element volume, excessive mesh ∑ ( )
distortion and/or mesh tangling within contact region which then
causes problems with the explicit time step and so on and so forth.
SPH is a Lagrangian based mesh-free method that can handle unlimited Conservation of Energy
plastic deformation. The rate of change of the field variables for a given
particle “i”, with N “j” neighbors in the support domain is given by Lui
∑ ( )(
and Lui [3]: is the smoothing function (interpolation kernel) and
can take on many different forms depending on the type of problem
being studied (e.g., the cubic spline function is popular), is an )
artificial viscosity term and is an artificial heating term. The general

idea is to use a finite number of neighbors within a radius of influence
(also known as the smoothing length) on the central element. The
graphic on the right depicts how the smoothing length can be visualized
for a SPH mesh.
The method converts a set of partial differential equations (PDE) into a

set of ordinary differential equations (ODE). The ODE’s can be
integrated in time with many different schemes; in LS-DYNA a multi-
step (fractional step) explicit method is used. This means that there is a
stability condition on the time-step size (CFL):
( )
is a constant and is typically 0.2 to 0.4. Implicit time integration a

possibility in the future in LS-DYNA, but for now, only explicit is
available. For the truly die-hard SPH gear-heads, I recommend the
book by Damien Violeau [4] (you gotta really love math to enjoy this
book) or for a less math intensive approach, see the book by William G.
Hoover [5].
Lastly, for modeling constitutive relationships in SPH, one can use many
of the same material cards (i.e., laws) as a regular Lagrangian analysis.
14.1.2 LAGRANGIAN VS EULERIAN
The two most common frames of reference for

numerical simulations is Lagrangian or Eulerian.
You can think of Lagrangian reference like you are
sitting on a plane and Eulerian like you are on the
ground (not moving) and watching the plane go
by. Lagrangian makes following the history of
each material point very easy compared with the
Eulerian description. CFD codes (finite difference,
finite volume and finite element) are written from
an Eulerian formulation.
The Lagrangian nature of SPH makes it a very

powerful numerical method, this opens many
doors that Eulerian method closes (or makes it
difficult to open the door).
14.1.3 TYPES OF SIMULATIONS WITH SPH
• Impact and ballistics (e.g., bird strike)
• Fracture and fragmentation
• Fluid structure interaction (e.g., sloshing)
• Linear and non-linear vibrations
• Microstructure evolution
• Heat transfer
• And many more…
14.1.4 COMMON KEYWORDS FOR SPH
 *CONTROL_SPH
 *SECTION_SPH
 *CONTACT_AUTOMATIC_NODES_TO_SURFACE with SOFT=1 is recommended
 All *EOS_ and most *MAT_ (see Keywords manual for details, [6] and [7]) cards can be directly used
 Most of the standard keywords work for SPH
 Node sets need to be used to define contact, boundary conditions, etc.
14.2 SPH WORKSHOP I: GETTING STARTED
This example (courtesy of Kirk Fraser) is geared toward demystify the process of creating
and running a SPH model. The concept is that any box, sphere, cylinder, etc. can be used to
create a SPH mesh. Once the mesh is created, the model is setup and analyzed.
Start-Up
• Open the Femap model SPH / SPH Workshop 1 / SPH 1 Ball and Plate Example (Setup
for SPH).modfem and then fire up the movie file.
Goals
• Get familiar with the concept of creating a SPH mesh.
• The procedure is pretty simple and offers flexibility
• Visualize your SPH spheres
• Convert model to full SPH analysis
o *CONTROL_SPH IDIM=3
o *SECTION_SPH
o *PART
o *CONTACT_AUTOMATIC_NODES….
o *TERMINATION 0.025 s
• View results
• At the end of this workshop, see Class
Reference Notes / SPH / LSTC SPH Short Course
Notes.pdf
14.3 SPH WORKSHOP II: FLUID MODELING
In this example (courtesy of Kirk Fraser), the plate is made

rigid and the ball is turned into water by using two cards:
*MAT_NULL and *EOS_GRUNEISEN.
Learning Objective
• Material modeling of fluids is not that difficult. It is

equivalent to modeling an isotropic solid material
with E and . For a fluid at its simplest form, there
is speed of sound “c”, density and viscosity.
• The cwater is 342 m/s and has a density of 1,000

kg/m3. The viscosity of water is 0.001 Pa*s.
• Run, explore and then change mu=1,000.
• SPH can also be used for a sloshing analysis.
Analyst’s Note: SPH doesn’t need hourglass control but if

a default value is applied (see *CONTROL_HOURGLASS) it
will get used! Since the Null material has no shear
stiffness, the default hourglass coefficient can lead to
significant energy losses. Given this feature, avoid the use
of *CONTROL_HOURGLASS for SPH simulations and apply
hourglass control directly to the part of interest.
14.4 SPH WORKSHOP III: VERIFICATION
This workshop (courtesy of Kirk Fraser) compares the SPH method against the standard FEM method and discusses a
common numerical difficulty with SPH method known as “Tensile Instability”. This introduction is not meant to be
anywhere complete but just to get the student started in how to debug their SPH models and what they might want to
look out for while interrogating the model. If this subject is of greater interest, LSTC provides a two-day course on the
SPH method and the course notes can be viewed within Class Reference Notes / SPH / LSTC SPH Short Course Notes.pdf.
Three models are provided: IIIA FEM Method, IIIB Tensile Instability and IIIC Recommended for the student to compare
and explore the SPH method. The workshop movie file provides all the steps and a bit of discussion. Along with learning
about tensile instability, the student will learn how to post-processes SPH results and create comparison graphs between
principal stress and contact force.
FEA Mesh Benchmark SPH Default (FORM=0) SPH (FORM=7)
14.4.1 SPH VERIFICATION III: WORKSHOP BONUS
At the end of the prior workshop, the student should have made graphs of the maximum principal stresses and contact
forces between the FEM, SPH Tensile Instability and SPH Recommended. In particular, the contact force graph will have
quite a bit of numerical noise. LSPP provides a Filter option to remove such noise. The student should explore this
option since it is invaluable in the interpretation of dynamic results.
14.5 SPH WORKSHOP IV: BIRD STRIKE
The Bird Strike workshop essentially leverages all the prior workshops and
explores in a bit greater detail the *CONTROL_SPH control card options ISHOW,
IEROD and ICONT to make the simulation more efficient. Lastly, the student is
encouraged to set the fan blade failure criterion using *MAT_ADD_EROSION
(MXEPS=0.05) while enforcing contact between the parts.
Hands-On Script: The procedure is to open the starting file (Workshops / SPH /
SPH Workshop IV – Bird Strike / Default / SPH Workshiop IV – Default Start.dyn,
inspect the model and then use this baseline model to create the other two
simulations. This is an interactive instructor/student Workshop and questions
should be asked if any of the operations don’t feel right. Please note that file
folders exist with the completed, final *.dyn model within the folder. Lastly, if
one is bored or has time, change the failure criterion for the bird.
Defaults ISHOW=1, IEROD=1 & ICONT=1 BLADE FAILURE
14.5.1 BIRD STRIKE MODELS
If one ever wonders what a spherical chicken looks like. Image courtesy of Aerospace_MGD_v12-1.pdf, page 101 (see
Class Reference Notes / Aerospace Working Group)
14.6 REFERENCES
[1] L. D. Libersky, A. G. Petschek, T. C. Carney et al., “High Strain Lagrangian Hydrodynamics: A Three-Dimensional SPH
Code for Dynamic Material Response,” Journal of Computational Physics, vol. 109, no. 1, pp. 67-75, 11//, 1993.
[2] J. L. Lacome, “Smooth Particle Hydrodynamics (SPH): A New Feature in LS-DYNA.”
[3] G.-R. Liu, and M. B. Liu, Smoothed particle hydrodynamics : a meshfree particle method, Hackensack, New Jersey:
World Scientific, 2003.
[4] D. Voileau, Fluid Mechanics and the SPH Method: Theory and Applications, Oxford, UK: Oxford University Press,
2012.
[5] W. G. Hoover, Smooth Particle Applied Mechanics: The State of the Art (Advanced Series in Nonlinear Dynamics),
Singapore: World Scientific Publishing, 2006.
[6] LSTC, “LS-DYNA Keywords User Manual Volume 1,” no. Version 971, July 12, 2012, 2012.
[7] LSTC, “LS-DYNA Materials User Manual Volume 2,” no. Version 971, July 12, 2012, 2012.
[8] S. Marrone, A. Colagrossi, D. Le Touzé et al., “Fast free-surface detection and level-set function definition in SPH
solvers,” Journal of Computational Physics, vol. 229, no. 10, pp. 3652-3663, 2010.
15. EXPLICIT MODEL CHECK-OUT AND RECOMMENDATIONS

If you think you might have a simulation that is working, here’s a short list of things to check for and review.
Here’s an order of checking: Units | Mesh | d3hsp File | History Plots | Material Modeling | Contact Behavior | Etc.
15.1 UNITS
It is recommended to settle on one unit system for as much of your LS-DYNA work as possible to avoid unit problems
when one is unfamiliar with a specific system. A commonly recommend system for dynamic events is the kN, mm, ms, kg
system. Stresses are then in GPa. We have covered this before but it is hard to overstate the importance of getting your
units straight. In a dynamic analysis, the mass of the system should always be checked.
15.2 MESH
When looking at your mesh, it should look good and if it looks good, it will generate a smooth stress contour. This is
never more so important than for an explicit analysis. If this sounds odd, please see Class Reference Notes / Stress
Visualization / Desktop Engineering Stress Visualization Article March 2011.pdf.
Besides this Zen of meshing statement, here are some bulleted items to consider:
• Is the mesh density sufficient to capture the mechanical response? (Remember one-point Guass Integration)
• If Hourglassing is significant, remesh and likewise, if contact is poor (high SLEOUT), remesh, etc.
• Lastly, check the explicit time step. Seriously, a couple bad elements can completely explode the analysis
(personal experience that cost me a weekend). This can be done easily via LSPP and viewing the D3hsp file under
100 smallest timesteps.
15.3 D3HSP FILE (LS-DYNA EQUIVALENT TO THE NASTRAN F06 FILE)

• The d3hsp file summarizes the input in descriptive terms and can be viewed via a text editor or from within LSPP
via Misc. and then D3hsp View.
Within the d3hsp, one should review:
o Verify mass of system (find "summary'')
o 100 elements listed in ascending order of time step (find "smallest'')

o Review rigid bodies for any that might be deleted or with super-small mass
o Warning and error messages (find "Warning" or "Error'')
15.4 HISTORY PLOTS
The GLSTAT file is your first stop for checking the analysis. A fundamental
check is that your Energy Ratio should be 1.0 (+/- 0.01). This can be quickly
checked within the ASCII, GLSTT file under Energy Ratio. An example of a
more complex energy plot is shown on the right. The high Sliding Energy is
because the model is simulating a burst containment of a fragmenting X-ray
target with friction. For more information on Energy Data, see
www.dynasupport.com under LS-DYNA User’s Guides and then Energy Data
also see Class Reference Notes / Energy Balance / Total energy LS DYNA
Support.pdf
If the GLSTAT checks out, it is time to go over the MATSUM file via individual entries. The internal energy should be
plotted against the hourglass energy and margins of 5% or less of the peak internal energy should be observed.
Sliding energy (GLSTAT - global) and sliding interface energy (ASCII – local via) say a lot about the numerical validity of
your contacts. If friction is zero, than the GLSTAT value should be 1% or less of the internal energy. Interrogate local
sliding energy by plotting all of the SLEOUT values. If they are negative greater than 5% of the peak internal energy, be
worried and start digging.
15.5 MATERIAL MODELING ERRORS

Although much-ado is made about strain rate sensitivity, for most engineering applications, the only really strain rate
sensitive materials are carbon based materials (rubbers, elastomers, plastics and foams) and to obtain such data is not
that difficult from any good material testing laboratory. What is more common is just plain screwing up the material
model. Thus, it is almost mandatory to demonstrate with a virtual coupon test model that one can match test data with
the LS-DYNA model. Such correlation should be within every engineering report. Never underestimate the power of
“KISS” and always attend to the basics before making your life more difficult.
15.6 CONTACT OPTIONS WITH RECOMMENDATIONS
SOFT=2 Soft=2 is really a quite good standard contact option

For handling small penetrations that are not worth the time to fix, this option is very useful.
IGNORE=1
Please don’t expect it to do wonders or correctly fundamental contact modeling problems
Contact is often noisy and adding 20% damping can be a nice option to smooth things out. It is
VDC=20
one of those tweaks that is often times worth investigation once you have the model working.
ISYM=1 If you have symmetry faces via SPC’s, please be aware of this option.
15.7 CONTROL CARDS WITH RECOMMENDATIONS

Commonly used options are shown in the table below
For dynamic structural-only analyses, this option provides a
*CONTROL_BULK_VISCOSITY TYPE=-2 mild smoothing of oscillations with little performance or
energy cost.
If you have small penetrations in your model, this option
IGNORE=1 makes good sense but don’t expect to help you with large
*CONTROL_CONTACT modeling contact errors.
ISYM=1 Symmetry Control
Calculates hourglass energy and sliding interface energy.
*CONTROL_ENERGY HGEN=2 & SLNTEN=2 Although it adds computational expense it should be added
and checked and then one can delete it later.
Improved formulations and the default value of 0.1 can often
*CONTROL_HOURGLASS IHQ=4 & QH=0.05
be lowered if needed.
Ensures highly twisted elements don’t cause harm to your
ESORT=1
simulation
*CONTROL_SHELL
Deletes highly distorted elements prior to them causing harm
NFAIL1=1 & NFAIL=1
to your simulation
Automatic sorting of tetrahedron and pentahedron elements
to treat degenerate tetrahedron and pentahedron elements as
*CONTROL_SOLID ESORT=1
tetrahedron (formulation 10). However, most LS-DYNA models
don’t have tetrahedrals but it is included for completeness.
Just admit it, you’ll going to use some mass scaling. If so, don’t
DT2MS (negative) forget to contour the added mass to verify it didn’t get to crazy
*CONTROL_TIMESTEP (see below for recommendation on contouring added mass).
ERODE=1 Just good practice when using solid elements. If the element
becomes highly distorted to the point of a negative volume,

it’ll be deleted without killing the simulation.
If one is using RB2 or 3 elements from a Nastran model or
directly using CNRB’s, this command will add plot only
*CONTROL_RIGID PLOTEL=1 or 2
elements to the d3plot file and allow visualization of the
elements.
15.8 DATABASE CARDS WITH RECOMMENDATIONS
This is the minimum recommended set. Please note if you

GLSTAT, MATSUM,
*DATABASE_ASCII have rigidwalls in your simulation, you should also have
SLEOUT & SPCFORC
RWFORC enabled.
Beam elements are isoparametric elements and have
integration points, setting BEAMIP=4 covers the default beam
formulation. STSSZ=3 is to dump out added mass information
and then also check MSSCL=1 or =2 to indicate incremental or
percentage increase of added mass (my choice is MSSCL=2).
*DATABASE_EXTENT_BINARY BEAMIP (4) & STSSZ=3 This is very useful to check your model for mass scaling
effects. To contour the added mass, see LSPP, Fcomp / Misc /
Time Step Size (whereas in this case, it is the added mass).
Please note, it is really mandatory if you are going to be
aggressive with your mass scaling to contour this item and be
aware of where you are adding mass to your structure.
15.9 ETC
I’m a big fan of building stupid, simple, itty-bitty test models to evaluate a proposed behavior. A standard downfall of
many simulations is an attempt to model all the physics out-of-the-gate without prior evaluation of the effects of
individual items, in brief, the more complex the model, the more heinous is the debugging.
16. IMPLICIT CHECK-OUT AND RECOMMENDATIONS

Implicit analyses are generally much more difficult to obtain convergence and that is often the downfall of its usage. To
truly get spun up to speed on implicit, the student should start by reading (see Class Reference Notes / Implicit Analysis /
LSTC Implicit Class Notes by Morten Jensen 2012 Draft.pdf).
For implicit analyses used for the initialization of explicit runs, it is recommended to use the *CONTACT_ birth option
(BT) to turn off all non-essential contacts until the start of the explicit run. Likewise use *BOUNDARY_SPC_NODE (or
SET)_BIRTH_DEATH to lock down any parts of the structure that are not relevant to the implicit initialization. These two
steps will greatly facility a fast and efficient implicit kick-off.
For troubleshooting LS-DYNA implicit analyses don’t be shy about locking down (SPC’ing) large parts of your structure
and ripping out contacts and nonlinear material laws. Once you have something running, it is a lot easier to add in
complexity step-by-step-by-step than struggle with behemoth that is taking 30 minutes to finally error out.
LS-DYNA also can automatically switch to explicit if the implicit solver is struggling to find convergence (see Class
Reference Notes / Implicit Analysis / LSTC Implicit Class Notes by Morten Jensen 2012 Draft.pdf, page 42). It may sound
attractive but don’t really expect it to perform miracles since it most likely has to return to the implicit solution to finalize
the run and most likely it would just fail to find convergence again (please note that most typical LS-DYNA runs are so
highly nonlinear that often times it is a miracle that the implicit run even works).
A little thing to note is that LS-DYNA has near identical element formulations to that of standard implicit codes but as of
this writing (Fall 2013), stresses are only provided at the centroid and integration points. Hence, one will note significant
differences in stresses due to lack of Jacobian extrapolation from the integration points out to the nodal points. Of
course, if your mesh is fine enough the differences will be small. A trick to overcome this limitation in solid meshes is to
“skin” the solid with very thin layer of plate elements and then take your stresses from this layer. This technique is well
known and leveraged within the implicit community for obtaining high-quality stresses for strain gauge and fatigue work.
16.1 IMPLICIT KEYWORD CARDS AND RECOMMENDATIONS
This algorithm was essentially developed for implicit analyses. It is super expensive
for explicit and not recommended. However, keep in mind that with the BT and DT
*CONTACT_ -- _MORTAR
contact card option one can always turn it off and switch to another contact
formation for the explicit run.
BCSLIB-EXT (Boeing) solver that is recommended for
LSOLVR=6 implicit solutions for your standard plate, beam, CNRB,
*CONTROL_IMPLICIT_SOLVER hex solid model.
For large tetrahedral models, the iterative solver
LSOLVR=10
performs well (just like Nastran).
Not a big fan of auto stepping. It will sometimes help but
*CONTROL_IMPLICIT_AUTO IAUTO=1 often is slower than just specifying a small DTO within the
*CONTROL_IMPLICIT_SOLUTION card.
Recommended Only for Pure Implicit Analyses where Stress Recovery in the Elastic Range is Important Since for Explicit
Analyses it Generates 4 to 8x larger D3Plot Files due to Integration Point Stresses
A negative number dumps out all integration point
MAXINT=-3
stresses.
*DATABASE_EXTENT_BINARY This dumps out all the integration point stress data for
solid elements. Very useful for implicit work if you would
NINTSLD=8
like to something that approaches a normal linear stress
result.
This command will correctly extrapolate the integration
point stresses for ELFORM=20 (shells) ELFORM=18 (solids)
LSPP (LS-PrePost) v4.1 Extrapolate -1 and aligns well with Nastran-type element results (with
MAXINT=-3). Note: For shells one can use -1 or 1, only
for solids is -1 required.
17. DISCRETE ELEMENT METHOD
See Class Reference Notes / DEM / Predictive Engineering Discussion of LS-DYNA Meshfree Methods.pptx
18. FLUID STRUCTURE INTERACTION AND MULTI-PHYSICS IN LS-DYNA
See Class Reference Notes / Multi-Physics / LS-DYNA Multi-Physics.ppsx
19. LS-DYNA RECENT DEVELOPMENTS
See Class Reference Notes / LS-DYNA Recent Developments / LSTC_2012_Conference_Recent Developments.pdf
END
End of LS-DYNA Analysis for Structural Mechanics Class Notes

Predictive Engineering PDF

Uploaded by

Copyright:

Available Formats

Predictive Engineering PDF

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

Predictive Engineering PDF

Uploaded by

Copyright:

Available Formats

What is the course outline about?

What is the course outline about?

What are some general applications of LS-DYNA?

What are some general applications of LS-DYNA?

LS-DYNA® Analysis for Structural Mechanics

Technical Support Team at Livermore Software Technology Corporation (LSTC)

External to LSTC, we have:

4.6 WORKSHOP I: LS-DYNA GETTING STARTED EXAMPLE ..........................................................................................................................................33

7.8 MATERIAL FAILURE SIMULATION ........................................................................................................................................................................55

END ........................................................................................................................................................................................ 112

1.2 WHAT WE COVER

1.4 GENERAL APPLICATIONS

Crashworthiness Driver Impact Train Collisions

Earthquake Engineering Metal Forming Military

1.5 SPECIFIC APPLICATIONS (COURTESY OF PREDICTIVE ENGINEERING)

Drop Test Consumer

Drop Test of Nuclear Waste

Plastic Thread Design PSD / Modal Analysis Digger Tooth Failure

Electron Beam Welding Pyro-Shock Analysis Medical Equipment

Ballistic Shock Loading of Optical Hyperelastic Medical Seal

Discrete Element Method for the Drop-Test of Hand Held

Ballistic Penetration of Al/Foam High-Speed Spinning Disk Locomotive Fuel Tank

Airplane 16g Crash Analysis Impact Analysis of Safety Block Device

2. IMPLICIT VERSUS EXPLICIT ANALYSIS

2.1 WHAT WE ARE SOLVING

2.2 EXPLICIT (DYNAMIC)

2.3 IMPLICIT (DYNAMIC)

3. FUNDAMENTAL MECHANICS OF EXPLICIT ANALYSIS

3.1.1 EXPLICIT TIME INTEGRATION

3.2 TIME STEP SIGNIFICANCE (COURANT-FRIEDRICHS-LEWY (CFL) CHARACTERISTIC LENGTH)

• In the simplest case (small, deformation theory), the timestep is

Explicit Time Step Mass Scaling (*Control_Timestep)*

3.3.1 WORKSHOP: FEMAP TO LS-DYNA WITH MASS SCALING BASICS

What You Will Learn Units: kN-mm-ms-kg

Femap to LS-DYNA with Mass Scaling Basics

Femap to LS-DYNA with Mass Scaling Basics (continued)

Femap to LS-DYNA with Mass Scaling Basics (continued)

Femap to LS-DYNA with Mass Scaling Basics (continued)

Model Time Step % Mass Added by Mass Scaling Max. Displacement

3.3.2 INSTRUCTOR LED WORKSHOP I: MASS SCALING ADVANCED

3.4 IMPLICIT MESH VERSUS EXPLICIT MESH CHARACTERISTICS

Meshing for Accuracy

4. FIRST LS-DYNA MODEL: GETTING STARTED

LS-DYNA has perhaps one of the most basic learning methods. It is

4.2 KEYWORD SYNTAX

• Commands are strings of words separated by an underscore, e.g.,

Consistent Unit Sets for LS-DYNA Analysis

g mm ms N MPa N-mm 7.83e-03 2.07e+05 9.806e-03

Ton (1,000 kg) mm s N MPa N-mm 7.83e-09 2.07e+05 9.806e+03

Lbf-s2/in (snail) in s lbf psi lbf-in 7.33e-04 3.00e+07 386

4.4 REFERENCE MATERIALS AND PROGRAM DOWNLOAD

The first site to visit: www.lsdynasupport.com LSTC Program Download Site

4.5 SUBMITTING LS-DYNA ANALYSIS JOBS AND SENSE SWITCHES

4.6 WORKSHOP I: LS-DYNA GETTING STARTED EXAMPLE

This workshop uses the LSTC Getting Started Example

5. EXPLICIT ELEMENT TECHNOLOGY

There are many different element types in LS-DYNA:

Extremely Brief Recommendations

5.2 ONE GUASSIAN POINT ISOPARAMETRIC SHELL ELEMENTS AND HOURGLASSING

Explicit Time Step Mass Scaling (Control_Timestep)

Nastran CBUSH LS-DYNA Equivalent SECTION_BEAM & MAT_66

7.7.1 MODELING WATER WITH EOS_GRUNEISEN AND MAT_NULL

SECTION_SHELL (Thickness=50) CONTACT_AUTOMATIC_ *CONTACT_AUTOMATIC_