INDUSE Report PDF

Structural safety of
industrial steel tanks, pressure vessels

and piping systems under seismic loading
(INDUSE)
Research and
Innovation EUR 26319 EN
EUROPEAN COMMISSION
Directorate-General for Research and Innovation
Directorate G — Industrial Technologies
Unit G.5 — Research Fund for Coal and Steel
E-mail: rtd-steel-coal@ec.europa.eu
RTD-PUBLICATIONS@ec.europa.eu
Contact: RFCS Publications
European Commission
B-1049 Brussels
European Commission
Research Fund for Coal and Steel

Structural safety of industrial steel tanks,
pressure vessels and piping systems
under seismic loading
(INDUSE)
P. Pappa, G. E. Varelis, M. Vathi, P. C. Perdikaris, S. A. Karamanos

University of Thessaly
Argonafton & Filellinon, 38221 Volos, GREECE
J. Ferino, A. Lucci, E. Mecozzi, G. Demofonti
Centro Sviluppo Materiali S.p.A.
Via di Castel Romano, 100/102, 00128, ITALY
A. M. Gresnigt, G. J. Dijkstra (Tebodin)
Delft University of Technology
Stevinweg 1, 2628 CN Delft, NETHERLANDS
Md. S. Reza, A. Kumar, F. Paolacci, O. S. Bursi
University of Trento
Via Belenzani 12, 38100 Trento, ITALY
M. Kopp, M. Pinkawa, M. Wieschollek, B. Hoffmeister
RWTH Aachen University
Templergraben 55, 52062 Aachen, GERMANY
A. Stamou, K. Diamanti, T. Papatheocharis
EBETAM (MIRTEC) S.A.
A’ Viomichaniki Periochi, 38500 Volos, GREECE
C. Botsis, I. Chandrinos, I. Doukas
TECHNIPETROL HELLAS S.A.
Pavlou Bakogianni Street 42, 14452 Metamorfosis, Athens, GREECE
Grant Agreement RFSR-CT-2009-00022

1 July 2009 to 31 December 2012
Final report
Directorate-General for Research and Innovation
2013 EUR 26319 EN

LEGAL NOTICE
Neither the European Commission nor any person acting on behalf of the Commission is
responsible for the use which might be made of the following information.
The views expressed in this publication are the sole responsibility of the authors and do not
necessarily reflect the views of the European Commission.
Europe Direct is a service to help you find answers

to your questions about the European Union
Freephone number (*):

00 800 6 7 8 9 10 11
(*) Certain mobile telephone operators do not allow access to 00 800 numbers or these calls may be billed.
More information on the European Union is available on the Internet (http://europa.eu).
Cataloguing data can be found at the end of this publication.
Luxembourg: Publications Office of the European Union, 2013
ISBN 978-92-79-34575-3
doi:10.2777/49423
© European Union, 2013

Reproduction is authorised provided the source is acknowledged.
Printed in Luxembourg
Printed on white chlorine-free paper

Table of contents
Table of contents....................................................................................................................................3
I. Final summary ..............................................................................................................................7
I.1. WP1: Comparison of European and American design standards for the seismic design of
industrial liquid storage tanks, pressure vessels and piping and development of available test data
base. ..................................................................................................................................................8
I.2. WP2: Determination of seismic forces on industrial structures and critical components and
preparation of experiments. ..................................................................................................................8
I.3. WP3: Experimental testing of structural components of critical importance under strong cyclic
loading. .................................................................................................................................................9
I.4. WP4: Numerical simulation of industrial component behaviour under strong cyclic loading.10
I.5. WP5: Seismic behaviour and special seismic design issues in liquid storage tanks;
development of relevant design guidelines/recommendations. ..........................................................12
I.6. WP6: Seismic behaviour and design of industrial pressure vessels; development of seismic
design guidelines/recommendations...................................................................................................13
I.7. WP7: Seismic behaviour and design of industrial piping systems and components;
development of seismic design guidelines/recommendations. ...........................................................14
I.8. WP8: Coordination of the project...........................................................................................15
I.9. Conclusions and main results .................................................................................................15
II. Scientific description of the results ............................................................................................17
Objectives of the project.....................................................................................................................17
Comparison of initially planned activities and work accomplished ...................................................17
II.1. Work package 1 – Comparison of European and American design standards for the seismic
design of industrial liquid storage tanks, pressure vessels and piping and development of available
test data base.......................................................................................................................................17
II.1.1. Task 1.1: Collection and evaluation of available relevant experimental data ................17
II.1.2. Task 1.2: Collection and evaluation of reported earthquake damages in industrial
facilities ........................................................................................................................................18
II.1.3. Task 1.3: Basic seismic structural design of typical liquid storage tanks.......................19
II.1.4. Task 1.4: Basic seismic structural design of typical cylindrical and spherical pressure
vessels ........................................................................................................................................19
II.1.5. Task 1.5: Basic seismic structural design of typical piping system and its support
structure ........................................................................................................................................20
II.2. Work package 2 – Determination of seismic forces on industrial structures and critical
components and preparation of experiments ......................................................................................22
II.2.1. Task 2.1: Seismic forces in vertical cylindrical liquid storage tanks .............................22
II.2.2. Task 2.2: Seismic forces in pressure vessels ..................................................................22
II.2.3. Task 2.3: Seismic forces in industrial piping systems ....................................................23
II.2.4. Task 2.4: Dissipative support systems ............................................................................24
II.2.5. Task 2.5 Determination of loading amplitudes on critical components .........................24
II.2.6. Task 2.6 Specification of procedures for specimen manufacturing and base material
supply ........................................................................................................................................24
II.3. Work package 3 – Experimental testing of structural components of critical importance under
strong cyclic loading ..........................................................................................................................25
II.3.1. Task 3.1: Basic material testing and characterization ....................................................26
3
II.3.2. Task 3.2: Tests on tank openings/nozzles under strong cyclic loading ..........................27
II.3.3. Task 3.3: Tests on bolted piping connections under strong cyclic loading ....................29
II.3.4. Task 3.4: Tests on piping elbows under strong cyclic loading .......................................31
II.3.5. Task 3.5: Tests on T-branch pipe connections under strong cyclic loading ...................35
II.4. Work package 4 – Numerical simulation of industrial components behaviour under strong
cyclic loading .....................................................................................................................................37
II.4.1. Task 4.1 Definition of inelastic material modelling under cyclic loading .....................37
II.4.2. Task 4.2 Numerical simulation of the experimental tests on components of critical
importance, tested in WP3..............................................................................................................41
II.4.3. Task 4.3 Parametric numerical study of critical component behaviour .........................55
II.5. Work package 5 – Seismic behaviour and special seismic design issues in liquid storage tanks;
development of relevant design guidelines/recommendations ...........................................................65
II.5.1. Task 5.1 Special issues in seismic behaviour of liquid storage tanks.............................65
II.5.2. Task 5.2 Design recommendations / guidelines for the seismic analysis & design of
liquid storage tanks .........................................................................................................................68
II.6. Work package 6 – Seismic behaviour and design of industrial pressure vessels; development
of seismic design guidelines/recommendations .................................................................................69
II.6.1. Task 6.1 Seismic behaviour of vertical-cylindrical pressure vessels..............................69
II.6.2. Task 6.2 Seismic behaviour of horizontal-cylindrical pressure vessels .........................69
II.6.3. Task 6.3 Seismic behaviour of pressure vessels of spherical shape ...............................70
II.6.4. Task 6.4 Development of design recommendations / guidelines for the seismic design &
analysis of industrial pressure vessels ............................................................................................70
II.7. Work package 7 – Seismic behaviour and design of industrial piping systems and
components; development of seismic design guidelines/recommendations ......................................71
II.7.1. Task 7.1 Seismic behaviour of industrial piping and their support systems...................71
II.7.2. Task 7.2 Development of design recommendations / guidelines for the seismic analysis
& design of industrial piping systems ............................................................................................72
II.8. Work package 8 – Coordination of the project.......................................................................73
II.8.1. Task 8.1 Efficient management process .........................................................................73
II.8.2. Task 8.2 Definition of the standard protocols ................................................................74
II.8.3. Task 8.3 Monitoring of project activities and evaluation of the project .........................74
II.8.4. Task 8.4 Progress, Mid-Term and Final reports .............................................................75
II.9. Conclusions ............................................................................................................................76
II.10. Exploitation and impact of the research results ..................................................................76
III. Appendices .................................................................................................................................77
III.1. Appendix of WP1 ...............................................................................................................77
III.1.1. Task 1.1: Collection and evaluation of available relevant experimental data ................77
III.1.2. Task 1.3: Basic seismic structural design of typical liquid storage tanks.......................77
III.1.3. Task 1.4: Basic seismic structural design of typical cylindrical and spherical pressure
vessels ........................................................................................................................................78
III.1.4. Task 1.5: Basic seismic structural design of typical piping system and its support
structure ........................................................................................................................................78
III.2.1. Task 2.1: Seismic forces in vertical cylindrical liquid storage tanks .............................79
4
III.2.2. Task 2.2: Seismic forces in pressure vessels ..................................................................80
III.3.1. Task 3.1 Basic material testing and characterization .....................................................81
III.3.2. Task 3.2: Tests on tank openings/nozzles under strong cyclic loading ..........................85
III.3.3. Task 3.3: Tests on bolted piping connections under strong cyclic loading ....................88
III.3.4. Task 3.4: Tests on piping elbows under strong cyclic loading .......................................94
III.3.5. Task 3.5: Tests on T-joint pipe connections under strong cyclic loading ....................103
III.4. Appendix of WP4 .............................................................................................................106
III.4.1. Task 4.3 Parametric numerical study of critical component behaviour .......................106
III.5.1. SEISMIC RESPONSE OF LIQUID STORAGE TANKS ...........................................113
III.6.1. Task 6.1 Seismic behaviour of vertical-cylindrical pressure vessels............................116
III.6.2. Task 6.2 Seismic behaviour of horizontal-cylindrical pressure vessels .......................118
III.6.3. Task 6.3 Seismic behaviour of pressure vessels of spherical shape .............................120
III.7.1. General analyses procedure for seismic design acc to EN 1998-1 and 4 .....................124
III.7.2. Seismic linear elastic analyses methods .......................................................................125
III.7.3. Investigation of seismic input .......................................................................................125
III.7.4. Influence of supporting structure..................................................................................127
III.7.5. Example of ultimate limit state for elbows ...................................................................128
IV. List of acronyms and abbreviations ..........................................................................................131
V. List of figures ...........................................................................................................................133
VI. List of tables .............................................................................................................................139
VII. References ................................................................................................................................141
5
I. Final summary
Liquid storage tanks, pressure vessels and industrial piping are steel structural systems, present in all
industrial facilities (chemical and petrochemical industries, and power plants). Safeguarding their
structural integrity against earthquakes constitutes a key issue towards increased safety and unhindered
operation of the industrial facility. Their structural behaviour, and in particular seismic design, is quite
different than steel buildings and has several particularities, requiring a combined civil and mechanical
engineering expertise. The particularities stem from their shape and geometry, the presence of high
internal pressure, and the dynamic behaviour of the enclosed liquid, which may affect significantly their
load and deformation capacity.
The design of those structures has been dominated by the use of American standards (API 650, ASME
VIII and B31.3). Nevertheless, for the case of seismic design, those standards (especially ASME
standards for pressure vessels and piping) contain very limited provisions, referring mainly to structural
design codes and specifications. European specifications (EN 14015, 13445, 13480) also contain
limited provisions for the earthquake-resistant design of industrial equipment. An effort has been made
in Eurocode 8 (EN 1998-4), which concerns almost exclusively vertical-cylindrical liquid storage tanks.
Nevertheless, those rules do not cover all possible limit states and, furthermore, they need significant
improvement to reach a level of applicability for design practice.
The INDUSE program consists of an interdisciplinary research effort that combines civil and
mechanical engineering expertise, for the purpose of developing guidelines, which can be used for the
seismic design of liquid storage tanks, pressure vessels and piping, within the Eurocode design
framework. Towards this purpose, extensive experimental, analytical, and numerical work has been
conducted within the INDUSE project with the synergy of academic units, research centres and
industrial partners. The guidelines are novel and unique, incorporating modern aspects of earthquake-
resistant design, and are aimed at:
 expanding EN 1998-4 provisions towards an integrated seismic design of liquid storage tanks
and attached equipment, incorporating some special features and all possible failure modes
(WP5) and
 extending the applicability of Eurocode 8 (EN 1998) concepts for the cases of industrial
pressure vessels (WP6), and piping systems (WP7).
The following intermediate goals have been achieved within the INDUSE project:
 A basic comparison has been performed between current seismic design provisions in
European and American standards together with an assessment of seismic damages in
industrial facilities (WP1).
 Taking into account the particularities of each structural system, seismic actions have been
determined (WP2).
 Extensive experimental testing has been conducted on key piping components (e.g. nozzles,
pipe connections/branches and elbows) under strong cyclic loading; furthermore a piping
system has been tested under pseudo-dynamic loading (WP3).
 Finite element analyses on these components have been performed, simulating the
experiments and covering a wide range of geometric, material and loading parameters (WP4).
The results of the above investigations are summarised below, demonstrating the achievements of
INDUSE project objectives, work package per work package.
7
I.1. WP1: Comparison of European and American design standards for the
seismic design of industrial liquid storage tanks, pressure vessels and piping and
development of available test data base.
The objectives of this workpackage consist of (a) collecting and evaluating available cyclic loading test
data on critical industrial components, (b) collecting and evaluating damages in industrial equipment
due to strong earthquakes and (c) comparing relevant European and American standards in terms of
seismic design. In task 1.1, the following experimental testing programs on piping components and
piping systems are described, evaluated and extensively discussed:
1. Berkeley National Laboratory tests, UK
2. University of Liverpool tests (1st and 2nd part)
3. EPRI-NRC Piping and Fitting Dynamic Reliability Program
4. ETEC System Tests
5. Westinghouse Hanford Test
6. EPRI Prototype Tests
7. Japanese High Level Tests
8. University of Akron Studies
9. Japanese Vibration Tests
The collection and evaluation of typical damages in industrial equipment structures subjected to strong
earthquake events has been performed in task 1.2, for the following earthquake events:
1. Imperial Valley Earthquake, California, USA, October 15, 1979
2. Coalinga Earthquake, California, USA, May 2, 1983
3. Northridge Earthquake, Los Angeles, USA, January 17, 1994
4. Kocaeli Earthquake, Turkey, August 17, 1999
5. Chile Earthquake, March 3, 1985
6. Kobe Earthquake, January 17, 1995
Finally, the main effort of WP1 has concentrated in task 1.3, on the selection and preparation of the
seismic design for eleven (11) case-studies on tanks, pressure vessels and piping, namely:
 Case Study #1: 27.4-m-diameter liquid storage tank
 Case Study #2a: 18-m-diameter liquid storage tank
 Case Study #2b: 6-m-diameter liquid storage tank
 Case Study #2c: liquid storage tank with floating roof
 Case Study #3: broad liquid storage tank
 Case Study #4: spherical pressure vessel
 Case Study #5: vertical pressure vessel
 Case Study #6: horizontal pressure vessel
 Case Study #7: vertical pressure vessel on a steel structure
 Case Study #8: piping system on a pipe-rack
 Case Study #9: piping system on a steel support structure
The above case-studies refer to existing industrial structures, and were provided by our industrial
partners. Their seismic design has been performed using European and American standards. Seismic
actions have been calculated, and the necessary verification checks were conducted. Inconsistencies and
issues that require further consideration have also been identified.
I.2. WP2: Determination of seismic forces on industrial structures and critical

components and preparation of experiments.
The objectives of this workpackage (WP2) have been: (a) the development of systematic procedures to
calculate seismic action on tanks, pressure vessels and piping, using advanced analysis methods,
considering the particularities of the above systems; (b) investigation of the effects of nonlinear support
system response on the seismic behaviour of pressure vessels and piping; examination of the influence
of dissipative support systems towards seismic force reduction; (c) Determination of cyclic loading
earthquake action on selected critical components, considering strong seismic input, and employing the
8
above analysis procedures; (d) Specification of specimen geometry to be tested in the subsequent
experimental program together with the detailed specification of base material and welding procedures.
The above work has been accomplished in the course of WP2. In particular,
 In Tasks 2.1, 2.2, and 2.3, procedures to calculate seismic action on industrial structures have
been presented, using advanced analysis methods, for liquid storage tanks, pressure vessels and
piping, respectively. The methodologies are based on advanced numerical methods, taking into
account the dynamic interaction between the structure and the enclosed liquid (in liquid storage
tanks) and the interaction between the piping and the supporting structure (in piping systems).
 The investigation of nonlinear support system response on the seismic behaviour of pressure
vessels, for the purpose of reducing of the seismic force, has been investigated in Task 2.4.
 A typical piping system has been considered, and simulated through advanced numerical tools in
order to determine seismic action in various components, as described in Task 2.5.
 Finally, in Task 2.6 specifications for the ordering and manufacturing of specimens to be tested
in the subsequent experimental program have been specified. Detailed specifications of base
material and welding procedures have been stated to be followed by all partners.
I.3. WP3: Experimental testing of structural components of critical importance

under strong cyclic loading.
In this workpackage, extensive experimental testing has been performed on steel material, on piping
components and on a piping system.
Testing has been performed for the mechanical characterization of both base material and weld
material. Base material characteristics has been evaluated for elbows; tee-fittings; plates for nozzles;
straight pipes connected to fittings in full-scale specimens. The base material characterization consists
of (a) monotonic tensile testing; (b) cyclic loading with imposed strains to obtain hysteresis loops
(cyclic hardening/softening) of elbows both longitudinally and transversely; (c) cyclic loading with
imposed stresses to obtain ratcheting characteristics of elbows both longitudinally and transversely; (d)
cyclic loading with imposed deformations to obtain cyclic hardening/ softening of straight pipes, to be
connected weld neck flanges in full scale testing; (e) ring crush testing to evaluate the possible material
anisotropy of straight pipes, to be connected to flanges and elbows in full scale testing. In addition, tests
on butt welded specimens in 8-inch pipes, to check the validity of the WPS (developed within the
INDUSE project). First, non-destructive tests were carried out. Then, destructive tests were performed:
(a) macroscopic examination of welds (EN1321); (b) hardness tests (EN1043-1); (c) bend test on welds
(EN910); (d) transverse tensile test on welds (EN895, EN6892-1); (e) Charpy V-notch impact test on
welds (EN875, EN10045-1). Low cycle fatigue tests of welded specimens were also carried out and
an S-N curve was derived.
Subsequently, a number of medium-scale experiments in key components of industrial structural
systems have been conducted under strong cyclic loading, to determine their mechanical behaviour and
failure modes, and to obtain adequate measurements for the calibration of the numerical models, to be
used in subsequent work packages. The components are 6-inch and 8-inch pipes, and the material of all
specimens is P355N, which is equivalent to X52. In addition to the tests on specific components, a
piping system, containing 3 elbows, 1 Tee junction and 1 flange, has been tested tests under pseudo-
dynamic loading at 4 seismic levels. More specifically,
 six (6) tests have been conducted on 6-inch tank openings (nozzles). Three different shell nozzle
reinforcements are considered, namely (P1) internal reinforcing plate, (P2) external reinforcing plate
and (P3) thickened nozzle reinforcement. The tests were carried out in the range of ultra-low-cycle
fatigue considering longitudinal and transverse load directions for each nozzle reinforcement type.
Increasing amplitude loading was applied according to ECCS No.45 protocol. Failure occurred in
form of fatigue crack in either side of the weld between the 6-inch pipe and the reinforced shell. All
three types of reinforcement exhibited similar strength in terms of loading cycles. Furthermore,
transverse loading has been more severe than longitudinal loading.
 fifteen (15) tests have been conducted on full-scale bolted (flanged) piping connections as follows:
(a) Six tests have been conducted on standard PN40 and PN63 bolted flanged connections, as
9
specified in EN 1092, subjected to monotonic and cyclic bending load conditions, in presence of
internal pressure. (b) An experimental test program on eight (8) non-standard Bolted Flange Joints
has also been conducted under monotonic and cyclic loading with increasing amplitude (in the
presence of low internal pressure) in order to investigate their capacity and performance and assess
their capacity under seismic loading. Two non-standard bolted connection geometries have been
considered, comparatively thinner than the Standard ones, connected with an 8-inch connecting pipe
under regular operating conditions of a petrochemical industry.
 A significant number of tests [thirteen (13) tests] have been performed on 8-inch (D=219.1mm)
long-radius (R/D=1.5) SCH40 (t=8.2mm) elbow specimens, from P355N material, connected to
straight pipes of the same dimensions at either side (L/D=5), under constant-amplitude in-plane
loading. Eight non-pressurized specimens have been tested at various loading amplitudes (covering
the entire range from quasi-elastic to severely-plastic behaviour). In addition, five specimens have
been tested under pressure (at pressure levels up to 42% of yield pressure) at strong cyclic loading,
causing severe plastic deformation. In all specimens, failure occurred in the form of a longitudinal
crack at the central part of the elbow, at the flank location. It was found that the number of cycles to
failure depends significantly on the loading amplitude, whereas the presence of internal pressure
resulted in a small decrease of the number of loading cycles to failure.
 Eight (8) tests have been performed on 8-inch-to-6-inch SCH40 Tee-branch pipe connections,
considering two configurations: (a) four specimens with Tee fittings (TJF) and (b) four specimens
with reinforcement plates at the connection of its two main parts (TJWR). The specimens have been
tested under monotonic loading (1 test for each type) and cyclic out-of-plane bending loading (3
tests for each type). Under monotonic loading conditions, the TJF specimen showed about 93%
higher load capacity. Under cyclic loading conditions, the TJF specimen exhibited higher fatigue
resistance of about 7-8 times that of the TJWR specimen. For the increasing amplitude loading, the
TJF specimen resisted 36% more loading cycles than the TJWR one. The TJWR specimens failed
along the circumference of the branch-to-reinforcement plate weld-toe either on the branch side or
on the welded reinforcement plate side of the connection. The TJF specimens failed in the region of
the joint fitting, at a distance of about 60 to 80 mm away from the branch-to-fitting weld-toe.
Finally, in addition to testing individual components under strong cyclic loading, a piping system has
been tested through pseudo-dynamic loading, at full-scale. The piping system contained several critical
components, mainly elbows, also 1 flange, and 1 Tee, and it is aimed at investigating its seismic
performance at four seismic levels, offering a good alternative to more expensive shaking table tests.
The present testing employs a new hybrid testing technique which enables the seismic testing of a
structural system by using a limited number of actuators. Several pseudo-dynamic tests on the piping
system were performed under different levels of earthquake PGAs corresponding to serviceability and
ultimate limit states suggested by Standards. The experimental and numerical results showed a good
agreement in all tests. Moreover, a good seismic performance of the piping system was observed during
the experiments. It was found that, even under the Collapse Limit State level earthquake, both the
piping system and its critical components did not fail or yield. Therefore, it is concluded that present
seismic design rules for piping systems and components are rather conservative and amendments are
necessary to overcome some degree of conservatism.
I.4. WP4: Numerical simulation of industrial component behaviour under strong

cyclic loading.
The main purpose of this work package is the development of an advanced numerical simulation
procedure, and its use for the prediction of mechanical behaviour of critical structural components, in a
rigorous, efficient and accurate manner. The specific objectives of this workpackage (WP4) have been:
(a) the definition and calibration of an appropriate cyclic-loading constitutive material model to use for
a realistic simulation of elasto-plastic behaviour of tubular components; the constitutive material model
should be able to simulate material degradation effects, as well as low-cycle fatigue, (b) the
development of a modelling procedure using finite element that simulates accurately the behaviour of
industrial components; the modelling procedure has been validated through the experimental data of the
previous workpackage (WP3) and (c) the performance of an extensive parametric study, using the
numerical modelling methodology developed above, to examine the behaviour of structural
components, within a wide range of geometric and material parameters.
10
Task 4.1 has been devoted on the choice, implementation and calibration of a suitable material
(constitutive) model for simulating the material behaviour of steel material. Following significant
investigation, it has been concluded that the classical plasticity models based on isotropic and kinematic
hardening rules may not be adequate in simulating cyclic response of steel material under both strain-
controlled and stress-controlled conditions. Under this perspective, the Armstrong-Frederick (A-F)
model has been selected among other models to be the main model for the numerical simulations within
the INDUSE project. It is an advanced plasticity model, significantly more powerful than linear
kinematic models, available in many commercial finite element codes (e.g. ABAQUS, MARC etc.).
The model adopts a von Mises description of the yield surface, and assumes a nonlinear kinematic
hardening rule for the evolution of the back stress (i.e. the centre of the yield surface). The calibration
procedure is based on material test results in Task 3.1, from coupon specimens for the material testing
were extracted from the longitudinal and, mainly, from the hoop direction of elbow specimens. The A-
F model is calibrated to fit the material cyclic stress-strain curves and the appropriate values for the
corresponding material parameters are obtained, considering the expected strain range at the critical
locations of the specimens under consideration. In addition to the A-F model, the Tseng-Lee model,
based on the bounding surface concept is also used and calibrated from the coupon tests. Such a model
is not incorporated in the commercial finite element codes (e.g. ABAQUS, MARC etc.) and, therefore,
a special-purpose user subroutine has been developed.
The simulation of piping component experiments conducted in WP3 (nozzles, elbows, flanges and
Tees), has been performed in Task 4.2, for the purses of developing and validating rigorous numerical
tools. To achieve a good level of accuracy, nonlinear finite element models are employed, which
employ shell elements or solid elements, and the nonlinear constitutive models for cyclic plasticity,
calibrated in Task 4.1. In particular,
(a) NOZZLES: The finite element models were developed in ABAQUS using 8-node reduced-
integration solid elements (C3D8R), considering geometric and material non-linearity. Each shell
nozzle reinforcement type was investigated under longitudinal and transverse loads, as in tests. By
comparing the cyclic load-displacement curves as well as the points of failure in the test with those of
the simulations, a good accordance was achieved. To estimate the number of cycles until failure, a quite
simple approach was chosen, in terms of the equivalent plastic strains.
(b) FLANGED (BOLTED) JOINTS: A three-dimensional model of the flanged joint including the
gasket and the pipes using MARC. Both weld neck flange ring (ring and hub) and non-standard flanges,
have been modelled by means of 8-node brick element, while the gasket sealing element employs 8-
node composite brick elements that allow adopting specific material model able to describe the gasket
behaviour. The pipe has been modelled by means of shell elements, each bolt is modelled with a beam
element, the head of the bolt is modelled as rigid and pre-tension is taken into account. The spiral
wound gasket has also been modelled. The comparison of the experimental and numerical results
demonstrated a good correlation in terms of both general load-versus-stroke evolution and specimen
deflections.
(c) ELBOWS: For the simulation of each conducted test, a numerical model has been developed, with
the exact dimensional characteristics of the specimen based on detailed dimensional measurements
conducted prior to testing (thickness variability of the curved and the straight parts of the elbows has
been taken into consideration). Weld over-thickness is also modeled. The finite element model has been
developed in ABAQUS with eight-node incompatible mode (C3D8I) solid elements. The material
models employed are the Tseng-Lee and the Armstrong-Frederick models. The numerical results are
reported in the form of load-displacement and flattening-displacement curves, and are in very good
agreement with the corresponding test data. Fatigue life has been estimated very successfully in terms
of local strain variation at the crack location.
(d) TEE-JUNCTIONS: Numerical models with solid elements in ABAQUS were used to simulate the
tests of task 3.5. The comparison of the predicted load-displacement curves with the corresponding
experimental curves shows that the adopted model is able to provide numerical results in good overall
agreement with the experimental findings. Some differences can be attributed to minor slips at the end
supports of the test specimens. These slips are more evident in the TJF tests due to their increased load
resistance. In any case, the numerical models can predict the maximum and minimum resistance loads
fairly well.
11
Finally, in Task 4.3, the above numerical tools have been employed in performing an extensive
parametric numerical study of the components under consideration, considering a wide range of
parameters. The main issues related to this numerical investigation are described below:
(a) NOZZLES: The defined parameters are (i) plate thicknesses, (ii) diameter and thickness of the
attached nozzle and (iii) geometry of the reinforcements. Parametric studies have been performed for
both monotonic and cyclic loading. In the first step, monotonic simulations for the 4 different load
directions for each type of reinforcement were conducted, which resulted in 240 monotonic simulations.
The influence of internal hydrostatic pressure has also been taken into account. Subsequently, a
parametric numerical study has been conducted for cyclic loading in both longitudinal and transverse
load direction and each type of reinforcement, considering an increasing amplitude loading (ECCS
No.45 protocol). The results have been presented in terms of relative resistance ratios versus partial
ductility of the component.
(b) FLANGED (BOLTED) JOINTS: The finite element models employed for the comparison with full
scale tests of the flanged joints have also been employed to perform a sensitivity analysis of various
load scenarios and geometrical parameters on flanged joints. Both standard and non-standard flanges
have been analysed: (a) A numerical parametric study on EN1092-1 type 11 welding neck standard
flanged joints subjected to bending loading aimed at investigating the effect of the parameters such as
internal pressure level, axial load and geometry on the behaviour of the bolted flanged connections. (b)
In non-standard flanged connections, special attention was paid on the thickness of the flange, in terms
of the relevant limit state, in both bending axial loading conditions. It has been demonstrated that non-
standard flanges can be used in several industrial piping applications for piping systems operating under
both normal operating and seismic conditions.
(c) ELBOWS: The behaviour of steel pipe elbows under strong cyclic in-plane loading has been studied
numerically, based on numerical models developed to simulate the experiments, and using two
appropriately calibrated plasticity models. Applying various ranges of end-displacement, strong cyclic
in-plane bending loading has been introduced and the number of cycles until failure has been
determined. The numerical results are reported in the form of load-displacement and flattening-
displacement curves. The effects of elbow geometrical characteristics on its mechanical behaviour have
been studied. Furthermore, the case of out-of-plane bending has been examined; it was found that the
formation of local buckling results in local strain concentration, which would affect significantly
the failure life of the elbow.
(d) TEE-JUNCTIONS: Following the calibration of the numerical models, an extensive parametric
study has been conducted. For the Tee-junctions under consideration the following parameters of
significant importance are taken into account: (a) tee-junction configurations, comparing those with a
reinforcement plate with those that use a Tee-junction fitting, (b) load direction, namely in-plane and
out-of-plane bending and axial loading of the branch, (c) level of internal pressure, (d) radius of Tee
fitting, (e) geometric parameters, such as diameter and thickness of branch and main pipe, (f)
reinforcement plate geometry (width and thickness variations). Both configurations are capable of
enduring high seismic loadings with the fitting being the strongest. The modes of failure have been
identified and the capacity for the different geometries and loading directions determined.
I.5. WP5: Seismic behaviour and special seismic design issues in liquid storage
tanks; development of relevant design guidelines/recommendations.
The main objective of this work package (WP5) is the development of a complete set of design
guidelines / recommendations for the seismic analysis and design of industrial liquid storage tanks. The
guidelines are compatible with the existing Eurocode design framework, and extend the existing rules
of Eurocode 8 (EN 1998-4).
The work conducted in Task 5.1 consisted of examining several special issues on the seismic response
of liquid storage tanks. The results of this investigation can be considered as a background document
for the seismic design guidelines of liquid storage tanks to be presented in a companion report. In
particular, the following issues have been addressed in detail: (a) the number of sloshing modes to be
considered in convective action and their combination with impulsive action; (b) the effects of shell
deformation on the seismic response; (c) calculations of hoop hydrodynamic stresses; (d) uplifting
12
behaviour of unanchored tanks; (e) buckling at the top of the tank; (f) elephant’s foot buckling and
behavior factor of liquid storage tanks; (g) nonlinear sloshing effects; (h) floating roof behaviour.
Considering typical liquid storage tanks, provided by our industrial partners, the investigation of the
above issues was aimed at evaluating EN 1998-4 provisions (in comparison with those of API 650) for
the seismic design of liquid storage tanks, towards proposing possible improvements/amendments of
EN 1998-4. The main conclusions from this investigation can be stated as follows:
 For the majority of tanks, consideration of only one sloshing (convective) mode is adequate
for the calculation of the total seismic force. Furthermore, the combination of the convective
force with the impulsive force should be conducted according to the SRSS rule.
 Using a systematic two-step methodology for analyzing the seismic response of uplifting
tanks has shown that the EN 1998-4 provisions provide a reasonable yet conservative
approach for the increase of axial compression due to uplifting.
 A simple methodology is presented in order to design tanks against buckling at the top, due to
liquid sloshing, using the concept of “equivalent” external pressure.
 The behavior (reduction) factor for the seismic design of liquid storage tanks is a controversial
issue that has not been examined thoroughly so far. Because of the relatively low post-
buckling strength of the shell tank, the authors would be quite cautious in using a behavior
factor greater than 1.5 for the impulsive component of motion.
 There exists little information on current design practice for floating roofs; to avoid complete
tank destruction, one should avoid seal damage to prevent fire ignition.
 Nonlinear sloshing effects on the seismic behavior have not been thoroughly investigated so
far. Limited numerical investigations, reported elsewhere, indicated that their effects on the
total seismic force may not be important, but they can have significant effects on freeboard
considerations.
The design guidelines for the seismic design of liquid storage tanks have been developed in the course
of Task 5.2, at an attempt to present a document that improves the existing provisions of EN 1998-4.
The structure of those guidelines is: 1 Introduction; 2 Relevant Standards and Guidelines; 3 List of
Symbols; 4 Seismic Actions in Liquid Storage Tanks; 5 Seismic Resistance of Liquid Storage Tanks; 6
Special seismic design provisions; ANNEX A: Hydrodynamic loading due to seismic action; ANNEX B:
Unanchored tanks; ANNEX C: Shell Buckling Resistance.
The guidelines can be employed by designers of liquid storage tanks in seismic areas, providing
background information to existing standards, and identification of ultimate limit states under seismic
action, as well as amending and improving the existing provisions of EN 1998-4.
I.6. WP6: Seismic behaviour and design of industrial pressure vessels;

development of seismic design guidelines/recommendations.
The main objective of this workpackage (WP6) is the development of design guidelines /
recommendations for the seismic analysis and design of industrial pressure vessels (vertical cylinders,
horizontal cylinders and spheres). The guidelines concern various geometries and are compatible with
the existing Eurocode 8 (EN 1998) framework for earthquake-resistant design.
The seismic behaviour of vertical and horizontal pressure vessels has been analyzed in Tasks 6.1, 6.2
and 6.3, using pushover and incremental dynamic analysis techniques. The failure mechanisms were
discussed and possible failure criteria were proposed. The behavior factor q determined for the vertical
vessels was found to be 1.5, lower than the value of 2 suggested by ASCE7 for such pressure vessel
geometries. The effect of internal pressure was insignificant up to the point of maximum capacity of the
vessel but beneficial for post-buckling behavior. For the horizontal vessels the effect of reinforcing
rings, the effects of length, thickness and those of internal pressure were studied using pushover
analysis. The effect of internal pressure was more significant when loading was applied in the
transverse rather than in the longitudinal direction due to the change of failure mode. Reinforcing rings
are recommended for large pressure vessels or pressure vessels located in high seismicity areas. Caution
shall be taken when deciding the area of the saddle reinforcement as the adjacent shell area is expected
to fail in the case of no internal pressure.
13
Moreover, behaviour factors q for spherical pressure vessels have been determined and showed to be
bigger than 2, evaluating conservatively the most critical earthquake from a set of 10 accelerograms. A
q-factor of 2.7 was found for the non-braced system, which should be restricted to this value since
dynamic instability phenomena can occurs suddenly using higher values. For the braced supporting
structure a q-factor of 2.3 was identified, when limitation of the plastification to the bracings is desired.
Otherwise a higher value of 3 has been suggested.
The guidelines for seismic design of pressure vessels are developed in Task 6.4 considering cylindrical
pressure vessels that are vertical and horizontal or of spherical shape. They consider ground level
supports, elevated supports and supports by large structures. Recommendations for achieving ductile
response are given by means of global design concepts and local detailing rules. The guidelines for
seismic design are structured as follows: Introduction; Relevant standards; List of symbols; Definition
of limit states; Seismic Input; Seismic analysis; Seismic design; Design Examples; ANNEX A: Supports
by large structures.
The developed guidelines are aimed for use by pressure vessel designers. They are meant to be
considered in addition to existing seismic design rules for steel structures which - to large extend -
remain valid for pressure vessels and in particular for their supporting structures. The guidelines are
also accompanied by specific design examples.
I.7. WP7: Seismic behaviour and design of industrial piping systems and
components; development of seismic design guidelines/recommendations.
The main objective of WP7 is the development of design guidelines for the seismic analysis and design
of industrial piping, including their support systems. Special emphasis has been given to the design of
an adequately dissipative support system, to reduce the seismic action on the piping. The guidelines are
compatible with the existing Eurocode 8 (EN 1998) framework for earthquake-resistant design.
Extensive work has been conducted in task 7.1, to develop background documents for the guidelines.
The study aimed at identifying all the design steps of piping systems and all design aspects that need to
be clarified. This was made through a representative case study, analysed according to both European
(EN13480:3) and American standards (ASME B31.3) devoted to piping systems. With reference to a
typical piping system the comparison between EN13480 and ASME B31.3 standards showed that the
two codes provide similar indications for the evaluation of the seismic response of pipelines supported
by a pipe-rack. The usual way of designing piping systems is based on the allowable stress approach, so
that the structure is considered elastic. Consequently only an operating basis earthquake condition
(OBE) can be taken into account for design. Conversely, the modern approach to the seismic design of
structures is to differentiate serviceability from ultimate limit states. This latter condition could be
represented by the Safe Shutdown Earthquake (SEE). It was identified that a proper definition of the
deformation capability of the pipes and fittings (ovalization, etc.) in the plastic range would be
necessary, and on this aspect the research has not reached solid conclusions. However, the advantages
would be several and first of all the design optimisation. For the case study under consideration, the
acting moment is well below from the moment capacity of the bolted joint of the piping. This
conservatism seems to be in contrast with the modern performance based-design approach, for which a
certain level of yielding in the system is allowed according to a specific performance.
The guidelines have been developed in task 7.2, and consist of a concise document compatible with
the existing Eurocodes design framework, which includes some useful indications on seismic
analysis of piping systems. The document follows the basic principles of earthquake engineering
expressed in EN1998:1, extending the existing rules to industrial piping components and systems.
The guidelines are organized in the following chapters: Scope; Definitions; Reference documents,
Symbols and Abbreviations, Methods and relevant parameters for seismic analyses; Seismic input;
Modeling of piping systems; Dynamic seismic analyses; Limit states (including results obtained from
the testing programs); Recommendations.
The guidelines will be used by designers of industrial piping systems, providing background
information to the existing standards, description of critical steps and parameters in seismic design of
piping systems, recommendations for modelling of piping systems and the interaction with supporting
substructure and identification of ultimate limit states under seismic action.
14
I.8. WP8: Coordination of the project.
The main objectives of WP8 have been the establishment of an efficient management process; the
definition of standard protocols; the performance of SWOT analysis; and the continuous monitoring of
the project. All coordination meetings were regularly carried out, the FTP server directory was created
by UThessaly and continuous exchanges of information favoured objectives and relevant project
outcomes. Regarding testing of components in WP3, common standard testing protocols have been
decided for assuring quality of test results. In addition, the range of geometric, material and loading
parameters to be used in the numerical studies in WP4 have been determined. Furthermore, the
evaluation of the project in terms of Strength, Weakness, Opportunity and Threat allowed a global
evaluation of the research project both in terms of activity coordination and work effectiveness of each
research unit. Intermediate annual reports and deliverables enabled internal and external monitoring of
the project advances, and of the degree of fulfilment of objectives set for various phases.
I.9. Conclusions and main results

The INDUSE project has been a pioneering project within the RFCS research program. It combined
experimental, numerical and analytical tools to conduct high-level research in the inter-disciplinary area
of seismic safety of industrial facilities. Extensive experimental and numerical work has been
conducted, to investigate the mechanical behavior of critical components; their behaviour under strong
cyclic loading has been shown well above the requirements of existing design provisions. In addition,
the INDUSE project resulted in the development of design guidelines/recommendations (a) expanding
EN 1998-4 provisions towards an integrated seismic design of liquid storage tanks and (b) extending the
applicability of Eurocode 8 seismic design concepts for the cases of industrial pressure vessels and
attached equipment and piping systems including their support systems. The results and deliverables of
INDUSE are both novel and unique, resulting in
 better understanding of structural behaviour of tanks, pressure vessels and piping under strong
cyclic (seismic) loading,
 more reliable definition of ultimate limit states for critical industrial components
 significant improvement of seismic design state-of-the-art
and offering the necessary scientific background for Code Drafting Committees for improving existing
design standards, towards safer design of industrial power plants and chemical/petrochemical facilities.
15
II. Scientific description of the results
Objectives of the project
 Critical evaluation and comparison between currently available seismic design provisions in
European and American standards (WP1), considering specific case studies.
 Determination of seismic actions on industrial systems (WP2), taking into account the
dynamic behaviour of each structural system.
 Testing of critical piping components (nozzles, flanged connections, Tees and elbows) under
strong cyclic loading (WP3). Pseudo-dynamic testing of a piping system.
 Performance of extensive numerical studies on critical piping components, based on rigorous
FE simulations, covering a wide range of geometric, material and loading parameters (WP4).
 Development of design guidelines/recommendations, expanding EN 1998-4 provisions
towards an integrated seismic design of liquid storage tanks (WP5) and extending the
applicability of Eurocode 8 (EN 1998) concepts for the cases of industrial pressure vessels
and attached equipment (WP6), and piping systems including their support systems (WP7).
Comparison of initially planned activities and work accomplished

All the above objectives, stated in the Technical Annex, have been successfully accomplished in the
course of the relevant work packages of INDUSE project.
II.1. Work package 1 – Comparison of European and American design standards

for the seismic design of industrial liquid storage tanks, pressure vessels and
piping and development of available test data base
The objectives of WP1 are summarized as follows:
 Collection and evaluation of cyclic loading (static and dynamic) test data on critical industrial
components of industrial systems available in the literature.
 Collection and evaluation of damages in industrial equipment structures subjected to strong
earthquake events
 Comparison of currently used European (EN and national) standards and American (API or
ASME) standards, in typical industrial equipment structures, namely (a) liquid storage tanks,
(b) pressure vessels and (c) industrial piping, to identify main differences in seismic design
procedures.
II.1.1. Task 1.1: Collection and evaluation of available relevant experimental data
II.1.1.1. Experiments on piping components and piping systems
In this critical evaluation of data, experiments on piping components and piping systems are described
and discussed. The following experimental testing programs on piping components and piping systems
are described, evaluated and extensively discussed.
1. Berkeley National Laboratory tests, UK,
2. University of Liverpool tests I
3. University of Liverpool tests II
4. EPRI-NRC Piping and Fitting Dynamic Reliability Program
5. ETEC System Tests
6. Westinghouse Hanford Test
7. EPRI Prototype Tests
8. Japanese High Level Tests
17
9. University of Akron Studies
10. Japanese Vibration Tests
II.1.1.2. Experiments on liquid storage tanks

The dynamic behavior of ground-based liquid storage tanks had been the subject of extensive studies at
University of California Berkeley during 70’s-80’s. Theoretical and experimental investigations had
been conducted to seek possible improvements in the design of such tanks to resist earthquakes [46],
[15], [37], [36]. These experiments were mainly directed towards studying sloshing behavior and
buckling of the shells.
More recently, in the same university, a well-organised series of tests took place [48]. Experimental and
analytical studies were conducted on seismically isolated and fixed-base liquid storage tanks. Shaking
table tests were performed on vertical cylindrical open-top steel tanks [48]. Extensive parameter studies
were performed, including comparisons of fixed-base and isolated model response to simulated ground
motions of varying magnitude, time scale, and combination of components.
Finally, in [43] the effectiveness of the base isolation on steel storage tanks has been investigated
through numerical models and then checked by shaking table tests on a reduced scale model of a real
steel tank, typically used in petrochemical plants. In the experimental campaign the floating roof has
also been taken into account.
II.1.2. Task 1.2: Collection and evaluation of reported earthquake damages in

industrial facilities
The examination of the seismic impact on industrial facility systems, and it is extremely valuable as one
attempt to assess the probable risks in a future earthquake. The following review is in the framework of
Task 1.2, and can serve as a guide to identify potential problems and weaknesses. In this report a few
notable examples of seismic damages are given where the impact was more severe.
The collection and evaluation of typical damages in industrial equipment structures subjected to strong
earthquake events has been performed in task 1.2, for the following earthquake events:
1. Imperial Valley Earthquake, California, USA, October 15, 1979
2. Coalinga Earthquake, California, USA, May 2, 1983
3. Northridge Earthquake, Los Angeles, USA, January 17, 1994
4. Kocaeli Earthquake, Turkey, August 17, 1999
5. Chile Earthquake, March 3, 1985
6. Kobe Earthquake, January 17, 1995
The main conclusions from this investigation can be briefly summarized as follows:
 Earthquakes may result in very significant damages to the industrial facilities.
 Main damages occurred to liquid storage tanks.
 The principal tank damage is elephant’s foot buckling at the tank bottom.
 In addition, damages to the top of the tanks may occur due to sloshing effects.
 Piping equipment may also exhibit significant damages, due to inadequate flexibility.
 In many instances, fire that occurred (especially in tanks) resulted in a complete destruction of
the tank.
 Failure of pipe support equipment has also occurred during the earthquake.
The above evaluation is taken into account in the next stages of the project
18
II.1.3. Task 1.3: Basic seismic structural design of typical liquid storage tanks
The present task focuses on the seismic design of liquid storage tanks. Five case studies are considered
and designed against seismic loads according to the following international design standards for liquid
storage tanks.
 Case Study #1: 27.4-m-diameter liquid storage tank
 Case Study #2a: 18-m-diameter liquid storage tank
 Case Study #2b: 6-m-diameter liquid storage tank
 Case Study #2c: liquid storage tank with floating roof
 Case Study #3: broad liquid storage tank
Their seismic design has been performed using European and American standards. Seismic actions have
been calculated, and the necessary verification checks were conducted. Inconsistencies and issues that
require further consideration have also been identified. The conclusions from this study are used in
WP2 and WP5 and can be summarized as follows:
1. There is a significant revision of API 650 11th edition with respect the 10th edition.
2. EN 14015 in seismic design follows the old edition of API 650, needs also revision.
3. Both specifications API 650 (2007) and EN 1998-4 follow closely the hydrodynamic solution
(convective/impulsive) for the contained liquid.
4. API 650 is a more design-oriented code, whereas EN 1998-4 is a document that contains very
good information, but sometimes difficult to apply.
5. Section A.3 on deformable tanks needs significant revision. The extensive rules stated in this
section are almost impossible to apply for design purposes.
6. There are significant differences in the overturning moment (ring-wall and foundation) and
the base shear, because of the significant difference in the reduction (behavior) factor on the
seismic design acceleration.
7. There is a difference in the formula for sloshing wave height (freeboard).
8. API 650 has explicit formulae for hoop stresses whereas EN 1998-4 does not. The accuracy of
those formulae needs to be investigated.
9. API 650 adopts a SRSS impulsive/convective combination, whereas EN 1998-4 adopts
summation of design values.
10. EN 1998-4 needs some revisions in section A.9 (uplifting of unanchored tanks) to reach a
more user-friendly form.
11. Direct comparison between EN 1998-4 section A.9 and API 650 E.6.2.1 for the effects of
uplifting is not possible; a more in-depth investigation is necessary.
II.1.4. Task 1.4: Basic seismic structural design of typical cylindrical and spherical
pressure vessels
In this Task, a comparison of currently used European and American standards for pressure vessels
design is being carried out. In particular, four case studies are considered, which constitute typical
pressure vessels of petrochemical facilities.
 Case Study #4: spherical pressure vessel
 Case Study #5: vertical pressure vessel on a skirt
 Case Study #6: horizontal pressure vessel on saddle supports
 Case Study #7: vertical pressure vessel on a steel structure
The analysis and design of the vertical pressure vessel supported on a skirt and horizontal pressure
vessel supported on saddles indicated the following issues:
1. Both specifications EN 13445 and ASME VIII do not contain specific provisions for the seismic
analysis of pressure vessels and the determination of seismic forces. Such guidance should be
developed and incorporated in the proposed guidelines.
2. Both specifications EN 13445 and ASME VIII indicate an increase of stress allowables for seismic
loading. In EN13445 the safety factor can be increased up to the one used for the testing load cases
for both tension and compression. In ASME VIII Div.1 a 20% increase is allowed in tension, while
19
in ASME Div.2 no such increase is specified apart from the compression allowable used in Zick’s
analysis.
3. The main conclusion from the above design and analysis of the two pressure vessels is a lack of
guidance for choosing an appropriate value of the behavior factor q . The only source for choosing a
value for the behavior factor for pressure vessels is Table 15-4.2 of ASCE 7-05, but these values are
not justified. More research work is necessary towards determining appropriate values of the
behavior factor .
4. The calculation of the natural period of the vessel depends on the flexibility of the supports, and may
affect significantly the seismic force. This causes some uncertainty in calculating the natural period
especially in the horizontal vessel on saddles supports. Therefore, it would be helpful to specify one
method to calculate the natural period to be used for design.
5. The design procedure in ASME VIII is based on local stresses calculated from Zick’s analysis. On
the other hand, the design procedure in EN 13445 is based on stress resultants.
The analysis and design of the vertical pressure vessel on a steel structure leaded to the following
conclusions.
1. First observation is that the analysis following EN1998-1 Clause 4.3.5 for appendages arrives to
more conservative Total Base Shear Values.
2. On the contrary analysis following EN1998-1 Clause 3.2.2 for structures supported on ground level,
lead to lower Total Base Shear Values. This result seam logic and it is something that was expected
to arrive to higher Base Shear value on a certain height from the ground due to higher acceleration
value.
3. Comparing the results from the above two with the results from integrated method (Col. 2), we see
that Base Shear value is between the above two cases (Col. 4 &5). We are not in position to say
which values are more realistic even though with Integrated Methods Analysis we could have more
realistic parameters concerning structure general geometry, stiffness that may affect on Floor
Acceleration. The matter should be further investigated.
4. Additionally, are the results following ASCE7-05 in between the results for Total Base Shear from
the above two methods. That method does not take into consideration the interaction between
pressure vessel and structure natural frequency.
5. Concerning the significant difference in shear that appears in –y direction at the analysis using
Integrated Method we assume that phenomenon could be explained because building overall
geometry (B.R.F. on –y direction), or perhaps local eccentricities of the vessel with respect of on –x
and –y axes.
II.1.5. Task 1.5: Basic seismic structural design of typical piping system and its
support structure
In this Task, a comparison of currently used European and American standards for piping system design
is being carried out. In particular, two case studies are considered.
 Case Study #8: piping system on a pipe-rack
 Case Study #9: piping system on a steel support structure
First, a piping system on a pipe-rack is considered and designed against seismic loads according to
European [25] and American standards (ASME B31.3) for piping systems. The comparison between
EN13480 and ASME B31.3 standards has shown that the two codes provide similar indications for the
evaluation of the seismic response of pipelines supported by a pipe-rack. From the analysis of the
results the following conclusions, considered valid for both the standards, can be drawn. Moreover,
some recommendations are provided.
1. Dynamic coupling. ASCE-07 [5] provides a simple rule to establish when the dynamic coupling
between pipes and supporting structure would be considered. It is based only on a weight ratio.
Actually, other ingredients should be included; for example the vibration periods range of the pipe.
In the analyzed case the single pipe have a limited weight and therefore short vibration periods. This
means that, as already shown, the dynamic coupling between the pipes and the pipe-rack is limited,
20
whereas the relative displacements between the pipe supports would represent a more relevant
effect. The pipes provide a sort of static coupling between the transverse frames. In fact, the models
with and without the pipe stiffness contribution have more or less the same period, whereas the
excited mass is differently distributed between the several vibration modes. Therefore, the
consequence of considering the structural contribution of the pipes is the introduction of a sort of a
rigid floor effect that couples the horizontal movements of the nodes of the same floor.
2. In-Structure spectra. The in-structure (floor) spectra suggested by prEN1998:1 and ASCE-07 differs
for a term that depends on the pipe-structure frequency ratio, included only in the European code.
This aspect is strictly related to the previous observation on the dynamic coupling. The exclusion of
the dynamic interaction could introduce a high error in the evaluation of the amplification factor of
in-structure spectra. Because usually the frequency of the piping system is relatively high with
respect to that of the supporting structure, the amplification effect would be relatively low. In any
case, to cover all the possible cases, the use of the European formula for in-structure spectra is
highly recommended.
3. Modeling of the pipes. In the analysis both beam element and shell elements have been used to
model elbows. The comparison has shown the reliability of the more simple elements based on the
beam theory, at least for the simple case of elbows. For more complicated situations the use of finite
shell elements should be considered.
4. Behavior factor. The behavior factor q is usually indicated by the codes. In particular the values
provided for piping systems by ASCE-07 seems to be too much high (R=612). This in contrast
with the idea of avoiding yielding phenomena in the pipes, in order to maintain operational
conditions also in case of strong seismic events. Certainly, the contribution of the supports in
providing dissipation capability could be taken into account, but on the basis of several investigation
found in the literature this exclude the possibility of having so high values of R.
5. Performance based-design of pipes. The usual way to design piping systems is based on allowable
stress approach. This means that the structure is considered elastic and the seismic action in reduce
of a safety coefficient equal to 1.4. Consequently only operating basis earthquake condition (OBE)
can be taken into account for the design. On the contrary, the modern approach to the seismic design
of structure is to differentiate serviceability from ultimate limit states. This latter could be
represented for example by the Safe Shutdown Earthquake (SEE).
In case study no 9, two different piping and support structures has been selected from a project, situated
in an area of moderate seismic activity (Algeria), which was completed by Tebodin in 2009. Both
simplified static equivalent (‘uniform load method’) and dynamic calculations were made in accordance
with American (ASCE7) and European (EN 1998-1/3 and EN 13480) earthquake design standards to
identify differences in approach, design seismic response spectra and results. The analysis of the piping
systems leaded to the following conclusions:
1. The design seismic spectra according to EN1998-1 and ASCE-are similar and therefore yield similar
stresses, forces and bending moments. The peaks occur roughly in the same periodicity.
2. The influence of the modeled steel support structure on the calculation results was limited.
3. For typical plant piping systems, having periods up to 0.3 sec., the uniform load method yield more
conservative results than the dynamic seismic calculations.
4. Long above ground pipeline systems on many supports may however have many modes of vibration
including lower frequencies and the a dynamic analysis may reveal higher stresses then calculated
with the uniform load method.
5. The existing earthquake standards are mainly different with respect to the design response seismic
spectra. Design and correct modeling however could prove to be of greater importance on the
dynamic calculation results then the difference in response spectra. It is recommended to include
guidelines for design and modeling in the next revisions of these standards.
21
II.2. Work package 2 – Determination of seismic forces on industrial structures and
critical components and preparation of experiments
The objectives of this workpackage (WP2) have been: (a) the development of systematic procedures to
calculate seismic action on tanks, pressure vessels and piping, using advanced analysis methods,
considering the particularities of the above systems; (b) investigation of the effects of nonlinear support
system response on the seismic behaviour of pressure vessels and piping; examination of the influence
of dissipative support systems towards seismic force reduction; (c) Determination of cyclic loading
earthquake action on selected critical components, considering strong seismic input, and employing the
above analysis procedures; (d) Specification of specimen geometry to be tested in the subsequent
experimental program together with the detailed specification of base material and welding procedures.
II.2.1. Task 2.1: Seismic forces in vertical cylindrical liquid storage tanks
The work in this task examines some special issues on the structural behaviour of upright-cylindrical
liquid storage tanks, with emphasis on seismic loading issues. The tanks under consideration are widely
used in industrial facilities and for water storage. Two main design standards are considered: EN 1998-
4 and Appendix E of API 650. There are significant differences between the two specifications, which
are due to the fact that there exist several controversial issues on this subject, open to further research.
These issues are (a) the number of modes necessary to estimate accurately the convective seismic force
due to the hydrodynamic behaviour of the liquid containment; (b) the appropriate combination of the
impulsive and the convective component of seismic force; (c) the uplifting behaviour of unanchored
tanks, with emphasis on the base plate behaviour and the increase of meridional compression; (d) the
choice of an appropriate reduction (behaviour) factor for calculating both the impulsive and the
convective force; (e) the calculation of hydrodynamic hoop stresses due to liquid hydrodynamic motion;
(f) the design of tanks against buckling at the top due to liquid sloshing; (g) the importance of nonlinear
wave sloshing effects.
In this study, the main conclusions can be stated as follows:
 For the majority of tanks, consideration of only one sloshing (convective) mode is adequate for the
calculation of the total seismic force.
 The combination of the convective force with the impulsive force should be conducted according to
the SRSS rule, which in accordance with the design provisions of API 650, Annex E.
 Numerical results show that the EN 1998-4 provisions provide a reasonable yet conservative of the
increase of axial compression due to uplifting.
 A simple methodology is presented in order to design tanks against buckling at the top, due to liquid
sloshing, using the concept of “equivalent” external pressure.
 The behavior (reduction) factor for the seismic design of liquid storage tanks is a controversial issue
that has not been examined thoroughly so far. The striking difference between the values of API 650
and EN 1998-4 is notable. Because of the unstable post-buckling behavior of thin-walled shells, the
authors would be quite cautious in using a behavior factor greater than 1.5 for the impulsive
component of motion.
 Nonlinear sloshing effects on the seismic behavior have not been thoroughly investigated so far.
Limited numerical investigations, reported elsewhere, indicated that their effects on the total seismic
force may not be important, but they may affect freeboard considerations.
II.2.2. Task 2.2: Seismic forces in pressure vessels

This task focuses on the seismic analysis and design of horizontal-cylindrical and vertical-cylindrical
pressure vessels. First, a comparison of the seismic provisions in ASME B&PV code and in EN 13445-
3 is conducted for two specific case studies. The two case studies are considered in the present task; (a)
a vertical-cylindrical slender pressure vessel on skirt and (b) a horizontal-cylindrical pressure vessel on
saddles. Both pressure vessels are typical vessels for petrochemical facilities. In addition, a numerical
investigation is conducted, aimed at proposing a reliable value of the behavior factor for those
22
structures. Those issues are examined numerically, simulating the above pressure vessels with nonlinear
finite element models. Finally, a methodology for determining seismic forces due to liquid sloshing
horizontal-cylndrical vessels is outlined. The results of the present study can be used towards a safer
and more reliable design of industrial pressure vessels under seismic action.
Two case studies are considered in the present task; (a) a vertical-cylindrical slender pressure vessel on
skirt and (b) a horizontal-cylindrical pressure vessel on saddles. Both pressure vessels are typical
vessels for petrochemical facilities.
The seismic analysis and design of these pressure vessels indicated that more detailed information
regarding the estimation of the behavior factor for these structures should be included in the design
codes, as well as more detailed guidelines for the determination of the seismic forces. An attempt to
estimate the behavior factor for these structures has been conducted using advanced numerical tools, for
pushover analysis and incremental dynamic analysis. The pushover analysis indicated that the capacity
of the presure vessels is quite higher than the required strength in seismic loading. On the other hand,
using incremental dynamic analysis for the vertical-cylindrical pressure vessel, the value of behavior
factor calculated is somewhat below the corresponding value suggested by American standard ASCE 7.
II.2.3. Task 2.3: Seismic forces in industrial piping systems

In this Task, the seismic analysis of a piping system has been studied giving particular emphasis on the
calculation of the seismic action.
To analyze the main differences of the possible seismic action models, an application to the case study
8, widely studied using the current codes for seismic analysis of piping systems, has been used. In
particular, both static and dynamic analyses have been carried out, using the response spectrum method
and time-history analysis. From the results the following main conclusions can be drawn:
 The problem of dynamic interaction between pipes and pipe-rack could play an important role in the
evaluation of the seismic analysis of a piping system. The classical primary-secondary systems may
be applied with difficulty to a piping system because they are usually based on single multi-support
secondary systems, whereas a piping represents a multi multi-support secondary systems. In
literature few works have been published in the past on this subject and more effort should be spent.
 The in-structure spectra suggested by the code are used as mean to evaluate the maximum
acceleration to apply to pipes for evaluating the maximum stress. This approach in most of cases is
certainly on the safe side, because overestimates the stress level of the pipes. But in case of complex
structural situation in which a piping system may be realized, this could not be strictly true. To avoid
problem, the generation of floor-spectra is recommended. This approach, often followed in design of
nuclear plant equipments, consists in performing a time-history analysis and evaluating the absolute
acceleration at each floor and the floor-spectrum is carried out. Thus, it is a direct approach difficult
to be generalized. This approach has also been followed for the case study #8 using a set of three
accelerograms, which has highlighted the important role of the support structure in filtering the
seismic action. Therefore, further analysis will be dedicated to the study of formulas for the
evaluation of floor-spectra.
 The analyzed case study #8 has evidenced that in some cases the problem of a correct evaluation of
the piping stress level is not univocally related to a correct determination of the acceleration at floor
where the pipes are placed. In fact, the relative displacements between supports of the same structure
could play an important role on the stress level. This means that for a correct evaluation of the stress
level in the pipes a multi-support excitation should be used, instead of the classical uniform
excitation. To clarify this aspect further analysis will be carried out.
 Another problem directly related to the previous ones is the determination of the response factor q
either if we use response spectra analysis or floor-spectra analysis. The results shown in literature
indicate that the q factor imposed by the American code is highly overestimated and that the value
suggested by other standards and guidelines as American Lifelines Alliance proposal seem to be
more reasonable: the value of q is comprised within the range 1.-3.5. Therefore, further analysis will
be dedicated to this particular aspect that has a strong influence on the seismic design of the
structures.
23
II.2.4. Task 2.4: Dissipative support systems
The investigation of nonlinear support system response on the seismic behaviour of pressure vessels, for
the purpose of reducing of the seismic force, has been investigated in this task, using a simplified
model. The investigations performed within the study focus on: verification of the dependency of the
floor response spectra on the fundamental period of the supporting structure and dependency on the
nonlinear (dissipative) behaviour of the supporting structure.
The first stage of the verification of the floor response spectra provided by the codes comprises the
calculation of the responses of ideal SDOF-models representing structures supporting industrial
components. Additional results focused on the dependency of the extreme response spectra values on
the amplification factor of the acceleration and on the maximum ductility ratio achieved during the
response. For each frequency two examples, representing the most different responses for different
accelerations, were obtained.
The results obtained by the simulations generally confirm the assumptions that the fundamental period
(resp. corresponding ground acceleration spectrum) and the dissipative behaviour of the supporting
structures do have a significant influence on the floor response spectra. Dissipative behaviour of the
supporting structure can reduce the floor spectra significantly; also the response spectra of long period
structures with low spectral ground acceleration are lower than those of short period structures.
Compared to the provisions given by EC 8, the determined spectra of short period supporting structures
are partially significantly larger. On the other hand the provisions of EC 8 for long-period structures
achieving a moderate ductility ratio are significantly higher than the expected floor response spectra. In
this context however it needs to be noted, that in order to benefit from the dissipative behaviour of the
supporting structure this structure must undergo plastification, i.e. possible overstrength of the material
may play a significant role.
In conclusion it can be stated that the floor response spectra shall undergo a systematic revision in order
to enhance both – the safety and the effectiveness of the code provisions. A proposal for a new function
for the floor response spectra cannot be provided on the basis of the results presented above only. For
the time being the only possibility for the achievement of reasonable results is the modelling of the
supporting structure together with the components. Such a possibility is explicitly included in the ASCE
7-05.
II.2.5. Task 2.5 Determination of loading amplitudes on critical components

The present task is aimed at using advanced numerical tools in terms of nonlinear finite elements for
determining loading on critical piping components (nozzles, T-junctions, elbows, flanged connections)
in a piping system, in the course of strong seismic input. Towards this purpose, a typical industrial
piping system is considered, which comprises all the above components (nozzles, T-junctions, elbows,
flanged connections), to be used later in WP3 for experimental testing. The main purpose of developing
this model is the correlation of the loading amplitudes applied in WP3 on the piping component
specimens, with the level of seismic input applied on this typical system.
II.2.6. Task 2.6 Specification of procedures for specimen manufacturing and base
material supply
For uniformity among the various experiments of the partners, it has been decided to use P355N quality
material for pressure pipes and fittings according to EN10216.03 or equivalent. For the plates P355N
quality according to EN10028.03 was selected. The selection of this material against other materials
with lower grade (e.g. grade 235 or grade 265) used in those applications, is based on the fact that the
results of the present project will be mainly applicable for new structures to be conducted in the next
years, and therefore, the use of an “obsolete” material may not be appropriate.
The material has been purchased and manufacturing of the test specimens for WP3 has started. Each
partner was responsible for the supply of the base material and manufacturing of the specimens.
However, for achieving uniformity between the specimens, it has been agreed that, UThessaly and
EBETAM have a common supplier and manufacturer for the Tee-joint piping connections. In addition,
TUDelft and CSM will have a common supplier of the elbows. An extra Tee joint was ordered by
24
UThessaly/EBETAM and sent to UTrento for the pseudo-dynamic testing. Furthermore, a number of
elbows from the TUDelft/CSM provider and manufacturer are also sent to UTrento for the same
purpose (pseudo-dynamic testing). The straight pipe elbows have been provided by the corresponding
provider as well, for the sake of uniformity.
Uniform welding procedures have been agreed where possible, similar to common industrial practices.
EBETAM was responsible to write the welding procedure specification for the various geometries.
Welding of the non-standard flanges manufactured by the UTrento in WP3 will be detailed in the
corresponding chapter.
Welding procedure will be certified according to EN 15614.01 and welder’s performance will be
certified according to EN287.01 for manufacturers of the test specimens. All samples will be inspected
for surface defects to ensure the quality of the welded samples. Visual Examination will be performed
according to EN 970. Magnetic Particle Testing (EN 1290) or Liquid Penetrant Testing (EN 571-1) will
also be performed and a test report will be submitted. Where possible butt welds will be radiographed to
ensure no defects are present at the weld that could initiate damage during the experiments, as this is
outside the scope of this project.
The WPS (Welding Procedure Specifications), as well as typical material deliverables for the straight
pipe sections, the Tee joints and the elbows are included in the corresponding section of deliverable D2.
II.3. Work package 3 – Experimental testing of structural components of critical

importance under strong cyclic loading
The work within WP3 consists of conducting a number of experiments in key components of industrial
piping systems. The specimens are typical piping components from industrial facilities, compatible with
the case studies examined in the first two work packages. Pipe diameter of the specimens is equal to 6
and 8-inch, with standard thickness (SCH40) for all specimens. A brief summary of the tests is shown
in Table II.3-1, whereas a detailed presentation of the tests is offered in the next paragraphs. For
uniformity the material of all specimens is P355N according to EN 10216-3 standard [21] which is
equivalent to API 5L Grade X52. This choice has been based on the fact that this steel grade is being
currently considered for the design and construction of new industrial facilities, instead of lower steel
grades. The P355 material is quite ductile and therefore, the test results can be also applicable to piping
components made of lower steel grade (e.g. P235, P275). Strong cyclic loading has been applied in
order to examine the structural behavior of the components under repeated excursions well into the
plastic range, simulating the loading conditions of a severe seismic event.
25
Table II.3-1: Summary of experimental tests
Tests on 6-inch tank openings (nozzles) of 3 different shell
Six (6) experiments on tank openings
nozzle reinforcement type, under 2 different load directions
(nozzles) – Task 3.2
(longitudinal and transverse)
Tests on 8-inch SCH40 long-radius induction bends ( R D 
Fifteen (15) experiments on bolted 1.5), connected to two straight pipe segments of 5 D length.
(flanged) piping connections – Task 3.3 Eight (8) tests on non-pressurized elbows and five (5) tests on
pressurized elbows.
Thirteen (13) experiments on pipe Pseudo-dynamic test of piping system with 8 and 6-inch pipes,
elbows – Task 3.4 3 elbows, 1 flange and 1 Tee- junction. Four (4) levels of
earthquake excitation with internal pressure 3.2 MPa.
Seven (7) bending tests (2 monotonic and 5 cyclic) have been
Pseudo-dynamic experiments of a conducted on pressurized standard joints (PN63, PN40).
typical piping system (4 levels) – Task Eight bending and axial tests (3monotonic and 5 cyclic tests)
3.4 have been also conducted on pressurized non-standard flanged
joints.
Eight (8) tests on 6-to-8-inch Tee-junctions specimens of two
types of branch to main steel pipe connections; connection
Eight (8) tests on Tee-branch piping
with a fitting and Tee-junction with a welded reinforcement
connections – Task 3.5
plate. Four (4) tests on Tee fittings and four (4) tests on weld
reinforced Tees.
Base material characterization.
Monotonic tensile testing for all pipe/plates.
Cyclic loading (strain and stress controlled) on coupons, from
Testing of base steel material and welds
elbows and plates.
- Task3.1
Charpy V-notch testing on elbow material.
Low cycle fatigue tests of butt-welded coupon specimens.
Weld characterization.
II.3.1. Task 3.1: Basic material testing and characterization

The objective of this task was to investigate the behavior of the base material of piping components
selected for full scale testing and the properties of butt welded joints employed in construction of the
full-scale specimens. The results obtained by the characterization campaign have been introduced as
input in FE models (described in WP4) in order to simulate as realistically as possible both monotonic
and cyclic loading of the piping components. Whenever suggested by the stress-strain field expected in
full scale testing, specimens have been extracted in longitudinal and transverse direction. The following
Table II.3.1-1 reports a summary of the small scale testing activity performed by EBETAM and CSM
within work package 3, task 3.1. It can be concluded that the performed material characterization
campaign is fully adequate with the aims of the project and the initial planned activity. Main results of
this activity can be found in the Appendix III.3.1, while the whole set of results are extensively reported
in the deliverable D3.1.
26
Table II.3.1-1: Small scale testing performed to support full scale activity and FEA
Number of corresponding full-

Component Material test type
tests scale testing
tensile-monotonic 4
cyclic/imposed strain 6
8" x 8.18mm Elbow X52
(equiv. to cyclic/imposed stress 6 Elbows
P355)
charpy V-notch 8
8" x 8.18mm Pipe tensile-monotonic 4
8" x 8.18mm Pipe P355 tensile-monotonic 4 Flanges
tensile-monotonic 4
20mm Plate P355 Nozzles
cyclic/imposed strain 3
tensile-monotonic 2
8"x8.18mm Pipe
Ring test monotonic 2
P355 Tee joints
6"x7.1 mm Pipe tensile-monotonic 2
8"x6" Fitting tensile-monotonic 2
magnetic particle inspection 1
radiographic inspection 1
macroscopic examination 1
hardness tests 1 Validation of WPS
8"x8.18mm Pipe,
P355 employed in full
butt weld joint bend test 4 scale testing
transverse tensile test 2
charpy v-notch 6
cyclic/imposed stress 9
II.3.2. Task 3.2: Tests on tank openings/nozzles under strong cyclic loading
The test program comprised in total 6 cyclic tests with 3 different shell nozzle reinforcement types and
2 different load directions (longitudinal and transverse):
 P1: Nozzle with internal reinforcing plate
 P2: Nozzle with external reinforcing plate
 P3: Thickened nozzle reinforcement
The 6 specimens have been manufactured by a company which is specialized in manufacturing of flat-
bottomed steel tanks according to the state-of-the-art. Different examples for flat-bottomed steel tanks
have been compiled from literature as well as from direct dialogue with manufacturers to mirror the
state-of-the-art. Following these examples one reference example of a complete tank has been agreed on
and the necessary size (cut-out from the complete tank) of the test specimens has been determined using
numerical calculations by comparing the stresses and deformations at the nozzle and the surrounding
area of the complete tank with those at the nozzle of the test specimen. Further on plate thickness and
joint details have been assigned according to the codes provisions of EN 14015 [26].
For the test setup the boundary conditions representing the cut-out from the complete tank were
designed and realized in order to represent in a best possible way the real load-deformation behaviour
of the complete tank. The most critical parameter so far was assumingly the possibility of the tank
bottom plate to lift up. Free rotation of the flanges was allowed for by a load application using
pendulum rods. The final complete test setup with test specimen 3-1 can be seen in Figure II.3.2-1.
27
Figure II.3.2-1: Test setup in the laboratory with test specimen P3-1.
According to the ECCS-recommendation [20] monotonic tests are required to define the test procedure
for cyclic testing. Instead of these monotonic tests, here numerical simulations were used. The results of
these simulations are given in III.3.2. Since the displacements ey are similar for all 3 reinforcement
types, the average values ey = 10 mm for all tests with longitudinal load direction and ey = 20 mm for all
tests with transverse load direction were taken for the laboratory cyclic tests. This procedure also
allowed for a good comparison of the results. The tests were run displacement-controlled, whereby the
loading was applied as given in Figure II.3.2-2 in accordance to ECCS [20].
Figure II.3.2-2: Schematic representation of the loading history according to ECCS [20]
The test results in terms of cyclic load-displacement curves and pictures of the specimens at the end of
the loading history are shown in III.3.2. Table II.3.2-1 summarized the number of cycles until failure
detected during the test execution.
Table II.3.2-1: Number of cycles until failure Nf for longitudinal and transverse load direction
load
Specimen
direction
Nf
P1-1 Internal 24
P2-1 External longitudinal 24
P3-1 Nozzle 16
P1-2 Internal 9
P2-2 External transverse 9
P3-2 Nozzle 11
28
II.3.3. Task 3.3: Tests on bolted piping connections under strong cyclic loading
II.3.3.1. Tests on standard bolted piping connections under strong cyclic loading
CSM performed a total of seven full scale tests on standard P355N steel grade circular bolted flanged
connections according to Euro code EN 1092-1 [23], Figure II.3.3-1.Two types of flanges have been
tested: PN63 DN200 and PN40 DN 200 weld neck flanges, type 11. The aim of the testing activity was
to investigate the behavior of P355N steel piping flanged joint both in monotonic than in cyclic bending
loading, with internal pressure. Testing have been performed adopting the 4-point bending
configuration, Figure II.3.3-2, in order to guarantee constant bending moment over the test section, thus
allowing to study the critical parts of the connection under the same load conditions. Internal pressure
Geometry has1:been obtained by means of water, andGeometry
the axial force
2: due to pressure acting on pipe end caps is
developed.
N63/DN200, material P355 A picture of the test device
PN40/DN200, material P355modified by CSM on for the present
that has been extensively
test campaign is reported in Figure II.3.3-3.Two bending displacement amplitudes have been adopted:
3 tests: 3 tests:
two times and four times the displacement causing yielding of the pipe section. Calculated according
1 monotonic theaxial
procedure reported in the ECCS no 45 1 monotonic axial[20], and two levels of internal pressure: 50
loading protocol
ending + constant axial
and 32 barsforce 2 bending
equal to 80% of nominal + constant
connection operativeaxial force
pressure.
Figure II.3.3-1: Weld neck flange geometry
Figure II.3.3-2: Test configuration, four point bending Figure II.3.3-3: Testing device
The test program on standard bolted flanged joints and main results / observations are reported in Table
II.3.1-1. Load vs. stroke actuator diagram, and pictures of the specimens after testing are reported in
III.3.3.1.
29
Table II.3.3-1: CSM test program on standard bolted flanged joints and main results / observations
Leakage Cycles Observations
Test to
Test type Load failure Pipe Flange Gasket Bolts Welding
Plastic No Little plastic No

PN63-1 Monotonic 157 kNm - No deformation
deformation deformation deformation deformation
Constant Wrinkling/ No Little plastic No

PN63-2 Amplitude 147 kNm 26th No deformation
cracking deformation deformation deformation
=+/-54mm
Constant
No Crush due to No
PN63-3 Amplitude 144 kNm 323th Cracking No deformation
deformation compression deformation
=+/-27mm
No Little plastic No
PN40-1 Monotonic 109 kNm - Buckling No deformation
deformation deformation deformation
Constant Buckling/ No Little plastic No
=+/-54mm
Constant Wrinkling/ No Little plastic No
=+/-27mm
Constant Buckling/
PN40-4 No Little plastic No
Amplitude 110 kNm 6th No deformation
(extra) cracking deformation deformation deformation
=+/-54mm
Large plastic deformation of the pipes is observed, wrinkling or local buckling is developed in cases
were bending amplitude =+/-54mm is applied. The welds were not affected by cracking in any of the
analyzed cases. Flanged joints shown high resistance to bending loading ensuring leak tightness far
beyond the pipe elastic limit, and no damage to bolts or flange at final unloading. It is thus concluded
that Standard welding neck joints are suitable to be employed in operating conditions where a seismic
event causes pipe bending present. On the other hand the life of the join is influenced by the pipe that
has to be carefully taken into account.
II.3.3.2. Tests on non-standard bolted piping connections under strong cyclic loading
An experimental test campaign on some non-standard Bolted Flange Joints was conducted by the
University of Trento (UTrento) in order to assess their capacity and performance under seismic loading.
A detail report on the experimental activities can be found in Deliverable D3. Although Bolted Flange
Joints (BFJs) -one of the most critical components of a piping system- are vulnerable to earthquake
loading, they do not have any design Standard that takes into account seismic loading. Current design
Standards, e.g., EN1092-1 [23], EN1591-1 [28], ASME B16.5 [6], mainly ensure joint integrity and
leak tightness under operating conditions but their high thicknesses make them unsuitable for seismic
applications. In this respect, two non-standard BFJs -Design 01 and Design 02- comparatively thinner
than the Standard ones were designed by UTrento to achieve a better performance under seismic
loading. Initial dimensions of the flanges were taken from Eurocode EN 1092-1 [23] to be used with an
8” connecting pipe under regular operating conditions of a petrochemical industry. The thickness of this
standard flange was then reduced based on mode #1 and mode #2 failures according to structural
Eurocode EN 1993-1-8 [29]. Dimensions of other components of the BFJs were selected from relevant
European standards. Materials compatible with design operating conditions were chosen for different
components of the flange joints. A calculation on the designed BFJs according to the new European
standard EN 1591-1 [28] was performed to check their mechanical integrity and leak-tightness
performance. Dimensions of the non-standard BFJs are presented in Figure III.3.3-17 and Table
III.3.3-1 (all figures and tables referred in this Section are presented in III.3.3.2).
A number of bending and axial experimental tests as listed in Table III.3.3-2 were carried out on the
designed non-standard BFJs under monotonic and cyclic loading in order to investigate their capacity
and performance. A moderate internal pressure of 1.5 MPa was used in all tests. The monotonic tests
were mainly performed to find required parameters to build ECCS45 [20] loading protocols to be used
for the cyclic tests. Test specimens were constructed by placing the flange joints in the middle of two
30
connecting pipes as depicted in Figure III.3.3-18 of III.3.3.2. Loading were applied to the test
specimens by means of MOOG actuators with a capacity of 1000 kN. During bending tests, loading
were applied till the failure of the test specimens, whereas, the maximum level of loading during the
axial tests was limited by the capacity of the actuators. The experimental set-ups are illustrated in
Figure III.3.3-19 of III.3.3.2. Experimental results were compared with the allowable, yield and
ultimate loads suggested by codes. A comparison was also made between the capacity of the designed
joints and the seismic demands on such joints coming from a case study performed on a petrochemical
piping system.
Both of the non-standard BFJs exhibited favourable performance under bending and axial loading as
can be noted from Figure III.3.3-20 and Figure III.3.3-21 of III.3.3.2. A good capacity of the joints in
terms of strength, ductility and energy dissipation were found in all experimental tests. Almost no
degradation of the joints in cyclic loading was observed and no failure occurred in any of the joints. In
all tests, leakage was observed with a high level of loading. During bending tests, failure took place in
the pipe wall where local buckling was observed as shown in Figure III.3.3-22. During axial tests,
loading were applied by means of two MOOG actuators each with a capacity of 1000 kN. Hence, a
maximum axial load of 2000 kN could be applied during axial tests. With this level of maximum load,
test specimens remained within its elastic limit while only small levels of flange displacement were
found. To assess low cycle fatigue behaviour of the BFJs in axial loading, once the limit load of the
actuators was reached, we continued several cycles of loading with the maximum displacement
attained. Nevertheless, no failure or yielding occurred in the flanges or in any part of the joints.
The comparison between experimental results and Coded loads displayed a favourable performance of
the non-standard BFJs. Allowable seismic design loads calculated according to Codes were
considerably lower than experimental leakage and yielding loads as can be noted from Table III.3.3-3 –
Table III.3.3-4. The levels of earthquake demands on a piping system found from a case study, see
Table III.3.3-5, were significantly lower than the capacity of the designed BFJs. Thus, the non-standard
BFJs were considered suitable for regular seismic applications. It was recommended that these types of
joints could be used in piping systems operating both under normal conditions and under seismic
events.
II.3.4. Task 3.4: Tests on piping elbows under strong cyclic loading
II.3.4.1. Tests on non-pressurized piping elbows under strong cyclic loading
Eight tests have been conducted on 8-inch SCH40 long-radius induction bends (nominal diameter and
thickness equal to D  219.1 mm and t  8.18 mm respectively, and R D  1.5), connected on two
straight pipe segments with length equal to 5 D and the same nominal diameter and thickness. The
material of the specimens is P355N according to EN 10216 standard [21], equivalent of API 5L X52.
The uniaxial and cyclic steel material properties have been examined in detail through an extensive
experimental program [13].
The experimental campaign took place at Delft University of Technology. The experimental set-up
consists of a steel frame with a hydraulic cylinder positioned horizontally over the lower part of the
frame. The cylinder’s one end is pin-connected on the frame while the other end is used for the
application of end-displacements on the elbow. The elbow specimen is pin-supported on the same frame
through a pair of steel plates attached to the end of the specimen. The hydraulic cylinder is connected to
the other end of the specimen using a similar support, as shown in Figure II.3.4-1. Strain gauges and
displacement transducers are used to monitor the evolution of local strains at critical location of the
specimens and the displacements of the elbow ends respectively (Figure II.3.4-2). The change of the
elbow vertical diameter is monitored at its middle cross-section and at a cross-section near the weld
region using two special-purpose frame devices. Thickness measurements on the specimens have been
obtained prior the execution of every test and differences with respect to the nominal value up to 1.9
mm (23.3 %) were recorded. The specimens are filled with water before testing with low water pressure
equal to 1.2 bar (0.5% of the yield pressure), which is kept constant throughout the test and used as an
indicator of through-thickness crack opening.
31
Figure II.3.4-1: Experimental set-up: Front view Figure II.3.4-2: Experimental set-up: Instrumentation
Test Displacement Number of cycles Local strain

No. range l (mm) until failure N f range* (%)
(Exp.)
1 ±25 13160 0.33
2 ±70 444 1.23
3 ±100 171 -
4 ±150 61 2.61
5 ±200 28 -
6 ±250 17 3.84
7 ±300 10 4.02
8 ECCS protocol 16 -
*: calculated in the elbow hoop direction at the critical
location
Table II.3.4-1: Displacement range and number of cycles Figure II.3.4-3: Cracked elbow specimen
configuration
The specimens have been subjected to intense cyclic in-plane bending through the application of cyclic
end-displacements on the moving support, keeping the other support pin-connected to the supporting
steel frame. The seven first tests followed a constant amplitude end-displacement, whereas a variable-
amplitude loading sequence has been applied to the eighth specimen. The number of cycles until failure
of each specimen is tabulated in Table II.3.4-1. The values of correspond to the development of
through-thickness crack and leakage of containment, as shown in Figure II.3.4-3. It is noted that this
failure pattern is similar for all specimens. In Figure II.3.4-4 a summary of the stabilized load-
displacement curves for the five tests conducted is presented.
The eighth test has been conducted following an increasing end-displacement pattern, shown in Figure
III.3.4-3 (Appendix III.3.4.1), according to the loading protocol for cyclic loading proposed by the
ECCS Recommendations for cyclic testing of structural components [20]. Note that in the case of an
elbow, because of its significantly different behavior under closing and opening bending, the
displacement pattern is not symmetric, shown in Figure II.3.4-5. Using this end-displacement pattern,
the specimen has failed in the form of fatigue cracking after 16 completed loading cycles.
32
Figure II.3.4-4: First load-displacement cycles for Figure II.3.4-5: Test 8 – ECCS load - displacement
the 5 tests cycles
II.3.4.2. Tests on pressurized piping elbows under strong cyclic loading

Five full scale tests on API 5L X52 grade, 8” piping sections containing an elbow connected by welding
to two straight pipes and subjected to cyclic bending moment and internal pressure. The geometry of the
specimen is reported in Figure II.3.4-6 and in Table II.3.4-2, while Table II.3.4-3 reports the test
program performed. One extra test with respect to the original test plane has been performed adopting a
spare specimen available. The experimental activity has been performed adopting the CSM Full scale
test machine for offshore tubular components, located in the CSM laboratory in Sardinia. A picture of a
specimen installed in the experimental device is reported in Figure II.3.4-7.
Table II.3.4-4 summarized the main outcome of the test campaign. All specimens have been tested up to
final failure, occurred by through wall cracking at the elbow mid-section, about 20mm below the flank
line, Figure II.3.4-8. Relevant ovalization occurred in all tests on the elbow cross section, which caused
large plastic strain accumulation. Results of Tests 1 to 4 show that the displacement amplitude has a
marked effect on the number of cycles to failure, while the internal pressure level, raise from 32 to 70
bar has a weaker effect. A further increase of internal pressure to 120 bar with a displacement amplitude
of +/-200mm (test 5) resulted in a reduction of cycles to failure form 27 to 22. For all levels of internal
pressure and displacement amplitude, although large bending displacement has been applied, no local
buckling has been observed either on pipes and elbows before to cracking occurrence. No cracking is
observed on welds to the straight pipe segments or in other locations.
D = 219.1 mm 8.625 inch

WT = 8.18 mm 0.322 inch
R/D = 1.5 – – inch
L = 1100 mm
Grade = API X52
5L
Figure II.3.4-6: Specimen geometry and test configuration Table II.3.4-2: Specimen properties
33
Table II.3.4-3: Full scale test program
Internal
Test ID End displacement (mm) Δδ (mm) 1st cycle
Pressure
1 ±200 400 3.2 MPa
2 ±200 400 7.0 MPa
3 ±300 600 Closing mode 3.2 MPa
4 ±300 600 7.0 MPa
5 (extra test) ±200 400 10-12 MPa
Figure II.3.4-7: Specimen inside the experimental Figure II.3.4-8: Specimen failure for cracking at flank
device position
Crack
Test Applied Internal Cycles at position
ID displacement Pressure failure
below flank
1 ±200 mm 3.2 MPa 26th 17 mm
2 ±200 mm 7.0 MPa 27th 15 mm
3 ±300 mm 3.2 MPa 10th 18 mm
4 ±300 mm 7.0 MPa 10th 24 mm
5 ±200 mm 12 MPa 22th 20 mm
Table II.3.4-4: Summary of Elbows test results performed by Figure II.3.4-9: Typical Force vs. displacement
CSM result, Test 2
II.3.4.3. Pseudo-dynamic loading on piping system

The University of Trento (UTrento) undertook a pseudo-dynamic test campaign on a typical full-scale
piping system containing several critical components in order to assess its seismic performance.
Pseudo-Dynamic testing with Dynamic substructuring (PDDS) is a novel hybrid experimental technique
in which a heterogeneous model of the emulated system is created by combining a Physical
Substructure (PS) with a Numerical Substructure (NS) that describes the remainder of the system. The
overall response of a structure is evaluated by combining the experimental response of the PS -which is
generally the most critical part of the structure- with the numerical response of the NS. By the use of
actuators, computers, finite element software and controllers, this tests can potentially reproduce a
shaking table test. A number of PDDS as reported in Table III.3.4-2 (all figures and tables referred in
34
this section are presented in III.3.4.3) were carried out on the piping system under different levels of
earthquake loading corresponding to serviceability and ultimate limit states suggested by Standards.
The piping system experimentally tested was a typical industrial piping system placed on a support
structure as shown in Figure III.3.4-24(a). It contained 8” and 6” straight pipes, several elbows, a Tee
joint and an EN 1092-1 [23] Standard PN 40 weld-neck bolted flange joint. Material of all straight pipes
and elbow elements was API 5L Gr. X52. The pipes were filled with water with an internal pressure of
3.2 MPa. Preliminary numerical analyses were carried out through a 3D Finite Element (FE) model of
the piping system developed in SAP2000 [50] software. The inputs earthquakes corresponding to
serviceability and ultimate limit states suggested by Standards, see Table III.3.4-3, were found on the
elevated floor of the support structure through time history analyses by imposing an earthquake to its
base. An elastic seismic analysis was carried out on the piping system with and without taking into
account a Rayleigh damping constructed considering a 0.5% damping found through Identification
Tests (IDTs) on the PS. The Rayleigh’s damping considerably reduced the stress level as can be noted
from Table III.3.4-4.
Implementation of the PDDS algorithm on the piping system was challenging mainly because it was a
structure with distributed mass –for which the PDDS has been considered inadequate– and it was
subjected to distributed earthquake forces. Two coupling points were chosen to divide the piping system
into a PS and NS based on the locations of bending moments close to zero in the x-y plane, see Figure
III.3.4-24(b), found from a time history analysis. Since the PDDS were carried out at an elongated time-
scale compared to the real-time of an earthquake, the inertia and damping forces had to be modelled
numerically. Additionally, earthquake forces were to be condensed to the two coupling nodes since it
was not feasible to apply an actuator in each node of the PS to produce the earthquake forces. In order
to condense the PS to the two coupling nodes, the so called Craig-Bampton (CB) [18] mode synthesis
or model reduction technique –which was found to be the most effective among others- was adopted.
The PDDS test algorithm was developed in Matlab and Simulink using the LSRT2 integration
algorithm developed by Bursi et al. [9]. Experiments were carried out using an xPC target, two MOOG
actuators and an MTS FT60 controller [44]. The experimental set-up was realized in the Materials and
Structural testing laboratory of UTrento. The schematic and actual test set-up is presented in Figure
III.3.4-25 and Figure III.3.4-26, respectively, whereas, some components of the PS are shown in Figure
III.3.4-27.
Several PDDS as listed in Table III.3.4-2 were carried out on the piping system under different levels of
earthquake PGAs corresponding to serviceability and ultimate limit states prescribed by an Italian
Performance-Based Earthquake Engineering Standard as listed in Table III.3.4-3. Some IDTs were
carried out to estimate the damping to be used during experimental tests. A favourable seismic
performance of the piping system was observed during the experiments. It was found that, even under
the Collapse Limit State level earthquake, both the piping system and its critical components did not fail
or yield. Small levels of rotations and strains were found in different components; see Figure III.3.4-28
and Figure III.3.4-29. In all tests, a good agreement between the experimental and numerical results was
observed as can be noted from Figure III.3.4-30 in this respect. Note that such behaviour of piping
system is desirable. In fact, many researchers reported similar performance of piping systems under
earthquake loading based on shaking table tests. Therefore, we concluded that present seismic design
rules for piping systems and components are conservative and amendments are necessary to overcome
some degree of conservatism.
II.3.5. Task 3.5: Tests on T-branch pipe connections under strong cyclic loading
This experimental investigation consists of eight (8) tests on T-joint pipe connections. Two types of
branch to main pipe connections are considered: connection with a fitting, denoted as TJF (Figure
II.3.5-1a) and T-joint with a welded reinforcement plate, denoted as TJWR (Figure II.3.5-1b). The pipes
are of P355NH/NL1 steel (Grade X52) with a yield stress of 412 MPa and the steel fitting has a nominal
yield stress of 355 MPa. The specimens have been subjected to monotonic and low cycle fatigue tests
until a through-crack failure occurred. For the fully-reversing cyclic tests three (3) imposed
displacement histories were chosen, two with constant displacement amplitude and one according to
ECCS No45 [20] testing procedure. The test results are presented in Table II.3.5-1.
35
The branch pipe has a nominal cross-section 168.3x7.1 mm and the main pipe 219.1x8.2 mm. The same
setup arrangement was used for both the static monotonic and cyclic tests, as shown in Figure II.3.5-2.
The load was applied at the free-end of the branch through a special ball-joint hinge, as shown in Figure
II.3.5-2b. Both ends of the main pipe of the specimens were bolted on the testing frame. The specimens
were filled with air (0.5 bar pressure) in order to determine the moment that the through-thickness crack
occurred. For all the T-joint tests, strain gages were placed at critical locations to determine the strain
distribution in the region of the joint studied. In addition, wire position transducers were used for the
load-point displacement (LPD) measurements.
Table II.3.5-1: Test results.

type of type of static peak load, displacement, loading cycles
specimen loading Pmax (kN) δ (mm) to failure, Nf
1M monotonic 158.0(*) 200.0 -
TJF 2C cyclic 131.2 60 42
WR
3C cyclic 117.9 45 135
WR
4C cyclic 159.9 ECCS No45 15
WR
1M monotonic 81.9(*) 200.0 -
WR
2C cyclic 78.3 60 6
TJWR WR 3C cyclic 64.9 45 17
WR
4C cyclic 77.9 ECCS No45 11
(*) WR
load corresponding to an imposed LPD of 200 mm
(a) (b)
Figure II.3.5-1: The two types of T-joint branch connection studied: (a) fitting, (b) welded reinforcement ring-
plate.
(a) (b)
Figure II.3.5-2: Test setup for the T-joint tests: (a) overall view, (b) side view.
36
(a) TJF (fitting) tests: one specimen was subjected to static monotonic loading and three (3) to low-
cycle fatigue. For an imposed LPD of 200 mm, the corresponding carrying capacity of the specimen
was 158.0 kN. The through-thickness crack occurred in the region of the fitting of the T-joint
connection (see deformed specimen in Figure II.3.5-3a). Two specimens were subjected to low-cycle
fatigue under displacement-control at an imposed LPD value of δ1=60mm, δ2=45mm and one
specimen to a cyclic loading history according to the ECCS No45 testing procedure, at a frequency of
0.010.02 Hz. The specimens failed after Nf = 42, 135 and 15 cycles, respectively. All specimens failed
when a through-thickness crack occurred (at the same location as in the test under monotonic load).
(b) TJWR (welded reinforcement) tests: one specimen was subjected to static monotonic loading and
three (3) specimens to low-cycle fatigue. For an imposed LPD of 200 mm, the corresponding load
carrying capacity of the specimen was 81.9 kN. The failure occurred in the branch region of the
connection (see deformed specimen in Figure II.3.5-3b). Two specimens were subjected to low-cycle
fatigue under displacement-control at an imposed LPD value of δ1=60mm, δ2=45mm and one
specimen to a cyclic loading history according to ECCS No45 testing procedure, at a frequency of
0.010.02 Hz. The specimen tested at an imposed displacement of 60 mm, failed after Nf= 6 cycles at
the same location as in the monotonic test. The other two specimens failed after Nf=17 and 11 cycles,
respectively, in the welded reinforcement plate region of the connection.
(a) (b)
Figure II.3.5-3: Deformed T-joint specimens at failure: (a) TJF, (b) TJWR.
II.4. Work package 4 – Numerical simulation of industrial components behaviour

under strong cyclic loading
The objectives of WP4 are:
• The definition and calibration of an appropriate cyclic-loading constitutive material model to use
for a realistic simulation of elasto-plastic behaviour of tubular components; the constitutive
material model should be able to simulate material degradation effects, as well as low-cycle
fatigue.
• The development of a modeling finite element procedure that simulates accurately the behaviour of
industrial components; the modeling procedure will be validated through the experimental data of
the previous workpackage (WP3).
• The performance of an extensive parametric study, using the numerical modeling methodology
developed above, to examine the behaviour of structural components, within a wide range of
geometric and material parameters
II.4.1. Task 4.1 Definition of inelastic material modelling under cyclic loading
II.4.1.1. Introduction
The behavior of metal structures that undergo plastic deformations can be effectively described
adopting a suitable plasticity model. Several models are available nowadays in the relevant literature.
Some of these are already implemented in commercial finite element codes, while others require the
37
development of user-subroutines for their implementation and use. In the following paragraphs the
proposed plasticity models for the simulation of cyclic loading of metals are presented. These advanced
models have been selected over other more simplified models (e.g. linear kinematic hardening model)
because of their ability to describe more accurately the Bauschinger effect, which is evident upon load
reversal of the material stress-strain curve in the plastic range and may affect the overall behavior of the
elbow specimens.
In the present study, the following models will be used (a) the Armstrong-Frederick model and (b) the
Tseng-Lee model. In most of the cases, the Armstrong-Frederick model will be employed.
Nevertheless, in a few cases, the Tseng-Lee model will be used, mainly for comparison purposes.
II.4.1.2. The Armstrong-Frederick model and calibration

The Armstrong-Frederick (A-F) model has been selected among other models to be used in the
numerical simulations within the INDUSE project. It is an advanced plasticity model, significantly
more powerful than linear kinematic models, available in many commercial finite element codes (e.g.
ABAQUS, MARK etc.). Within J2-plasticity framework, the Armstrong - Frederick model in its most
general version allows the size of the yield surface to change together with translation. The equation for
the yield surface is written as:
1 k 2 ( q )
F (s  a)  (s  a)  0 (1)
2 3
where s is the deviatoric stress tensor, a is the so-called “backstress” tensor and k ( q ) is a function
that defines the size of the yield surface. In general can be a function of the equivalent plastic
strain  q or it can be assumed to be constant, as in the present case. The model assumes a nonlinear
kinematic hardening rule for the evolution of the back stress written in the form:
a  Cε p   a q (2)
where C and  are model parameters that need to be calibrated.
The calibration procedure is based on material test results obtained by CSM. The coupon specimens for
the material testing were extracted from the longitudinal and hoop direction elbow of specimens
planned to be tested under cyclic loading conditions. The behavior of the elbows is governed mainly by
the material properties in the hoop direction, therefore the corresponding test results have been used for
the calibration of the material model parameters and . A summary of these tests is presented in
Figure II.4.1-1.
Lemaitre and Chaboche have proposed a systematic calibration method based on cyclic stain controlled
material tests, as those proposed in Figure II.4.1-2. According to this method, by using the three
available test stress-strain curves, a mean size of the yield surface k is defined. Subsequently, for each
test curve, one  Δσ 2 -k, Δε 2  point is defined and the points are plotted in a graph. In the same graph a
p
curve is fitted following the equation:
Δσ C  Δε p 
-k= tanh  γ  (3)
2 γ  2 
Parameters and are selected to provide the best fit of the curve to the experimental points (Figure
II.4.1-2). The resulting and by this method are reported as Set 1 in Table II.4.1-1.
38
Figure II.4.1-1: Material test results – elbow hoop Figure II.4.1-2: Lemaitre-Chaboche calibration method
direction
Figure II.4.1-3: Calibration results – Test E1B T-4 Figure II.4.1-4: Calibration results – Test E1B T-9
Figure II.4.1-5: Calibration results – Test E1B T-5 Figure II.4.1-6: Test on elbows – Simulation results
using Set 1 and Set 3
Alternatively, the A-F model can be calibrated to fit each material cyclic stress-strain curve. The
resulting parameter sets by this approach are denoted as Set 2, Set 3 and Set 4 for the tests E1B-T-4,
E1B-T-9 and E1B-T-5 respectively (Figure II.4.1-3 to Figure II.4.1-5). Based on the expected strain
range at the critical locations of the examined elements, the appropriate A-F model parameters (those
taken from tests at similar strain levels) should be selected. Nevertheless, as it has been proven by the
simulation results on elbows presented in Figure II.4.1-6, the adoption of different parameter sets does
not affect the simulation accuracy significantly. Therefore, for simplicity the parameter Set 3 could be
used for all the simulations.
39
Table II.4.1-1: Material model parameter values
E (Mpa) ν
210000 0.3
σy (MPa) C γ
Set 1 317.83 18000 75
Set 2 300 41000 200
Set 3 300 46000 215
Set 4 300 46000 185
II.4.1.3. Tseng-Lee model and calibration

In the classical plasticity models as those presented in the previous paragraphs, only one surface is used,
namely the yield surface (YS), for the determination of the limits of the elastic regime. Based on this
formulation, when plastic loading takes place, the hardening modulus is defined through the consistency
condition. As an alternative to this approach, the “Bounding Surface” concept can be adopted. In the
plasticity models following this concept, the hardening modulus is defined by a given expression, not
through the consistency condition.
The Tseng-Lee model [53] is an effective model to describe loading histories including cyclic plasticity
adopting the so-called “Bounding Surface” concept. In this model, in addition to the yield surface, a
“memory surface” is introduced, which plays the role of a bound and obeys isotropic hardening (Figure
II.4.1-7). The yield (inner) surface (Y.S.) is free to translate and change shape within the memory
surface (M.S.) through a mixed (combined) hardening rule. The memory surface is centered at the
origin and hardens only isotropically every time its stress level is exceeded. Thus, it represents the
highest level of stress developed in the loading history.
During initial plastic loading, the flow rule is based on the yield surface and the hardening modulus
depends on the relative distance of the current stress on the yield surface and an appropriately chosen
stress point on the memory surface. When the yield surface reaches the memory surface during further
loading, it stays attached to the memory surface at the specific stress point. The two surfaces lose
contact when the first reverse loading occurs associated with plastic deformation. The expressions
describing the yield surface F=0 and the memory surface F=0 are respectively:
1 K 2 (εq )
F  s  a  s  a   0 (4)
2 3
1 K 2 ( εq )
F s  s 0 (5)
2 3
where s is the deviatoric part of the stress tensor  , a is the deviatoric part of backstress tensor  , s
is the deviatoric stress on the memory surface and K (εq ) , K (εq ) are functions of equivalent plastic
strain εq , representing the size of the yield and the memory surface respectively.
In this model, similar to all uncoupled models, the hardening modulus H is defined directly through an
appropriate function. Motivated by experimental observations this definition of plastic modulus should
account for smooth transition from the elastic to the inelastic stage. The model accounts for the gradual
decrease of the value of H as hardening proceeds, which is of particular importance for the successful
modeling of strong low-cycle loading conditions associated with reverse plastic loading and the
Bauschinger effect.
The basic concept of this model is that the value of the hardening modulus H depends on the "distance"
 in stress space between the yield surface and the bounding surface. More specifically H is described
as follows:
  
H(  , in )  ˆ 1  h  (6)
  in   
40
where ̂ is the hardening modulus of the memory surface when the two surfaces get in contact. As a
first approximation one may assume a constant value of Ĥ , but can also be a function of equivalent
plastic strain. The parameter h affects the steepness of the strain-strain curve during plastic
deformation. In the initial formulation of the Tseng – Lee model, h is assumed to be constant.
The material model with the TL hardening rule has been implemented in a user-subroutine (UMAT) for
ABAQUS, using an Euler-forward numerical integration scheme. The corresponding material model
parameters have been properly defined based on the material testing data from CSM [13], presented in
another work package of the project. The values of and h have been selected equal to 5,000 MPa,
and 10, to match the experimental results. In addition, an exponential function for controlling the size of
the yield surface has been adopted:

k ( q )  k0  Q 1  e
b q
 (7)
where k0 is the initial yield stress, Q and b are appropriately defined parameters. In the present
analysis, the initial size of the yield surface is equal to 280 MPa, Q is -50 MPa (cyclic softening)
and b is equal to 100. The initial size of the memory surface is equal to 430 MPa, allowing for the
simulation of the smooth transition from the elastic to the plastic regime observed in material testing.
The prediction of the stress-strain curve with the TL model is presented in Figure II.4.1-8.
memory
nn
surface s
n

s
k 23
a
k 23
origin
yield
surface
Figure II.4.1-7: A schematic view of the two-surface Figure II.4.1-8: Material curve and model predictions
model.
II.4.2. Task 4.2 Numerical simulation of the experimental tests on components of

critical importance, tested in WP3
II.4.2.1. Numerical simulation of elbow behaviour under extreme monotonic and cyclic
conditions
A detailed finite element model has been developed to support the experimental results in the general-
purpose finite element code ABAQUS. Eight-node incompatible mode (C3D8I) solid elements are used
for the simulations. The model takes into account geometrical and material nonlinearities through a
large-strain formulation and a von Misses description of the yield surface. The exact configuration of
the two support systems at the specimen ends have also been modeled, to account for their flexibility in
a rigorous manner.
In Figure II.4.2-1 a general view of the numerical model is depicted. The curved part of the elbow and
the two straight parts have been modeled as separate parts of the model, so that an appropriate value of
thickness is assigned at each part, according to thickness measurements. For each test, a detailed
numerical model has been developed, adopting the mean thickness for each part of the specimen
(straight pipe segments and elbow) according to thickness measurements. The weld regions have been
also modeled separately considering the local increase of thickness due to the cap of the weld.
41
For the description of the nonlinear elastic-plastic material behavior two advanced plasticity models, (a)
the Armstrong-Frederick model and (b) the Tseng-Lee model, have been adopted, both based on von
Mises plasticity.
Figure II.4.2-1: General view of the numerical model Figure II.4.2-2: Concentrations of plastic deformations
– closing bending loads
II.4.2.1.1. Numerical results and comparison with test data of non-pressurized piping elbows
Eight models have been developed on 8-inch SCH40 long-radius induction bends, tabulated with the
amplitude end-displacement and the number of cycles until failure in Table II.4.2-1 for each specimen.
Simulation results for Test No4 are compared with the corresponding experimental measurements in
Figure II.4.2-3. The comparison between the experimental results and the numerical predictions is
considered satisfactory for all the tests and is presented in Deliverable4. Using the finite element model
and adopting the most appropriate set of material model parameters for each case, the load-
displacement and flattening-displacement experimental curves can be reproduced in a precise and
effective way. In Figure II.4.2-4 the predictions of the numerical model concerning the evolution of the
cross-section flattening in “Position 1” are compared with the experimental measurements. The
numerical predictions are in good agreement with the measured cross-sectional flattening. The
concentration of plastic deformations reaches its highest value below the central axis of the elbow
(Figure II.4.2-2).
The evolutions of maximum and minimum local strains are of particular importance. It is presented in
Figure II.4.2-5 and Figure II.4.2-6, based on the experimental and numerical results of the Test No.4.
Since the behavior of the elbow is not symmetrical under opening and closing bending loads, the local
mean strain is nonzero causing ratcheting phenomena. Due to these phenomena, despite the fact that the
strain range remains almost constant, there is an increase of the value of the maximum strains and
consequently an increase of the mean strain accelerating fatigue failure.
The numerical predictions for the evolution of local strains using the NLKH and the TL model are
compared with the measured values and are found to be in reasonable agreement with the
experimentally measured values (Table II.4.2-1). It can be observed that the NLKH model
overestimates the ratcheting behavior of the material, while the TL model provides strain predictions
much closer to the measured values. This can be attributed to the fact that, under increasing plastic
deformations, the hardening modulus in the NLKH model constantly decreases resulting to higher strain
predictions for a small increase of the stress level.
42
Table II.4.2-1: Displacement range, number of cycles and local strain range.
Test Displ. range Number of Local strain Local strain Local strain
No. l (mm) cycles until range* (%) range* (%) range* (%)
failure N f (Exp) (NLKH) (TL)
1 ±25 13160 0.33 0.46 0.33

2 ±70 444 1.23 1.07 1.25
3 ±100 171 - 1.59 1.59
4 ±150 61 2.61 2.21 2.55
5 ±200 28 - 2.66 2.77
6 ±250 17 3.84 3.30 3.75
7 ±300 10 4.02 4.29 4.03
ECCS -
8 16 - -
protocol
*: calculated in the elbow hoop direction at the critical location
Figure II.4.2-3: Numerical simulation of Test 4 - Load- Figure II.4.2-4: Numerical simulation of Test 4 -
displacement cycles Flattening- displacement cycles
Figure II.4.2-5: Numerical simulation of Test 4 - Figure II.4.2-6: Numerical simulation of Test 4 -
Evolution of maximum local strains Evolution of minimum local strains
II.4.2.1.2. Numerical results and comparison with test data of non-pressurized piping elbows
Based on the properly calibrated numerical models presented in the previous sections, the tests reported
in Table II.4.2-2 have been simulated. Indicative simulation results for tests 1 and 2 are compared with
the corresponding experimental measurements in Figure II.4.2-7 and Figure II.4.2-9. As in the
43
experimental set of the non-pressurized elbows, the developed numerical models are able to simulate
accurately the tests on pressurized elbows as well.
Table II.4.2-2: Displacement range and number of cycles

Test Displ. range l Pressure load [MPa] Number of cycles
No. (mm) until failure N f
1 ±200 3.2 26
2 ±200 7 27
3 ±200 12 22
4 ±300 3.2 10
5 ±300 7 10
Figure II.4.2-7: Numerical simulation of Test No1- Figure II.4.2-8: Numerical simulation of Test No1 -
Load - displacement Ovalization - displacement curve
Figure II.4.2-9: Numerical simulation of Test No2: Load - Figure II.4.2-10: Concentration of strains at the
displacement elbow critical region
The critical position where cracking took place in the experiments reported in the present study is
verified by the numerical model as well. Bellow the elbow mid-axis at the elbow flank region, there is a
localization of strains mainly in the hoop direction (Figure II.4.2-10). This justifies the fact that all the
cracks formed along the longitudinal direction of the elbow.
44
II.4.2.1.3. Fatigue life and fatigue curves
Non-pressurized specimens Nos.2-7 have been subjected to constant amplitude end-displacement range
and failed under a small number (less than 105) (see Table II.4.2-1). Test No.1 is not considered
because the corresponding number of cycles ( N f  13,160) falls at the transition region between low-
cycle and high-cycle fatigue. In addition, test No.8 refers to variable amplitude loading and, therefore,
is not considered in this analysis.
To examine the low-cycle fatigue behavior of the elbows under consideration, it is necessary to express
the number of cycles N f until failure in terms of the hoop strain amplitude at the critical location
where cracking takes place. The numerical results have been in good agreement with the available
measured strains. For each test, the hoop strain range  at the critical location computed from the
numerical simulation using either the NLKH model or the TL model is shown in Table II.4.2-1.
Using these values for tests Nos. 2-7, the following fatigue equation can be derived that correlates the
number of cycles to failure N f with the local hoop strain range  . According to the values reported
in Table II.4.2-1, it is possible to derive equations result from the experimentally measured values and
the corresponding values from the numerical analysis using the NLKH and the TL model respectively.
The corresponding equations are:
Experimental measured values
3.11
 9.124 
Nf   (4)
 ε 
Numerically predicted values - NLKH model

2.88
 9.057 
Nf   (5)
 ε 
Numerically predicted values - TL model

3.13
 8.693 
Nf   (6)
 ε 
The resulting fatigue curves are presented in Figure II.4.2-11. In this graph, the ASME BPVC standard
[4] which provides design fatigue curves for elbows failing in the low-cycle fatigue regime is also
included.
Figure II.4.2-11: Local hoop strain range versus number of cycles to failure.
45
II.4.2.1.4. Numerical simulation of pseudo-dynamic loading on piping system
As reported in Task 3.4, a pseudo-dynamic test campaign was carried out by UTrento in order to
investigate the seismic performance of a typical full-scale industrial piping system and some of its
critical components. In a Pseudo-Dynamic test with Dynamic Substructuring (PDDS), the overall
response of a structure is evaluated by combining the experimental response of a Physical Substructure
(PS) -which is generally the most critical part of the structure- with the numerical response of a
Numerical Substructure (NS) that contains the remainder of the structure. A typical full-scale industrial
piping system, see Figure III.3.4-24, containing 8 and 6 inch straight pipes, several elbows, a Tee joint
and a bolted flange joint was considered for the experimental tests. A 3D Finite Element (FE) model of
the piping system was developed in SAP2000 [50] software in order to perform numerical analyses.
Two coupling nodes were chosen in the locations of bending moments close to zero found from a time-
history seismic analysis; see Figure III.3.4-24(b) for the two substructures and coupling nodes. It was
shown that by satisfying the compatibility condition only in the horizontal x direction, see Figure
II.4.2-12, we were able to reproduce the response of the original system; details can be found in
Deliverable D3. The input earthquake accelerogram for analyses and experimental tests were generated
on the elevated floor of the support structure of the piping system by imposing an earthquake to its
base. PGAs of different earthquake levels were adjusted according to the serviceability and ultimate
limit states suggested by an Italian Standard. An elastic seismic analysis on the piping system was
carried out in SAP2000 [50] with and without considering damping. The Rayleigh damping model was
constructed for the analysis considering a 0.5% damping for the piping system; it was estimated
through identification tests on the test specimen. It was found that the Rayleigh’s damping considerably
reduced the stress levels in the piping system.
(a) Specifications of the piping system (b) PS, NS and coupling nodes
Figure II.4.2-12: The piping system and the two substructures
Implementation of the PDDS on the piping system was challenging mainly because it was a structure
with distributed mass –for which PDDS has been considered inadequate- and it was subjected to
distributed earthquake forces. The earthquake forces had to be condensed to the two coupling nodes
owing to the infeasibility of applying earthquake forces in each node of the PS through actuators.
Additionally, mass and damping matrices of the PS were also to be reduced to 2 by 2 matrices to be
numerically modelled in the system of equations of motion of the piping system. The so called Craig-
Bampton [18] mode synthesis technique -which was found to be the most effective among others- was
adopted to accomplish this task. The LSRT2 integration algorithm developed by Bursi et al. [9] was
used to perform the PDDS. Tests were carried out by using two MOOG hydraulic actuators, an MTS
FT60 controller [44], an xPC target and a Host PC. The hardware-software scheme for the PDDS is
presented in Figure II.4.2-13.
46
Figure II.4.2-13: Hardware-Software configuration for the PDDS
Several PDDS were carried out on the piping system under different levels of earthquake PGAs
corresponding to serviceability limit states suggested by Standards; refer to Task 3.4 for the
experimental program and input earthquake PGAs. All experimental results showed a good agreement
with relevant simulation results as can be noted from Figure II.4.2-14 in which experimental and
numerical displacements in Coupling Node #1 under earthquake PGAs corresponding to the safe life
limit state (SLVT) and collapse limit state (SLCT) are compared.
(a) (b)
Figure II.4.2-14: Displacement histories and relevant spectra of the Coupling Node #1 at (a) SLVT and (b) SLCT
(Blue: Experimental; Magenta: Numerical)
II.4.2.2. Numerical simulation of tank nozzles

Subject of these simulations was numerical recalculation and verification of 6 tests under strong cyclic
loading, as carried out and described in Task 3.2. The geometries of the FE-models were chosen equal
to the measured dimensions of the test specimens (including tolerance deviations). Figure II.4.2-15
shows an overview of the FE-models, whereby the numerical simulations were conducted taking into
account advantages of symmetry conditions.
47
(P1) (P2) (P3)
Figure II.4.2-15: FE-models overview: (P1) internal-, (P2) external- and (P3) nozzle-reinforcement.
The FE-program ABAQUS (version 6.12.1) was used and the FE-models were created using 8-node
reduced-integration solid elements (type C3D8R). Both geometric and physical non-linearities were
considered and moreover the FE-mesh as well as the FE-mesh refinements around the critical points
were calibrated to get stable and adequate results.
To have accurate input data for the material characteristics, some small scale specimens were taken
from the material of the shell plates of the nozzle specimens (P355NH). These small scale specimens
were tested at CSM under monotonic tensile (4 tests) and cyclic loading (3 tests). For the large scale
simulations under monotonic loading conditions, in order to determine the test procedure for cyclic
loading according to ECCS [20] (see Task 3.2), non-linear isotropic hardening was considered using the
extrapolated true stress-strain curve obtained from the test results of the small scale tests, whereas for
cyclic loading conditions linear isotropic and non-linear kinematic hardening according to Lemaitre
Chaboche [42] were considered. The required parameters were determined analytically and were
verified numerically using numerical simulations of the small scale tests under cyclic loading.
With regard to the test setup the lateral supports as well as the load application were incorporated by
adequate boundary conditions. Also the possibility of the bottom plates to lift up was allowed for by
using a “surface to surface” contact definition between the bottom plates and support. The numerical
simulations were run displacement-controlled with the measured loading history of the tests.
The numerical verification of the tests was aiming at two main aspects:
 Validation of the numerical model with regard to its accordance with the experimental
behaviour and thus its applicability for parametric studies;
 Identification of damage and formulation of equivalent damage criterion transferrable from
damages observed in the tests to the numerical model.
The experimental and numerical load-displacement curves show a very good agreement confirming the
capability of the FE-models to represent the overall performance of the shell nozzle reinforcements. For
the identification of failures (cracks) in the numerical FE-models an approach has been selected, in
which the accumulated plastic strains εp were compared to a limit value εp,limit. The limiting values have
been calibrated by the test results such that for each tested specimen respectively its FE-model the
location and the accumulated plastic strains have been determined from the simulations at the cyclic
deformation stage of the tested specimen corresponding to the first visually detected crack. From this
evaluation the number of cycles until failure Nf and the location (point of failure) Pf can be obtained.
The limiting plastic strain is not universally applicable. By applying this method two different values
have been obtained – εp,limit = 8.1 % for longitudinal and εp,limit = 2.2 % for transverse load direction. For
each load direction however one value matched all three types of shell nozzle reinforcement.
The experimental and numerical results for Nf and Pf are summarized in Table II.4.2-3.
48
Table II.4.2-3: Experimental and numerical results for Nf and Pf
Nf,exp Nf,num Pf,exp1) Pf,num1)
Reinforcement
[-] [-] [-] [-]
P1-1 Internal 24.0 23.1 1,2 1
P2-1 External 24.0 24.4 1,2 1
P3-1 Nozzle 16.0 19.5 3 3
P1-2 Internal 9.0 8.4 1 1
P2-2 External 9.0 8.7 1 1
P3-2 Nozzle 11.0 10.1 3 3
1)
Point of failure: 1 - nozzle; 2 - reinforcing plate; 3 - shell
plate
II.4.2.3. Numerical simulation of flanged joints
II.4.2.3.1 Numerical simulation of standard flanged joints
In this section are reported the results of the FE numerical simulations of the tests on standard EN1092-
1 [23], bolted weld neck flanged joints. The aim of these simulations is to obtain a model that
reproduces with accuracy the behavior of the specimen subjected to monotonic and cyclic bending with
internal pressure.
A FE model has been developed by adopting the commercial software MSC.Marc©. Some details of the
mesh are reported in Figure II.4.2-16. In order to obtain a good balance between the accuracy required
and the computational resources request, it was thus decided, exploiting the symmetry, to model only ¼
of the entire geometry, and to adopt different element types for different regions of the model. The
flange has been modeled with 3D brick elements, the bolt shank by means of beam elements and the
bolt head is modeled as a surface that goes in contact with the flange ring. Solid-shell elements have
been chosen for the mesh of the pipe body This type of element uses the 8 node brick topology, but
adopts a shell element formulation which is particularly adequate to represent the behavior of pipes
subjected to bending in combination with internal pressure and provides higher computational
efficiency if compared to 3D brick elements. The gasket has been modeled in detail by means of 3D
brick elements. In particular the spiral wound sealing element adopts a “gasket” material model
provided in Marc, which allows the use as input of pressure vs. closure relationships which describe the
trough thickness behavior of the gasket.
Figure II.4.2-16: Mesh details
The P355 N material true stress true strain curve obtained by full thickness specimens extracted from
pipe in longitudinal direction have been employed for the simulation of the monotonic tests and is
reported in Figure II.4.2-17a. For the simulation of the cyclic tests, nonlinear kinematic hardening
model have been adopted. The material uniaxial behaviour introduced as input in FEA is reported in
49
Figure II.4.2-17b. This model can better represent the actual material behaviour when cyclic loading is
present as it is capable, even with some limitations, to reproduce the Bauschinger effect, which on the
contrary is totally neglected in isotropic or linear kinematic hardening material descriptions.
700 600
600 400
500
True Stress [MPa]

True Stress [MPa]
200
400
0
300 -0.03 -0.02 -0.02 -0.01 -0.01 0.00 0.01 0.01 0.02 0.02 0.03
200 -200
100 -400
0
-600
0.00 0.04 0.08 0.12 0.16
True strain [-] True strain [-]
(a) (b)
Figure II.4.2-17: Monotonic and cyclic material properties employed in FEA
Two typical numerical vs. experimental results comparison are reported in Figure II.4.2-18 for PN40
monotonic bending test, and Figure II.4.2-19 for the PN63 connection subjected to a bending
displacement of +/-27mm. It is possible to observe that the model is able to capture the bending with
very good accuracy all phases of the monotonic case: yielding, plastic behavior, and local buckling of
the pipe which causes a load drop due to collapse of the section. In case of the cyclic bending loading a
good agreement is obtained although the material model adopted in FEA provides a sharper transition
between elastic to plastic regime with respect to smooth and continuous transition experimentally
observed.
500 250
400 200
Deflection (mm)
Load [kN]
300 150
200 100
Experimental load
FEA Load
100 50
Experimental Deflection
FEA Deflection
0 0
0 50 100 150
Stroke [mm]
Figure II.4.2-18: PN63-1 FE model deformed shape and plastic strains at maximum bending deflection (right) and
specimen at the end of the test (left)
600
FEA
Experimental 400
200
Load [kN]
0
-30 -20 -10 0 10 20 30
-200
-400
-600
Stroke [mm]
Figure II.4.2-19: PN63-1 FE model deformed shape and plastic strains at maximum bending deflection (right)
and specimen at the end of the test (left)
50
II.4.2.3.2 Numerical simulation of non-standard flanged joints
With reference to bolted flanged joints, the aim was to calibrate an accurate numerical model of non-
standard flanged joints on the basis of experimental results. In order to obtain these results a finite-
element software ABAQUS 6.11-2 ([1], [2]) has been used. In fact, this software allowed to build a 3D
model of pipes, flanges and bolts by means of an eight-node hexahedron brick elements called C3D8,
and allowed to understand the behaviour of joints under monotonic loads. UTrento designed two
different non-standard joints for testing. The thicknesses were chosen according to the failure modes 1
and 2 of the Eurocode EN 1993-1-8 [29]. The thicknesses of the designed non-standard flanges were
taken as 18 mm (Design 01) and 27 mm (Design 02), respectively. Apart from the thicknesses, all other
dimensions of the designed flanges were the same as those of a PN 40 (for a DN 200 pipe size) plate
flange given in the Eurocode EN 1092-1 [23].
The actual joint system was modelled with the ABAQUS software, as shown in Figure II.4.2-20. For
bending tests a fixed-roller end constraints scheme was adopted and the load was applied by means of
two plates that surrounded pipes; instead, for axial test, a free-fixed end constraints scheme was used,
while the load was applied on the free end through the upper steel plate.
(a) (b)
Figure II.4.2-20: Geometrical models for the: (a) bending test, and (b) axial test
Meshing was obtained through C3D8R - with reduced integration - and C3D8I - with full integration
and incompatible modes- brick elements. The non-linearity contacts were shaped by the so-called
Interactions that identified a Master surface and a Slave; this wording recognizes that the surfaces were
in contact, requiring that the nodes on the slave surface could not penetrate each other in the master
surface. The sliding and the contact responded to the Coulomb’s friction law. The welds, which
guaranteed the connection between the flange and the tube, were shaped by a Tie contact, as well as the
contact between the stem of the bolt and nuts and between these and the flange. The analysis of the
responses of the Bending Test Design 01 and 02 – see Figure II.4.2-21(a) and Figure II.4.2-21(b) – and
of the Axial Test Design 02 – see Figure II.4.2-21(c) – shows that responses obtained with the FE
model appear to be in agreement with experimental curves. However, for the Bending test, the
behaviour in the elastic range was not completely caught by the model; this phenomenon is mainly due
to the fact that the boundary conditions of the experiments are difficult to be approximated in the FE
models. For the Axial test, instead, the behaviour of the numerical model in the elastic range is very
close to that experimental one. Beyond the elastic limit the two curves diverge, and bricks with reduced
integration elements provide a response less strong than the elements with incompatible models.
The gasket between flanges has not been taken into account in the model. The reason is because this
element does not significantly influence the overall response of the FE models, though determines
leakage. The difference between the experimental curve and the responses of the numerical models is
also linked to the fact that all various imperfections due to working or the alteration of properties of the
materials, like residual stresses, were not considered in the numerical analysis.
51
(a) (b) (c)
Figure II.4.2-21: Comparison between experimental and numerical data for bending tests: (a) Design 01, (b)
Design 02; and axial tests: (c) Design 01
A comparison between the model that contained C3D8R elements and the one with C3D8I elements
was done. The first ones are the preferable choice for the bending model, while the latter are the best
choice for the axial model, according to results shown in Figure II.4.2-21.
(a) (b) (c)

Figure II.4.2-22: Deformed shapes and stress levels in the joint for the: (a) bending test – Design 01 –, (b) bending
test – Design 02 –, and (c) axial test – Design 01 –.
Finally, relationships are used to calculate the Normal Root Mean Square Deviation (NRMSD) and the
Root Mean Square Deviation (RMSD) of the error between experimental results and numerical
simulations. They read:
 F  F 
2
1
 F  Fnum 
2
NRMSD  ex num
; RMSD 
 F 
2 ex
num
N
where Fex is the force applied in the experimental case, Fnum the force provided by the numerical model
and N the number of load steps. Table II.4.2-4 summarises all relevant results which appear to be
satisfactory.
Table II.4.2-4: Error between experimental results and numerical simulations calculated with NRMSD and RMSD
methods
Test Model NRMSD (-) RMSD (kN)

Bending Design 01 0.13 29.91
Bending Design 02 0.12 25.94
Axial Design 01 0.10 132.42
52
II.4.2.4. Numerical simulation of Tee pipe joints
The experimental investigation of the junction performance (WP3) is supported by detailed finite
element models, developed to simulate the experiments. Two finite element models have been
developed in ABAQUS FE code for each specimen type using 8-noded reduced integration solid
elements (C3D8R). The symmetry of the problem was taken into consideration allowing for the
simulation of the half specimen. At the symmetry plane, the appropriate symmetry boundary conditions
were applied. In addition, the extension part attached to the branch end facilitating the connection of the
specimens to the loading device was also modeled accurately in order to take into consideration its
flexibility. Finally, a rigid beam (B31) element is attached at the end of the tubular member by a hinge
interaction to simulate the connection between the hydraulic cylinder to the specimens. The loading is
applied at the free end of the rigid beam which is allowed to be translated only in the vertical direction
to the tee-junction main plane. In the tubular members of the junction three elements along the
thickness where used for the main pipes connected and four elements for the fitting component, while
the mesh density varies along the length of the member as presented in Figure II.4.2-23 for the Tee
branch junction with the fitting and the reinforced branch junction. The nonlinear behavior of the
material is described adopting the advanced cyclic plasticity material constitutive model proposed by
Tseng-Lee [53].
The developed numerical model is used to simulate the tests reported in Table II.4.2-5. The comparison
of the predicted load-displacement curves with the corresponding experimental curves shows that the
adopted model is able to provide numerical results in good agreement with the experimental findings.
Indicative results are presented in Figure II.4.2-24 and Figure II.4.2-26. The observed differences can
be attributed to minor slips at the end supports of the test specimens. These slips are more evident in the
TJF tests due to their increased load resistance. Nevertheless, the numerical models can predict the
maximum and minimum resistance loads fairly well.
In addition, the stress and strain concentration areas according were examined using the developed
numerical model. As it is presented in Figure II.4.2-25 in the TJF specimens there is a concentration of
strains at the curved part of the fitting connecting the main pipe to the branch. Due to the local
curvature, the distribution of the developing strains is spread in a bandwidth of about 2 cm. On the
contrary, in the TJWR specimens due to the existence of the welds, the strain concentration takes place
at the weld area of the branch pipe (Figure II.4.2-27). The above observations are in excellent
accordance with the experimental findings.
Cycles
Loading Pmax Displacement to
Specimen type failure
type (kN) δ (mm)
Nf
1M M* 158.0 200a -
TJF 2C C* 131.2 60 42
3C C 117.9 45 135
4C C 159.9 ECCS No45 15
1M M 81.9 200a -
2C C 78.3 60 6
TJWR 3C C 64.9 45 17
4C C 77.9 ECCS No45 11
*: M=monotonic, C=cyclic,
a
: Maximum stroke displacement reached
Figure II.4.2-23: Numerical models: General Table II.4.2-5: Experiments results
view
All the Tee-junction specimens failed due to a through-thickness crack. The TJWR specimens failed
along the circumference of the branch-to-reinforcement plate weld-toe. The specimens examined under
monotonic loading and under an imposed LPD of 60 mm, failed in the branch side of the connection
53
while the other two specimens in the welded reinforcement plate side of the connection. On the other
hand, the TJF specimens failed in the region of the joint fitting, at a distance of about 60 to 80 mm away
from the branch-to-fitting weld-toe.
The developed numerical models were able to simulate fairly well the behavior of both types of Tee-
junctions. The areas where cracking took place are also indicated in the models. While in the case of
TJWR the critical areas are located at the welds of the Tee-junction, in the case of the TJF type, there is
significant stress and strain concentration at an area located about 80 mm from the weld region of the
fitting component to the secondary tube branch.
Figure II.4.2-24: Finite element analysis results: Figure II.4.2-25: Numerical TJF model
TJF 2C test
Figure II.4.2-26: Finite element analysis results: TJWR Figure II.4.2-27: Numerical TJWR model
2C test
The current numerical analysis focused on the strength and fatigue resistance behavior of two
configurations of Tee-junctions, one using a fitting component and a second one using reinforcement
plates at the connection of its two main parts. Under monotonic loading conditions, the specimen with
the Tee fitting showed about 93% higher load capacity at an imposed displacement of 200 mm. Under
cyclic loading conditions, for the imposed displacement value of 60 mm, the TJF specimen exhibited
higher fatigue resistance of about 7 times that of the TJWR one. For the imposed cyclic displacement of
45 mm, the TJF specimen exhibited a higher fatigue resistance of about 8 times that of the TJWR one.
For the low-cycle fatigue tests according to ECCS No45, the TJF specimen resisted 36% more loading
cycles than the TJWR one.
54
II.4.3. Task 4.3 Parametric numerical study of critical component behaviour
II.4.3.1. Parametric numerical study of elbow behavior under extreme monotonic and cyclic
conditions
The behaviour of 8-inch elbows under in-plane and out-of plane bending is investigated through a
parametric study. The complete list of their geometrical characteristics is reported in Table II.4.3-1. A
detailed finite element model has been developed to support the above experimental results and conduct
the parametric study in the general-purpose finite element code ABAQUS. Both eight-node
incompatible mode (C3D8I) solid elements and four-node reduced integration (S4R) are used for the
simulations. Both types of elements have shown a very good performance in simulating the
experiments. The model takes into account geometrical and material nonlinearities through a large-
strain formulation and a von Mises yield surface. For the description of the nonlinear elastic-plastic
material behavior two advanced plasticity models have been adopted, both based on von Mises
plasticity. These advanced models have been selected among other more simplified models (e.g. linear
kinematic hardening model) because of their ability to describe more accurately the Bauschinger effect,
which is evident upon load reversal of the material stress-strain curve in the plastic range and may
affect the overall behavior of the elbow specimens. In this parametric study, the two pipe ends are
capped, and a bending moment is applied. The aim of this study is to examine the effects of the
variation of the elbow geometrical characteristics on its mechanical behavior.
Table II.4.3-1: Geometrical characteristics of examined cases

Geometrical Diameter D Thickness t Bend radius
R/D
Characteristics (mm) (mm) R (mm)
8-inch SCH80
219.1 12.7 304.8 1.391
Long radius
8-inch SCH80
219.1 12.7 203.2 0.927
Short radius
8-inch SCH40
219.1 8.2 304.8 1.391
Long radius
8-inch SCH40
219.1 8.2 203.2 0.927
Short radius
8-inch SCH20
219.1 6.4 304.8 1.391
Long radius
II.4.3.1.1. In-plane bending
The pipe elbows have been subjected to monotonic in-plane bending first and subsequently to severe
cyclic in-plane bending through the application of cyclic end-displacements on the moving support,
keeping the other support pin-connected to the supporting steel frame. A general view of the numerical
model is presented in Figure II.4.3-1. The specimens followed a constant amplitude end-rotation equal
to ±0.02, ±0.04, ±0.08 and ±0.12 rad.
In the monotonic loading cases, the elbow is subjected to opening or closing moment loading (Figure
II.4.3-2). In this case, the two plasticity models predict similar behavior. Nevertheless, some differences
between two models become evident at large rotation levels, where significant plastic deformations are
involved, due to the different constitutive equations formulation of the two models. The same trend in
the predictions of the two models is also observed in the cyclic loading at different end-rotation levels
presented in Appendix III.4.1.1. Despite the fact that the resulting moment-rotation curves by the two
models are close to each other, the measured local strain ranges differ significantly. Consequently, the
predicted fatigue life for these cases may differ. The reason for the above observations is that under
55
increased plastic deformations, the hardening modulus predicted by the NLKH model tends to zero,
while the TL model provides a more realistic prediction for it.
Figure II.4.3-1: Numerical model for the elbows used Figure II.4.3-2:Numerical parametric analysis
for the parametric analysis results – 8-inch SCH40 long radius - Monotonic
analysis
II.4.3.1.2. Out-of-plane bending
The specimens have been also subjected to out-of-plane bending through the application of out-of-plane
end-rotations on the free section, keeping the other section fully restrained. A general view of the
developed numerical model is presented in Figure II.4.3-3. The monotonic loading response is reported
in Figure II.4.3-5 for the case of 8-inch SCH40 long radius. The elbow is subjected to out-of-plane
moment loading with three levels of pressure (0MPa, 3.2 MPa and 7.0 MPa). It should be noted that
under out-of-plane bending, the elbow is subjected to a combination of bending and torsion. The
presence of torsion is responsible for the direction of the buckle, whereas the additional compressive
stresses at the “flank” of the elbow due to bending result in excessive strain development at this
location. Apparently, the buckle occurs at the “flattened” area of the cross-section, where buckling
resistance is minimal due to the small value of local curvature (Figure II.4.3-4). For pressure equal to
10% of py the buckled shape is similar to the one of the non-pressurized case. On the other hand, in the
case of pressure level equal to 20% of py, in cases of SCH80 and SCH20, the elbow distortion is
minimized, and buckle may appear at the straight part of the pipe. Based on those monotonic loading
results, the specimens were subjected to out-of-plane cyclic following a constant amplitude end-rotation
equal to ±0.05, ±0.1, ±0.25 and ±0.4 rad.
Figure II.4.3-3: General view of the numerical model Figure II.4.3-4: Concentrations of plastic deformations
56
Figure II.4.3-5:Moment versus end-rotation curve for Figure II.4.3-6: Moment versus end-rotation curve
pressure 0, 3.2 and 7.0 MPa for displacement ranges ±0.05, ±0.1, ±0.25 and ±0.4
rad, (pressure 0 MPa).
II.4.3.2. Parametric numerical study of tank nozzles

Within this task the findings from the tests as well as from the simulations of the tests were broadened
by further parametric numerical study. On the basis of the verified FE-models of the 3 investigated shell
nozzle reinforcements the influence of component parameter on the cyclic behaviour of the different
shell nozzle reinforcements was assessed. The reference geometries have been represented by the
nominal dimensions of the specimens. While all dimensions were kept unchanged, only one parameter
was varied. The defined parameters are shown in Figure II.4.3-7 and summarized in Table II.4.3-2.
Figure II.4.3-7: Overview of the geometrical parameters for the parametric study
Table II.4.3-2: Geometrical parameters for the parametric study

Variations of parameters
Parameter
1 2 Ref 3 4
P1 Internal and P2 External [mm]
Shell thickness t s 10 15 20 25 30
Bottom plate thickness t b 5 8 10 15 20
Nozzle thickness t n 8.8 10 11 12.5 14.2
Reinf. plate thickness t r 5 8 10 15 20
Reinf. plate diameter d r 250 275 300 350 400
P3 Nozzle [mm]
as above t s , t b , t n similar to P1 and P2
Reinf. nozzle thickness t nr 17.5 20 22.2 25 28
Reinf. nozzle length l r 50 62.5 75 100 150
In the first step monotonic simulations for the 4 different load directions (see Task 3.2) for each type of
reinforcement and each variation of parameter were conducted, thus the total number of monotonic
simulations resulted in 240 (+ 12 for the reference models). In order to assess the influence of internal
57
hydrostatic pressure, the same monotonic simulations were carried out under consideration of internal
hydrostatic pressure (2 bar corresponding to a tank height of 20 m). By comparing all parametric
numerical results for monotonic loading it could be observed that the differences between the results
obtained with and without consideration of internal hydrostatic pressure are small but also remain
nearly constant for each varied parameter. This observation led to the conclusion that the cyclic
simulations without considering of internal hydrostatic pressure would provide sufficient information
on the dependency of the performance as function of the varied parameter.
In the second step cyclic simulations for longitudinal and transverse load direction and each type of
reinforcement were conducted. Contrary to the monotonic simulations, here only simulations for
parameter variation 1 and 4 (+ reference) according to Table II.4.3-2 were considered, thus the total
number of cyclic simulations resulted in 60 (+ 6 for the reference).
The numbers of cycles until failure Nf and the corresponding amplitudes of displacement emax as well as
the observed points of failure Pf determined as described in Task 4.2.3 are summarized in Table
II.4.3-3 for longitudinal load direction and in Table II.4.3-4 for transverse load direction.
Table II.4.3-3: Parametric numerical results for cyclic loading in longitudinal load direction
P1-1 Internal P2-1 External P3-1 Nozzle
Parameter
Nf e max P f 1) Nf e max P f 1) Nf e max P f 1)
[mm]
[-] [mm] [-] [-] [mm] [-] [-] [mm] [-]
Ref. 20.9 70.0 1 22.6 80.0 1 18.5 60.0 3
t s = 10 24.5 80.0 2 28.3 100.0 2,3 18.4 60.0 3
t s = 30 14.3 50.0 1 15.6 50.0 1 18.7 60.0 3
tb = 5 21.3 70.0 1 23.2 80.0 1 18.1 60.0 3
t b = 20 20.0 70.0 1 20.8 70.0 1 19.6 70.0 3
t n = 8.8 18.1 60.0 1 19.2 70.0 1 18.4 60.0 3
t n = 14.2 22.3 80.0 2 23.8 80.0 2,3 18.6 60.0 3
t r = 5 2) 21.4 70.0 2 23.9 80.0 2 18.6 60.0 3
t r = 20 2) 19.8 70.0 1 21.2 70.0 1 18.5 60.0 3
d r = 250 3) 21.1 70.0 1 22.8 80.0 1 18.4 60.0 3
d r = 400 3) 20.2 70.0 1 22.4 80.0 1 19.1 70.0 3
1)
Point of Failure: 1 - nozzle; 2 - reinforcing plate; 3 - shell plate
2)
P3-1: t nr = 17.5 / 28 m m instead of t r = 5 / 20 mm
3)
P3-1: l r = 50 / 150 mm instead of d r = 250 / 400 mm
Table II.4.3-4: Parametric numerical results for cyclic loading in transverse load direction
P1-2 Internal P2-2 External P3-2 Nozzle
Parmeter Nf e max Pf 1)
Nf e max Pf 1)
Nf e max P f 1)
[-] [mm] [-] [-] [mm] [-] [-] [mm] [-]
Ref. 14.4 50.0 1 15.1 50.0 1 17.6 60.0 3
t s = 10 22.1 80.0 2 22.6 80.0 2,3 19.2 70.0 3
t s = 30 11.2 40.0 1 10.7 40.0 1 17.0 60.0 3
tb = 5 15.6 50.0 1 16.1 60.0 1 17.5 60.0 3
t b = 20 13.5 50.0 1 14.1 50.0 1 18.2 60.0 3
t n = 8.8 13.1 50.0 1 13.7 50.0 1 17.4 60.0 3
t n = 14.2 17.2 60.0 1 19.1 70.0 1 17.9 60.0 3
t r = 5 2) 16.0 60.0 1 16.1 60.0 1 17.7 60.0 3
t r = 20 2) 13.7 50.0 1 14.2 50.0 1 17.6 60.0 3
d r = 250 3) 14.6 50.0 1 15.5 50.0 1 17.4 60.0 3
d r = 400 3) 14.2 50.0 1 14.7 50.0 1 17.8 60.0 3
1)
Point of Failure: 1 - nozzle; 2 - reinforcing plate; 3 - shell plate
2)
P3-1: t nr = 17.5 / 28 m m instead of t r = 5 / 20 mm
3)
P3-1: l r = 50 / 150 mm instead of d r = 250 / 400 mm
58
The observations and conclusions given hereafter summarize the results of the numerical parametric
study but also consider the results of the experimental tests and monotonic simulations. The first
observation regards the experimental tests on the 3 types of reinforcements. Despite differences in detail
regarding the load-deformation characteristics and the localisation of failure (crack formation), all types
showed a clearly ductile behaviour and resulted to remarkably high number of cycles at high
displacement amplitudes.
Fundamental informations on the behaviour characteristics are available already from monotonic load-
displacement curves. They provide information on stiffness and yield load. Under the condition of
displacement-controlled cyclic loading increased stiffness leads to lower cyclic performance and
respectively lower number of cycles until failure; the cycles however are coupled with higher load
amplitudes.
Summarizing the performance of the 3 types of reinforcement observed from the tests and simulations
the following conclusions can be drawn:
 Type 1 and 2 (internal and external reinforcement) provide a comparable performance under
cyclic loading;
 Type 1 and 2 show a favourable performance under longitudinal loading;
 Type 1 and 2 exhibit the best cyclic performance in combination with lower thickness of the
tank shell and of the bottom plate; this fact makes them a preferable solution for smaller tanks;
 The detailing requirements for type 1 and 2 provided by EN 14015 [26] leave some space for
the selection and combination of parameters within defined limits; larger diameter dr and
smaller thickness tr of the reinforcing plate leads in general to higher cyclic performance. The
location of the damage however moves from the nozzle (thick reinforcement) to the reinforcing
plate (thin reinforcement); this can be of disadvantage for the stability of the tank shell and for
the reparability of the damage.
 Type 3 (thickened nozzle) show a favourable performance under transverse loading;
 Type 3 perform in particular better than type 1 and 2 when a thick tank shell and bottom plate is
present; this fact makes it more favourable for large tanks;
 The possible increasing of thickness tnr and length lr of the thickened nozzle of type 3 has no
significant influence on the cyclic performance; for each investigated parameter however the
damage was always observed in the shell plate of the tank;
 Regarding the stability of the tank shell a solution with a larger nozzle thickness tnr and shorter
nozzle length lr in combination with a significantly thinner attached pipe is preferable.
The provisions of EN 14015 [26] for the reinforcement of shell nozzles -which are intended to be used
for static design - can be applied to shell nozzles under seismic actions as well. Additional measures,
e.g. increase of reinforcing thicknesses, does not necessarily increase the performance; quite the
contrary the performance can decrease.
II.4.3.3. Parametric numerical study of flanged joints
II.4.3.3.1 Parametric numerical study of standard flanged joints
A FEA parametric study has been carried out on EN1092-1 [23] type 11 welding neck standard flanged
joints subjected to bending loading. The objective of this study is to investigate the effect of the
parameters such as internal pressure level, axial load and geometry on the behaviour of the bolted
flanged connections. Three geometries, as reported in Table II.4.3-5, have been taken into account,
while the load cases considered are reported in Table II.4.3-6. A schematic of the load application on
FE mode is depicted in Figure II.4.3-8.
59
Table II.4.3-5: Flanged joints considered in parametric study
Pipe size Pipe D/t Steel grade
Joint type Geometry
ODxWT [mm] - -
DN200 - PN40 219.1 x 6.35 35 P355N
EN1092-1 Standard
DN200 - PN63 219.1 x 8.18 27 P355N
weld neck type 11
DN200 - PN100 219.1 x 12.7 17 P355N
P=0.5*PN, N=+0.12*Ny
P=0.5*PN, N=+0.24*Ny
P=0, P=PN,
P=0.5*PN, N=0
N=0 N=0
applied
Internal pressure
P=0.5*PN, N=-0.24*Ny forces
applied
P=0.5*PN, N=-0.12*Ny rotation
Ny = pipe section yield axial load

notes
PN =maximum allowable pressure
Table II.4.3-6: Load cases analyzed for each Figure II.4.3-8: FE model and load scheme considered in
geometry parametric study
The parametric study confirmed that for standard weld neck flanged joints subjected to bending the
resistance of the pipe and flange assembly is governed by the pipe characteristics. On the flange, the
only large strain spots can be reported on the neck, close to the weld, where neck thickness approaches
the pipe thickness. On the pipe some large strain spots can be found on the extrados (tensile strains) and
on the bulged sections on the intrados (compressive strains).
Internal pressure, the axial load and the geometry have marked effect in the joint bending resistance,
which has been identified by the maximum of the bending moment and the corresponding rotation.
Internal pressure has a beneficial effect on the sustainable rotations before instability of the joint, while
its effect on the maximum bending moment is always negligible, up to the maximum operative pressure
allowable by EN1092-2-1 code. On the contrary local tensile strains on pipe extrados and flange neck
tend to increase with internal pressure.
Applying a compression load at pipe end, the maximum applied end moment and the corresponding
rotation decrease. For tension load up to +0.12*Ny are applied rotations and moments increase for all
geometries, while for N=0.24*Ny no further increase is observed for PN40 case and a slight decrease
for N=0.24*Ny.
Tensile strains on pipe are influenced by axial loading conditions. This is particularly true at pipe
extrados where axial tensile load produce large plastic strain spots that can result in wall crack.
II.4.3.3.2 Parametric numerical study of non-standard flanged joints
With reference to bolted flanged joints, the aim of Task 4.3 was to use the numerical FE models of
calibrated bolted flange joints presented in the Deliverable D4.2, in order to understand the behaviour of
joints for a change of flange thickness.
A parametric study of joints proposed was performed, in order to evaluate the different behaviours
under bending and axial loading. The flanges used in Design 01 and 02 have the same geometrical
characteristics of the flanges of Type 1, as indicated in the standard EN 1092-1 [23], except for the
thickness flange. Hence, a comparison among the 36 mm thickness flanges (proposed in the EN 1092-
1:2007) and the 18 and 27 mm thickness flanges (proposed by UTrento) was done. Moreover, in order
60
to understand the behaviour of the joints without necks (instead of the ones suggested by the EN 1092-
1:2007, Flanges Type 11) a 40 mm thickness flanges was studied. The results obtained for different
thickness flanges are reported in Figure II.4.3-9 for bending tests and in Figure II.4.3-10 for axial tests,
respectively.
600
500
400 Standard Flange 36 mm Type 1

300
Load F (kN)
Standard Flange 40 mm Type

200 11
Non-Standard Flange 18 mm
100 Type 1
0 Type 1
0 100 200 300
Displacement Δ (mm)
Figure II.4.3-9: Comparison among different thickness flanges for bending tests
3500
3000
2500 Standard Flange 36 mm Type 1
Load F (kN)
2000
Standard Flange 40 mm Type 11
1500
1000 Non-Standard Flange 27 mm
500 Type 1
0 Type 1
0 2 4 6 8
Displacement Δ (mm)
Figure II.4.3-10: Comparison among different thickness flanges for axial tests
In bending tests, as expected, the response of joints with thicker flanges is stronger than in non-standard
cases. Since mainly pipes are stressed in the elastic range, FE simulations of pipes are easier than the
one of bolted flange joints; all numerical responses are almost identical, where only pipes are stressed.
Furthermore, for the same load F applied, the displacement Δ is greater for the thinner flanges; this
happens because they are subject to greater strains than the ones of standard design, and thus subject to
major deformations. In Design 01 non-standard joint, as shown in Figure II.4.3-11, pipes are more
stressed than flanges, and subject to a buckling phenomenon. Instead, in Type 01 standard joint, see
Figure II.4.3-12, the thickness increases, so flanges are more stressed; these stresses are transferring by
bolts that result the weakest elements. In axial tests, at the same load the opening between the flanges is
reduced in the joints with the greater thickness; this reason explains because there are different slopes of
curves in the elastic range. This behaviour is related to the fact that the thicker flanges are able to
absorb and distribute stresses in a more favourable manner, thus decreasing deformations. The standard
joints are therefore able to withstand more axial loads. In Design 01 non-standard case pipes show a
buckling phenomenon along the contact circumference. Instead, in Type 01 standard case, stresses in
flanges become relevant, while pipes remain in the elastic range.
The Stress Modified Critical Strain (SMCS) fracture criterion was based to assess whether the limits
imposed by the ABAQUS convergence were realistic with respect to the limits imposed by a reliable
failure criterion. The critical plastic strain εpcrit can be expresses as:
61
( )
where α is a function of steel properties that is related to a critical void growth rate, and gives an
indication of the material resistance to ductile crack initiation, σm is the mean normal stress and σeq is
the effective flow stress. The triaxiality is defined as the ratio σm/σeq. In our case, the parameter is equal
to 2.59 for steel pipes and flanges (S355), and equal to 5.00 for bolts (Grade 8.8). The application of the
failure criterion was capable of identifying some elements that were deemed to be critical – as shown in
Figure II.4.3-13 and Figure II.4.3-14, i.e. the elements that were most susceptible than others, and
therefore subject to ductile failure upon reaching a certain load limit. The results obtained from the
application of the SMCS failure criterion are in agreement with experimental data. In particular, Table
II.4.3-7 and Table II.4.3-8 show the most critical components connected to the flange thickness for
bending and axial tests, respectively. A careful reader can notice that a thickness increase entails a
failure shift from pipes to bolts.
Table II.4.3-7: Fracture load for bending tests with SMCS criterion
Fracture load (kN)
Test type Design Pipe Flange Bolt Failure ele.
Bending Design 01 464.41 485.46 501.11 Pipe
Bending Design 02 476.52 510.90 521.21 Pipe
Bending Type 01 - - 476.30 Bolt
Bending Type 11 - - 471.30 Bolt
Table II.4.3-8: Fracture load for bending tests with SMCS criterion
Fracture load (kN)
Test type Design Pipe Flange Bolt Failure ele.
Axial Design 01 2274.00 2367.90 2351.60 Pipe
Axial Design 02 2633.01 - 2779.29 Pipe
Axial Type 01 - 2886.57 1853.43 Bolt
Axial Type 11 3510.68 - 2150.00 Bolt
(a) (b)
Figure II.4.3-11: Stress distribution for the bending test: (a) 18 mm thickness joint, and (b) 36 mm thickness
joint
62
(a) (b)
Figure II.4.3-12: Stress distribution for the axial test: (a) 18 mm thickness joint, and (b) 36 mm thickness joint
(a) (b) (c)

Figure II.4.3-13: Critical elements for the bending test: (a) pipe, (b) flange, and (c) bolt
(a) (b) (c)

Figure II.4.3-14: Critical elements for the axial test: (a) pipe, (b) flange, and (c) bolt
II.4.3.4. Parametric numerical study of Tee pipe joints

Following the calibration of the numerical models (Task4.2) with the experimental test results of the
medium scale tests of critical components (Task3.5) a parametric study has been conducted. For the tee-
junction connections the following parameters are considered of importance: (a) tee-junction
configurations, with a reinforcement plate or with a fitting, Figure II.4.3-15, (b) load direction, in-plane
and out-of-plane bending and axial loading of the branch, (c) internal pressure variation, (d) fitting
radius, (e) geometric parameters, diameter and thickness of branch and main pipe, (f) reinforcement
plate geometry (width and thickness variations).
Finite element models have been developed in ABAQUS version 6.11 for each specimen configuration
for the tee-junctions with a reinforcement plate and fitting. 8-noded reduced integration solid elements
63
(C3D8R) were used. Three elements through the thickness are considered adequate to capture the stress
and strain variation through the thickness. The main pipe ends were fixed and a prescribed displacement
was applied at the branch end. The distance from the branch to the main pipe and was at least 3 times
the pipe diameter so that the effect from the supports was negligible. Both isotropic hardening and LKH
material behaviors were considered for the monotonic analyses.
 f s  c Afs  Apl   pc Ap , l pl  ls , eapl  eas

 p 
 2 
Figure II.4.3-15: Reinforcement pad (EN13480-3 [25]) and tee-fitting geometry (EN10253-2 [22]).
II.4.3.4.1 Effect of loading
The capacity of the tee-junctions in various loading directions and internal pressure levels was studied.
Axial loading, in-plane and out-of-plane bending of the branch was considered. The internal pressure
values selected were 0, 30 and 60 bar. Isotropic material hardening and the full junction configuration
were considered.
The capacity in the axial direction is larger than bending, with out-of-plane bending being the least
strong. The internal pressure has no effect in the response of the junction up to yield but increases the
ultimate capacity. The effect is more pronounced for axial loading rather than bending.
A comparison of the results of the two different tee-junction configurations (see appendix Figure
III.4.1-18 and Figure III.4.1-19) shows that the tee junction with the fitting is considerably stronger than
the tee-junction with the reinforcing plate in all loading direction. In addition, the deformability of the
junction is lower compared to the junction with the reinforcing plate, except the case of in-plane
bending. There is 93% difference in the out-of-plane loading capacity, whereas the difference drops to
55% in the in-plane bending capacity. In the axial direction the difference is more pronounced when the
branch is being pushed towards the main pipe. The fitting is 22% stronger in the axial “pulling” and
48% stronger in the axial “pushing” than the reinforced tee-junction. There is also big stiffness variation
between the two configurations for the various loading directions. The stiffness of the fitting is twice
the stiffness of the reinforced tee-junction in the out-of-plane bending and is also significant for in-
plane bending. Finally, there is no significant change in stiffness between the two configurations when
loading is axial to the branch.
II.4.3.4.2 Effect of fitting radius
Standard EN10253-2 [22] for pipe welded pipe fittings specifies that the crotch radius is less or equal to
¾ of the height of the branch (rc≤0.75(G-D/2)) or 0.5 of the height (rc≤0.5(G-Ds/2)), Figure II.4.3-15.
Because no tolerance is provided but rather guidance, the effect of radius on the fitting capacity was
studied here. Considering the limitation of ¾ of branch height the radius was varied between 35, 40 and
45 mm. Only slight changes in the overall capacity of the tee-junctions are observed in terms of load
resistance with the variation in radius of curvature considered.
64
II.4.3.4.3 Effect of branch to pipe diameter
Three fitting geometries, with varying branch to pipe diameters were compared; an 8”x4”, 8”x6” and
8”x8” SCH40 configuration. The dimensions, as they appear in Figure II.4.3-15, were taken from
EN10253-2 [22]. The radius of curvature for the fitting was 30, 40 and 50 mm respectively for the three
geometries of 8”x4”, 8”x6” and 8”x8” based on the standard maximum allowed curvature. Out-of-plane
and in-plane bending as well as axial loads in the direction of the branch were applied. Internal pressure
of 60 bar was applied in all cases. Linear kinematic hardening material behavior was assumed.
The behavior of the three configurations varies greatly in all loading directions both in capacity as well
as mode of failure. The capacity of the 8x6 fitting is 1.4 times greater than that of the 8x4 fitting, and
the 8x8 fitting has a capacity 84% greater than that of the 8x6 fitting. For in-plane bending the
comparative figures become 1.6 and 100% respectively. The yield and ultimate strength variation is
shown in Figure III.4.1-20.
In the case of axial loading, when the branch is pulled away from the junction fracture is expected to
occur at the branch. The strength of the junction increases as the branch diameter increases. Whereas,
when the branch is pushed towards the fitting the change in capacity is affected by the strength of the
branch, the pipe and the fitting and the mode of failure changes in each configuration. For the 8”x8”
tee-junction failure occurs at the main pipe, for the 8”x6” tee-junction failure occurs at the fitting,
while, for the 8”x4” tee-junction failure occurs at the branch.
II.4.3.4.4 Effect of reinforcing pad geometry
For the 8”x6” SCH40 configuration and assuming 60 bar internal pressure a number of possible pad
geometries have been selected based on the designed criteria of EN13480-3, Figure II.4.3-15. The width
and thickness varied between the minimum and maximum allowable values. Namely, the thickness
varied between 4, 6, and 8 mm for pad width of 40mm and for a thickness of 6 mm the pad width varied
between 20, 30 and 40 mm.
The effect of thickness is more significant than the width of the pad in terms of bending strength as
shown in Figure III.4.1-21. When the branch is pulled axially away from the pipe there is no significant
effect of thickness and width of the pad for most cases studied. Similarly, when pushing the branch
towards the pipe the effect of pad geometry is not clear. However, when the maximum allowed
thickness of the pad is used, being equal to that of the pipe thickness, the axial positive capacity
decreases. This happens because the reinforcement is too stiff and stress concentration appears in the
centerline of the main pipe. On the other hand, in the negative axial loading, the significantly stronger
reinforcement pad increases strength and reduces the instability of the branch which causes buckling in
the branch in all other four pad geometries modeled.
II.5. Work package 5 – Seismic behaviour and special seismic design issues in
liquid storage tanks; development of relevant design guidelines/recommendations
The main objective of this work package (WP5) is the development of a complete set of design
guidelines / recommendations, accompanied by design examples, for the seismic analysis and design of
industrial liquid storage tanks. The guidelines are compatible with the existing Eurocode design
framework, and extend the existing rules of Eurocode 8 (EN 1998-4).
II.5.1. Task 5.1 Special issues in seismic behaviour of liquid storage tanks
The work conducted in Task 5.1 consisted on examining several special issues on the seismic response
of liquid storage tanks. The results of this investigation can be considered as a background document
for the seismic design guidelines of liquid storage tanks developed in Task 5.2. In particular, the
following issues have been addressed in detail:
(a) the number of sloshing modes to be considered in convective action and their combination
with impulsive action;
65
Three tanks under consideration have been subjected to base-ground seismic acceleration from 10
artificial earthquakes, generated from the EN 1998-1 design spectrum. The impulsive and convective
forces have been calculated for each time increment of the accelerogram. The maximum forces have
been compared then to each other to see how important the convective forces are to the total action and
to determine how many sloshing modes are necessary to be considered in an analysis. The two
maximum forces were also combined with two ways: a) with the addition of the maximum convective
and maximum impulsive forces b) the SRSS rule and compared with the maximum of the addition of
the two forces. The analysis aimed at determining which of the two ways of combination is more
accurate.
(b) the effects of shell deformation on the seismic response;
Steel tanks are not rigid, they deform due to their small thickness with respect to their diameter. A finite
element analysis, which accounted for the dynamic interaction between the liquid and the deformable
tank, has been reported by Scharf [51], where it is assumed that the motion of the tank-liquid system is
the sum of three contributions, referred to as: ‘rigid impulsive’, ‘sloshing’ and ‘flexible’. The third
component of liquid motion satisfies the condition that the radial velocity of the fluid along the wall
equals the deformation velocity of the tank wall, as well as the conditions of zero vertical velocity at the
tank bottom and zero pressure at the free surface of the fluid. More simplified models have also been
developed for deformable liquid containers, such as the Veletsos model [54]. Similar to Scharf’s
formulation, the Veletsos’ model further assumes that the convective motion of the liquid remains
unaffected by tank shell deformation. Therefore, the hydrodynamic solution for the rigid container is
still applicable, and the tank shell deformation affects only the impulsive motion. This methodology has
been adopted by the API 650 provisions, whereas the simplified methodology for deformable containers
in section A.3.2.2 of EN 1998-4 is in-line with this approach. The accuracy of this approach has been
examined.
(c) calculations of hoop hydrodynamic stresses;
Thickness design of liquid storage tanks is based on hydrostatic pressure and the corresponding
allowable stress of the tank material, as described in API 650. During an earthquake event, additional
hoop stresses develop due to hydrodynamic loading, which need to be added to the stresses from the
hydrostatic pressure and compared with the allowable stresses. A methodology for calculating hoop
stresses has been proposed in the paper of Wozniak & Mitchell [55] based on the work of Veletsos [54].
Consideration of this methodology for hoop hydrodynamic stresses is included in the new Appendix E
of API 650, 2007 edition [3]. On the other hand, there is no such methodology in EN 1998-4 [17].
Using finite element models of three tanks, the hoop hydrostatic and hydrodynamic stresses have been
computed and compared to the ones predicted by API 650.
(d) uplifting behaviour of unanchored tanks;
In many practical applications, relatively broad aboveground liquid storage tanks are constructed
unanchored, in the sense that their bottom plate is in simple contact with the ground without anchors. In
such a case, under strong seismic loading, the tank may exhibit uplifting of its base plate, when the
magnitude of the overturning moments exceeds a threshold value. Although uplift does not necessarily
result in the collapse of the tank, its consequences may lead to serious damage to any piping
connections that are incapable of accommodating severe vertical displacements. In addition, possible
damage at the uplifted bottom plate may occur due to its separation from the ground and an increase in
the axial stress acting on the tank wall part close to the ground, leading to a precipitation of “elephant’s
foot” buckling. For this purpose, base uplifting mechanics is examined numerically through a two-step
methodology: a) a detailed finite element shell model of the tank for incremental static analysis, capable
of describing the state of stress and deformation at different levels of loading and b) a simplified
modeling of the tank as a spring-mass system for dynamic analysis, enhanced by a nonlinear spring at
its base to account for the effects of uplifting. Three cylindrical liquid storage tanks of different aspect
ratios are modeled and examined both as anchored and unanchored. The results are aimed at comparing
with the relevant seismic design provisions of EN 1998-4 and API 650.
(e) buckling at the top of the tank;
A sloshing motion of the tank contents occurs during earthquake motion. Observations following strong
earthquakes indicated that the actual amplitude of liquid surface elevation may exceed several meters in
66
some cases. For full or near-full tanks, resistance of the roof to the impact of the sloshing wave results
in an upward (internal) pressure distribution on the roof and buckling of the upper courses of the shell
walls called “sloshing buckling” attributed to the alternating sign of hydrodynamic (sloshing) pressure
on the tank wall during the seismic excitation. More specifically, when – in the course of a seismic
event – pressure becomes negative, i.e. directed inwards, it may overtake the internal (outward)
hydrostatic pressure, so that the thin-walled tank top is locally under external pressure leading to shell
buckling. Damage may also occur at the frangible joints between walls and cone roofs, with
accompanying spillage of tank contents over the top of the wall. To quantify this effect, a tank is
considered and the impulsive and convective pressures are computed at   0 and    for 7
artificial earthquakes, corresponding to a maximum ground acceleration of 0.25g. Resistance of the
thin-walled tank shell against this type of buckling can be provided by one or more ring stiffeners at the
top of the tank.
(f) elephant’s foot buckling;
When subjected to earthquake loading, tanks may fail due to excessive compression at the bottom of the
tank shell. This is a typical buckling mode of a thin-walled shell under the simultaneous action of
meridional compression and hoop tension due to the presence of internal pressure; it is referred to as
“elephant’s foot buckling” and is considered as the major source of liquid storage tank failure under
seismic action. Elephant’s foot buckling in a tank has been examined numerically, with emphasis on
post-buckling strength, through the consideration of an axisymmetric finite element model, mainly to
investigate post-buckling behavior.
(g) behavior factor of liquid storage tanks;
The API 650 provisions adopt totally different values of R than the values of q in EN 1998-4. The
smaller values of behavior factor for the convective motion in comparison with the impulsive motion
are attributed to the fact that the convective response may not be capable at dissipating seismic energy.
However, there exist striking differences on the values of and for the impulsive component of the
seismic force, which may not be reasonable. The ductility of the structural system constitutes the main
parameter for the reduction of the seismic force. In particular, considering elephant’s foot buckling as
the dominant mode of buckling, there is limited capability of the tank shell of absorbing and dissipating
seismic energy due to its unstable post-buckling behavior. Communication of the authors with the API
650 drafting committee indicated that the relatively high values of reduction factors adopted by the
American standards is based both on previous experience, as well as on other dissipation mechanisms
such as radiation damping at the tank foundation (soil-tank interaction), the ductility of the anchor bolts,
or the inelastic behavior of the base plate in the case of uplifted unanchored tanks. Nevertheless, the
unstable post-buckling behavior of elephant’s foot buckling is not in favour of a high value of behavior
factor.
(h) nonlinear sloshing effects;
The calculation of hydrodynamic forces is based on linear liquid hydrodynamic theory, assuming small-
amplitude of sloshing waves. The question is whether this assumption is valid or not in the course of a
strong earthquake. This requires a more refined analysis. A first analysis has been conducted by Chen et
al. [14], for rectangular liquid storage tanks with finite elements. In addition, an asymptotic solution of
the nonlinear problem has been reported in the recent study of Katsikogiannis [41]. In that study, a wide
range of tank geometries have been subjected to real earthquake events.
(i) floating roof behaviour.
For the case of a Floating Roof (FR) storage tank, the seismic loads can induce a large displacement of
the FR due to the liquid sloshing. This FR motion possibly leads to severe structural damage and
catastrophic fire accidents as experienced in Japan and Turkey during past earthquakes. The main issues
that should be taken into account in designing floating roofs in seismic-prone areas are: adequate
freeboard, the stress conditions in outer pontoon, seal friction, damping due to seal friction,
modification of wall pressure due to floating roof motion, flexible nature floating roofs and effects of
base isolation on FR motion.
Considering typical liquid storage tanks, provided by our industrial partners, the investigation of the
above issues has been aimed at evaluating EN 1998-4 provisions (in comparison with those of API 650)
67
for the seismic design of liquid storage tanks, towards proposing possible improvements/amendments of
EN 1998-4.
The main conclusions from the investigation of these tanks can be stated as follows:
 For the majority of tanks, consideration of only one sloshing (convective) mode is adequate for
the calculation of the total seismic force. Furthermore, the combination of the convective force
with the impulsive force should be conducted according to the SRSS rule.
 Using a systematic two-step methodology for analyzing the seismic response of uplifting tanks
has shown that the EN 1998-4 provisions provide a reasonable yet conservative approach of the
increase of axial compression due to uplifting.
 The Veletsos methodology for deformable containers can be used for design purposes.
 A simple methodology is presented in order to design tanks against buckling at the top, due to
liquid sloshing, using the concept of “equivalent” external pressure.
 The behavior (reduction) factor for the seismic design of liquid storage tanks is a controversial
issue that has not been examined thoroughly so far. Because of the relatively low post-buckling
strength of the shell tank, the authors would be quite cautious in using a behavior factor greater
than 1.5 for the impulsive component of motion.
 There exists little information on current design practice for floating roofs; to avoid complete
tank destruction, one should avoid seal damage to prevent fire ignition.
 Nonlinear sloshing effects on the seismic behavior have not been thoroughly investigated so far.
Limited numerical investigations, reported elsewhere, indicated that their effects on the total
seismic force may not be important, but they can have significant effects on freeboard
considerations.
II.5.2. Task 5.2 Design recommendations / guidelines for the seismic analysis & design
of liquid storage tanks
The design guidelines for the seismic design of liquid storage tanks have been developed in the course
of Task 5.2, in an attempt to present a document that improves the existing provisions of EN 1998-4.
The structure of those guidelines is: 1 Introduction; 2 Relevant Standards and Guidelines; 3 List of
Symbols; 4 Seismic Actions in Liquid Storage Tanks; 4.1 Spectrum for seismic analysis and design; 4.2
Behavior factor; 4.3 Importance factor; 4.4 Calculation of Seismic Forces and Moments; 4.4.1
Impulsive/ convective decomposition of motion; 4.4.2 Vibration periods and damping coefficients; 4.4.3
Seismic Hydrodynamic pressures; 4.4.4 Horizontal seismic forces; 4.4.5 Ringwall and slab overturning
moment; 4.4.6 Combination of impulsive / convective action; 4.4.7 Vertical Seismic Action; 4.4.8 Hoop
Hydrodynamic Stresses; 4.4.9 Meridional Stresses; 4.4.10 Forces on anchors; 4.4.11 Elevation of
liquid free surface; 4.4.12 Seismic Action on nozzles (pipe attachments); 5 Seismic Resistance of Liquid
Storage Tanks; 5.1 Limit States (General); 5.1.1 Shell Buckling Mode; 5.1.2 Anchorage Failure; 5.1.3
Roof Damage; 5.1.4. Hoop Hydrodynamic Stresses; 5.1.5 Base Plate Connection Failure; 5.1.6 Failure
at Attached Piping Locations; 5.2 Tensile Strength of Tank Shell; 5.3 Tank shell buckling; 5.4 Stability
against sliding; 5.5 Design of Nozzles; 5.6 Resistance of Anchorage; 5.7 Sloshing; 5.8 Foundation; 6
Special seismic design provisions; 6.1 Bunding; 6.2 Floating Roof Requirements; 6.3 Constructional
Details; ANNEX A: Hydrodynamic loading due to seismic action; A.1 Introduction; A.2 Horizontal
seismic action; A.2.1 General; A.2.2 Impulsive pressure / resultants / heights; A.2.3 Convective
pressure / resultants / heights; A.3 Vertical seismic action in deformable tanks; A.4 Sloshing wave
height; A.5 Soil-Tank interaction; ANNEX B: Unanchored tanks; B.1 General – uplifting; B.2 Effects
on meridional compression; B.3 Size of uplift and length; B.4 Base plate and deformation; ANNEX C:
Shell Buckling Resistance; C.1 EN 1998-4 Shell buckling provisions; C.1.1 Elastic Buckling; C.1.2
Elastic-plastic collapse; C.2 API 650 Shell buckling provisions.
The guidelines can be employed by designers of liquid storage tanks in seismic areas, providing
background information to existing standards, and identification of ultimate limit states under seismic
action, as well as amending and improving the existing provisions of EN 1998-4. Design examples of
three liquid storage tanks provided by our partners have been considered and designed against seismic
loads to illustrate the use of those guidelines. These tanks are:
68
(a) Tank I is a moderately-broad tank. This is a 27.4-meter-diameter tank with total height 16.5
meters. The filling height H is 15.7 m (   H R  1.131). The tank thickness varies from 6.4
mm at its top course to 17.7 mm at its bottom course; it is unanchored with bottom plate of 6
mm thick and an 8-mm-thick annular plate.
(b) Tank II is a tall 18-meter-diameter tank with a total height of 20 meters and with filling height
equal to 19 m (   H R  2.1). Tank thickness varies from 6 mm at its top course to 10
mm at its bottom course and it is anchored.
(c) Tank III, is a very broad 68-meter-diameter tank with a total height of 20 meters. The filling
height is equal to 19 m (   H R  0.558). Tank thickness varies from 34 mm at its top
course to 38.5 mm at its bottom course.
(d) All three of the above liquid storage tanks are filled with water (   1000 kg/m3) and the
material of the tank shell, the bottom plate and the roof is structural steel S235 (equivalent to
A36 steel) with yield stress  y  235 MPa. The tanks are depicted in Figure II.5.2-1.
Figure II.5.2-1: The three liquid storage tanks used in the design examples.
II.6. Work package 6 – Seismic behaviour and design of industrial pressure

vessels; development of seismic design guidelines/recommendations
The main objective of this workpackage (WP6) is the development of design guidelines /
recommendations for the seismic analysis and design of industrial pressure vessels (vertical cylinders,
horizontal cylinders and spheres). The guidelines will concern various geometries and support
conditions and will be compatible with the existing Eurocode 8 (EN 1998) framework for earthquake-
resistant design.
II.6.1. Task 6.1 Seismic behaviour of vertical-cylindrical pressure vessels

The seismic behavior of different geometries of vertical vessels on skirt supports has been analyzed
using pushover and incremental dynamic analysis techniques. The failure mechanisms were discussed
and possible failure criteria were proposed. Two regions of the vertical vessel have been identified as
critical, namely the skirt and the 1st shell course. The behavior factor q determined for the vertical
vessels was found to be 1.5 lower than the value of 2 suggested by ASCE7 [5] for such pressure vessel
geometries. The effect of internal pressure was insignificant up to the point of maximum capacity of the
vessel but beneficial for post-buckling behavior.
II.6.2. Task 6.2 Seismic behaviour of horizontal-cylindrical pressure vessels

The seismic behavior of different geometries of horizontal pressure vessels on saddle supports has been
analyzed using pushover analysis techniques. The effect of reinforcing rings, length, thickness and
internal pressure was studied using pushover analysis. The mode of failure is buckling at the saddle
support. The exact location in the saddle varies with load direction. The effect of internal pressure was
more significant when loading was applied in the transverse rather than in the longitudinal direction due
to the change of failure location to the shell area adjacent to the saddle reinforcement. Reinforcing rings
are recommended for large pressure vessels or pressure vessels located in high seismicity areas. Caution
69
shall be taken when deciding the area of the saddle reinforcement as the adjacent shell area is expected
to fail in the case of no internal pressure.
II.6.3. Task 6.3 Seismic behaviour of pressure vessels of spherical shape
Within this Task one example of an existing pressure vessel supported by non-braced columns with
circular hollow sections has been investigated, extending the work carried out in Deliverable 1.3b for
case study 4. Further investigations were carried out using the same vessel supported by braced
columns as well as parametric geometries, taking into account both vessels as references. For all vessels
modal and pushover analyses have been performed with the aim to assess the influence of different
geometrical parameters on the non-linear behaviour as well as to identify critical failure modes (limit
states). In addition a method for calculating the fundamental period using a simplified beam model were
derived and verified on the basis of the numerical investigations. An excellent correlation between the
above mentioned methods has been observed. The deviations lie below 2% on average and 6.2% on
maximum for non-braced systems and also below 2% on average and 8.0% on maximum for braced
systems, whereby the maximum deviations refer to systems where the recommended limit values for the
dimensions are not complied. In order to determine exemplary the behaviour factor q non-linear
dynamic analyses were performed, with the result that for both examples the behaviour factor was
determined and showed to be bigger than 2.0. A behaviour factor of 2.7 was found for the non-braced
system, which should be restricted to this value since dynamic instability phenomena can occurs
suddenly using higher values. For the braced system a behaviour factor of 2.3 was identified, when
limitation of the plastification to the bracings is desired. Otherwise higher values than 3.0 were defined,
if additional plastification of column head is allowed. Moreover the parametric numerical study
(pushover analyses) has been shown that by compliance of the limit values for the dimensions of
spherical pressure vessels the displacement ductility μ = eult/ey (where eult and ey represent the
displacements by reaching the ultimate load and yield load respectively) for different braced and non-
braced systems with 4, 8 or 12 columns is relatively stable. For the non-braced systems slightly higher
values than μ = 2.0 were defined, whereas for the braced systems the displacement ductility showed to
be bigger than μ = 4.0.
However these recommendations apply only for spherical pressure supported by an even number of
braced or non-braced columns with circular hollow sections as well as only on the condition that the
limit values for the dimensions of the vessels given within the guidelines (Deliverable D6.2) are
complied.
II.6.4. Task 6.4 Development of design recommendations / guidelines for the seismic
design & analysis of industrial pressure vessels
The developed guidelines contain design recommendations for industrial pressure vessels in seismic
regions. The considered types of pressure vessels are vertical pressure vessels on skirt supports,
horizontal pressure vessels on saddle supports and spherical pressure vessels supported by an even
number of braced or non-braced columns with circular hollow sections. For each type of vessel rules for
seismic analysis and design are provided. Furthermore recommendations for achieving ductile response
are given by means of global design concepts and local detailing rules. The recommendations comprise
common rules valid for any type of pressure vessels as well as rules depending on the type of the
pressure vessel. They are meant to be considered in addition to existing seismic design rules for steel
structures which - to large extend - remain valid for pressure vessels and in particular for their
supporting structures.
In particular the determination of fundamental periods, application of adequate behaviour factors,
determination of seismic forces taking into account effects of liquid sloshing and recommendations for
the seismic design as well as the definition of limit states (failure modes) and dimension limits for the
pressure vessels and the supporting structures are in the focus of these guidelines. The
recommendations and seismic design rules are also illustrated by means of design examples given
within the background document (Deliverable D6.1). Furthermore some recommendations regarding
the influence of large supporting buildings in terms of modified floor response spectra and floor
acceleration characteristics are given.
70
The documentation of the scientific and engineering background for these guidelines as well as the
extensive representation of the design examples are provided within the background document
(Deliverable D6.1), which composed of the extended work of Task 6.1, 6.2 and 6.3.
II.7. Work package 7 – Seismic behaviour and design of industrial piping systems
and components; development of seismic design guidelines/recommendations
The main objective of WP7 is the development of design guidelines / recommendations for the seismic
analysis and design of industrial piping, including their support systems. Special emphasis will be given
to the design of an adequately dissipative support system, to reduce the seismic action on the piping.
The guidelines will be compatible with the existing Eurocode 8 (EN 1998) framework for earthquake-
resistant design.
II.7.1. Task 7.1 Seismic behaviour of industrial piping and their support systems
II.7.1.1. Analysis of a Representative Piping System
Piping systems represent key structural systems for an industrial facility. Few contributions in the
literature are available, in order to clarify the seismic requirements of piping systems. In addition,
aspects like seismic action and structural modeling have not yet been treated in a satisfactorily manner.
Moreover, current American and European standards do not contain adequate information and relevant
provisions for a proper seismic analysis and design of piping systems. Along these lines, some problems
relevant to seismic analysis criteria of piping systems are addressed in this investigation. In particular,
several aspects are treated: 1) modeling of pipes and pipe-racks; 2) selection of the analysis method; 3)
definition of the seismic action; 4) dynamic analysis of the system; 5) stress analysis of pipes; 6)
definition of the ultimate capacity of pipes and joints between pipes.
Figure II.7.1-1: The piping system considered in the present study.
To this end, in the first part Work package 7 (WP7, Task 7.1) a representative case study of an actual
piping system is analysed, as shown in Figure II.7.1-1. The requirements of American and European
codes are compared and important aspects are highlighted: a) definition of numerical models for
seismic analysis, b) seismic input and analysis methods, c) piping stress analysis and checks d)
evaluation of the ultimate capacity of pipes and joints necessary for a correct design of a structure, as
suggested by modern approaches like Performance-based Engineering. A discussion of the main results
with respect to experimental data for a flange joint, carried out at University of Trento, allowed
highlighting the conservatism of current standards.
In particular, the present investigation has been aimed at clarifying all the design steps of this type of
structural systems, analyzing the representative case study according to both European [25] and
American standards (ASME B31.3). The comparison between EN13480 and ASME B31.3 standards
71
showed that the two codes provide similar indications for the evaluation of the seismic response of
pipelines supported by a pipe-rack. The usual method of designing piping systems is based on the
allowable stress approach, considering that the structure is elastic. Consequently only an operating basis
earthquake condition (OBE) can be taken into account for design. Conversely, the modern approach to
the seismic design of structures is to differentiate serviceability from ultimate limit states. This latter
condition could be represented, for example, by the Safe Shutdown Earthquake (SEE). However,
appropriate definition of the deformation capability of the pipes and fittings (ovalization, etc.) in the
inelastic range would be necessary. The advantages of such an approach would be several and would
lead to design optimization. For example in the case study under consideration, seismic action results in
a bending moment far below the moment capacity of a bolted flange joint located in the piping system.
This conservatism seems to be in contrast with the modern performance based-design approach, for
which a certain level of yielding in the structure may be allowed according to a specific performance.
For instance, it would be possible to accept a certain level of yielding of pipes under the condition that
leakage would not occur, leading to a reduction of the thickness of the flange and, therefore, at a more
efficient and economical design.
II.7.2. Task 7.2 Development of design recommendations / guidelines for the seismic
analysis & design of industrial piping systems
The second part of work package 7 (WP7, Task 7.2) is dedicated to the development of guidelines
for the seismic design of industrial piping systems that synthesize the work conducted in the previous
work packages on the same topic (WP1, WP2, WP3 and WP4). The guidelines consist of a concise
document compatible with the existing Eurocodes design framework, which includes some useful
indications on seismic analysis of piping systems. The developed document follows the basic principles
of earthquake engineering expressed in EN1998:1, extending the existing rules to industrial piping
components and systems. The guidelines are meant to be used by designers of industrial piping
systems and provide:
 background information to the existing standards, including calculation examples, identifying
differences between these standards and evaluation of these differences;
 identification of critical steps and parameters in seismic design of piping systems;
 recommendations for modeling of piping systems for seismic analyses;
 identification and calculation example of the interaction between piping systems and the
supporting substructure;
 identification of bearing resistance (ultimate limit states) for piping components, as obtained
from experimental programs , performed within the framework of the INDUSE project, such as
bends (elbows), tee-joints , flanges and tank nozzles, including recommendations for the use of
these components in piping systems, subject to seismic action.
The guidelines are organized as follows: Scope; Definitions; Reference documents, Symbols and
Abbreviations, Methods and relevant parameters for seismic analyses; Seismic input; Modeling of
piping systems; Dynamic seismic analyses; Limit states (including results obtained from the testing
programs); Recommendations. The use of these guidelines is explained in appendices, presenting
calculation examples on:
A. seismic analyses of different types of piping systems
B. interaction between piping systems and the supporting pipe rack
C. calculation of connections between piping and tank wall and selection of nozzle types
D. Example for load combinations and load factors for ultimate lmit states, using the LRFD method
Flow charts are also developed (see Appendix II.7), to facilitate their use.
The Guidelines deals with the results of a comparative study of the relevant European and American
Codes and Guidelines investigating possible differences in seismic input (elastic response spectra) are
evaluated and conclusions drawn (Appendix II.7). A direct comparison between the equivalent static
analyses versus linear dynamic analyses indicated that the uniform load seismic calculations yielded
more conservative results than the dynamic seismic calculations, provided the natural frequencies are
high enough. However, for long and slender piping systems, a dynamic analysis may be of fundamental
importance especially if the natural frequency is lower than 4-5Hz. In such cases, the uniform-load
72
method might substantially underestimate the stresses and forces. The design of the piping system is
more important than the differences between codes in seismic design spectra. This is the decisive factor
on whether the system is capable of withstanding a particular seismic event, resulting in a trade-off
between satisfying the requirements of static and dynamic analysis.
II.7.2.1. Influence of Supporting Structure

The results of earlier work packages, dealing with dynamic coupling between piping system and the
supporting structure (pipe rack) are evaluated in the design guideline, giving suggestions for the
behavior factor for a moment-resistant pipe-rack frame, as well as a criterion based on the WR ratio on
whether the pipe weight and stiffness should be taken into account. This is illustrated in the calculation
example in Annex B of the document (Appendix II.7).
II.7.2.2. Recommendations for Piping Modeling

When performing dynamic analyses it is very important to apply the correct modeling practice in order
to obtain reliable results, however this is not addressed in the present codes. The Guidelines provide
recommendations for modeling, in order to obtain the required calculation accuracy, as well as point of
attention when using commercially available piping software for dynamic analyses. Also some
guidance is given for the design forces on restraints and the influence of gaps. It is advised to include
those recommendations for modeling in the next revision of EN 1998-3 and ASCE 7-05
II.7.2.3. Limit States of Industrial Piping

Available piping codes and standards, such as ASME B31.3 and EN 13480-3, appear to be rather
conservative for the case of seismic design. The results of literature study have been combined with the
results of extensive numerical analyses of piping components and validated against extensive testing of
piping components and parametric analyses. From this evaluation, proposals for limit states of piping
components, such as bends (elbows), Tee-junctions, flanges, and tank nozzles have been developed and
further recommendations for the use of these components are given.
II.8. Work package 8 – Coordination of the project

The main objectives of the WP8, as described in Tasks 8.1, 8.2, 8.3 and 8.4 were to establish an
efficient management process, to define standard protocols, to carry out both a SWOT evaluation and
the monitoring of activities of the project, towards successful completion.
II.8.1. Task 8.1 Efficient management process

INDUSE project comprises several partners and specific objectives. The co-ordinator (University of
Thessaly) has been responsible to manage the project activities to complete successfully on time. In
order to manage the project work certain necessary steps were taken.
(i) The representatives of each partner meet once in approximately every six month for work-
progress, discussing the technical issues of the project, and planning the next activities. The
meetings provided an opportunity for formal discussion in detail among partners. Seven meetings
were held among partners of the project during the period of concern, i.e. 01 July 2009-31 Dec
2012.
(ii) An FTP server directory (ftp://ftp.mie.uth.gr/) was created by the University of Thessaly. Each
partner could access this directory and it could download and upload necessary information and
documents related to the project. The documents related to the project work, i.e. research papers,
thesis report etc., the agenda and minutes of the meeting, experimental as well as simulation data
and reports were uploaded on the server. This provided an efficient distribution of the urgent and
necessary information helpful to the progress of the project.
(iii) A website of the INDUSE project (www.mie.uth.gr/induse) was created by the University of
Thessaly, to disseminate the results of the project.
73
(iv) The co-ordinator and all the partners were in regular contact via e-mail. The necessary issues
were also discussed by phone or Skype. E-mail was the communications medium that was used at
the large and preliminary scale. If more than two partners were involved in discussion,
teleconferencing was used.
Partners who were conducting similar activities, for example RWTH Aachen and EBETAM in the
course of WP6, discussed together technical and administration issues several times by teleconference.
This helped for the smooth progress of the work. For very critical and urgent issues two partners for
example UThessaly and EBETAM, met together in Volos in order to discuss the production of
specimens. UThessaly and Technip have also met several times in Athens.
II.8.2. Task 8.2 Definition of the standard protocols

A protocol is a document that describes the objective(s), design, methodology, statistical considerations
and organization of a trial. In the course of INDUSE project, the coordinator, together with partners, set
standard protocols for the tests regarding specimen order/manufacturing and test performance. INDUSE
project involved several types and number of laboratory tests to be performed by the different partners.
It is necessary, in this situation to follow a common set of procedures to prepare the tests. This reduces
the discrepancy in the test results as well as increases the efficiency. The following steps have been
taken in this respect:
In particular,
 The pipe material for all specimens (straight pipes and fittings) was decided to be P355, which is
equivalent to X52.
 Elbow and flange specimens have been decided to be 8-inch pipes.
 Nozzles have been decided to correspond to 6-inch pipes
 Tee-junction specimens were 8-inch-to-6-inch pipe branches.
 Elbows (to be used in the experiments of 3 different partners) were provided by the same
manufacturer.
 The flange and Tee used in the pseudo-dynamic loading test, has been the same as the ones used
in the individual tests in CSM and UThessaly respectively.
 The final drawings of the specimens by the manufacturer were provided and verified by the
different partners to ensure compatibility between the users and the manufacturer.
 Several analyses have been performed in order to examine the ultimate states of the specimens
and compare with the testing capabilities of each partner.
 The welding procedure specifications (WPS) were chosen for the entire set of specimens
following the same criteria.
 The test protocols regarding performance of tests were made and the information in detail about
the types of tests, loadings, number of cycles, strain rate and instrumentation were described in
WP3.
II.8.3. Task 8.3 Monitoring of project activities and evaluation of the project
The monitoring of activities was regularly performed through the co-ordination meetings in about every
six month period. Seven meetings were held among partners of the project during the period of concern
(01 July 2009 – 31 Dec 2012).
In the kick-off meeting held in Athens, Greece, on 24-25 September 2009, the objectives of the project
was recalled together with the distribution and content of the work in different WPs related to the
concerned period. The case studies in WP1 were also chosen, some preliminary discussion on the
experiments was made, and planning work in WP2 was presented. In the second meeting held in Rome,
Italy, on 21-22 January 2010, the design and analysis of the case studies were discussed and issues
related to preparation of specimens were discussed. In the third meeting held in Delft, The Netherlands,
on 08-09 July 2010, the specimens geometry were finalized and the set-up has been discussed in terms
of several initial numerical simulations. In the fourth project meeting, held in Trento, Italy, on 20-21
February 2011, the first experimental results were conducted, and the results from WP1 and WP2 were
summarized, for the development of the corresponding deliverables D1 and D2. The fifth meeting was a
74
short one, held in Corfu, Greece, on 25 May, 2011, in the course of the COMPDYN 2011 conference.
In that meeting, a brief overview of the progress (mainly the experimental work) has been made and
several papers were presented by the partners in dedicated conference sessions. In the sixth project
meeting, held in Aachen, Germany, on 23-24 September 2011, the progress of the experimental and
numerical results has been discussed and a first discussion on the design guidelines has been made. In
the seventh project meeting, held in Volos, Greece, on 23-24 April 2012, apart from summarizing
experimental and numerical work, the design guidelines in WP5, WP6 and WP7 were extensively
discussed. Furthermore, the improved simulation models of piping components were presented and
discussed. Finally, in the last (eight) project meeting, held in Thessaloniki, Greece, on 6-7 November
2012, the entire experimental results and the corresponding numerical simulations were discussed and
the guidelines under preparation has been extensively presented by the corresponding partners;
furthermore, the format of the draft final report, the deliverables and the contents of the design
guidelines were presented and discussed.
A SWOT analysis has also been conducted for evaluating the Strengths, Weaknesses, Opportunities,
and Threats involved in the INDUSE project. It involves specifying the objective of the project and
identifying the internal and external factors that are favorable and unfavorable for achieving the
prescribed objectives. The SWOT analysis of the INDUSE project can be outlined as follows:
Strengths
(i) A good partnership between universities, research centres and industrial partners,
(ii) A multi-discipline expertise from structural, mechanical and material engineering,
(iii)Earlier experience of most of the partners on similar (European) projects,
(iv) Very good facilities for experimental testing,
(v) Excellent connections of EBETAM, CSM and TU Delft with pipe manufactures
(vi) Selection of case-studies facilitated by the participation of Technip and Tebodin.
(vii)
Experience on manufacturing of steel piping components by CSM, EBETAM, TU Delft and
RWTH Aachen.
Weaknesses
(i) The coordinator had to manage a large consortium (7 partners and 1 subcontractor), with
different specializations, expertise and background.
(ii) European specifications for industrial facilities (tanks, pressure vessels and piping) have
several limitations, and have not been used in everyday design practice
(iii) Harmonization of experimental activities, in terms of specimen size, geometry, loading
schemes.
(iv) Ordering of material; fittings and pipes from P355 material was not easy to find in pipe
manufacturers, especially for the small quantities required in the present project.
Opportunities
(i) INDUSE is a pioneering project of its kind, offering excellent opportunity for state-if-the-art
scientific work.
(ii) INDUSE is a great opportunity to develop novel and unique design guidelines, very helpful
for current design practice.
Threats
(i) A small delay of several experiments has delayed the numerical work and, by consequence, it
delayed the preparation of the guidelines.
(ii) The difficult adaptation of several rules from API and ASME standards to the EN framework
of the present guidelines.
II.8.4. Task 8.4 Progress, Mid-Term and Final reports

The following reports have been prepared and uploaded to CIRCABC:
 1st annual report – March 2010.
 2nd Mid-Term report – March 2011.
 3rd annual report – March 2012.
 4th Final report – March 2013.
75
The deliverables of the project have also been uploaded to CIRCABC (see list of deliverables below).
In addition, minutes of each meeting have been held, and are available upon request.
II.9. Conclusions
The INDUSE project has been a pioneering project within the RFCS research program. Combining
experimental, numerical and analytical tools, high-level research has been performed in the inter-
disciplinary area of seismic safety of steel industrial facilities. Extensive experimental and numerical
work has been conducted, to investigate the mechanical behavior of critical steel components; their
behaviour under strong cyclic loading has been shown well above the requirements of existing design
provisions. In addition, the INDUSE project resulted in the development of design
guidelines/recommendations (a) expanding EN 1998-4 provisions towards an integrated seismic design
of liquid storage tanks and (b) extending the applicability of Eurocode 8 seismic design concepts for the
cases of industrial pressure vessels and attached equipment and piping systems including their support
systems.
The project has been executed smoothly, without any major delays or adjustments, and successfully.
The results and deliverables of INDUSE are both novel and unique, resulting in
 better understanding of structural behaviour of steel tanks, pressure vessels and piping under
strong cyclic (seismic) loading,
 more reliable definition of ultimate limit states for critical industrial components,
 significant improvement of seismic design state-of-the-art of industrial facilities, through the
developed design guidelines,
and, most importantly,
 offering the necessary scientific documents for Code Drafting Committees for improving
existing design standards within the EN design framework, towards safer design of industrial
power plants and chemical/petrochemical facilities.
II.10. Exploitation and impact of the research results

The main outcome of INDUSE has been the development of design guidelines for the seismic design of
industrial tanks, PVs and piping systems. Those guidelines have been developed in close cooperation
with our end-user partners, Technip and Tebodin (subcontractor of TU Delft), improve the design
provisions for industrial tanks, PVs and piping systems that exist within the EN framework, and are
expected to contribute significantly in more reliable seismic design of such structural systems.
Furthermore, the scientific research in this topic, associated with state-of-the-art experimental and
numerical work in the research laboratories of the partners, has improved significantly the know-how at
an international level. The high-quality of the research within INDUSE is reflected in the numerous
scientific publications of the research group, and the special sessions organized in COMPDYN 2011
and PVP 2013 conferences.
As a result, it is believed that all the research work conducted in INDUSE has been novel and
innovative, and has made a significant impact in both the scientific and the professional community.
76
III. Appendices
III.1. Appendix of WP1
III.1.1. Task 1.1: Collection and evaluation of available relevant experimental data
III.1.1.1. Experiments on piping components and piping systems
Tests indicated that the main failure mechanisms are (a) pipe collapse, (b) pipe low-cycle fatigue and
(c) pipe hoop ratcheting. All tests indicated that there is some significant attenuation of due to inelastic
material response, resulting in inelastic energy dissipation. It should be noted that the piping design
specifications are based on elastically predicted stresses. Therefore, the elastically predicted stresses,
strains, accelerations, and support loads from the piping analysis for the SSE seismic level will be over
predictions of the actual response quantities. The attenuation of elastically predicted response quantities
are demonstrated by these tests.
It is noted that, for the interpretation of tests results, where material properties of tested components
were measured and reported, the actual measured yield stress has been significantly above the code
minimum value in most cases. Therefore, it is reasonable and appropriate to evaluate the load level in
terms of the material yield stress not the code minimum value. In these test programs different methods
are used to specify the intensity of the dynamic input. It appears more appropriate to specify the
intensity of the seismic input in terms of the elastic response.
The tests results indicate a substantial conservatism of design provisions for dynamic response of
piping. As shown by these tests, the dynamic primary bending stress can significantly exceed the 2Sy
limit adjusted for actual material properties without a collapse failure. However, the amount of
conservatism in the limit for all possible piping configurations may not be possible to establish from
those tests and propose a relevant amendment of the specification limit.
It also noted that almost all the tests have been conducted on piping with a frequency of 4 Hz or greater.
The tests have demonstrated that piping frequency has a significant effect on response; the lower the
frequency, the greater the response. Therefore, direct applicability of the test results is limited to piping
with a frequency of 4 Hz or greater. The test data clearly shows the superiority of stainless steel piping
in comparison to carbon steel piping in terms of the cyclic life (fatigue failure mode) for dynamic loads.
However, for large deformation or collapse modes of failure, stainless steel may not be superior to
carbon steel. More discussion of the failure mode is offered in the paragraphs below.
Finally, the safety margins (factors of safety) indicated that the above component and piping system
tests need to be evaluated for establishing safe seismic stress/strain levels for the piping systems of
concern in the present research project, within the design framework of ASME B31.3 and EN 13480-3.
This evaluation should be based on the various failure modes (mainly collapse and fatigue) and account
for the influence of pipe diameter, schedule, D/t, material, material properties, hoop pressure stress, and
fabrication details. Additional testing and relevant nonlinear finite element parametric studies of the
component tests are necessary towards this effort, and will be conducted in the framework of the
present project.
III.1.2. Task 1.3: Basic seismic structural design of typical liquid storage tanks
The present task focuses on the seismic design of liquid storage tanks. The main purposes of this study
are
 the identification of significant differences between the American and the European standards.
 to pin-point inconsistencies and missing links in design
 to identify issues that require further consideration
77
Figure III.1.2-1: Three liquid storage tanks which are used in the design examples
For a detail description of the analysis and design of the structures the reader is referred to deliverable
D1.3.
III.1.3. Task 1.4: Basic seismic structural design of typical cylindrical and spherical
pressure vessels
The present task focuses on the seismic design of pressure vessels. The main purposes of this study are
 outlining the current methodology for pressure vessel design with regard to seismic loading
 summarizing the provisions of European and American standards for the treatment of earthquake
loads
 pin-pointing important design issues (e.g. missing links, inconsistencies) that require further
investigation
(a) (b)
Figure III.1.3-1: Sketch of (a) vertical pressure vessel on skirt support and (b) horizontal pressure vessel on saddle
supports (drawings made by software PV Elite 2010).
For a detail description of the analysis and design of the structures the reader is referred to deliverable
D1.4.
III.1.4. Task 1.5: Basic seismic structural design of typical piping system and its
support structure
The main goal of this study is to illustrate in detail the seismic design of piping systems, typical of
petrochemical refineries. Both the available American and European standards have been applied to an
existing piping system, performing the prescribed verification steps. In particular, the following issues
have been illustrated and discussed:
 Calculation of seismic forces using the available codes/specifications, using both static and
dynamic analyses.
78
 Verification checks at some critical components
 Comparison between European and American code results
 Inconsistencies, issues that require further consideration; to be used in the following WPs.
1 2
z 3 4
x y 5 6 7
B
A
Figure III.1.4-1: The analyzed piping system on a pipe-rack
Figure III.1.4-2: Configuration of 10” ammonia Figure III.1.4-3: Configuration of 20” gas transmission
transmission line. pipeline.

III.2.1. Task 2.1: Seismic forces in vertical cylindrical liquid storage tanks
Three liquid storage tanks are existing tanks, constructed in seismic regions, and have been designed
against hydrostatic pressure following the API 650 methodology “Calculation of Thickness by the 1-
Foot Method” in section 5.6.3, considering a liquid containment of unit gravity (water). The tanks are
depicted in Figure III.2.1-1.
Figure III.2.1-1: Tank I (left), tank II (center) and tank III (right) used for the parametric studies in the present
paper.
79
uplifting
w
(a) L
Elephant’s
foot
formation
(b)
Figure III.2.1-2: Finite element analysis of unanchored liquid storage tank under lateral loading; (a) uplifting of
base plate, (b) development of elephant’s foot at the tank bottom.
Figure III.2.1-3: Elephant’s foot buckling, as obtained by finite element simulation.
III.2.2. Task 2.2: Seismic forces in pressure vessels

In this task, the seismic analysis and design indicated that more detailed information regarding the
estimation of the behavior factor for these structures should be included in the design codes, as well as
more detailed guidelines for the determination of the seismic forces. An attempt to estimate the
behavior factor for these structures is presented using advanced numerical tools, for pushover analysis
and incremental dynamic analysis. The pushover analysis (see Figure III.2.2-1 and Figure III.2.2-2)
indicated that the capacity of the presure vessels is quite higher than the required strength in seismic
loading. On the other hand, using incremental dynamic analysis for the vertical-cylindrical pressure
80
vessel, the value of behavior factor calculated is somewhat below the corresponding value suggested by
American standard ASCE 7.
Figure III.2.2-1: Von Mises stress distribution at maximum capacity (approximately 300mm displacement)
(a) (b)
Figure III.2.2-2: Von Mises stress distribution for the horizontal pressure vessel in the (a) longitudinal x-direction
(approx. 10 mm displacement) and (b) transverse z-direction (approx. 50 mm displacement).

III.3.1. Task 3.1 Basic material testing and characterization
III.3.1.1. Base metal testing of X52 and P355 material extracted from elbows, pipes and plates
Longitudinal and transversal specimens have been extracted from X52 elbows at flank position, where
large plastic strain accumulation is present during bending. Full thickness strip specimens have also
been extracted in longitudinal direction from X52 pipes to be connected to elbows in full scale testing
The results are summarized in Table III.3.1-1 and Table III.3.1-2 and shows that actual material
properties fulfil the requirements of ISO3183 standard [33], for X52 steel elbows and EN 10261-3 for
P355 steel grade pipes [21]. It can be observed that very little difference exists between these two
materials.
81
Table III.3.1-1: Monotonic tensile properties of API X52 specimen extracted from elbow
Pipe type Elbow
Specimen id. E2B-L10 E2B-L11 E2B-T1 E2B-T2
Direction Longitudinal Transversal
Rp0.2% Yield Strength (MPa) 412 425 400 399
Rm Tensile Strength (MPa) 569 577 580 578

A% Elongation (%) 40 39 41 39
Table III.3.1-2: Monotonic tensile properties of API5L X52 specimens extracted from pipes to be used in elbow
piping specimens
Pipe type X52 Pipe
Specimen id. ASTM Strip-1 ASTM Strip-2
Direction Longitudinal
Rp0.2% Yield Strength (MPa) 397 396
Rm Tensile Strength (MPa) 543 537
A% Elongation (%) 36 31
Table III.3.1-3: Summary of monotonic tensile properties of P355N extracted from straight pipes
Pipe type P355N
Direction Longitudinal
Cyclic testing on X52 extracted from elbow, reported in Figure III.3.1-1 and Figure III.3.1-2, shown
that little hardening or softening can be reported for this material, even if some softening seems present
and its level reduces when increasing the strain cyclic range, Figure III.3.1-1. Ratcheting testing,
reported in Figure III.3.1-2, revealed that some difference in the accumulation of the strain can be
reported between longitudinal and transversal direction being the longitudinal direction more prone to
accumulation of strains. Furthermore it can be reported that the evolution of strain accumulation is
linear for the transversal direction while is always increasing in the longitudinal direction.
600 600
400 400
200 200
Stress - MPa
Stress - MPa
0 0
-2.5 -1.5 -0.5 0.5 1.5 2.5 -2.5 -1.5 -0.5 0.5 1.5 2.5
-200 -200
-400 -400
E1A-L-3 strain +/- 2.25% E1B-T-5 strain +/- 2.25%
-600 E1A-L-2 strain +/- 1.5% -600 E1B-T-9 strain +/- 1.5%
Strain - % E1A-L-4 strain +/- 0.75% Strain - % E1B-T-4 strain +/- 0.75%
Figure III.3.1-1: API X52 from elbows cyclic properties at various imposed strain levels, longitudinal direction
(left) transverse direction (right)
82
16 E2A-L-7 - Stress 3 R = -0.8 16 E1B-T-8 - Stress 3 R = -0.8
E2A-L-6 - Stress 2 R = -0.6 14 E1B-T-7 Stress 2 R = -0.6
14
E1B-T-6 - Stress 1 R = -0.4
E2A-L-9 - Stress 1 R = -0.4
12 12
10 10
Strain - %
Strain - %
8 8
6 6
4 4
2 2
0 0
0 100 200 300 400 0 100 200 300 400
Cycle number Cycle number
Figure III.3.1-2: API X52 from elbows cyclic properties at various imposed stress levels, longitudinal direction
(left) transverse direction (right)
Finally, results of small scale testing on P355 specimen extracted from plates to used in the nozzles full
scale testing, are reported in the following Table III.3.1-4, for monotonic testing and in Figure III.3.1-3,
for cyclic testing at various levels of imposed strain. Again little hardening or softening is observed in
strain controlled cyclic tests on plates, thus confirming that P355 and X52 material properties show The
measured cyclic properties have been used by RWTH to calibrate the nonlinear kinematic hardening
model employed in FE simulations of nozzle components.
Table III.3.1-4: Summary of monotonic tensile properties of P355 extracted from plates
Material P355
Direction-1
Direction-2
600
400
200
Stress - [MPa]
0
-0.03 -0.02 -0.01 0 0.01 0.02 0.03
-200
-400
P355 plate Δε= +/-2.25%
P355 plate Δε= +/-1.5%
-600
P355 plate Δε=+/-0.75%
strain [-]
Figure III.3.1-3: P355 from elbows cyclic properties at various imposed stress levels
83
III.3.1.2. Base metal testing of material to be used for the tee-joint specimens
Monotonic tensile tests on coupon specimens, extracted from the 8” and 6” pipes and from the tee-joint
8”x6” fitting to be used in WP3 Task 3.5, were carried out to define the true stress-strain curve of the
material. The specimens were taken along the length of the pipes and along the 8” straight part of the
fitting. The tensile tests and geometry selection are in accordance with EN ISO 6892.01 [34]. The
material standard specifies a minimum yield point of 355 MPa, ultimum strength of 490 MPa and
elongation 22% for the P355NH/NL1 material. The data, presented in the table below, are in
agreements with both the EN standard and the material certificates.
Table III.3.1-5: Summary of monotonic tensile properties of the base material.

8” pipe 6” pipe 8”x6” fitting
Yield strength (MPa) Re 410 414 418 409 410 412 378 369
Tensile strength (MPa) Rm 541 542 549 538 540 543 571 548
Elongation (%) A 23.3 23.4 26.3 26 26.1 26.4 30.4 27.1
III.3.1.3. Testing of welded specimens from P355NH/NL1 material

Butt welded specimens were manufactured from 8” pipe supplied for the needs of Tee-joint medium
scale tests in Task 3.5. The WPS No 02 (given in D2.3) was used. The test program defined for welding
certification (EN15614.01 [27]) was performed in order to check the validity of the WPS for butt welds
that will be used in the medium scale tests.
Firstly, non-destructive tests were carried. No defects were detected by magnetic particle and
radiographic inspection. Destructive tests were also performed. Namely, macroscopic examination,
hardness measurement, bend, transverse tensile and impact tests were carried out. Overall, the test
results for the certification of the butt-welded specimens using the TIG method were satisfactory.
III.3.1.4. Cyclic testing of welded specimens from P355NH/NL1 material

Uni-axial cyclic tests of welded specimens have been carried out at varying stress amplitudes. First a
monotonic test was carried out to confirm the yield point of the joint which was measured to be 400
MPa. Then 6 coupon specimens were tested at different levels of amplitude and a mean stress value of
σm=Re/2=200 MPa. The stress range was gradually increased at specified multiples of the yield stress
Δσ=a% x Re. The stress amplitude versus the number of cycles until failure is plotted in Figure
III.3.1-4. Two more specimens were tested at different levels of mean stress and amplitude and the
results are also plotted in Figure III.3.1-4. The first was tested from σmin = 0 MPa to σmax = 464 MPa (σm
= 232 MPa, Δσ = 464 MPa) and the second from σmin = -410 MPa to σmax = 410 MPa (σm = 0, Δσ = 810
MPa).
800
Stress amplitude Δσ (MPa)
600
400
σm=200MPa
200 σm=0
σm=232MPa
Λογαριθμική (σm=200MPa)
0
0 10.000 20.000 30.000 40.000 50.000 60.000 70.000 80.000
Number of cycles
Figure III.3.1-4: Stress amplitude versus number of cycles until failure.
84
In all specimens apart from the last one, with the largest compressive stress applied, cracks initiated at
the HAZ near the weld root. In some specimens cracks also started from the weld cap. When cracks
initiated from the weld root the always propagated through the weld metal. In two specimens cracks
developed in the HAZ of both the weld root and cap. In those the crack that started from the cap
propagated quicker through the base metal of the thinner part rather than the ones that initiated from the
root and propagated though the weld. For a mean value of stress equal to 0 and stress amplitude 820
MPa the mode of failure was quite different. The crack initiated in the HAZ area near the weld cap and
quickly propagated through the base metal of the thinner part of the joint. Failure occurred at only 880
cycles.
III.3.2. Task 3.2: Tests on tank openings/nozzles under strong cyclic loading
III.3.2.1. Numerical results for monotonic loading
For all 3 types of shell nozzle reinforcements 4 different loading conditions were investigated: (1)
tension, (2) compression, (3) shear upward and (4) shear downward acting onto the nozzle (see Figure
III.3.2-1), where hereafter tension and compression are also referred to longitudinal load direction (-1)
and the shear loadings are referred to transverse load direction (-2). They were used the determination
of the cyclic test procedure.
Figure III.3.2-1: Load directions for monotonic loading
Table III.3.2-1 summarizes the yield loads Fy and the corresponding displacement ey obtained from the
monotonic load-displacement curves regarding the ECCS-recommendation [20].
Table III.3.2-1: Yield loads and corresponding displacements of the internal-, external- and thickened nozzle
reinforcement for different load directions
1 Tension 2 Compression 3 Shear up 4 Shear down
Specimen F y e y F y e y F y e y F y e y
[kN] [mm] [kN] [mm] [kN] [mm] [kN] [mm]
P1 Internal 677.2 9.9 481.1 8.1 178.5 24.5 178.8 24.2
P2 External 583.1 9.9 427.8 8.6 148.2 23.1 146.4 22.5
P3 Nozzle 455.0 8.1 494.4 10.6 101.1 18.2 101.1 18.1
III.3.2.2. Test results for cyclic loading

The test results in terms of cyclic load-displacement curves referring to the point of load application and
pictures of the specimens at the end of the loading history are shown in Figure III.3.2-2 and Figure
III.3.2-3. Due to slippage in the bolts of the load application the test specimens were loaded with
slightly smaller displacements than indicated by the true movement of the jack.
85
P1-1 Internal-longitudinal
cycle 1-4 1200
cycle 5-7
cycle 8-10 1000
cycle 11-20
cycle 21-23 800
cycle 24-26
cycle 27-38 600
400
Load [kN]
200
-200
-400
-600
-100 -80 -60 -40 -20 0 20 40 60 80 100
Load-point displacement [mm]
P2-1 External-longitudinal
cycle 1-4 1200
cycle 5-7
cycle 8-10 1000
cycle 11-20
cycle 21-23 800
cycle 24-26
cycle 27-38 600
400
Load [kN]
200
-200
-400
-600
-100 -80 -60 -40 -20 0 20 40 60 80 100
P3-1 Nozzle-longitudinal
cycle 1-4 1200
cycle 5-7
cycle 8-10 1000
cycle 11-20
cycle 21-23 800
cycle 24-26
cycle 27-30 600
400
Load [kN]
200
-200
-400
-600
-100 -80 -60 -40 -20 0 20 40 60 80 100
Figure III.3.2-2: P1-1, P2-1 and P3-1– Cyclic load-displacement curve and pictures of the specimens at the end of
the loading history (longitudinal load direction)
86
P1-2 Internal-transverse 300
cycle 1-4
cycle 5-7
cycle 8 200
cycle 9-11
cycle 12-13
100
Load [kN]
0
-100
-200
-300
-100 -80 -60 -40 -20 0 20 40 60 80 100
P2-2 External-transverse 300
cycle 1-4
cycle 5-7
cycle 8 200
cycle 9-12
cycle 13-15
100
Load [kN]
-100
-200
-300
-100 -80 -60 -40 -20 0 20 40 60 80 100
P3-2 Nozzle-transverse 300
cycle 1-4
cycle 5-7
cycle 8-10 200
cycle 11-17
cycle 18-27
100
Load [kN]
-100
-200
-300
-100 -80 -60 -40 -20 0 20 40 60 80 100
Figure III.3.2-3: P1-2, P2-2 and P3-2– Cyclic load-displacement curve and pictures of the specimens at the end of
the loading history (transverse load direction)
87
III.3.3. Task 3.3: Tests on bolted piping connections under strong cyclic loading
III.3.3.1. Tests on standard bolted piping connections under strong cyclic loading
III.3.3.1.1 Test instrumentation
Full scale testing instrumentation has been employed to measure, monitor and control the main test
parameters: forces, displacements, local strains and pressure. Measured data have been employed in
order to numerically simulate (by means of FEA) the experimental tests and to provide a tuning of the
numerical tools. A plan of test instrumentation positioning is reported in Figure III.3.3-1 and Figure
III.3.3-2. It included:
 Load cell to control and measure the applied bending load;
 Strain gauges to measure strains on the pipe at critical locations, Figure III.3.3-1;
 LVDT: to control and measure applied displacements and rotations, Figure III.3.3-2;
 Pressure transducers to monitor and control the internal pressure level.
A dedicated Digital acquisition system has been developed on purpose by CSM.
Figure III.3.3-1: Strain gauges position mapping
Figure III.3.3-2: Measured deflection at mid sections and rotation at load application points
III.3.3.1.2 Test results
Test results in terms of actuator’s load vs. stroke diagrams pictures of the specimen after testing are
reported in the following.
88
Figure III.3.3-3: Actuator load vs. displacement Test
Figure III.3.3-4: PN63-1 specimen after testing
PN63-1
450 Start of leakeage 371kN 200

400 180
350 160
300 140
Deflection [mm]
250
120
buckling
Load [kN]
100
200
80
150 Load
60
100 40
Deflection
50 20
0 0
0 50 100 150
Stroke [mm]

PN40-1
800
600
400
Load [kN]
200
0
-60 -40 -20 -200 0 20 40 60
-400
-600
-800
Stroke [mm]

PN63-2
89
600
400
200
Load [kN]
0
-30 -20 -10 0 10 20 30
-200
-400
-600
Stroke [mm]

PN63-3
500
400
300
200
Load [kN]
100
0
-60 -40 -20 -100 0 20 40 60
-200
-300
-400
-500
Stroke [mm]

PN40-2
400
200
Load [kN]
0
-30 -20 -10 0 10 20 30
-200
-400
Stroke [mm]

PN40-3
500
400
300
200
Load [kN]
100
0
-60 -40 -20 -100 0 20 40 60
-200
-300
-400
-500
Stroke [mm]

Figure III.3.3-16: PN40-4specimen after testing
PN40-4
90
III.3.3.2. Tests on non-standard bolted piping connections under strong cyclic loading
Figure III.3.3-17: (a) Designed non-standard BFJs and dimensions; (b) spiral wound gasket; (c) stud bolt
Table III.3.3-1: Dimensions of the designed non-standard BFJs (mm)

No Stud
Pipe
O W G K T J A B H P Q M N of Bolt
Size
Bolts Size
18 (Design 202.74
221.5
219.1
DN 200 01) M 27 x
8.18
375
320
290
216
228
248
290
12
30
SCH 40 27 (Design 3.00

02)
Table III.3.3-2: Test program on non-standard BFJ carried out by UTrento

No. Test type Test name Loading type Internal pressure Specimen
1 Bending BSML18 Monotonic 1.5 MPa BS1
2 Bending BSML27 Monotonic 1.5 MPa BS2
3 Bending BSCL18 Cyclic 1.5 MPa BS3
4 Bending BSCL27 Cyclic 1.5 MPa BS4
5 Axial ASML18 Monotonic 1.5 MPa AS1
6 Axial ASCL18 Cyclic 1.5 MPa AS2
7 Axial ASCL27-1 Cyclic 1.5 MPa AS3
8 Axial ASCL27-2 Cyclic 1.5 MPa AS4
Figure III.3.3-18: (a) Bending test specimen, BS1; (b) axial test specimen, AS1; (c) a spiral wound gasket; (d) a
stud bolt.
91
(a) (b)
Figure III.3.3-19: Test set-up for (a) bending tests; (b) axial tests.
450 200
Push
400 150
BSML18
350
100
300
50
Moment, kNm
Load, kN
250
0
200
-50
150
BSCL18
-100
100
50 -150
Pull
0 -200
0 50 100 150 200 250 -80 -60 -40 -20 0 20 40 60 80 100
Displacement, mm Rotation, millirad
(a) (b)
250 200
BSML27
Push
150
200
100
50
Moment, kNm
Moment, kNm
150
100
-50 Pull BSCL27
-100
50
-150
0 -200
0 10 20 30 40 50 60 70 80 -40 -20 0 20 40 60
Rotation, millirad Rotation, millirad
(c) (d)
Figure III.3.3-20: (a) Actuator load-displacement curve from test- BSML18; moment-rotation curves from (b)
test- BSCL18, (c) test- BSML27 and (d) test- BSCL27.
92
2000 2000
1800
ASML18
1500
1600
1400
1000
1200
Load, kN
Load, kN
1000 500
ASCL18
800
0
600
400
-500
200
0 -1000
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6
Displacement, mm Displacement, mm
(a) (b)
2000 2000
1500
1500
ASCL27-2
1000
ASCL27-1
1000
Load, kN
Load, kN
500
500
0
0
-500
-1000 -500
0 0.5 1 1.5 2 0 0.5 1 1.5 2
Displacement, mm Displacement, mm
(c) (d)
Figure III.3.3-21: Load-flange displacement curve from (a) test ASML18, (b) test ASCL18, (c) test ASCL27-1
and (d) test ASCL18-2.
Figure III.3.3-22: Failure of the pipe and bending of a bolt during test BSCL27
93
Table III.3.3-3: Comparison between experimental leakage moment and allowable moments suggested by codes
Experimental moment Allowable moments by codes
Minimum leakage moment obtained
EN 13480, 2002 ASME B31.1 & B31.3
from bending tests
80.24 kNm 51.23 kNm 57.08 kNm
Table III.3.3-4: Comparison between experimental leakage load and allowable loads suggested by codes
Experimental load Allowable loads by codes
Minimum leakage load obtained
EN 13480, 2002 ASME B31.1 & B31.3
from axial tests
1170 kN 885.20 kN 885.20 kN
Table III.3.3-5: Maximum moment, Mmax, axial force, Nmax and shear force, τmax in the piping system obtained
from a case study
Mmax Nmax τmax
16.79 kNm 180.5 kN 5.08 N
III.3.4. Task 3.4: Tests on piping elbows under strong cyclic loading
III.3.4.1. Tests on non-pressurized piping elbows under strong cyclic loading
The evolution of the load-displacement cycles for each one of the eight tests conducted is presented in
detail in Figure III.3.4-4 to Figure III.3.4-11. For the lower end-displacement amplitudes (Test 1 to Test
4), the resulting load-displacement loops are not changing in shape except for the cycle where failure
takes place. On the contrary, in Tests 5, 6 and 7, the initial shape and size of the load-displacement
cycles changes significantly indicating a loss of resistance (structural degradation), as the number of
cycles increases. This phenomenon can be attributed to the accumulation of plastic deformation at the
most strained locations of the elbow resulting to the formation of plastic hinges at the elbow middle
section and finally to cracking of the elbow, see Figure III.3.4-2.
Cross-sectional “flattening” is measured at two positions along the specimen. The first position is at the
middle section of the elbow, where maximum deformation occurs, while the second location is near the
weld region, as shown in Figure III.3.4-2. Two representative flattening-displacement curves are shown
in Figure III.3.4-8 and Figure III.3.4-12.
Figure III.3.4-1: Cracked elbow specimen No. 4 with Figure III.3.4-2: Experimental set-up:
displacement range ±150 mm: Failure stage with through Instrumentation
thickness crack development
94
Figure III.3.4-3: Increasing amplitude – Loading pattern Figure III.3.4-4: Test 1 - l  25 mm Load –
according to ECCS recommendations. displacement cycles
Figure III.3.4-5: Test 2 - l  70 mm Load – Figure III.3.4-6: Test 3 - l  100 mm Load –
displacement cycles displacement cycles
Figure III.3.4-7: Test 4 - l  150 mm Load – Figure III.3.4-8: Test 4 - Flattening

displacement cycles – displacement cycles
95
Figure III.3.4-9: Test 5 - l  200 mm Load – Figure III.3.4-10: Test 6 - l  250 mm Load –
displacement displacement
Figure III.3.4-11: Test 7 - l  300 mm Load – Figure III.3.4-12: Test 7 -

displacement cycles Flattening – displacement cycles
III.3.4.2. Tests on pressurized piping elbows under strong cyclic loading

CSM Full Scale Test Machine (FSTM) Figure III.3.4-13, dedicated to the testing of tubular components
has been used for elbow testing, some modifications have been made in order to perform the test, in
particular the specimen is connected to the machine by means of specially designed hinged clamping
system that allows for rotations of both ends, while one end is horizontally displaced by actuator. Main
machine capabilities in terms of loads and displacements are reported in Table III.3.4-1.
Clamping system
2500 tons, tensile

Load
Tension and
Compression : 3000,
load capacity compressive
+/- 350mm
Actuator
: 700 in one
Bending jacks displacement direction
Frame Internal
: 1200 bar
pressure
Figure III.3.4-13: CSM FSTM in full scale testing
Table III.3.4-1:FSTM main capabilities
machine
96
The test the following parameters of the test have been monitored:
 elbow displacements by LVDT’s as for Figure III.3.4-14;
 Mid-section vertical and horizontal ovalization, by means of the equipment of Figure III.3.4-15;
 Strains, Figure III.3.4-16.
 Applied force by means of a load cell.
Figure III.3.4-17 to Figure III.3.4-21 show the load displacement curves for all tests and the through
wall crack on elbows. Figure III.3.4-22 and Figure III.3.4-23 show, as an example, the ovalization
diagrams for tests 2 and 4 with the same pressure (70 bar). Whole test data is reported in deliverable
D3.4.
LVDT 2
LVDT 1
Figure III.3.4-14: Measure of elbow displacements Figure III.3.4-15: Measure of ovalization
Figure III.3.4-16: Strain gauges installation

80
40
0
Force - kN
-220 -160 -100 -40 20 80 140 200

-40
Test-1 CSM
-80 +/-200mm
p=32bar
-120
Displacement - mm
Figure III.3.4-17: Test 1 load displacement curve (left) and final crack position (right)
97
80
40
0
Force - kN
-220 -160 -100 -40 20 80 140 200

-40
Test-2 CSM
-80 +/-200mm
p=70bar
-120
Displacement - mm
80
40
0
Force - kN
-320 -240 -160 -80 0 80 160 240 320

-40
-80
Test-3 CSM
+/-300mm
-120 p=32bar
-160
Displacement - mm
80
40
0
Force - kN
-320 -240 -160 -80 0 80 160 240 320

-40
-80
Test-4 CSM
+/-300mm
-120 p=70bar
-160
Displacement - mm
98
120
80
40
Force - kN
0
-220 -160 -100 -40 20 80 140 200
-40
Test-5 CSM
-80 +/-200mm
p=120bar
-120
Displacement - mm
270
OD vert. actual Test-2 CSM
+/-200mm p=70bar 260
OD horiz actual Test-2 CSM
250
+/-200mm p=70bar
240
Diameter - mm
230
220
210
200
190
180
-320 -240 -160 -80 0 80 160 240 320
Applied displacement - mm
Figure III.3.4-22: Test 2 elbow ovalization
270
OD vert. actual Test-4 CSM +/-
300mm p=70bar 260
OD horiz actual Test-4 CSM
+/-300mm p=70bar 250
240
Diameter - mm
230
220
210
200
190
180
-320 -240 -160 -80 0 80 160 240 320
Applied displacement - mm
Figure III.3.4-23: Test 4 elbow ovalization
99
III.3.4.3. Pseudo-dynamic loading on piping system
Table III.3.4-2: Experimental program on a piping system carried out by UTrento
Excitation Excitation PGA PGA
Test Case
Wave Direction m/s2 g
Identification test of the
Identification Hammer Hammer
physical substructure, - - -
tests Test impact
IDT
Elastic tests Elastic tests, ET PDDS Seismic Horizontal 0.4128 0.0421
Operational limit state
Serviceability PDDS Seismic Horizontal 0.7719 0.0787
test, SLOT
limit state
Damage limit state test
tests PDDS Seismic Horizontal 1.0980 0.1119
SLDT
Ultimate Safe life limit state test
PDDS Seismic Horizontal 4.1278 0.4208
limit state SLVT
tests Collapse limit state tests
PDDS Seismic Horizontal 5.8779 0.59918
SLCT
Table III.3.4-3: PGA corresponding to different limit states
Limit States PGA, m/sec2 PGA, Ratio
Serviceability SLO Operational limit state 0.77 0.187 SLV
Limit States SLD Damage limit state 1.1 0.266 SLV
Ultimate SLV Safe life limit state 4.13 1.000 SLV
Limit States SLC Collapse limit state 5.88 1.424 SLV
Table III.3.4-4: Maximum moment, Mmax stress, σmax and shear, τmax in the piping system under reference
seismic loading
Response without damping
Component Mmax (kNm) σmax (MPa) Stress level
Elbow 2 69.15 503.56 Above yield strength
Straight pipe 81.56 296.03 Below yield strength
Response with Rayleigh damping

Component Mmax (kNm) σmax (MPa) Stress level
Elbow 2 43.63 317.74 Below yield strength
Straight pipe 45.62 288.12 Below yield strength
τmax in the straight pipe (kN)

τmax (without damping) 95.48
τmax (with damping) 48.50
100
(a) (b)
Figure III.3.4-24: (a) A 3D model of the piping system and support structure; (b) PS, NS and coupling nodes
Figure III.3.4-25: Schematic of the experimental set-up for the hybrid tests
Figure III.3.4-26: Actual test set-up for the PDDS
101
An elbow The tee joint The bolted flange joint
Figure III.3.4-27: Some components of the test specimen
(a) (b)
Figure III.3.4-28: Rotation histories and relevant spectra of the two end nodes of Elbow #1 at (a) SLVT and (b)
SLCT
(a) (b)
Figure III.3.4-29: Strain histories of Elbow #2 at (a) SLVT and (b) SLCT
102
(a) (b)
Figure III.3.4-30: Displacement histories and relevant spectra of the Coupling Node #1 at (a) SLVT and (b) SLCT
(Blue: Experimental; Magenta: Numerical)
III.3.5. Task 3.5: Tests on T-joint pipe connections under strong cyclic loading
For the experimental study of branch-to-main-pipe T-joint connections under strong fully-reversing
cyclic loading, two (2) connection types were considered, a connection with a fitting (TJF) and a
connection with a welded reinforcement ring-plate (TJWR). In total, eight (8) specimens were tested,
one (1) under static monotonic and three (3) under fully reversing cyclic loading for each connection
configuration.
To conduct these tests, the main steel pipe of the specimen was fixed at both of its ends to the steel
reaction testing frame. Both ends of the main pipe were bolted to steel columns at a large enough
distance from the T-joint so that the fixity at the supports would not affect the behavior at the critical
region of the connection. The load was applied through a hydraulic actuator hinged to the free-end of
the branch through a steel ball-joint hinge. The applied load was perpendicular to the pipe-branch plane.
Regarding the instrumentation setup for both types of specimens, two (2) wire position transducers were
used to measure the load-point displacement (LPD) and a number of strain gages were placed at critical
locations of the T-joint connection. In particular, axial 5-element strip strain gages were attached to the
specimen in the fitting and welded reinforcement plate region for determining the strain distribution in
the T-joint connection along the axis of the branch. The strain gages were at a distance of 3 mm, with
the first one being placed 5 mm and 5078 mm away from the weld toe of the connection for the TJWR
and TJF specimens, respectively.
Two (2) of the T-joint specimens were subjected to static monotonic loading, one for each connection
type. Under monotonic loading and for an imposed LPD of 200 mm the specimen with the fitting
showed about 93% higher load carrying capacity than that of the specimen with the welded
reinforcement plate (see Figure III.3.5-1).
Also, six (6) specimens were subjected to low-cycle fatigue three (3) for each connection type, up to a
through-thickness crack failure of the T-joint connection. Two (2) specimens were subjected to an
imposed LPD value of δ1=60 mm and δ2=45 mm and one (1) specimen was subjected to a fully-
reversing cyclic loading history according to the ECCS No45 [20] testing procedure. For the ECCS
testing procedure, an LPD value of δy=20 mm was considered corresponding to a conventional yield
load level Py based on the results of the tests under monotonic loading according to the recommended
method of ECCS No 45.
For the imposed LPD value of δ1=60 mm, the TJF specimen exhibited a higher fatigue strength of
about 7 times that of the TJWR one. The TJF specimen failed after Nf=42 fully-reversing loading
cycles, while the TJWR specimen after Nf=6 loading cycles (see Figure III.3.5-2). For the imposed
displacement of δ2=45 mm, the TJF specimen failed after Nf=135 fully-reversing loading cycles
exhibiting a higher fatigue resistance of about 8 times that of the TJWR one, which failed after N f=17
loading cycles (see Figure III.3.5-3). For the low-cycle fatigue tests according to ECCS No45, the TJF
103
specimen resisted 36% more loading cycles than the TJWR one. The TJF and TJWR specimen failed
after Nf=15 and 11 loading cycles, respectively (see Figure III.3.5-4).
All the T-joint specimens failed due to a through-thickness crack. The TJWR specimens failed along
the circumference of the branch-to-reinforcement plate weld-toe (see Figure III.3.5-5). The specimens
examined under monotonic loading and under an imposed LPD of 60 mm, failed in the branch region
of the connection while the other two specimens in the welded reinforcement plate region of the
connection. In the case of the TJF specimens, cracking initiated at the top or the bottom of the fitting
region in a bandwidth of about 20 mm transversely to the hoop direction of the main pipe at a distance
of about 70 mm away from the branch-to-fitting weld-toe (see Figure III.3.5-6).
180
TJF-1M
160 TJWR-1M
140
Applied load, P max (kN)
120
100
80
60
40
20
0
0 20 40 60 80 100 120 140 160 180 200 220 240
Load-point deflection, δ (mm)
Figure III.3.5-1: Load vs. LPD curves for the T-joint specimens under monotonic loading.
180
180 160
160 δ=60 mm
140 δ=60 mm
140 120
120 100
100 80
80 60
60
40
Load, P max (kN)
40
Load, P max (kN)
20
20
0
0
-80 -60 -40 -20 0
-20δ=60mm 20 40 60 80
-80 -60 -40 -20 -20 0 20 40 60 80
-40
-40
-60
-60
-80
-80
-100
-100
-120 δ=-60 mm -120
-140 -140
δ=-60 mm -160 -160

-180 -180
Load-point deflection, δ (mm) Load-point deflection, δ (mm)
(a) (b)
Figure III.3.5-2: Load vs. LPD curves under cyclic loading at an imposed LPD value of δ1=±60 mm: (a) TJF, (b)
TJWR specimen.
104
180 180
160 160
δ=45 mm 140
140
120 δ=45 mm
120
100 100
80 80
60 60
40 40
Load, P max (kN)

Load, P max (kN)
20 20
0 0
-80 -60 -40 -20 0
-20δ=60mm 20 40 60 80 -80 -60 -40 -20 -20 0 20 40 60 80
-40 -40
-60 -60
-80 -80
-100 -100
-120 δ=-45 mm -120
-140 -140
δ=-45 mm -160 -160
-180 -180
(a) (b)
Figure III.3.5-3: Load vs. LPD curves under cyclic loading at an imposed LPD value of δ2=±45 mm: (a) TJF, (b)
TJWR specimen.
180 180
160 160
140 140
120 120
100 100
80 80
60 60
40 40
Load, P max (kN)
Load, P max (kN)
20 20
0 0
-200 -150 -100 -50 -20 0 50 100 150 200 -200 -150 -100 -50 -20 0 50 100 150 200
-40 -40
-60 -60
-80 -80
-100 -100
-120 -120
-140 -140
-160 -160
-180 -180
(a) (b)
Figure III.3.5-4: Load vs. LPD curves under cyclic loading according to the ECCS No45 testing procedure: (a)
TJF, (b) TJWR specimen.
(a) (b)
Figure III.3.5-5: Through-thickness crack failure of TJWR specimen at the weld toe of: (a) branch region, (b)
main pipe region at the end of the welded reinforcement plate.
105
Figure III.3.5-6: Through-thickness crack failure in the region of the fitting for the TJF specimen subjected to
monotonic load.

III.4.1. Task 4.3 Parametric numerical study of critical component behaviour
III.4.1.1. Parametric numerical study of elbow behavior under extreme monotonic and cyclic
conditions
III.4.1.1.1. In-plane bending
For each case examined, moment-rotation curves are derived and the local strain ranges at the critical
locations are monitored. Analysis results for the case of 8-inch SCH40 long radius elbow are reported in
Figure III.4.1-1 to Figure III.4.1-4.
Figure III.4.1-1: Numerical parametric analysis Figure III.4.1-2: Numerical parametric analysis
results – 8-inch SCH40 long radius , results – 8-inch SCH40 long radius ,
Δr=±0.02 rad Δr=±0.04 rad
Figure III.4.1-3: Numerical parametric analysis Figure III.4.1-4: Numerical parametric analysis
results – 8-inch SCH40 long radius, results – 8-inch SCH40 long radius,
Δr=±0.08 rad Δr=±0.12 rad
106
Parametric analysis results for the local strain range are reported in Table III.4.1-1 to Table III.4.1-4 for
the complete set of cases. The strain ranges at the critical location of the elbow are reported for each
case showing that the increased elbow wall thickness results to increased fatigue life for the same end-
rotation range due to the reduction of the strain field range developing localy. In addition, higher R/D
ratio results to increased fatigue life for the same end-rotation amplitude and schedule.
8-inch SCH20 8-inch SCH40

Δε  %  Nf
Long radius Long radius
Rotation Rotation
NLKH TL NLKH TL NLKH TL NLKH TL
Amplitude (rad) Amplitude (rad)
±0.02 0.69 0.76 1651 1263 ±0.02 0.67 0.68 1838 2848
±0.04 1.66 1.70 133 123 ±0.04 1.61 1.58 143 207
±0.08 4.87 3.18 5 20 ±0.08 6.00 2.95 5 29
±0.12 7.88 4.47 1 7 ±0.12 6.70 3.12 2 24
Table III.4.1-1: Parametric analysis results: 8-inch Table III.4.1-2: Parametric analysis results: 8-inch
SCH20 long radius SCH40 long radius
8-inch SCH80 8-inch SCH40

Long radius Short radius
Rotation Rotation
NLKH TL NLKH TL NLKH TL NLKH TL
Amplitude (rad) Amplitude (rad)
±0.02 0.59 0.61 2630 2399 ±0.02 0.77 0.81 1215 1712
±0.04 1.46 1.56 193 157 ±0.04 2.01 1.79 76 141
±0.08 3.80 1.67 12 130 ±0.08 6.02 3.42 3 18
±0.12 6.99 2.46 2 42 ±0.12 9.78 6.42 0 2
SCH80 long radius SCH40 short radius
III.4.1.1.2. Out-of-plane bending
For each case examined, moment-rotation curves where derived and the local strain ranges at the
critical locations where monitored. For the case of 8-inch SCH40 short radius results are reported in
Figure III.4.1-5 and Figure III.4.1-6 for three levels of pressure (0MPa, 3.2 MPa and 7.0 MPa). For the
case of 8-inch SCH40 long radius results are reported in Figure III.4.1-7 for 0MPa pressure. Parametric
analysis results are reported in Table III.4.1-5 to Table III.4.1-7 for the complete set of cases. The local
strain ranges at the critical location of the elbow are reported for each case.
107
Figure III.4.1-5: Moment versus end-rotation curve Figure III.4.1-6: Moment versus end-rotation curve
for displacement range ±0.25 rad, (for pressure 0 for displacement range ±0.4 rad, (for pressure 0
MPa, 3.2 MPa and 7.0 MPa). MPa, 3.2 MPa and 7.0 MPa).
8-inch SCH80 8-inch SCH80 8-inch SCH40 8-inch SCH40

Long radius Short radius Long radius Short radius
Rotation Rotation
Amplitude Δε  %  Nf Δε  %  Nf Amplitude Δε  %  Nf Δε  %  Nf
(rad) (rad)
±0.05 0.49 4424 0.87 849 ±0.05 0.55 3132 0.55 3164
±0.1 0.83 966 1.45 195 ±0.1 0.91 745 0.98 595
±0.25 7.83 1 12.72 0 ±0.25 3.29 18 3.41 16
±0.4 10.8 0 19.3 0 ±0.4 5.54 4 5.72 3
SCH20 long and short radius. SCH40 long and short radius.
8-inch SCH20
Long radius
Rotation
Amplitude Δε  %  Nf
(rad)
±0.05 0.87 839
±0.1 1.66 132
±0.25 8.78 1
±0.4 14.41 0
Table III.4.1-7: Parametric analysis results: 8-inch Figure III.4.1-7: Moment versus end-rotation curve
SCH20 long radius. for displacement ranges ±0.05, ±0.1, ±0.25 and ±0.4
rad, (pressure 0 MPa).
III.4.1.2. Parametric numerical study of flanged joints
III.4.1.2.1 Parametric numerical study of standard flanged joints
The typical deformed shape observed in the present FEA study is reported in Figure III.4.1-8. Local
buckling of the pipe in the form of inward directed bulging is reported. Within the buckle large
compressive plastic strains are present. Although these strains are the largest observed in the whole
model, it is not likely that they can cause crackling. Large tensile strains are observed in the neck of the
108
flange and on the pipe extrados, which have to be carefully considered as they could lead to through
wall cracks.
The moment rotation curves obtained as a result of the parametric FE study are reported in Figure
III.4.1-9, Figure III.4.1-10 and Figure III.4.1-11. The peak moments that can be sustained by the various
connections and the corresponding pipe end rotations are tabulated in Table III.4.1-8.
The effect of the load parameters on the joint and a comparison for various geometries can be observed
in Figure III.4.1-12 and Figure III.4.1-13.
Large tensile strains

risk of cracking
Inward directed bulging

large compressive strains
Figure III.4.1-8: typical deformed shape and plastic strain map of the flanged joints after bending
160 200
180
140
160
120
End moment - [kN*m]
End moment - [kN*m]
140
100
120
80 100
60 80
PN40, p=0.5*PN, N=0.24*Ny PN63, p=0.5*PN, N=-0.24*Ny
PN40, p=0.5*PN, N=-0.24*Ny 60 PN63, p=0.5*PN, N=-0.12*Ny
40 PN40, p=0.5*PN, N=-0.12*Ny PN63, p=0.5*PN, N=0.24*Ny
PN40, p=0.5*PN, N=0.12*Ny 40 PN63, p=0.5*PN, N=0.12*Ny
20 PN40, p=1*PN, N=0 PN63, p=1*PN, N=0
20
PN40, p=0.5*PN, N=0 PN63, p=0.5*PN, N=0
0 0
0 100 200 300 400 500 0 100 200 300 400 500
End rotation - [millirad] End rotation - [millirad]
Figure III.4.1-9: Moment vs. rotation curves for Figure III.4.1-10: Moment vs. rotation curves for
monotonic bending of PN40 flanged connections monotonic bending of PN63 flanged connections
109
300
250
End moment - [kN*m]

200
150
100 PN100, p=0.5*PN, N=-0.24*Ny

PN100, p=0.5*PN, N=-0.12*Ny
PN100, p=0.5*PN, N=0.12*Ny
50 PN100, p=100, N=0
PN100, p=0.5*PN, N=0.24*Ny
PN100, p=50, N=0
0
0 100 200 300 400 500
End rotation - [millirad]
Figure III.4.1-11: Moment vs. rotation curves for monotonic bending of PN100 flanged connections
Table III.4.1-8: Summary of buckling moment and rotation for monotonic behavior.
Internal pressure effect, with N = 0
PN40 PN63 PN100
P/PN end rotation end moment end rotation end moment end rotation end moment
- millirad kNxm millirad kNxm millirad kNxm
0 137 139 160 178 210 270
0.5 165 141 189 180 230 271
1 196 142 228 180 258 271
Axial load effect, with P = 0.5*PN
PN40 PN63 PN100
N/Ny end rotation end moment end rotation end moment end rotation end moment
- millirad kNxm millirad kNxm millirad kNxm
-0.24 100 122 110 153 115 229
-0.12 130 132 140 167 150 250
0 165 141 189 189 230 271
+0.12 195 146 195 185 270 280
+0.24 195 145 175 175 251 277
300 300 300 300
250 250 250 250


End moment - [kNxm]
End moment - [kNxm]
200 200 200 200
150 150 150 150
PN100 end rotation PN40 end rotation

100 PN63 end rotation 100 100 PN63 end rotation 100
PN40 end rotation PN100 end rotation
50 PN100 end moment 50 50 PN40 end moment 50
PN63 end moment PN63 end moment
PN40 end moment PN100 end moment
0 0 0 0
0% 20% 40% 60% 80% 100% -24% -18% -12% -6% 0% 6% 12% 18% 24%
Pressure - [% of PN] Axial load - [% of Ny]
Figure III.4.1-12: Effect of pressure for all flange Figure III.4.1-13: Effect of axial load for all flange
geometries geometries
110
XIII.4.1.3.2 Parametric numerical study of non-standard flanged joints
(a) (b)
Figure III.4.1-14: Stress distribution for the bending test: (a) 18 mm thickness joint, and (b) 36 mm
thickness joint
(a) (b)
Figure III.4.1-15: Stress distribution for the axial test: (a) 18 mm thickness joint, and (b) 36 mm thickness joint
(a) (b) (c)

Figure III.4.1-16: Critical elements for the bending test: (a) pipe, (b) flange, and (c) bolt
111
(a) (b) (c)
Figure III.4.1-17: Critical elements for the axial test: (a) pipe, (b) flange, and (c) bolt
III.4.1.3. Parametric numerical study of Tee pipe joints

140 140
120 120
100 100
Force (kN)
Force (kN)
80 80
60 60
40 40
20 Reinforced 20 Reinforced
Fitting
Fitting
0 0
0 20 40 60 80 100 0 20 40 60 80 100
(a) Imposed displacement (mm) (b) Imposed displacement (mm)
Figure III.4.1-18: Tee-junction capacity in (a) out-of-plane bending and (b) in plane bending.
1400 1000
1200
800
1000
Force (kN)
Force (kN)
600
800
600 400
400
200
200 Reinforced Reinforced
Fitting Fitting
0 0
0 20 40 60 80 100 0 20 40 60 80 100
(a) Imposed displacement (mm) (b) Imposed displacement (mm)
Figure III.4.1-19: Tee-junction capacity in axial loading (a) pulling branch away from the main pipe and (b)
pushing branch towards the pipe.
Yield; out-of-plane bending Ultimum; out-of-plane bending

Yield; in-plane bending Ultimum ; in-plane bending
300
250
200
Strength (kN)
150
100
50
0
8x4 8x6 8x8
pipe-to-branch diameter
Figure III.4.1-20: Variation of yield and ultimate strength in for the 3 branch-to-pipe diameters in bending.
112
Figure III.4.1-21: Comparison of out-of-plane and in-plane strength for the different pad geometries.

III.5.1. SEISMIC RESPONSE OF LIQUID STORAGE TANKS
Representative results of the behavior of Tank I are displayed in the present Appendix. Numerical
results for the unanchored tank are examined first, and compared with results from the anchored tank.
General-purpose finite element program ABAQUS is used for the simulation of the structural behavior
of the tank under static loading, assuming anchored and unanchored conditions.
For the case of unanchored conditions, the tank shell and the bottom plate are modeled with shell finite
elements, with special attention on the annular base plate of the tank. More specifically, the tank and the
bottom plate are modeled with four-node reduced-integration shell elements (S4R). In such a case, the
effects of uplifting of the unanchored tank are simulated with the use of appropriate contact conditions
between the bottom plate of the tank and the ground. The ground is simulated using solid eight-node
reduced-integration elements (C3D8R). The friction coefficient between the tank base plate and the
ground is assumed to be 0.3. The tank roof is modeled with the use of two-node (linear) beam-type
finite elements (B31).
An incremental non-linear static analysis is conducted in three steps. In the first step, gravity of the tank
is applied, whereas in the second step, hydrostatic pressure of the liquid is activated. Finally, in the third
step, horizontal hydrodynamic loading is applied, through an incremental elastic-plastic “pushover”
analysis with large displacements. The distribution of hydrodynamic pressures on the tank wall follows
the corresponding solution of the hydrodynamic problem, for both the impulsive and convective
component of liquid motion It is applied on the inner surface of the tank shell and the bottom plate
through a special-purpose load-user subroutine in the ABAQUS model.
The results of this analysis indicate a significant effect of uplifting in tank behavior, when compared
with the case of anchored tanks. Those results are aimed at better understanding the uplifting
phenomenon during seismic excitation, and may be used for improvement of the current design
practice.
A non-linear dynamic analysis is also performed by simulating the tank-liquid system with a simplified
model that accounts for the geometry of the liquid storage tank of radius R and filling height H and the
hydrodynamic response of the liquid. The tank is assumed to be excited by a horizontal base-ground
seismic acceleration motion, X . In the case of anchored tanks, the container is anchored to the
foundation and the impulsive and convective motion of the liquid are modelled by assuming two linear
oscillators with masses mI and mC at heights hI and hC respectively, as seen in Figure III.5.1-6(a). In
the case of unanchored tanks, the container lies on the foundation and its base plate without anchors.
The main difference between anchored and unanchored tanks is the uplifting mechanism which can be
modelled by adding a nonlinear spring at the tank (Figure III.5.1-6(b)). For accurate modeling, the
M  relationship of Figure III.5.1-4 is used. Finally, to facilitate the formulation, the convective mass
has not been considered.
113
(a) (b)
Figure III.5.1-1: Finite element analysis of unanchored liquid storage tank under lateral loading; (a) uplifting of
base plate, (b) development of elephant’s foot at the tank bottom.
Figure III.5.1-2: Top horizontal displacement u and maximum vertical uplift of Tank I versus overturning
moment M , subjected to finite element pushover analysis.
Figure III.5.1-3: Ratio of maximum compressive axial membrane forces for the unanchored over the anchored
tank versus the overturning moment; finite element results for Tank I and curves for   H / R  0.8, 1.0, 2.0,
adopted by EN 1998-4.
114
Figure III.5.1-4: Overturning moment in terms of the rotation of the tank.
Figure III.5.1-5: Length of uplifted part of the base as a function of the vertical uplift at the edge and bilinear
approximation to be used later in the dynamic analysis. The equations of the two lines are shown on the graph.
(a) (b)
Figure III.5.1-6: Simplified models for anchored and unanchored liquid storage tank.
115
Figure III.5.1-7: Time history of the uplifting axial membrane force in the case of (a) the anchored Tank I and (b)
the unanchored Tank I for the earthquake in Duzce, Turkey (1999).
The analysis of elephant’s foot buckling at the bottom of Tank I is examined, through the consideration
of an axisymmetric finite element model. It can be concluded that there is a quasi-uniform state of
deformation along a significant part of the tank perimeter, justifying the axisymmetry assumption. The
most important observation refers to post-buckling response of the tank, which appears to be very
unstable; after reaching a maximum stress (load) the capacity drops very rapidly, indicating that the
tank has not the ability of absorbing significant amounts of inelastic energy.
0.35
D=27.4 m
H=16.514 m
0.3 Hliq=15.7 m
0.25 δ/L=0.00052
normalized height
δ/L=0.00126
0.2 δ/L=0.0063
Tank shell
0.15
0.1
elephant’s
foot buckling
0.05
(a) (b) (c)
Bottom plate 0
0 0.01 0.02 0.03 0.04 0.05
normalized horizontal displacement (δ/R)
Figure III.5.1-8: Development of elephant’s foot, as predicted by the finite element analysis.

III.6.1. Task 6.1 Seismic behaviour of vertical-cylindrical pressure vessels
Three different geometries are considered for the case of vertical-cylindrical slender pressure vessels on
skirt supports, a short, a tall and a thick vessel as described in Deliverable 6.1, extending the work
carried out in Deliverable 2 for case study 5.
III.6.1.1. Pushover analysis results discussion

A non-linear static (pushover) analysis has been employed to identify the modes of failure and study the
progressive collapse after exceeding the yield limit. For the case of the vertical-cylindrical vessel No1
the base shear versus drift is plotted in Figure III.6.1-1 for both the rectangular and triangular
assumption for the seismic load distribution as well as the effect of internal pressure. Maximum
operating pressure (PS), design pressure (Pd) and zero internal pressure was considered.
116
The maximum capacity is reached very quickly after the first-yield point, whereas the vessel resistance
is significantly reduced beyond the maximum loading stage; this indicates that the vessel may not be
capable of absorbing significant amount of inelastic energy. The effect of internal pressure is
insignificant up to the point of maximum capacity of the vessel but is beneficial for post-buckling
behavior. By comparing the seismic forces considered in the initial design of these pressure vessels it
was observed that these lie within the elastic behaviour of the pushover curve and there is still reserve
to endure higher seismic loads.
P=0 triangular distribution P=PS triangular distribution P=Pd triangular distribution

P=0 unifrorm distribution P=PS unifrorm distribution P=Pd unifrorm distribution
5000
4000 PEEQ 4.7%

Base shear (kN)
3000
2000
1000 Shear Force 1.04MN q=2
0
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Drift (mm)
Figure III.6.1-1: Base shear versus roof displacement – pushover analysis results for the VV No1
The corresponding mode of failure is buckling at the first shell course immediately above the skirt for
vessels 1 and 2, whereas for the thick pressure vessel 3 the critical mode of failure is buckling on the
skirt, as shown in Figure III.6.1-2.
The base shear forces derived from the pushover analysis are plotted in Figure III.6.1-3 for vertical
vessel No3 together with the results of IDA analysis. Pushover analysis using the triangular load
distribution is conservative, while the rectangular distribution could overestimate the capacity of the
vessels.
No1 No3
Figure III.6.1-2: Equivalent plastic strain at maximum capacity for VVs No1 and 2
117
40000
Base shear (kN)

30000
uniform distribution
20000
triangular distribution
IDA Akz1
IDA Akz2
IDA Akz3
10000 IDA Akz4
IDA Akz5
IDA Akz6
IDA Akz7
0
0 100 200 300 400
No 3
Drift (mm)
Figure III.6.1-3: Comparison of base shear vs drift between pushover and IDA results for the VV No3
III.6.1.2. Incremental dynamic analysis results discussion

The vertical-cylindrical vessels have been analyzed under incremental dynamic analysis in order to
estimate the behavior factor (Deliverable 6.1). Based on the pushover analysis results presented in the
previous section, two regions of the vessel have been identified as critical, namely (a) the skirt and (b)
the 1st shell course. In order to estimate the q factor, two seismic input levels have to be identified in
each of the response curves; (a) the level where transition from linear to non-liner behavior occurs (λe)
and (b) the level at which a specific failure criterion is satisfied (λu). Three failure criteria have been
used: (a) Rupture of vessel wall when the equivalent plastic strain reaches a value equal to 0.4%, (b)
Buckling of vessel wall when the compressive strain in the longitudinal direction reaches a critical
value defined by [38] and (c) Instability when the maximum value of the principal structural strain
reaches a value equal to 5% according to EN13445-3 Annex B [24].
Using the three failure criteria, the estimated value of the q factor is given below in Table III.6.1-1.
The first two failure criteria, predict behavior factor values somewhat lower than the value of 2
suggested by ASCE 7 [5]. However, the failure criteria suggested from EN13445-3 gives much higher
values. The more realistic results are believed to be given by the empirical formulas for buckling of the
vessel wall giving lower behavior factor for the thin walled vertical pressure vessels but predicting
better the behavior factor for the case of buckling of the skirt.
Table III.6.1-1: Estimated q factor by IDA for the vertical pressure vessels
VV No1 VV No2 VV No3
FC 1 FC 2 FC 3 FC 1 FC 2 FC 3 FC 1 FC 2 FC 3
λe 1.5 1.5 1.5 2.5 2.5 2.5 1.5 1.5 1.5
λu 2.5 2.25 4 3.5 4 6 2.5 3.5 4.0
q 1.7 1.5 2.7 1.4 1.6 2.4 1.7 2.3 2.7
III.6.2. Task 6.2 Seismic behaviour of horizontal-cylindrical pressure vessels

Four different geometries are considered for the case of horizontal-cylindrical pressure vessel on saddle
supports. The first pressure vessel represents case study 6 extensively discussed in Deliverable 1.3b.
The other vessels are variations with regards to reinforcing rings, length and thickness as given in
Deliverable 6.1. In HV2 the reinforcing rings have been omitted, while in HV3 the length between the
saddle supports has been doubled. For HV4 an increase in thickness as well as increase in weight of the
vessel has been considered.
118
Force displacement diagrams (pushover curves) are depicted in Figure III.6.2-1 for all model variations
in the longitudinal and transverse directions. The pushover analysis shows that for the level of seismic
loads used for the design of those vessels, the behavior is elastic; in fact, the level of seismic design
loads is well below the ultimate capacity level indicated by the pushover curves. A “ductile” behavior is
observed, with the vessel capable of sustaining significant deformation in the inelastic range. The
capacity of the vessel in the transverse direction is double that in the longitudinal.
1000
2500
800
2000
Shear Force (kN)
Shear Force (kN)

600 1500
400 1000
HV1 PS HV1 PS
200 HV2 PS 500 HV2 PS
HV3 PS HV3 PS
HV4 PS HV4 PS
0 0
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
(a) Drift (mm) (b) Drift (mm)
Figure III.6.2-1: Base shear versus roof displacement for all horizontal vessels with internal pressure – pushover
analysis results (a) in the longitudinal and (b) in the transverse direction.
The location of failure is given in Figure III.6.2-2 for one representative case for loading in the
longitudinal and transverse directions respectively. It can be seen that the mode of failure is buckling at
the saddle support. The exact location in the saddle varies with load direction as depicted in the figures.
The effect of internal pressure when pushover loading is applied in the longitudinal direction is not
significant. This is true for all pressure vessels modeled. Whereas, when loading is applied in the
transverse direction the effect is more pronounced for no internal pressure. This can be explained
because for loading in the longitudinal direction stress concentration is limited in the saddle, however
when loading is applied in the longitudinal direction and there is no internal pressure then there is also
stress concentration in the shell next to the saddle, making the effect of internal pressure significant.
The reinforcing rings as well as increased thickness of the vessel increases the capacity of the vessel to
endure seismic loading as expected. Reinforcing rings are recommended for large pressure vessels or
pressure vessels located in high seismicity areas. Caution shall be taken when deciding the area of the
saddle reinforcement as the adjacent shell area is expected to fail in the case of no internal pressure. A
simplified model of a single support and part of the vessel could be analysed instead of the full
structure.
(a) (b)
Figure III.6.2-2: PEEQ distribution for the Horizontal Vessel No1 with internal pressure – (a) loading in the
longitudinal and (b) in the transverse direction
119
1000
2000
800
1500
Shear Force (kN)
Shear Force (kN)

600
1000
400
500
HV2 PS HV2 PS
200
HV2 P=0 HV2 P=0
HV2 Pd HV2 Pd
0
0
0 10 20 30 40 50 60 70 80 90 100
0 10 20 30 40 50 60 70 80 90 100
(a) (b) Drift (mm)
Drift (mm)
Figure III.6.2-3: Effect of internal pressure on base shear versus roof displacement for HV2– pushover analysis
results (a) in the longitudinal and (b) in the transverse direction
III.6.3. Task 6.3 Seismic behaviour of pressure vessels of spherical shape

Two different types of pressure vessels (braced and non-braced pressure vessels with 12 Columns) have
been investigated using modal and pushover analyses as well as incremental dynamic analyses.
Furthermore more on the basis of both examples the investigations have been extended by a parametric
numerical study with 78 different geometries of braced and non-braced 4-, 8- and 12-column systems.
For the calculation of fundamental periods a simplified beam method were derived. Using this method
an excellent agreement with the results of the performed modal analyses has been observed.

For the pushover analyses the same FE-models were used as for the modal analyses with exception of
some crucial details; geometrical non-linearities and inelastic material behaviour in terms of a von
Mises plasticity material model with linear isotropic hardening (slope of about 500 times smaller than in
the elastic range), based on the nominal values of P355 GH were considered. In the first load step the
structure was loaded by the operating load including the dead weight and the maximum weight of
filling (kf = 90 %); and in the second step the structure was loaded displacement-controlled using a
sliding boundary condition at the centre of gravity, which was coupled to the shell. Furthermore the
influence of internal pressure on the structural behaviour of the vessel has been investigated.
The results of the pushover analyses for the braced and non-braced system are given in Figure III.6.3-1
in terms of a force-displacement curve (pushover curve) referred to the centre of gravity.
8.8 Operating load (kf = 90 %)

25.6 Operating load (without imperfections and εpl,limit)
Operating load + internal pressure Operating load (pre-curved braces)
7.7 22.4 Operating load (pre-curved braces) + internal pressure
Base shear force [MN]
6.6
Base shear force [MN]
19.2
5.5 Ultimate load
16.0
4.4 12.8
Yield load Ultimate load
3.3 9.6
Yield load
2.2 6.4
1.1 3.2 Non-braced PV
0.0 0.0
0 120 240 360 480 600 0 120 240 360 480 600
Drift [mm] Drift [mm]
Figure III.6.3-1: Pushover analysis results in terms of force-displacement curve referred to the centre of gravity
for the non-braced system (left) and the braced system (right).
The non-braced system behaved elastic without any significant plastifications up to a displacement
equal to 132 mm. The first plastifications occurred directly below the shell-column connections of the
columns which are loaded under the greatest compressive stress due to the overturning moment.
120
Thereafter up to a displacement equal to 265 mm (ultimate state) all columns successively plastified
also directly below the shell-column connections. By reaching the ultimate state first local buckling
occurred (see Figure III.6.3-2, left). And finally up to a displacement of approx. 600 mm all other
columns buckled locally, whereby for the columns loaded in tension due the overturning moment the
location of local buckling shifted in the range of compressive strains due to bending in the columns.
Furthermore it should be mentioned that no plastifications occurred in the sphere up to the end of the
complete loading history (1000 mm).
260 mm displacement 134 mm displacement

Non-braced PV Braced PV
Figure III.6.3-2: Pushover analysis results in terms of deformed shape of critical details showing failure modes
and equivalent plastic strains PEEQ.
The braced system behaved elastic without any significant plastifications in columns or bracings up to a
displacement equal to 45 mm. At the time of reaching the yield load the first plastifications occurred in
the most loaded bracings, which were directed parallel to the load direction and therefore loaded under
the greatest tension stress. Immediately after reaching the yield point all bracings in compression failed
rather quickly by formation of plastic hinges. Thereafter up to a displacement equal to 197 mm
(ultimate state) the strongest loaded bracings in tension failed due to reaching of the ultimate elongation
(see Figure III.6.3-2, right), which leaded in consequence to the first plastifications in the columns
loaded in compression due the overturning moment. After reaching the ultimate state up to a
displacement equal to 264 mm all bracings in tension failed as well as all columns successively
plastified directly below the shell-column connections, so that the post critical behaviour was equal to a
non-braced system.
III.6.3.2. Incremental dynamic analysis results discussion

In order to examine the range of possible behaviour factors q, which could be applied to braced and
non-braced spherical pressure vessel supporting structures, nonlinear incremental dynamic analyses
were conducted on simplified models. In these simplified models the pressure vessel sphere was
modeled as rigid framework since pushover analysis with the detailed model showed no influence of
sphere flexibility up to the displacements of interest. A set of 10 artificial accelerograms using
SIMQKE program is considered for the seismic input. The selection and generation of the artificial
accelerograms takes into account the following objectives and requirements:
1. comparability of time step calculations with practical design calculations using seismic spectrum
2. consideration of seismic action randomness
3. computational efficiency
Incremental dynamic analyses have been conducted and appropriate failure criteria have been defined
and evaluated to determine behaviour factors. A conservative estimate using the most critical
earthquake from the set of 10 artificial accelerograms showed that the behaviour factors q are larger
than 2, what confirms current code provisions.
121
For both supporting structures dynamic instability was evaluated as upper limit failure criterion.
Dynamic instability is defined as large increase of lateral drift of the structure with only a small increase
in applied earthquake forces and was determined by means of Ballio-Setti method [7]. It was observed
that in the case of non-braced systems severe dynamic instability could be reached suddenly whereby
for the braced structure it occurred more smoothly and at higher earthquake intensities. The increase of
lateral drift as function of seismic level for all ten accelerograms can be seen in Figure III.6.3-3 for both
systems.
1,4 1,4
1,2 1,2
maximum lateral drift umax [m]

maximum lateral drift umax [m]
1 ACC 1 1 ACC 1
ACC 2 ACC 2
0,8 ACC 3 0,8 ACC 3
ACC 4 ACC 4
ACC 5 ACC 5
0,6 ACC 6 0,6 ACC 6
ACC 7 ACC 7
0,4 ACC 8 0,4 ACC 8
ACC 9 ACC 9
ACC 10 ACC 10
0,2 0,2
0 0
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
seismic level λ seismic level λ
(a) (b)
Figure III.6.3-3: Maximum lateral displacements versus seismic level for all ten accelerograms for (a) the non-
braced supporting structure and (b) the braced supporting structure
Because the columns were modeled as fibre elements in the simplified model, failure criteria like local
buckling phenomena could not directly be accounted for. Therefore the displacements, which
corresponded to these failure modes, were identified in the pushover analyses on the detailed model and
then used as limit displacements in the incremental dynamic analyses. For the non-braced supporting
structure the following failure criteria have been used: (a) dynamic instability according to Ballio-Setti
method as mentioned above; (b) achievement of 1st local buckling in column which corresponds well
with achievement of ultimate load. This state occurred in the pushover analysis with the detailed FE-
model at a displacement of 265 mm; (c) buckling of all columns. The accompanying displacement in
the detailed model was determined to 363 mm. The resulting behaviour factors for each of the failure
criterion are given in Table III.6.3-1 for the most critical earthquake as well as the calculated mean of all
10 earthquakes.
Table III.6.3-1: Most critical as well as mean behaviour factor for all failure criteria for the non-braced supporting
structure
Failure criterion Most critical q value Mean q value
(a) Dynamic instability 3.0 5.10
st
(b) 1 buckling of any column 2.7 3.30
(c) Buckling of all columns 3.3 4.69
Based on the observation that after the first column buckled, which was also the state of ultimate load in
the detailed model, buckling of all other columns followed very quickly and was also followed quickly
by severe and sudden dynamic instability problems in the simplified model, a behaviour factor q of 2.7
is conservatively suggested for the non-braced supporting structure investigated herein.
For the braced supporting structure the following failure criteria have been used: (a) as for the non-
braced system dynamic instability; (b) acceptance criterion for braces in tension according to [35],
where the maximum plastic elongation of bracings is limited to seven times the axial deformation at
tensile yielding load; (c) full plastic behaviour of an arbitrary column head; (d) limit displacement of
122
197 mm at which ultimate load was achieved in the pushover curve of the detailed model; and (e) first
buckling of a column which was observed at an accompanying displacement of 284 mm The resulting
behaviour factors for each criterion are listed in Table III.6.3-2.
Table III.6.3-2: Most critical as well as mean behaviour factor for all failure criteria for the braced supporting
structure
Failure criterion Most critical q value Mean q value
(a) Dynamic instability 3 to 4 -
(b) Plastic elongation of bracing 4.2 5.77
(c) Full column head plastification 2.3 2.58
(d) Ultimate load from Pushover 3.3 3.94
(e) 1st buckling of any column 3.9 5.32
Engineering judgment was necessary to define the dynamic instability failure criterion because although
the maximum lateral drift was above the elastic reference line at almost all earthquake intensities, which
would result in failure when the Ballio-Setti method was strictly applied, no sudden increase in
deformation occurred up to seismic levels of approximately 4 to 4.5. Therefore a factor of 4 seems
reasonable for this failure criterion. If columns are designed and detailed in such a way, that sufficient
ductility is ensured, preventing premature failure of columns due to strong plastification, a behaviour
factor q of 3.3 is suggested for the investigated structure. Otherwise, if energy dissipation is desired to
take place basically in the bracings, the q factor however should be restricted to a value of 2.3.
123
III.7.1. General analyses procedure for seismic design acc to EN 1998-1 and 4
Collection of design data: Establish design seismic action criteria Evaluate consequences of
failure:
- types of fluids -No-collapse requirement
- pressures & temperatures (Ultimate Limit State): PNCR, 50 Year ≤ 10% or TNCR= 475 years* - risk to the public
- piping data; -Damage limitation requirement: -risk to the environment
- seismic data (serviceability limit state) PDCR, 10 years≤ 10% or TDLR = 95 years * - loss of investment
- soil data - loss of production
*
- soil stratigraphy Or: Target reliabilities, to be established by National Authorities - other
- deep geology
(-determination seismic
zones, IBC ground Maps) da
Reference elastic peak ground Define Importance class

acceleration (reliability differentiation by
agR adjustment of return period TNCR
Determine ground type
(type A ground, TNCR= 475 years
Determine soil factor Design peak ground acceleration Determine Importance

Soil type → factor
soil factor S*
(response spectrum dependant!) ag =i* agR I
ag =i* agR
Very low seismicity area ag ≤ 0,04g (0,39 m/s2)

No seismic analyses acc. to EN yes
or
1998 required
ag*S ≤ 0,05g (0,49
m/s2)
no
Low seismicity area ag ≤ 0,08g (0,78 m/s2)

yes
simplified seismic analyses or
possible
ag*S ≤ 0,1g (0,98 m/s2)
no
Select method of seismic analysis:
Equivalent static analyses

Linear elastic
analyses
Modal reponse spectrum analyses
Non-Linear
Time –history analyses elastic-plastic
analyses
124
III.7.2. Seismic linear elastic analyses methods
Seismic input Modeling and analyses

Influence of sub structure
Design peak ground acceleration
Building the piping model:
 Modeling end restraints (anchors,
ag =i* agR
Tanks, Vessels)
ag =i* agR
 Mass distribution
Modal elastic ground acceleration response spectrum
 Representation of (horizontal and vertical)  Support type and- spacing
earthquake motion at a given point at the surface)
 Component flexibility and stress
 Same shape of response spectrum for both the Evaluate type of structure intensification
ultimate- and the damage limitation state - evaluation of non linear
energy dissipating capacity
 Number. of nodes between supports
 Seismic action described by two independent for required accuracy of dynamic
(horizontal) orthogonal components analyses
Determine behaviour
factor
da
q(piping structure) (preliminary) Static analyses:
q (substructure) a) Internal pressure
Design spectrum for elastic analyses b) Weight & sustained load
- Representation of (horizontal and vertical) c) Flexibility

earthquake motion at a given point at the surface)
yes Seismic dynamic

analyses?
Seismic Dynamic analyses (with modal Amplification factor

elastic response spectra for piping located at different
levels inside a structure no
Calculation of natural frequencies

(possible modes of vibration) of the
piping system: Seismic analysis by Equivalent Static (uniform
load) method :
Ultimate limit states a) Response to uniform load:
Calculation of system response
a) to modal elastic response spectra (EN 13480-3)
b) to Seismic Anchor movement
b) Response to Seismic Anchor Movement
da (SAM)
Evaluation of system reponse LRFD design

(SRSS method)
Testing response against limit states Allowable stress Testing response against limit states
design (ASD)
III.7.3. Investigation of seismic input

III.7.3.1. Shape of elastic response spectra
Within the framework of the INDUSE project a comparison has been made between spectra derived for
the same location according to available international codes (ASCE 7 vs EN 1998-1 and 4. Figure
III.7.3-1 gives a comparison between spectra, generated for the same case, according to both codes:
125
Figure III.7.3-1: Horizontal design spectrum data – EN 1998-1 vs ASCE 7
Analyzing Figure III.7.3-1 it can be observed that the horizontal design spectra generated according
ASCE 7 show approximately the same peak level, the peak of the ASCE spectra being reached at a
somewhat higher value of the response period and extending over a wider range of natural periods. One
could say that in this example the ASCE 7 spectra cover both the Type 1 and type 2 spectra of the
European code, except for very small periods.
III.7.3.2. Influence of natural frequency of the piping system

The design acceleration to be considered in the calculation depends on both the shape of the response
spectrum and the stiffness characteristics (possible modes of vibration) of the piping system under
consideration. This is illustrated in the calculation example of Annex A of the Guidelines:
Figure III.7.3-2: Configuration of 10” ammonia transmission line.
126
Figure III.7.3-2 shows is section of an above ground 10”, 25 bar Ammonia transportation pipeline
subject to considerable temperature difference, supported on sliding supports, with locally side
restraints.
Figure III.7.3-3: Horizontal and vertical seismic design spectra according EN 1998-1 and ASCE-7, with natural
frequency range (blue shaded) of above ground cross country ammonia pipeline system
In Figure III.7.3-3, the blue shaded region of the curve represents the range in which this particular
system’s natural frequencies lie; the bold lines are sample modes which have been highlighted (it is
observed that the pipe system under consideration is rather flexible , showing relatively low natural
frequencies). The region outside of the shaded area is above a certain cut-off period, which translates
for the ammonia line to 0.866 Seconds, which is 1.154 Hz – the 1st mode of the systems natural
frequency. Although the peak of the type II horizontal curve is higher, the maximum acceleration with
which the software calculates the system at the 1st mode of vibration is 0.15 ms -2 for the type I
horizontal curve. The pipe’s natural frequency is the main factor on how much the spectra influence the
results.
III.7.4. Influence of supporting structure

A key point in modeling a piping system is the possibility to neglect the interaction (static and dynamic)
between the pipes and the supporting structure. A crude rule is based on the ratio WR of the weight of
pipes over the weight of the supporting structure. In particular, if WR < 25% the interaction can be
excluded and the piping system can be considered as a non-building structure, loaded by a seismic
action coming from the supporting structure at pipes level. This rule was investigated and found to be
approximately useful. A recommendation has been taken up in the guidelines.
127
Figure III.7.4-1: Piping system on a pipe-rack
Figure III.7.4-2: Possible positions of pipes Figure III.7.4-3: Variation of the moment in the pipes for
the coupling effect
Figure III.7.4-4: Deformation of substructure
III.7.5. Example of ultimate limit state for elbows

From the test series carried out at TU Delft and at CSM, a fatigue model was developed, showing very
good conformance with the test results and able to predict the result of testing. From this a limit state
for bends under strong cyclic loading can be defined.
128
Test No. Displacement range Number of cycles Local strain
l (mm) until failure Nf range* (%)
(Exp.)
1 ±25 13160 0.33
2 ±70 444 1.23
3 ±100 171 -
4 ±150 61 2.61
5 ±200 28 -
6 ±250 17 3.84
7 ±300 10 4.02
8 ECCS protocol 16 -
*: calculated in the elbow hoop direction at the critical location
Figure III.7.5-1: Increasing amplitude – Loading Table III.7.5-1: Displacement range and number of cycles
pattern according to ECCS recommendations.
Figure III.7.5-2: Limit state for strong cyclic loading, as taken up in the design guideline
129
IV. List of acronyms and abbreviations
AF Armstrong – Frederick model
ASME American Society of Mechanical Engineers
CSM Centro Sviluppo Materiall
EC8 Eurocode 8
ECCS European Convention for Constructional Steelwork
FE Finite Element
LVDT Linear Variable Differential Transformer
MPS Modal percentage of excited mass
PGA Peak Ground Acceleration
RWTH Aachen University of Technology
SWOT Strengths, Weaknesses, Opportunities, Threats
TL Tseng-Lee
TPLH / Technip TECHNIPETROL HELLAS SA
TUDelft Delft University of Technology
UThessaly University of Thessaly
UTrento University of Trento
WP Work Package
A% elongation (%)
D diameter of tube
ey elastic deformation of a specimen in the ECCS procedure for a cyclic test
Nf number of cycles to failure
py yield pressure
Rm tensile strength
R p 0.2% yield strength
t thickness
Δl displacement amplitude
σy yield stress of a material
a “backstress” tensor
C material model parameter
E Young modulus
q equivalent plastic strain
H hardening modulus
Ĥ hardening modulus when two surfaces are in contact

h an appropriately calibrated parameter
131
ii the intensification factor given in Appendix D of ASME B31.3
k0 initial yield stress
k mean size of the yield surface

k ( q ) function that defines the size of the yield surface
M bending moment
N number of loading cycles
Nf number of cycles to failure
s deviatoric stress tensor

SA allowable local stress amplitude
SC basic allowable stresses at minimum metal temperature
Sh basic allowable stresses at maximum metal temperature
SL longitudinal stress
S L,Occ longitudinal stress produced by occasional loads
Z pipe cross-section section modulus

 material model parameter
 in initial distance between the two surfaces
 hoop strain range

Δl displacement amplitude
σy yield stress of a material
132
V. List of figures
Figure II.3.2-1: Test setup in the laboratory with test specimen P3-1. .................................................... 28
Figure II.3.2-2: Schematic representation of the loading history according to ECCS [20] ..................... 28
Figure II.3.3-1: Weld neck flange geometry ........................................................................................... 29
Figure II.3.3-2: Test configuration, four point bending .......................................................................... 29
Figure II.3.3-3: Testing device ................................................................................................................ 29
Figure II.3.4-1: Experimental set-up: Front view .................................................................................... 32
Figure II.3.4-2: Experimental set-up: Instrumentation ............................................................................ 32
Figure II.3.4-3: Cracked elbow specimen configuration ......................................................................... 32
Figure II.3.4-4: First load-displacement cycles for the 5 tests ................................................................ 33
Figure II.3.4-5: Test 8 – ECCS load - displacement cycles .................................................................... 33
Figure II.3.4-6: Specimen geometry and test configuration .................................................................... 33
Figure II.3.4-7: Specimen inside the experimental device ...................................................................... 34
Figure II.3.4-8: Specimen failure for cracking at flank position ............................................................. 34
Figure II.3.4-9: Typical Force vs. displacement result, Test 2 ................................................................ 34
Figure II.3.5-1: The two types of T-joint branch connection studied: (a) fitting, (b) welded
reinforcement ring-plate. ......................................................................................................................... 36
Figure II.3.5-2: Test setup for the T-joint tests: (a) overall view, (b) side view. ................................... 36
Figure II.3.5-3: Deformed T-joint specimens at failure: (a) TJF, (b) TJWR........................................... 37
Figure II.4.1-1: Material test results – elbow hoop direction .................................................................. 39
Figure II.4.1-2: Lemaitre-Chaboche calibration method ......................................................................... 39
Figure II.4.1-3: Calibration results – Test E1B T-4 ................................................................................ 39
Figure II.4.1-6: Test on elbows – Simulation results using Set 1 and Set 3 ............................................ 39
Figure II.4.1-7: A schematic view of the two-surface model. ................................................................. 41
Figure II.4.1-8: Material curve and model predictions............................................................................ 41
Figure II.4.2-1: General view of the numerical model ............................................................................ 42
Figure II.4.2-2: Concentrations of plastic deformations – closing bending loads ................................... 42
Figure II.4.2-3: Numerical simulation of Test 4 - Load-displacement cycles ........................................ 43
Figure II.4.2-4: Numerical simulation of Test 4 - Flattening- displacement cycles .............................. 43
Figure II.4.2-5: Numerical simulation of Test 4 - Evolution of maximum local strains ........................ 43
Figure II.4.2-6: Numerical simulation of Test 4 - Evolution of minimum local strains.......................... 43
Figure II.4.2-7: Numerical simulation of Test No1- Load - displacement ............................................. 44
Figure II.4.2-8: Numerical simulation of Test No1 - Ovalization - displacement curve ......................... 44
Figure II.4.2-9: Numerical simulation of Test No2: Load - displacement .............................................. 44
Figure II.4.2-10: Concentration of strains at the elbow critical region ................................................... 44
Figure II.4.2-11: Local hoop strain range versus number of cycles to failure......................................... 45
Figure II.4.2-12: The piping system and the two substructures .............................................................. 46
Figure II.4.2-13: Hardware-Software configuration for the PDDS ......................................................... 47
Figure II.4.2-14: Displacement histories and relevant spectra of the Coupling Node #1 at (a) SLVT and
(b) SLCT (Blue: Experimental; Magenta: Numerical) ............................................................................ 47
Figure II.4.2-15: FE-models overview: (P1) internal-, (P2) external- and (P3) nozzle-reinforcement. .. 48
Figure II.4.2-16: Mesh details ................................................................................................................. 49
Figure II.4.2-17: Monotonic and cyclic material properties employed in FEA ...................................... 50
Figure II.4.2-18: PN63-1 FE model deformed shape and plastic strains at maximum bending deflection
(right) and specimen at the end of the test (left) ...................................................................................... 50
Figure II.4.2-19: PN63-1 FE model deformed shape and plastic strains at maximum bending deflection
(right) and specimen at the end of the test (left) ...................................................................................... 50
Figure II.4.2-20: Geometrical models for the: (a) bending test, and (b) axial test .................................. 51
Figure II.4.2-21: Comparison between experimental and numerical data for bending tests: (a) Design
01, (b) Design 02; and axial tests: (c) Design 01..................................................................................... 52
Figure II.4.2-22: Deformed shapes and stress levels in the joint for the: (a) bending test – Design 01 –,
(b) bending test – Design 02 –, and (c) axial test – Design 01 –. ............................................................ 52
Figure II.4.2-23: Numerical models: General view................................................................................. 53
133
Figure II.4.2-24: Finite element analysis results: TJF 2C test ................................................................ 54
Figure II.4.2-25: Numerical TJF model .................................................................................................. 54
Figure II.4.2-26: Finite element analysis results: TJWR 2C test ............................................................ 54
Figure II.4.2-27: Numerical TJWR model .............................................................................................. 54
Figure II.4.3-1: Numerical model for the elbows used for the parametric analysis ................................ 56
Figure II.4.3-2:Numerical parametric analysis results – 8-inch SCH40 long radius - Monotonic analysis
................................................................................................................................................................. 56
Figure II.4.3-3: General view of the numerical model ............................................................................ 56
Figure II.4.3-4: Concentrations of plastic deformations ......................................................................... 56
Figure II.4.3-5:Moment versus end-rotation curve for pressure 0, 3.2 and 7.0 MPa .............................. 57
Figure II.4.3-6: Moment versus end-rotation curve for displacement ranges ±0.05, ±0.1, ±0.25 and ±0.4
rad, (pressure 0 MPa). ............................................................................................................................. 57
Figure II.4.3-7: Overview of the geometrical parameters for the parametric study ................................ 57
Figure II.4.3-8: FE model and load scheme considered in parametric study .......................................... 60
Figure II.4.3-9: Comparison among different thickness flanges for bending tests ................................. 61
Figure II.4.3-10: Comparison among different thickness flanges for axial tests..................................... 61
Figure II.4.3-11: Stress distribution for the bending test: (a) 18 mm thickness joint, and (b) 36 mm
thickness joint .......................................................................................................................................... 62
Figure II.4.3-12: Stress distribution for the axial test: (a) 18 mm thickness joint, and (b) 36 mm
thickness joint .......................................................................................................................................... 63
Figure II.4.3-13: Critical elements for the bending test: (a) pipe, (b) flange, and (c) bolt ...................... 63
Figure II.4.3-14: Critical elements for the axial test: (a) pipe, (b) flange, and (c) bolt ........................... 63
Figure II.4.3-15: Reinforcement pad (EN13480-3 [25]) and tee-fitting geometry (EN10253-2 [22]). ... 64
Figure II.5.2-1: The three liquid storage tanks used in the design examples. ................................... 69
Figure II.7.1-1: The piping system considered in the present study. ....................................................... 71
Figure III.1.2-1: Three liquid storage tanks which are used in the design examples .............................. 78
Figure III.1.3-1: Sketch of (a) vertical pressure vessel on skirt support and (b) horizontal pressure vessel
on saddle supports (drawings made by software PV Elite 2010). ........................................................... 78
Figure III.1.4-1: The analyzed piping system on a pipe-rack.................................................................. 79
Figure III.1.4-2: Configuration of 10” ammonia transmission line. ........................................................ 79
Figure III.1.4-3: Configuration of 20” gas transmission pipeline. .......................................................... 79
Figure III.2.1-1: Tank I (left), tank II (center) and tank III (right) used for the parametric studies in the
present paper............................................................................................................................................ 79
Figure III.2.1-2: Finite element analysis of unanchored liquid storage tank under lateral loading; (a)
uplifting of base plate, (b) development of elephant’s foot at the tank bottom. ...................................... 80
Figure III.2.1-3: Elephant’s foot buckling, as obtained by finite element simulation. ............................ 80
Figure III.2.2-1: Von Mises stress distribution at maximum capacity (approximately 300mm
displacement)........................................................................................................................................... 81
Figure III.2.2-2: Von Mises stress distribution for the horizontal pressure vessel in the (a) longitudinal
x-direction (approx. 10 mm displacement) and (b) transverse z-direction (approx. 50 mm
displacement)........................................................................................................................................... 81
Figure III.3.1-1: API X52 from elbows cyclic properties at various imposed strain levels, longitudinal
direction (left) transverse direction (right) .............................................................................................. 82
Figure III.3.1-2: API X52 from elbows cyclic properties at various imposed stress levels, longitudinal
direction (left) transverse direction (right) .............................................................................................. 83
Figure III.3.1-3: P355 from elbows cyclic properties at various imposed stress levels .......................... 83
Figure III.3.1-4: Stress amplitude versus number of cycles until failure. ............................................... 84
Figure III.3.2-1: Load directions for monotonic loading......................................................................... 85
Figure III.3.2-2: P1-1, P2-1 and P3-1– Cyclic load-displacement curve and pictures of the specimens at
the end of the loading history (longitudinal load direction) .................................................................... 86
Figure III.3.2-3: P1-2, P2-2 and P3-2– Cyclic load-displacement curve and pictures of the specimens at
the end of the loading history (transverse load direction) ....................................................................... 87
Figure III.3.3-1: Strain gauges position mapping .................................................................................... 88
Figure III.3.3-2: Measured deflection at mid sections and rotation at load application points ............... 88
Figure III.3.3-3: Actuator load vs. displacement Test PN63-1 ............................................................... 89
Figure III.3.3-4: PN63-1 specimen after testing ...................................................................................... 89
134
Figure III.3.3-10: PN63-3 specimen after testing .................................................................................... 90
Figure III.3.3-11: Actuator load vs. displacement Test PN40-2 ............................................................. 90
Figure III.3.3-16: PN40-4specimen after testing ..................................................................................... 90
Figure III.3.3-17: (a) Designed non-standard BFJs and dimensions; (b) spiral wound gasket; (c) stud
bolt........................................................................................................................................................... 91
Figure III.3.3-18: (a) Bending test specimen, BS1; (b) axial test specimen, AS1; (c) a spiral wound
gasket; (d) a stud bolt. ............................................................................................................................. 91
Figure III.3.3-19: Test set-up for (a) bending tests; (b) axial tests. ......................................................... 92
Figure III.3.3-20: (a) Actuator load-displacement curve from test- BSML18; moment-rotation curves
from (b) test- BSCL18, (c) test- BSML27 and (d) test- BSCL27. .......................................................... 92
Figure III.3.3-21: Load-flange displacement curve from (a) test ASML18, (b) test ASCL18, (c) test
ASCL27-1 and (d) test ASCL18-2. ......................................................................................................... 93
Figure III.3.3-22: Failure of the pipe and bending of a bolt during test BSCL27 ................................... 93
Figure III.3.4-1: Cracked elbow specimen No. 4 with displacement range ±150 mm: Failure stage with
through thickness crack development ..................................................................................................... 94
Figure III.3.4-2: Experimental set-up: Instrumentation........................................................................... 94
Figure III.3.4-3: Increasing amplitude – Loading pattern according to ECCS recommendations. ......... 95
Figure III.3.4-4: Test 1 - l  25 mm Load – displacement cycles ...................................................... 95
Figure III.3.4-5: Test 2 - l  70 mm Load – displacement cycles ....................................................... 95
Figure III.3.4-6: Test 3 - l  100 mm Load – displacement cycles ..................................................... 95
Figure III.3.4-7: Test 4 - l  150 mm Load – displacement cycles .................................................... 95
Figure III.3.4-8: Test 4 - l  150 mm Flattening – displacement cycles ............................................. 95
Figure III.3.4-9: Test 5 - l  200 mm Load – displacement ............................................................... 96
Figure III.3.4-10: Test 6 - l  250 mm Load – displacement .............................................................. 96
Figure III.3.4-11: Test 7 - l  300 mm Load – displacement cycles ................................................... 96
Figure III.3.4-12: Test 7 - l  300 mm Flattening – displacement cycles ........................................... 96
Figure III.3.4-13: CSM FSTM in full scale testing machine................................................................... 96
Figure III.3.4-14: Measure of elbow displacements ............................................................................... 97
Figure III.3.4-15: Measure of ovalization ............................................................................................... 97
Figure III.3.4-16: Strain gauges installation ............................................................................................ 97
Figure III.3.4-17: Test 1 load displacement curve (left) and final crack position (right) ........................ 97
Figure III.3.4-22: Test 2 elbow ovalization ............................................................................................. 99
Figure III.3.4-23: Test 4 elbow ovalization ............................................................................................. 99
Figure III.3.4-24: (a) A 3D model of the piping system and support structure; (b) PS, NS and coupling
nodes...................................................................................................................................................... 101
Figure III.3.4-25: Schematic of the experimental set-up for the hybrid tests ........................................ 101
Figure III.3.4-26: Actual test set-up for the PDDS................................................................................ 101
Figure III.3.4-27: Some components of the test specimen .................................................................... 102
Figure III.3.4-28: Rotation histories and relevant spectra of the two end nodes of Elbow #1 at (a) SLVT
and (b) SLCT ......................................................................................................................................... 102
Figure III.3.4-29: Strain histories of Elbow #2 at (a) SLVT and (b) SLCT .......................................... 102
135
Figure III.3.4-30: Displacement histories and relevant spectra of the Coupling Node #1 at (a) SLVT and
(b) SLCT (Blue: Experimental; Magenta: Numerical) .......................................................................... 103
Figure III.3.5-1: Load vs. LPD curves for the T-joint specimens under monotonic loading. ............... 104
Figure III.3.5-2: Load vs. LPD curves under cyclic loading at an imposed LPD value of δ1=±60 mm:
(a) TJF, (b) TJWR specimen. ................................................................................................................ 104
Figure III.3.5-3: Load vs. LPD curves under cyclic loading at an imposed LPD value of δ2=±45 mm:
(a) TJF, (b) TJWR specimen. ................................................................................................................ 105
Figure III.3.5-4: Load vs. LPD curves under cyclic loading according to the ECCS No45 testing
procedure: (a) TJF, (b) TJWR specimen. .............................................................................................. 105
Figure III.3.5-5: Through-thickness crack failure of TJWR specimen at the weld toe of: (a) branch
region, (b) main pipe region at the end of the welded reinforcement plate. .......................................... 105
Figure III.3.5-6: Through-thickness crack failure in the region of the fitting for the TJF specimen
subjected to monotonic load. ................................................................................................................. 106
Figure III.4.1-1: Numerical parametric analysis results – 8-inch SCH40 long radius , Δr=±0.02 rad .. 106
Figure III.4.1-2: Numerical parametric analysis results – 8-inch SCH40 long radius , Δr=±0.04 rad .. 106
Figure III.4.1-3: Numerical parametric analysis results – 8-inch SCH40 long radius, Δr=±0.08 rad .. 106
Figure III.4.1-4: Numerical parametric analysis results – 8-inch SCH40 long radius, Δr=±0.12 rad .. 106
Figure III.4.1-5: Moment versus end-rotation curve for displacement range ±0.25 rad, (for pressure 0
MPa, 3.2 MPa and 7.0 MPa). ................................................................................................................ 108
Figure III.4.1-6: Moment versus end-rotation curve for displacement range ±0.4 rad, (for pressure 0
MPa, 3.2 MPa and 7.0 MPa). ................................................................................................................ 108
Figure III.4.1-7: Moment versus end-rotation curve for displacement ranges ±0.05, ±0.1, ±0.25 and ±0.4
rad, (pressure 0 MPa). ........................................................................................................................... 108
Figure III.4.1-8: typical deformed shape and plastic strain map of the flanged joints after bending .... 109
Figure III.4.1-9: Moment vs. rotation curves for monotonic bending of PN40 flanged connections ... 109
Figure III.4.1-10: Moment vs. rotation curves for monotonic bending of PN63 flanged connections . 109
Figure III.4.1-11: Moment vs. rotation curves for monotonic bending of PN100 flanged connections 110
Figure III.4.1-12: Effect of pressure for all flange geometries .............................................................. 110
Figure III.4.1-13: Effect of axial load for all flange geometries ........................................................... 110
Figure III.4.1-14: Stress distribution for the bending test: (a) 18 mm thickness joint, and (b) 36 mm
thickness joint ........................................................................................................................................ 111
Figure III.4.1-15: Stress distribution for the axial test: (a) 18 mm thickness joint, and (b) 36 mm
thickness joint ........................................................................................................................................ 111
Figure III.4.1-16: Critical elements for the bending test: (a) pipe, (b) flange, and (c) bolt ................... 111
Figure III.4.1-17: Critical elements for the axial test: (a) pipe, (b) flange, and (c) bolt ........................ 112
Figure III.4.1-18: Tee-junction capacity in (a) out-of-plane bending and (b) in plane bending. .......... 112
Figure III.4.1-19: Tee-junction capacity in axial loading (a) pulling branch away from the main pipe and
(b) pushing branch towards the pipe. .................................................................................................... 112
Figure III.4.1-20: Variation of yield and ultimate strength in for the 3 branch-to-pipe diameters in
bending. ................................................................................................................................................. 112
Figure III.4.1-21: Comparison of out-of-plane and in-plane strength for the different pad geometries.113
Figure III.5.1-1: Finite element analysis of unanchored liquid storage tank under lateral loading; (a)
uplifting of base plate, (b) development of elephant’s foot at the tank bottom. .................................... 114
Figure III.5.1-2: Top horizontal displacement u and maximum vertical uplift of Tank I versus
overturning moment M , subjected to finite element pushover analysis. .............................................. 114
Figure III.5.1-3: Ratio of maximum compressive axial membrane forces for the unanchored over the
anchored tank versus the overturning moment; finite element results for Tank I and curves for
  H / R  0.8, 1.0, 2.0, adopted by EN 1998-4. .................................................................................. 114
Figure III.5.1-4: Overturning moment in terms of the rotation of the tank. ......................................... 115
Figure III.5.1-5: Length of uplifted part of the base as a function of the vertical uplift at the edge and
bilinear approximation to be used later in the dynamic analysis. The equations of the two lines are
shown on the graph................................................................................................................................ 115
Figure III.5.1-6: Simplified models for anchored and unanchored liquid storage tank......................... 115
Figure III.5.1-7: Time history of the uplifting axial membrane force in the case of (a) the anchored Tank
I and (b) the unanchored Tank I for the earthquake in Duzce, Turkey (1999). ..................................... 116
136
Figure III.5.1-8: Development of elephant’s foot, as predicted by the finite element analysis. ........... 116
Figure III.6.1-1: Base shear versus roof displacement – pushover analysis results for the VV No1 .... 117
Figure III.6.1-2: Equivalent plastic strain at maximum capacity for VVs No1 and 2 ........................... 117
Figure III.6.1-3: Comparison of base shear vs drift between pushover and IDA results for the VV No3
............................................................................................................................................................... 118
Figure III.6.2-1: Base shear versus roof displacement for all horizontal vessels with internal pressure –
pushover analysis results (a) in the longitudinal and (b) in the transverse direction. ............................ 119
Figure III.6.2-2: PEEQ distribution for the Horizontal Vessel No1 with internal pressure – (a) loading in
the longitudinal and (b) in the transverse direction ............................................................................... 119
Figure III.6.2-3: Effect of internal pressure on base shear versus roof displacement for HV2– pushover
analysis results (a) in the longitudinal and (b) in the transverse direction ............................................ 120
Figure III.6.3-1: Pushover analysis results in terms of force-displacement curve referred to the centre of
gravity for the non-braced system (left) and the braced system (right). ................................................ 120
Figure III.6.3-2: Pushover analysis results in terms of deformed shape of critical details showing failure
modes and equivalent plastic strains PEEQ. ......................................................................................... 121
Figure III.6.3-3: Maximum lateral displacements versus seismic level for all ten accelerograms for (a)
the non-braced supporting structure and (b) the braced supporting structure ....................................... 122
Figure III.7.3-1: Horizontal design spectrum data – EN 1998-1 vs ASCE 7 ........................................ 126
Figure III.7.3-2: Configuration of 10” ammonia transmission line. ...................................................... 126
Figure III.7.3-3: Horizontal and vertical seismic design spectra according EN 1998-1 and ASCE-7, with
natural frequency range (blue shaded) of above ground cross country ammonia pipeline system ....... 127
Figure III.7.4-1: Piping system on a pipe-rack ...................................................................................... 128
Figure III.7.4-2: Possible positions of pipes .......................................................................................... 128
Figure III.7.4-3: Variation of the moment in the pipes for the coupling effect ..................................... 128
Figure III.7.4-4: Deformation of substructure ....................................................................................... 128
Figure III.7.5-1: Increasing amplitude – Loading pattern according to ECCS recommendations. ....... 129
Figure III.7.5-2: Limit state for strong cyclic loading, as taken up in the design guideline .................. 129
137
VI. List of tables
Table II.3-1: Summary of experimental tests……………………………………………………….26
Table II.3.1-1: Small scale testing performed to support full scale activity and FEA ............................ 27
Table II.3.2-1: Number of cycles until failure Nf for longitudinal and transverse load direction ........... 28
Table II.3.3-1: CSM test program on standard bolted flanged joints and main results / observations .... 30
Table II.3.4-1: Displacement range and number of cycles ...................................................................... 32
Table II.3.4-2: Specimen properties ........................................................................................................ 33
Table II.3.4-3: Full scale test program .................................................................................................... 34
Table II.3.4-4: Summary of Elbows test results performed by CSM ...................................................... 34
Table II.3.5-1: Test results. ..................................................................................................................... 36
Table II.4.1-1: Material model parameter values .................................................................................... 40
Table II.4.2-1: Displacement range, number of cycles and local strain range. ....................................... 43
Table II.4.2-2: Displacement range and number of cycles ...................................................................... 44
Table II.4.2-3: Experimental and numerical results for Nf and Pf ........................................................... 49
Table II.4.2-4: Error between experimental results and numerical simulations calculated with NRMSD
and RMSD methods ................................................................................................................................ 52
Table II.4.2-5: Experiments results ......................................................................................................... 53
Table II.4.3-1: Geometrical characteristics of examined cases ............................................................... 55
Table II.4.3-2: Geometrical parameters for the parametric study ........................................................... 57
Table II.4.3-3: Parametric numerical results for cyclic loading in longitudinal load direction ............... 58
Table II.4.3-4: Parametric numerical results for cyclic loading in transverse load direction .................. 58
Table II.4.3-5: Flanged joints considered in parametric study ................................................................ 60
Table II.4.3-6: Load cases analyzed for each geometry .......................................................................... 60
Table II.4.3-7: Fracture load for bending tests with SMCS criterion ...................................................... 62
Table II.4.3-8: Fracture load for bending tests with SMCS criterion ...................................................... 62
Table III.3.1-1: Monotonic tensile properties of API X52 specimen extracted from elbow ................... 82
Table III.3.1-2: Monotonic tensile properties of API5L X52 specimens extracted from pipes to be used
in elbow piping specimens ...................................................................................................................... 82
Table III.3.1-3: Summary of monotonic tensile properties of P355N extracted from straight pipes ...... 82
Table III.3.1-4: Summary of monotonic tensile properties of P355 extracted from plates ..................... 83
Table III.3.1-5: Summary of monotonic tensile properties of the base material. .................................... 84
Table III.3.2-1: Yield loads and corresponding displacements of the internal-, external- and thickened
nozzle reinforcement for different load directions .................................................................................. 85
Table III.3.3-1: Dimensions of the designed non-standard BFJs (mm) .................................................. 91
Table III.3.3-2: Test program on non-standard BFJ carried out by UTrento ......................................... 91
Table III.3.3-3: Comparison between experimental leakage moment and allowable moments suggested
by codes ................................................................................................................................................... 94
Table III.3.3-4: Comparison between experimental leakage load and allowable loads suggested by
codes ........................................................................................................................................................ 94
Table III.3.3-5: Maximum moment, Mmax, axial force, Nmax and shear force, τmax in the piping
system obtained from a case study .......................................................................................................... 94
Table III.3.4-1:FSTM main capabilities .................................................................................................. 96
Table III.3.4-2: Experimental program on a piping system carried out by UTrento ............................. 100
Table III.3.4-3: PGA corresponding to different limit states ................................................................ 100
Table III.3.4-4: Maximum moment, Mmax stress, σmax and shear, τmax in the piping system under
reference seismic loading ...................................................................................................................... 100
Table III.4.1-1: Parametric analysis results: 8-inch SCH20 long radius ............................................... 107
Table III.4.1-2: Parametric analysis results: 8-inch SCH40 long radius ............................................... 107
Table III.4.1-3: Parametric analysis results: 8-inch SCH80 long radius .............................................. 107
Table III.4.1-4: Parametric analysis results: 8-inch SCH40 short radius .............................................. 107
Table III.4.1-5: Parametric analysis results: 8-inch SCH20 long and short radius. .............................. 108
Table III.4.1-6: Parametric analysis results: 8-inch SCH40 long and short radius. .............................. 108
Table III.4.1-7: Parametric analysis results: 8-inch SCH20 long radius. .............................................. 108
Table III.4.1-8: Summary of buckling moment and rotation for monotonic behavior. ......................... 110
Table III.6.1-1: Estimated q factor by IDA for the vertical pressure vessels ........................................ 118
139
Table III.6.3-1: Most critical as well as mean behaviour factor for all failure criteria for the non-braced
supporting structure ............................................................................................................................... 122
Table III.6.3-2: Most critical as well as mean behaviour factor for all failure criteria for the braced
supporting structure ............................................................................................................................... 123
Table III.7.5-1: Displacement range and number of cycles .................................................................. 129
140
VII. References
[1] ABAQUS Theory Manual, Vers. 6.11-2, 2011a
[2] ABAQUS User’s Manual, Vers. 6.11-2, 2011b
[3] American Petroleum Institute (2007), Seismic Design of Storage Tanks - Appendix E, Welded
Steel Tanks for Oil Storage, API 650, 11th Edition, Washington, D.C.
[4] American Society of Mechanical Engineers (2010), Boiler and Pressure Vessel Code Section III:
Rules for Construction of Nuclear Facility Components, Mandatory Appendix I.
[5] ASCE 7-05, Minimum Design Loads for Buildings and Other Structures, American Society of
Civil Engineers.
[6] ASME B16.5, 2003, Pipe Flanges and Flanged Fittings.
[7] Ballio, G., Mazzolani F.M., Plumier, A., Sedlacek, G., “Specific rules for steel structures”,
Background Documents for Eurocode 8 part 1, Volume 2 – Design rules, 1988
[8] Basic seismic structural design of typical piping system, Annual Report of INDUSE project,
Deliverable D1.3c, March 2011
[9] Bursi, O. S., Buelga, A. G., Vulcan, L., Neild, S. A., Wagg, D. J., 2008, “Novel Coupling
Rosenbrock-Based Algorithms for Real-Time Dynamic Substructure Testing,” Earthquake
Engineering and Structural Dynamics, 37:339360.
[10] Bursi, O.S., and Jaspart, J.P., Benchmarks for Finite Element Modelling of Bolted Steel
Connection, J. Construct. Steel Res., Vol 43, Nos. 1-3, pp. 17-42, 1997
[11] Bursi, O.S., and Jaspart, J.P., Calibration of a Finite Element Model for Isolated Bolted End-Plate
Steel Connections, J. Construct. Steel Res., Vol 44, No. 3, pp. 225-262, 1997
[12] Bursi, O.S., Reza, Md.S., and Kumar, A., Experimental tests on bolted flange joints - Final
report, February, 2012.
[13] Centro Sviluppo Materiali, (2011). Cyclic loading on P355N steel grade material coupons; CSM
Internal Report, INDUSE RFCS project.
[14] Chen W., Haroun M. A., and Liu F. (1996), “Large amplitude liquid sloshing in seismically
excited tanks”, Earthquake Engineering & Structural Dynamics, Vol. 25, No. 7, pp. 653-669.
[15] Clough, D. P., “Experimental Evaluation of Seismic Design Methods for Broad Cylindrical
Tanks”, University of California Earthquake Engineering Research Center, Report, No.
UC/EERC 77-10, May 1977.
[16] Coelho, A.M.G., “Rotation capacity of partial strength steel connections with three-dimensional
finite element approach”, Computers and Structures, Vol. 126, pp. 88-97, 2012
[17] Comité Européen de Normalization (2006), Silos, tanks and pipelines, Eurocode 8, part 4,
CEN/TC 250, EN-1998-4, Brussels.
[18] Craig, R., Bampton M., 1968, “Coupling of Substructures for Dynamic Analysis,” AIAA Journal,
Vo1.6, (7).
[19] Diamanti, K., Doukas, I., and Karamanos, S.A., “Seismic Analysis and Design of Industrial
Pressure Vessels”, COMPDYN 2011,Corfu, Greece, 26–28 May 2011
[20] ECCS “Recommended Testing Procedure for Assessing the Behaviour of Structural Steel
Elements under Cyclic Loading”, ECCS - Technical Committee 1 - Structural Safety and
Loadings, Technical Working Group 1.3 - Seismic Design, Brussels 1986, Belgium.
[21] EN 10216-3, “Seamless steel tubes for pressure purposes – Technical delivery conditions”
[22] EN 10253-2, “Butt-welding pipe fittings – Part 2: Non alloy and ferritic alloy steels with specific
inspection”, CEN, European Committee for Standardization, Brussels, 2008
[23] EN 1092-1:2007, Flanges and their joints – Circular flanges for pipes, valves, fittings and
accessories, PN designated – Part 1: Steel flanges.
[24] EN 13445-3:2010-12; “Unfired pressure vessels - Part 3: Design”, European Committee for
Standardization, Brussels, 2010.
[25] EN 13480-3, “Metallic Industrial Piping – Part 3: Design and Calculation”, CEN, European
Committee for Standardization, Brussels, 2011
141
[26] EN 14015:2004-11; “Specification for the design and manufacture of site built, vertical,
cylindrical, flat-bottomed, above ground, welded, steel tanks for the storage of liquids at ambient
temperature and above”, European Committee for Standardization, Brussels 2004.
[27] EN 15614.01, “Specification and qualification of welding procedures for metallic materials -
Welding procedure test - Part 1: Arc and gas welding of steels and arc welding of nickel and
nickel alloys”, CEN, European Committee for Standardization, Brussels, 2004.
[28] EN 1591-1, 2009. Flanges and their joints - Design rules for gasketed circular flange connections
- Part 1: Calculation method
[29] EN 1993-1-8, 2005. Eurocode 3: Design of steel structures - Part 1-8: Design of joints.
[30] EN 1998-1:2004 (E), Eurocode 8: Design of structures for earthquake resistance – Part 1:
General rules, seismic actions and rules for building.
[31] EN 1998-1:2010-12: “Design of structures for earthquake resistance – Part 1: General rules,
seismic actions and rules for buildings”, European Committee for Standardization, Brussels, 2010
[32] EN 1998-4:2006, Eurocode 8: Design of structures for earthquake resistance – Part 4: silos, tanks
and pipelines.
[33] EN ISO 3183 2007, “Petroleum and natural gas industries — Steel pipe for pipeline
transportation
[34] EN ISO 6892.01, “Metallic materials - Tensile testing - Part 1: Method of test at room
temperature”, CEN, European Committee for Standardization, Brussels, 2010.
[35] Federal Emergency Management Agency (FEMA), Prestandard and Commentary for the
Rehabilitation of Buildings (FEMA 356), Washington D.C., 2000
[36] G. C. Manos and R. W. Clough, ‘Further study of the earthquake response of broad cylindrical
liquid storage tank model’, Report No. UCB/EERC-82/07, Earthquake Engineering Research
Center, University of California, Berkeley, CA, 1983.
[37] G. C. Manos, ‘Earthquake tank-wall stability of unanchored tanks’, Journal of Structural
Engineering, ASCE 112, 1863-1880, 1986.
[38] Gresnigt, A. M., “Plastic Design of Buried Steel Pipelines in Settlement Areas”, Heron, 31 (4),
Delft, The Netherlands, 1986.
[39] Hancock, J.W., and Mackenzie, A.C., “On the mechanisms of ductile failure in high-strength
steels subjected to multi-axial stress-state”, J. Mech. Phys. Solid 24, pp. 147-169, 1976.
[40] Kanvinde, A.M., and Deielein, G.G., “Micromechanical Simulation of Earthquake-Induced
Fracture in Steel Structures”, The John A. Blume Earthquake Engineering Centre, Report
No.145, July, 2004.
[41] Katsikogiannis, G. (2013). “Numerical Simulation of Liquid Container Dynamics,” Diploma
Thesis, Mechanical Engineering, University of Thessaly, Volos, Greece.
[42] Lemaitre, J. and Chaboche, J-L.: “Mechanics of Solid Materials”, Cambridge University
Press, 1990.
[43] M. De Angelis, R. Giannini and F. Paolacci, “Experimental investigation on the seismic response
of a steel liquid storage tank equipped with floating roof by shaking table tests.”, Earthquake
Engineering and Structural Dynamics (2009).
[44] MTS Models FlexTest 40/60/100/200 Controller Hardware, 2008, MTS Systems Corporation.
[45] Natsiavas, S. and Babcock, C. D. (1987), “Buckling at the Top of a Fluid-Filled Tank during
Base Excitation”, ASME Journal of Pressure Vessel Technology, Vol. 109, No. 4, pp. 374-380.
[46] Niwa, A. and Clough, R. W., 1982, "Buckling of Cylindrical Liquid-Storage Tanks under
Earthquake Loading," Earthquake Engineering and Structural Dynamics, 10, pp. 107-122.
[47] Paolacci, F., Reza, Md. S., Bursi, O. S., (2011). Seismic design criteria of refinery piping systems.
COMPDYN 2011 -III ECCOMAS Thematic Conference on Computational Methods in Structural
Dynamics and Earthquake Engineering, Corfu, Greece, 26–28 May 2011.
[48] Research program: “Shakinng Table Testing of Fixed-Base and Seismically Isolated Liquid
Storage Tanks” by S. Mahin and V. Calugaru, University of California Berkeley, 2009.
142
[49] Rice, J.R., and Tracey, D.M., “On the ductile enlargement of voids in triaxial stress fields”, J.
Mech. Phys. Solid 17, pp. 201-217, 1969.
[50] SAP2000, 2004, “Linear and Nonlinear Static and Dynamic Analysis and Design of Three-
Dimensional Structures,” v9, Computers and Structures, Inc., Berkeley, California, USA.
[51] Scharf, K. (1990), “Beiträge zur Erfassung des Verhaltens von erdbebenerregten, oberirdischen
Tankbauwerken”, Fortschritt-Berichte VDI, Reihe 4. Bauingenieurwesen, Nr. 97, VDI Verlag,
Düsseldorf
[52] Seismic forces in industrial piping systems, Annual Report of INDUSE project Deliverable D2.1c,
March 2011
[53] Tseng, N. T., Lee, G. C., (1983).” Simple plasticity model of the two-surface type”. ASCE, J. of
Engineering Mechanics 109, 795-810.
[54] Veletsos, A. S. and Yang, J. Y. (1977), “Earthquake Response of Liquid Storage Tanks”, 2nd
ASCE Engineering Mechanics Conference, Raleigh, NC, pp. 1-24
[55] Wozniak, R. S. and Mitchell, W. W. (1978), “Basis of seismic design provisions for welded steel
oil storage tanks”, Advances in Storage Tank Design, API 43rd mid-year meeting, Toronto,
Canada.
143
European Commission
EUR 26319 — Structural safety of industrial steel tanks, pressure vessels and piping systems under
seismic loading (INDUSE)
Luxembourg: Publications Office of the European Union
2013 — 143 pp. — 21 × 29.7 cm
ISBN 978-92-79-34575-3
doi:10.2777/49423
145
HOW TO OBTAIN EU PUBLICATIONS
Free publications:
• one copy:
via EU Bookshop (http://bookshop.europa.eu);
• more than one copy or posters/maps:
from the European Union’s representations (http://ec.europa.eu/represent_en.htm);
from the delegations in non-EU countries (http://eeas.europa.eu/delegations/index_en.htm);
by contacting the Europe Direct service (http://europa.eu/europedirect/index_en.htm) or
calling 00 800 6 7 8 9 10 11 (freephone number from anywhere in the EU) (*).
(*) The information given is free, as are most calls (though some operators, phone boxes or hotels may charge you).
Priced publications:
• via EU Bookshop (http://bookshop.europa.eu).
Priced subscriptions:
• via one of the sales agents of the Publications Office of the European Union
(http://publications.europa.eu/others/agents/index_en.htm).
KI-NA-26319-EN-N
The INDUSE project combines seismic engineering concepts with mechanical
engineering practice, and is aimed at developing design guidelines/
recommendations for safeguarding structural integrity of industrial tanks,
pressure vessels and piping under strong seismic action. The guidelines improve
and extend current seismic design practice, considering the particularities of
those industrial systems. The following intermediate targets have been achieved
within the INDUSE project:
— A basic comparison has been performed between the currently available
seismic design provisions in European and American standards (WP1),
considering specific design examples.
— Seismic actions on those industrial systems have been determined (WP2),
taking into account the dynamic behaviour of each structural system.
— Critical piping components (nozzles, flanged connections, Tees and elbows)
have been tested under strong cyclic loading (WP3). Pseudo-dynamic testing
of a piping system has also been performed.
— Extensive numerical studies on critical piping components have been
conducted, based on rigorous FE simulations, covering a wide range of
geometric, material and loading parameters (WP4).
— Design guidelines/recommendations have been developed, expanding
EN 1998-4 provisions towards an integrated seismic design of liquid
storage tanks (WP5) and extending the applicability of Eurocode 8 (EN
1998) concepts for the cases of industrial pressure vessels and attached
equipment (WP6), and piping systems including their support systems (WP7).
The results and deliverables of INDUSE are both novel and unique, resulting
in better understanding of structural behaviour of tanks, pressure vessels and
piping under strong cyclic (seismic) loading, more reliable definition of ultimate
limit states for critical industrial components, significant improvement of
seismic design state of the art and offering the necessary scientific background
for Code Drafting Committees, towards safer design of industrial power plants
and chemical/petrochemical facilities.
Studies and reports
doi:10.2777/49423

INDUSE Report PDF

Uploaded by

Copyright:

Available Formats

INDUSE Report 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

INDUSE Report PDF

Uploaded by

Copyright:

Available Formats

Structural safety of

industrial steel tanks, pressure vessels

Contact: RFCS Publications

Research Fund for Coal and Steel

P. Pappa, G. E. Varelis, M. Vathi, P. C. Perdikaris, S. A. Karamanos

Grant Agreement RFSR-CT-2009-00022

2013 EUR 26319 EN

Europe Direct is a service to help you find answers

Freephone number (*):

More information on the European Union is available on the Internet (http://europa.eu).

Cataloguing data can be found at the end of this publication.

Luxembourg: Publications Office of the European Union, 2013

© European Union, 2013

Printed on white chlorine-free paper

I.2. WP2: Determination of seismic forces on industrial structures and critical

I.3. WP3: Experimental testing of structural components of critical importance

I.4. WP4: Numerical simulation of industrial component behaviour under strong

I.6. WP6: Seismic behaviour and design of industrial pressure vessels;

I.9. Conclusions and main results

Comparison of initially planned activities and work accomplished

II.1. Work package 1 – Comparison of European and American design standards

II.1.1.2. Experiments on liquid storage tanks

II.1.2. Task 1.2: Collection and evaluation of reported earthquake damages in

II.2.2. Task 2.2: Seismic forces in pressure vessels

II.2.3. Task 2.3: Seismic forces in industrial piping systems

II.2.5. Task 2.5 Determination of loading amplitudes on critical components

II.3. Work package 3 – Experimental testing of structural components of critical

II.3.1. Task 3.1: Basic material testing and characterization

Number of corresponding full-

Figure II.3.3-1: Weld neck flange geometry

Plastic No Little plastic No

Constant Wrinkling/ No Little plastic No

Test Displacement Number of cycles Local strain

II.3.4.2. Tests on pressurized piping elbows under strong cyclic loading

D = 219.1 mm 8.625 inch

II.3.4.3. Pseudo-dynamic loading on piping system

Table II.3.5-1: Test results.

II.4. Work package 4 – Numerical simulation of industrial components behaviour

II.4.1.2. The Armstrong-Frederick model and calibration

curve is fitted following the equation:

II.4.1.3. Tseng-Lee model and calibration

II.4.2. Task 4.2 Numerical simulation of the experimental tests on components of

1 ±25 13160 0.33 0.46 0.33

Table II.4.2-2: Displacement range and number of cycles

Numerically predicted values - NLKH model

Numerically predicted values - TL model

II.4.2.2. Numerical simulation of tank nozzles

II.4.2.3. Numerical simulation of flanged joints

II.4.2.3.1 Numerical simulation of standard flanged joints

Figure II.4.2-16: Mesh details

True Stress [MPa]

(a) (b) (c)

Test Model NRMSD (-) RMSD (kN)

Table II.4.3-1: Geometrical characteristics of examined cases

II.4.3.1.1. In-plane bending

II.4.3.1.2. Out-of-plane bending

II.4.3.2. Parametric numerical study of tank nozzles

Table II.4.3-2: Geometrical parameters for the parametric study

II.4.3.3. Parametric numerical study of flanged joints