Sei sulla pagina 1di 283

Z_SOIL.

PC 2003
USER MANUAL
THEORY
Copyright 1985-2003
Zace Services Ltd, Software engineering
P.O.Box 2, 1015 Lausanne, Switzerland
Tel.+41 21 802 46 05, Fax 802 46 06
http://www.zace.com, Hotline zsoil@zace.com
Soil, Rock and Structural Mechanics
in dry or partially saturated media

since 1985
GENEVA
SWEDEN
IRAN
LOETSCHBERG
since 1982

END-USER LICENSE AGREEMENT FOR ZACE's Z_SOIL.PC SOFTWARE
______________________________________________________________________________________
Read carefully this document, it is a legal agreement between you and Zace Services Ltd for the
software product identified above. By installing, copying, or otherwise using the software product
identified above, you agree to be bound by the terms of this agreement. If you do not agree to the
terms of this agreement, promptly return the unused software product to the place from which you
obtained it for full refund (less shipping).
ZACE SERVICES LTD OFFERS A 60 DAYS MONEY-BACK GUARANTEE ON Z_SOIL.PC.
______________________________________________________________________________________
Z_SOIL.PC (the Software) SOFTWARE PRODUCT LICENSE
The software Z_Soil.PC is protected by copyright laws and international copyright treaties, as well as other intellectual property
laws and treaties. The Z_Soil.PC software product is licensed, not sold.
1. GRANT OF LICENSE
a. Zace Services Ltd grants you, the customer, a non-exclusive license to use copies of Z_Soil.PC. you may install copies of
Z_ Soil.PC on an unlimited number of computers, provided that you use only one copy at the time.
b. You may make an unlimited number of copies of documents accompanying Z_Soil.PC, provided that such copies shall be
used only for internal purposes and are not republished or distributed to any third party.
2. COPYRIGHT
All title and copyrights in and to the Software product (including but not limited to images, photographs, text, applets, etc), the
accompanying materials, and any copies of Z_Soil.PC are owned by Zace Services Ltd. Z_Soil.PC is protected by copyright
laws and international treaties provisions. Therefore, you must treat Z_Soil.PC like any other copyrighted material except that
you may make copies of the software for backup or archival purposes or install the software as stipulated under 1.
3. OTHER RIGHTS AND LIMITATIONS
Limitations on Reverse Engineering, Decompilation, Disassembly. Yo u may not reverse engineer, decompile, or disassemble the
Software.
a. No separation of components. Z_Soil.PC is licensed as a single product and neither the Software's components, nor any
upgrade may be separated for use by more than one user at the time.
b. Rental. You may not rent or lease the software product.
c. Software transfer. You may permanently transfer all of your rights under this agreement, provided you do not retain any
copies, and the recepient agrees to all the terms of this agreement.
d. Termination. Without prejudice to any other rights, Zace Services Ltd may terminate this agreement if you fail to comply
with the conditions of this agreement. In such event, you must destroy all copies of the Software.
______________________________________________________________________________________
LIMITED WARRANTY
Zace Services Ltd. warrants that Z_Soil.PC will a)perform substantially in accordance with the accompanying written material
for a period of 90 days from the date of receipt, and b) any hardware accompanying the product will be free from defects in
materials and workmanship under normal use and service for a period of one year, from the date of receipt.
CUSTOMER REMEDIES
Zace Services Ltd entire liability and your exclusive remedy shall be at Zace's option, either a)return of the price paid, or b)
repair or replacement of the software or hardware component which does not meet Zace's limited warranty, and which is returned
to Zace Services Ltd, with a copy of proof of payment of Z_Soil.PC. This limited warranty is void if failure of the Software or
hardware component has resulted from accident, abuse, or misapplication. Any replacement of software or hardware will be
warranted for the remainder of the original warranty period or 30 days, whichever is longer.
______________________________________________________________________________________
NO OTHER WARRANTIES.
TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, ZACE SERVICES LTD DISCLAIMS ALL OTHER
WARRANTIES, EITHER EXPRESS OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, IMPLIED WARRANTIES OF
MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE, WITH REGARD TO THE SOFTWARE
PRODUCT, AND ANY ACCOMPANYING HARDWARE.
NO LIABILITY FOR CONSEQUENTIAL DAMAGES.
TO THE MAXIMUM EXTENT PERMITTED BY LAW, IN NO EVENT SHALL ZACE SERVICES LTD BE LIABLE FOR
ANY SPECIAL INCIDENTAL, INDIRECT, OR CONSEQUENTIAL DAMAGES WHATSOEVER (INCLUDING, WITHOUT
LIMITATION, DAMAGES FOR LOSS OF BUSINESS, PROFITS, BUSINESS INTERRUPTION, LOSS OF BUSINESS
INFORMATION, OR ANY OTHER PECUNIARY LOSS) ARISING OUT OF THE USE OF OR INABILITY TO USE THE
SOFTWARE PRODUCT, EVEN IF ZACE SERVICES LTD HAS BEEN ADVISED OF THE POSSIBITY OF SUCH
DAMAGES.
______________________________________________________________________________________
HOTLINE: During the first year following purchase, hotline assistance will be provided by Zace Services Ltd, by fax or E-
mail exclusively. This service excludes all forms of consulting on actual projects. This hotline assistance can be renewed, for
following years, at a cost of 10% of current full package price.
THIS AGREEMENT IS GOVERNED BY THE LAWS OF SWITZERLAND, LAUSANNE 3.1.2003







WARNING


Z_Soil.PC is regularly updated for minor changes .We recommend that you
send us your e_mail, as Z_Soil owner, so that we can inform you of latest
changes. Otherwise, consult our site regularly and download free upgrades
to your version.
Latest updates to the manual are always included in the online help, so that
slight differences with your printed manual will appear with time; always
refer to the online manual for latest version, in case of doubt.

HOTLINE: zsoil@zace.com

Upgrades: http://www.zace.com/zsoil.htm
































January 2003
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM2
Contents
1 INTRODUCTION 11
1.1 NOTATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1:2
1.2 SOME IMPORTANT FORMULAE IN TENSOR ALGEBRA AND ANALYSIS 1.2:2
2 PROBLEM STATEMENT 2.0:11
2.1 SINGLE PHASE, SOLID MEDIUM . . . . . . . . . . . . . . . . . . . . . . 2.1:2
2.2 TWO-PHASE PARTIALLY SATURATED MEDIUM . . . . . . . . . . . . . 2.2:2
2.3 TRANSIENT FLOW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3:2
2.4 HEAT TRANSFER . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4:2
2.5 HUMIDITY TRANSFER . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5:2
3 MATERIAL MODELS 3.0:4
3.1 ELASTICITY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1:2
3.2 CONSOLIDATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2:2
3.2.1 GENERALIZED DARCY LAW . . . . . . . . . . . . . . . . . . . . 3.2.1:2
3.2.2 FLUID MOTION . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.2:2
3.2.3 OEDOMETRIC TEST . . . . . . . . . . . . . . . . . . . . . . . . 3.2.3:2
3.3 PLASTICITY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3:2
3.3.1 SKETCH OF THE PLASTICITY APPROACH . . . . . . . . . . . . 3.3.1:2
3.3.2 MOHRCOULOMB CRITERION . . . . . . . . . . . . . . . . . . . 3.3.2:2
3.3.3 DRUCKER-PRAGER CRITERION . . . . . . . . . . . . . . . . . . . 3.3.3:2
3.3.4 CAP MODEL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.4:2
3.3.5 MOHR-COULOMB (M-W) . . . . . . . . . . . . . . . . . . . . . . 3.3.5:2
3.3.6 HOEKBROWN CRITERION (SMOOTH) . . . . . . . . . . . . . . 3.3.6:2
3.3.7 CUT-OFF CONDITION AND TREATMENT OF THE APEX . . . . 3.3.7:2
3.3.8 MULTILAMINATE MODEL . . . . . . . . . . . . . . . . . . . . . 3.3.8:2
3.3.9 MODIFIED CAM CLAY MODEL . . . . . . . . . . . . . . . . . . . 3.3.9:2
3.4 CREEP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4:2
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3
3.4.1 CREEP UNDER VARIABLE STRESS . . . . . . . . . . . . . . . . 3.4.1:2
3.4.2 CREEP PARAMETER IDENTIFICATION FROM EXPERIMENTS . 3.4.2:2
3.5 SWELLING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5:2
3.6 AGING CONCRETE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.6:2
3.7 APPENDICES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.7:2
3.7.1 SAFETY FACTORS AND STRESS LEVELS . . . . . . . . . . . . . 3.7.1:2
4 NUMERICAL IMPLEMENTATION 4.0.1:3
4.1 WEAK FORM AND MATRIX FORMS OF THE PROBLEM . . . . . . . . 4.1:2
4.1.1 SINGLE PHASE MEDIUM, TIME INDEPENDENT LOADING . . . . 4.1.1:2
4.1.2 TWO-PHASE MEDIUM, RHEOLOGICAL BEHAVIOUR . . . . . . . 4.1.2:2
4.1.2.1 PRESSURE STABILIZED FINITE FORMULATION FOR PAR-
TIALLY SATURATED TWO-PHASE MEDIA . . . . . . . 4.1.2.1:2
4.1.3 HEAT TRANSFER . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.3:2
4.2 ELEMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2:2
4.2.1 FINITE ELEMENTS FOR 2D/3D CONTINUUM PROBLEMS . . . . 4.2.1:2
4.2.2 NUMERICAL INTEGRATION . . . . . . . . . . . . . . . . . . . . 4.2.2:2
4.2.3 STRAINS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.3:2
4.2.4 STIFFNESS MATRIX . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.4:2
4.2.5 BODY FORCES AND DISTRIBUTED LOADS . . . . . . . . . . . 4.2.5:2
4.2.6 INITIAL STRESSES, STRAINS . . . . . . . . . . . . . . . . . . . . 4.2.6:2
4.3 INCOMPRESSIBLE AND DILATANT MEDIA . . . . . . . . . . . . . . . . 4.3:2
4.3.1 INCOMPRESSIBLE MEDIA : B-BAR STRAIN PROJECTION METHOD4.3.1:2
4.3.2 DILATANT MEDIA: ENHANCED ASSUMED STRAIN METHOD . . 4.3.2:2
4.3.2.1 INTRODUCTION TO ENHANCED ASSSUMED STRAIN
(EAS) APPROACH . . . . . . . . . . . . . . . . . . . . . 4.3.2.1:2
4.3.2.2 EXTENSION OF THE EAS METHOD TO NONLINEAR
ELASTOPLASTIC ANALYSIS . . . . . . . . . . . . . . . 4.3.2.2:2
4.3.2.3 REMARKS AND ASSESSMENT OF EAS ELEMENTS . . 4.3.2.3:2
4.4 FAR FIELD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4:2
4.5 OVERLAID MESHES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5:2
4.6 ALGORITHMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6:2
4.6.1 FULL/MODIFIED NEWTON-RAPHSON ALGORITHM . . . . . . . 4.6.1:2
4.6.2 CONVERGENCE NORMS . . . . . . . . . . . . . . . . . . . . . . 4.6.2:2
4.6.3 INITIAL STATE ANALYSIS . . . . . . . . . . . . . . . . . . . . . 4.6.3:2
4.6.4 STABILITY ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . 4.6.4:2
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4
4.6.5 ULTIMATE LOAD ANALYSIS . . . . . . . . . . . . . . . . . . . . 4.6.5:2
4.6.6 CONSOLIDATION ANALYSIS . . . . . . . . . . . . . . . . . . . . 4.6.6:2
4.6.7 CREEP ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6.7:2
4.6.8 LOAD FUNCTION AND TIME STEPPING PROCEDURE . . . . . 4.6.8:2
4.6.9 SIMULATION OF EXCAVATION AND CONSTRUCTION STAGES . 4.6.9:2
4.7 CONSTITUTIVE THEORIES . . . . . . . . . . . . . . . . . . . . . . . . . 4.7:2
4.7.1 DRUCKER-PRAGER PLASTICITY STRESS RETURN PROBLEM 4.7.1:2
4.7.2 MOHR-COULOMB PLASTICITY STRESS RETURN ALGORITHM 4.7.2:2
4.7.3 CAP PLASTICITY STRESS RETURN ALGORITHM . . . . . . . 4.7.3:2
4.7.3.1 TWO SURFACE STRESS RETURN FOR DP AND CAP
SURFACES . . . . . . . . . . . . . . . . . . . . . . . . . 4.7.3.1:2
4.7.4 MOHR-COULOMB (SMOOTH) PLASTICITY STRESS POINT AL-
GORITHM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.7.4:2
4.7.5 MULTILAMINATE MODEL INVISCID PLASTIC APPROACH . . . 4.7.5:2
4.7.6 MODIFIED-CAM CLAY MODEL STRESS RETURN ALGORITHM 4.7.6:2
4.7.7 AGING CONCRETE MODEL - IMPLEMENTATION SCHEME . . . 4.7.7:2
4.8 APPENDICES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.8:2
4.8.1 SHAPE FUNCTION DEFINITION AND REFERENCE ELEMENTS
FOR 2/3 D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.8.1:2
4.8.2 NUMERICAL INTEGRATION DATA FOR DIFFERENT ELEMENTS
IN 1/2/3D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.8.2:2
4.8.3 MULTISURFACE PLASTICITY CLOSEST POINT PROJECTION AL-
GORITHM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.8.3:2
4.8.4 SINGLE SURFACE PLASTICITY CLOSEST POINT PROJECTION
ALGORITHM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.8.4:2
5 STRUCTURES 5.0.4:3
5.1 TRUSSES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1:2
5.1.1 TRUSS ELEMENT . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.1:2
5.1.1.1 GENERAL IDEA OF TRUSS ELEMENT . . . . . . . . . 5.1.1.1:2
5.1.1.2 GEOMETRY AND DOF OF TRUSS ELEMENTS . . . . . 5.1.1.2:2
5.1.1.3 INTERPOLATION OF THE DISPLACEMENTS AND STRAINS
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.1.3:2
5.1.1.4 WEAK FORMULATION OF THE EQUILIBRIUM . . . . . 5.1.1.4:2
5.1.1.5 STIFFNESS MATRIX AND ELEMENT FORCE VECTOR 5.1.1.5:2
5.1.2 RING ELEMENT . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.2:2
5.1.2.1 GEOMETRY AND KINEMATICS OF A RING ELEMENT 5.1.2.1:2
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5
5.1.2.2 WEAK FORMULATION OF THE EQUILIBRIUM . . . . . 5.1.2.2:2
5.1.2.3 STIFFNESS MATRIX AND ELEMENT FORCES . . . . . 5.1.2.3:2
5.1.3 ANCHORING OF TRUSS AND RING ELEMENTS . . . . . . . . . 5.1.3:2
5.1.3.1 NUMERICAL REALIZATION OF ANCHORING . . . . . . 5.1.3.1:2
5.1.4 PRESTRESSING OF TRUSS AND RING ELEMENTS . . . . . . . 5.1.4:2
5.2 BEAMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2:2
5.2.1 GEOMETRY OF BEAM ELEMENT . . . . . . . . . . . . . . . . . 5.2.1:2
5.2.2 KINEMATICS OF BEAM THEORY . . . . . . . . . . . . . . . . . 5.2.2:2
5.2.3 WEAK FORMULATION OF THE EQUILIBRIUM . . . . . . . . . . 5.2.3:2
5.2.4 INTERPOLATION OF THE DISPLACEMENT FIELD . . . . . . . . 5.2.4:2
5.2.5 STRAIN REPRESENTATION . . . . . . . . . . . . . . . . . . . . 5.2.5:2
5.2.6 STIFFNESS MATRIX AND ELEMENT FORCES . . . . . . . . . . 5.2.6:2
5.2.7 MASTER-CENTROID (OFFSET) TRANSFORMATION . . . . . . 5.2.7:2
5.2.8 RELAXATION OF INTERNAL DOF . . . . . . . . . . . . . . . . . 5.2.8:2
5.2.9 BEAM ELEMENT RESULTS . . . . . . . . . . . . . . . . . . . . . 5.2.9:2
5.3 MEMBRANES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3:2
5.3.1 ELEMENT GEOMETRY MAPPING AND COORDINATE SYSTEM 5.3.1:2
5.3.2 DISPLACEMENTS AND STRAINS FIELD . . . . . . . . . . . . . . 5.3.2:2
5.3.3 CONSTITUTIVE MODELS . . . . . . . . . . . . . . . . . . . . . . 5.3.3:2
5.3.4 WEAK FORMULATION OF EQUILIBRIUM . . . . . . . . . . . . . 5.3.4:2
5.3.5 STIFFNESS MATRIX AND ELEMENT FORCES . . . . . . . . . . 5.3.5:2
5.4 APPENDICES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4:2
5.4.1 MASTER-SLAVE (OFFSET) TRANSFORMATION . . . . . . . . . 5.4.1:2
5.4.2 UNI-AXIAL ELASTO-PLASTIC MATERIAL MODEL . . . . . . . . 5.4.2:2
6 INTERFACE 6.0.2:4
6.1 CONTACT OF SOLIDS AND FLUID INTERFACE . . . . . . . . . . . . . . 6.1:2
6.1.1 GENERAL OUTLOOK . . . . . . . . . . . . . . . . . . . . . . . . 6.1.1:2
6.1.2 DISPLACEMENTS AND STRAINS . . . . . . . . . . . . . . . . . 6.1.2:2
6.1.3 CONSTITUTIVE MODEL . . . . . . . . . . . . . . . . . . . . . . 6.1.3:2
6.1.4 STIFFNESS MATRIX AND ELEMENT FORCE VECTOR . . . . . . 6.1.4:2
6.1.5 AUGMENTED LAGRANGIAN APPROACH . . . . . . . . . . . . . 6.1.5:2
6.1.6 CONTRIBUTION TO CONTINUITY EQUATION . . . . . . . . . . 6.1.6:2
7 GEOTECHNICAL ASPECTS 7.0.6:3
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM6
7.1 TWO-PHASE MEDIUM . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.1:2
7.2 EFFECTIVE STRESSES . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2:2
7.3 SOIL PLASTICITY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.3:2
7.3.1 DRUCKER-PRAGER VERSUS MOHR-COULOMB CRITERION . . . 7.3.1:2
7.3.2 CAP MODEL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.3.2:2
7.3.3 DILATANCY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.3.3:2
7.4 INITIAL STATE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.4:2
7.4.1 COEFFICIENT OF EARTH PRESSURE AT REST, K0 . . . . . . . 7.4.1:2
7.4.2 STATES OF PLASTIC EQUILIBRIUM . . . . . . . . . . . . . . . . 7.4.2:2
7.4.2.1 MOHR-COULOMB MATERIAL . . . . . . . . . . . . . . 7.4.2.1:2
7.4.2.2 DRUCKER-PRAGER MATERIAL . . . . . . . . . . . . . 7.4.2.2:2
7.4.3 INFLUENCE OF POISSONS RATIO . . . . . . . . . . . . . . . . 7.4.3:2
7.4.4 COMPUTATION OF THE INITIAL STATE . . . . . . . . . . . . . 7.4.4:2
7.4.5 INFLUENCE OF WATER . . . . . . . . . . . . . . . . . . . . . . . 7.4.5:2
7.5 SOIL RHEOLOGY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.5:2
7.6 ALGORITHMIC STRATEGIES . . . . . . . . . . . . . . . . . . . . . . . . 7.6:2
7.6.1 SEQUENCES OF ANALYSES . . . . . . . . . . . . . . . . . . . . 7.6.1:2
7.6.2 EXCAVATION, CONSTRUCTION ALGORITHM . . . . . . . . . . 7.6.2:2
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM8
Preface
Document THEORY MANUAL presents the constitutive model and the nite element im-
plementation. The discussion is limited to features which are actually used in the program.
No attempt is made to give a general overview of numerical methods in soil mechanics.
The proposed models include elasticity, various plasticity models, time dependent behavior
resulting from consolidation, and actual creep.
Sign convention are dierent in continuum and soil mechanics. Both sign conventions are
used in this text; variables which are positive in compression are underlined, in order to avoid
possible confusion.
INTRODUCTION
PROBLEM STATEMENT
MATERIAL MODELS
NUMERICAL IMPLEMENTATION
STRUCTURES
INTERFACE
GEOTECHNICAL ASPECTS
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM9
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM10
Preface
Chapter 1
INTRODUCTION
The following sections present the constitutive model and the nite element implementation.
The discussion is limited to features which are actually used in the program. No attempt is
made to give a general overview of numerical methods in soil mechanics.
The proposed models include elasticity, various plasticity models, time dependent behavior
resulting from consolidation, and actual creep.
Sign convention are dierent in continuum and soil mechanics. Both sign conventions are
used in this text; variables which are positive in compression are underlined, in order to avoid
possible confusion.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM11
Preface Introduction
1.1 NOTATION
General rules
underlined variables (like stress , pressure p, etc.) are positive in compression
overbarred symbols mean the parameters with prescribed (known) value ( p, u)
abbreviations in sub/superscripts
cr creep
e elastic part
e element
eq equilibrium
Ext external
F uid
L liquid
m mean
max maximum
min minimum
n normal direction
p plastic part
tot total
0 reference state
innity
Symbols
Latin symbol SI units Meaning
b
i
, b N/m
3
body force vector
C
ijkl
, C N/m
2
compliance constitutive tensor
D
ijkl
, D N/m
2
stiness
G N/m
2
Kirchho modulus
g m/s
2
earth acceleration
E N/m
2
Young modulus
e void ratio
e
ij
strain deviator
k
ij
, k m/s permeability tensor
K N/m
2
bulk modulus (solid)
n
i
, n normal vector
n porosity
u
i
, u m displacement vector
p N/m
2
pressure
Q kg/(s m
2
) mass source
V
i
, v m/s relative uid velocity
R
MC
N/m
2
radius of the MohrCoulomb circle
S saturation coecient
S
r
residual saturation coecient
s
ij
N/m
2
stress deviator
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM1.1:1
Preface Introduction
Greek symbol SI units Meaning

ij
, strain tensor
Ns/m
2
uid viscosity
N/m
3
specic weight
= m boundary of the domain

p
m boundary with imposed pressure conditions

q
m boundary with imposed ow conditions

s
m boundary with imposed seepage (i.e. pressure dependent) ow conditions

t
m boundary with imposed traction conditions

u
m boundary with imposed displacement conditions
N/m
2
Lame constant
Poisson coecient
kg/m
3
mass density

ij
, N/m
2
stress tensor

ij
,

N/m
2
eective stress tensor

ij
, N/m
2
shear stress tensor
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM1.1:2
Preface Introduction
1.2 SOME IMPORTANT FORMULAE IN TENSOR ALGEBRA AND
ANALYSIS
In mechanics several tensorial variables of dierent rank are used. Examples are:
scalar (zeroth rank tensor) density , temperature T, energy W, . . .
vector (rst rank tensor) displacement vector u, velocity vector v, . . .
dyad (second rank tensor) stress tensor , deformation tensor , . . .
fourth, sixth and higher rank tensor material tensor . . .
Some rules of calculations with tensors in the three-dimensional Euclidean space are presented
in this section. The direct (symbolic) and the component notation of tensor quantities are
used. For shorter writing we introduce the Einsteins summation convention (repeated index
in some term in the expression requires summation)
. . . +a
i
b
i
+. . . = +
3

i=1
a
i
b
i
+. . . = . . . +a
1
b
1
+a
2
b
2
+a
3
b
3
+. . .
i is a dummy index.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM1.2:1
Preface Introduction
Window 1-1: Scalars, vectors and tensors
SCALAR
Scalars are variables, which are fully independent on the choice of coordinate system (invariant
variables) because they have no orientation.
VECTOR
Vectors can be written as
a = a
1
e
1
+a
2
e
2
+a
3
e
3
, a = (a
1
, a
2
, a
3
), a = (a
i
), i = 1, 2, 3.
The a
i
are the coordinates of the vector, which are related to the vector basis e
i
with respect
to the given coordinate system. This vector basis is assumed to be an orthonormal basis
[e
i
[ = 1, e
i
e
j
=
_
1 i = j
0 i ,= j
The scalar product (inner product, dot product) of two vectors a and b is dened as
a b = a
i
e
i
b
j
e
j
= a
i
b
j
e
i
e
j
= a
i
b
j

ij
= a
i
b
i
= ;
ij
=
_
1 i = j
0 i ,= j
The dyadic product of two vectors a and b
ab = a
i
e
i
b
j
e
j
= a
i
b
j
e
i
e
j
= T
ij
e
i
e
j
= T.
In some textbooks for this product the following designation is used
a b ab.
SECOND RANK TENSOR With the help of the dyadic product the second rank tensor
T can be introduced
T = ab = a
i
b
j
e
i
e
j
= T
ij
e
i
e
j
.
For the second rank tensors T and S we dene the following products:
Scalar product e.g. tensor product with the contraction
T : S = T
ij
e
i
e
j
: S
kl
e
k
e
l
= T
ij
S
kl
e
i

jk
e
l
= T
ij
S
jl
e
i
e
l
= M
il
e
i
e
l
,
which leads to a second rank tensor.
Double scalar product e.g. tensor product with the double contraction
T :: S = T
ij
e
i
e
j
:: S
kl
e
k
e
l
= T
ij
S
kl

jk

il
= T
ij
S
ji
= ,
resulting in a scalar.
HIGHER RANK TENSOR
In a similar way (by dyadic product) we can dene tensors of higher ranks:
the 4th rank tensor
(4)
A = abcd = a
i
b
j
c
k
d
l
e
i
e
j
e
k
e
l
= TS =
T
ij
S
kl
e
i
e
j
e
k
e
l
= A
ijkl
e
i
e
j
e
k
e
l
the 6th rank tensor
(6)
B = abcdgh = TSP =
T
ij
S
kl
P
mn
e
i
e
j
e
k
e
l
e
m
e
n
= B
ijklmn
e
i
e
j
e
k
e
l
e
m
e
n
.
Also tensor products (from uni- to multi- scalar) can be dened.
Window 1-1
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM1.2:2
Preface Introduction
Window 1-2: Special for second rank tensors
Unit tensor (identity tensor) I
I =
ij
e
i
e
j
Transposed tensor T
T
T = ab = T
T
= ba ,
T = T
ij
e
i
e
j
= T
T
= T
ij
e
j
e
i
= T
ji
e
i
ej ,
Symmetric tensor
If T = T
T
(T
ij
= T
ji
) then tensor T is symmetric
Antisymmetric tensor
If T = T
T
(T
ij
= T
ji
) then tensor T is antisymmetric
Trace of the tensor
trT = I :: T = T
ii
= T
11
+T
22
+T
33
.
Tensor decomposition
The tensor T may be decomposed into two parts: axiator and deviator dened as follow:
Axiator (spherical tensor); denoted A
T
or T
A
A
T
=
1
3
(I :: T)I =
1
3
T
kk

ij
e
i
I
j
Deviator denoted D
T
or T
D
D
T
= TA
T
= (T
ij

1
3
T
kk

ij
)e
i
e
j
so
T = A
T
+D
T
Window 1-2
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM1.2:3
Preface Introduction
Window 1-3: Transformation rules for tensors
The rules of transformation from one coordinate system to a rotated system (marked with )
for tensors of the rank 2, 4 or 6 are (all indices range from 1 to 3)
a

ij
=
mi

nj
a
mn
,
b

ijkl
=
mi

nj

sk

tl
b
mnst
,
c

ijklop
=
mi

nj

sk

tl

uo

vp
c
mnstuv
.
The
ij
are the elements of the transformation matrix (direction cosines):

ij
= cos(e

i
, e
j
).
Window 1-3
Window 1-4: Eigenvalue problem for a second rank tensor
The eigenvalues and the eigenvectors (eigendirections) n for a second rank tensor T can
be obtained from the solution of the following equations
(TI) n = 0, (T
ij

ij
)n
j
= 0
The eigenvalues follow from the condition that nontrivial solutions are existing, which leads
to the characteristic equation:
det(TI) = 0; det(T
ij

ij
) = 0. (1)
The roots of this equation
()
, = 1, 2, 3 sort in the ascending order (e.g.
1

2

3
)
are called the principal values. It can be shown that in the case of symmetric second rank
tensors all principle values are real.
For each root we get the eigenvector (eigendirections, principal directions) n
()
j
, = 1, 2, 3
from the system
(T
11
)n
1
+T
12
n
2
+T
13
n
3
= 0,
T
21
n
1
+ (T
22
)n
2
+T
23
n
3
= 0,
T
31
n
1
+T
32
n
2
+ (T
33
)n
3
= 0,
n
2
1
+n
2
2
+n
2
3
= 1.
Window 1-4
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM1.2:4
Preface Introduction
Window 1-5: Invariants of a second rank tensor
Invariant terms are independent on the choice of the coordinate system. Such a system of
invariants can be related to the coecients of the characteristic equation 1 rewrite in the
form:
det(TI) =
3
I
1
(T)
2
+I
2
(T) I
3
(T) = 0. (1)
The I
i
are called principal invariants, dened as:
Linear principal invariant
I
1
(T) = trT T :: I T
ii
,
Quadratic principal invariant
I
2
(T) =
1
2
_
J
2
1
(T) J
1
(T
2
)

=
1
2
(T
ii
T
jj
T
ij
T
ji
),
Cubic principal invariant
I
3
(T) =
1
3
_
J
1
(T
3
) + 3J
1
(T)J
2
(T) J
3
1
(T)

=
1
3
J
1
(T
3
)
1
2
J
1
(T
2
)J
1
(T) +
1
6
J
3
1
(T)
= det(T
ij
).
In the stress space very often we use invariants of the stress deviator D

= s
ij
:
J
1
= 0
J
2
=
1
2
s
ij
s
ji
J
3
=
1
3
s
ij
s
jk
s
ki
Cylindrical invariants of stress tensor (Haigh and Westergaard):
=
1

3
I
1
cos 3 =
3

3
2
J
3
J
3/2
2
=
_
2J
2
Invariants commonly used in geotechnical material models are:
p =
1
3
I
1
q =
_
3J
2
Window 1-5
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM1.2:5
Preface Introduction
Window 1-6: Cayley-Hamilton theorem
The second rank tensor satises characteristic equation
T
3
I
1
(T)T
2
+I
2
(T)TI
3
(T)I = 0,
which enables the representation of T
n
(n 3) as a linear function of T
2
, T, T
0
= I, e.g.,
T
3
= I
1
(T)T
2
I
2
(T)T+I
3
(T)I.
Window 1-6
Window 1-7: Derivatives of the invariants of a second rank tensor
A scalar-valued function of a second rank tensor can be represented by
= (T) = (T
11
, T
22
, . . . , T
31
).
Then we can calculate the derivative by the following equation

,T
=

T
=

T
kl
e
k
e
l
.
On the other hand the derivatives of the invariants are
J
1
(T)
,T
= I, J
1
(T
2
)
,T
= 2T
T
, J
1
(T
3
)
,T
= 3T
2
T
,
J
2
(T)
,T
= J
1
(T)I T
T
,
J
3
(T)
,T
= T
2
T
J
1
(T)T
T
+J
2
(T)I = J
3
(T)(T
T
)
1
.
So, we nally get
[J
1
, J
2
, J
3
]
,T
=
_

J
1
+J
1

J
2
+J
2

J
3
_
I
_

J
2
+J
1

J
3
_
T
T
+

J
3
T
2
T
.
These calculations can be helpful for the use of the representation theorem of an isotropic
function
P = F(T) =
0
I +
1
T+
2
T
2
.
The coecients
i
itself are functions of the invariants

i
=
i
[J
1
(T), J
2
(T), J
3
(T)] . (1)
Window 1-7
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM1.2:6
Preface Introduction
Window 1-8: Transition from tensorial to matrix notation
It is assumed that stress/strain components are ordered in column vectors as follows:
=
_

xx

yy

xy

zz

xz

yz
_
T
=
_

xx

yy

xy

zz

xz

yz
_
T
The following table is used to extract vector components from appropriate tensorial objects:
I/J i/k j/l
1 1 1
2 2 2
3 1 2
4 3 3
5 1 3
6 2 3
With the above table we can set:
- I-th component of stress vector :
I
=
i(I)j(I)
- I-th component of strain vector :
I
=
i(I)j(I)
(shear terms have to doubled)
- I-th, J-th component of stiness matrix D : D
IJ
= D
i(I)j(I)k(J)l(J)
Window 1-8
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM1.2:7
Preface Introduction
Window 1-9: Example: 2nd rank tensor in 2D space
Transformation rule
The (u, v) coordinates system is rotated from (x, y) one by angle
t
uu
= t
xx
cos
2
+t
yy
sin
2
+ 2t
xy
sin cos =
t
xx
+t
yy
2
+
t
xx
t
yy
2
cos 2 +t
xy
sin 2
t
vv
= t
xx
sin
2
+t
yy
cos
2
2t
xy
sin cos =
t
xx
+t
yy
2

t
xx
t
yy
2
cos 2 t
xy
sin 2
t
uv
= (t
xx
t
yy
) sin cos +t
xy
_
cos
2
sin
2

_
=
t
xx
t
yy
2
sin 2 t
xy
cos 2
Principal coordinate system ( )
t
12
= t
21
def
= 0 =
1
2
arctan
_
2t
xy
t
xx
t
yy
_
Diagonal, principal (max/min) components
t
1,2
= t
max,min
=
t
xx
+t
yy
2

_
t
xx
t
yy
2
_
2
+t
2
xy
Maximal outof diagonal component
Position of the coordinate system:
t
12
max =
1
2
arctan
_
t
yy
t
xx
2t
xy
_
e.g. such system is rotated from principal one by 45

Maximal value of out-of diagonal component


t
max
uv
=
t
1
t
2
2
Window 1-9
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM1.2:8
Preface Introduction
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM1.2:9
Preface
Chapter 2
PROBLEM STATEMENT
In following sections formulations of problems available to be solved are given. In particular:
SINGLE PHASE
TWO PHASE
TRANSIENT FLOW
HEAT TRANSFER
HUMIDITY TRANSFER
They contain governing dierential equations and boundary conditions (strong formulation).
Despite this, for each problem variational formulations (weak form) are given which are the
basis for numerical solution.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM2.0:10
Preface Problem Statement
2.1 SINGLE PHASE, SOLID MEDIUM
Window 2-1: Strong form of the boundary value problem
Equilibrium of the single phase medium
Equilibrium equation:

ij,j
+b
i
= 0, in
Boundary conditions (BC):
traction BC:
ij
n
j
=

t
i
, on
t
displacement BC: u
i
= u
i
, on
u
=
t

u
Straindisplacement relation (analysis of the small strain tensor and the linear relation is
assumed):

ij
=
1
2
(u
i
,
j
+u
j
,
i
)
Constitutive equation (incremental form e.g. means a stress increment )

ij
= C
ijkl

kl
where C
ijkl
is the 4th rank constitutive tensor
Window 2-1
Related Topics
THEORY MANUAL: TWO-PHASE MEDIA
THEORY MANUAL: WEAK FORM
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM2.1:1
Preface Problem Statement
2.2 TWO-PHASE PARTIALLY SATURATED MEDIUM
The simulation of a twophase medium is necessary in order to account for time-dependent
behaviour resulting from consolidation and/or transient ow. Actual creep will be discussed
later on.
The boundaryvalueproblem to be solved requires the coupled solution of conservation of
mass and momentum in both the solid and the liquid phases, together with boundary and
initial conditions. The general transient case is considered here.
The Windows in section present:
Two phase medium model Window 2-2
Equilibrium of two phase medium Window 2-3
Strong form of BVP for two phase partially saturated medium Window 2-4
Strong form of BVP for two phase fully saturated medium Window 2-5
Window 2-2: Twophase medium model
(1 n)
n
nS
n(1 S)
n porosity
S saturation ratio
solid
uid
air
The two-phase medium is in fact an approximation of a three-phase medium where it is
assumed that the air bubbles are trapped in the liquid phase so that the mixture (uid + air)
forms a compressible uid obeying Darcys law.
Window 2-2
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM2.2:1
Preface Problem Statement
Window 2-3: Equilibrium of twophase medium
b
i
denote body force, p is the uid pressure (positive in tension).

ij
is used here for the
eective stresses (positive in tension) see problem statement for:
Partially saturated medium Window 2-4
Fully saturated medium Window 2-5
Equilibrium equation:

ij,j
+b
i
= 0 on T
Straindisplacement relation (analysis of the small strain tensor and the linear relation is
assumed):

ij
=
1
2
(u
i
,
j
+u
j
,
i
)
Constitutive equation (incremental form e.g. means a stress increment )

ij
= C
ijkl

kl
where C
ijkl
is the 4th rank constitutive tensor
Window 2-3
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM2.2:2
Preface Problem Statement
Window 2-4: Strong form of BVP for 2phase partially saturated medium problem
Eective stresses

ij
:

ij
=
ij
Sp
ij
,
where S is the saturation coecient
1
S = S(p) =
_

_
S
r
+
1 S
r
_
1 +
_

F
_
2
_
1/2
if p > 0
1 if p 0

F
stands for proper weight of uid, S
r
is the residual saturation ratio, a parameter,
Flow equation (the Darcy law)
q
i
= k

ij
_

F
+z
_
,
j
The permeability tensor k

ij
is obtained by scaling the k
ij
tensor for fully saturated medium
by scalar valued function k
r
dependent on the saturation ratio.
k

ij
= k
r
(S)k
ij
k
r
=
(S S
r
)
3
(1 S
r
)
3
=
1
_
1 +
_

F
_
2
_
3/2
Mass balance:
= nS
w
+ (1 n)
s
where n is the porosity dened by the void ratio e
n =
e
1 +e
, e =
void volume
solid volume
1
Van Genuchten (1980), A closed form of the equation for predicting the hydraulic conductivity of unsat-
urated soils. Soil Sciences Am. Soc., 44, 892898.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM2.2:3
Preface Problem Statement
Continuity equation:
S
kk
+q
k,k
= c p
with c the specic storage coecient
c = c(p) = n
_
S
K
F
+
dS
dp
_
K
F
uid bulk modulus
Boundary conditions:
on solid phase on uid phase

ij
n
j
=

t
i
on
t
u
i
= u
i
on
u
,
q
j
n
j
= q on
q
q
j
n
j
= q
s
on
s
p = p on
p
=
u
+
t
=
p
+
q
+
s
The seepage surface requires a special treatment as the applicable boundary condition is
unknown a priori. Boundary condition on
s
:
p = 0 on
s
if S = 1
p = p
Ext
on
s
if S = 1 and p
Ext
imposed
q = 0 on
s
if S < 1
This boundary condition is satised through penalization imposing
q = 0 if p 0 in the domain and p
Ext
= 0
q = k
v
p if p < 0 in the domain and p
Ext
= 0
q = k
v
(p p
Ext
) p in the domain and p
Ext
,= 0
with k
v
, a ctitious permeability (penalty parameter) and p
Ext
the pressure on the external
face of
s
; p
Ext
= 0 corresponds to atmospheric pressure.
Boundary conditions for typical ow problem (damping)
Initial conditions
u
i
(t = t
0
) = u
i0
on
p(t = t
0
) = p
0
on
Window 2-4
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM2.2:4
Preface Problem Statement
Window 2-5: Strong form of BVP for 2phase fully saturated medium problem
It can be treated as the particular case of partially saturated medium for which S
r
1 e.g.
S 1.
Eective stresses

ij
:

ij
=
ij
p
ij
Continuity equation:

kk
+q
k,k

n
K
F
p = 0 in T
K
F
uid bulk modulus
Flow equation (the Darcy law)
q
i
= k
ij

,j
= k
ij
_
=
p

F
+z
_
,j
k
ij
is the permeability tensor and
F
stands for proper weight of uid.
Boundary conditions:
on solid phase on uid phase

ij
n
j
=

t
i
on
t
u
i
= u
i
on
u
,
q
j
n
j
= q on
q
q
j
n
j
= q
s
on
s
=
u
+
t
= +
q
+
s
Initial conditions
u
i
(t = t
0
) = u
i0
on
p(t = t
0
) = p
0
on
Window 2-5
Related Topics
THEORY MANUAL: TRANSIENT FLOW
THEORY MANUAL: SINGLE PHASE, SOLID MEDIUM
THEORY MANUAL: TWOPHASE MEDIA. APPROXIMATION AND MATRIX FORM
THEORY MANUAL: TWOPHASE MEDIA. WEAK FORM
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM2.2:5
Preface Problem Statement
2.3 TRANSIENT FLOW
Transient ow problem formulation may be derived from twophase media formulation, see
section 2.2. The only primary state variable is pore pressures p. In continuity equation, term
resulting from skeleton volume changes
kk
should be neglected. Constitutive relation for the
ow (generalized Darcy law will take identical form). Also boundary contitions for uid phase
are identical as in the case of twophase media.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM2.3:1
Preface Problem Statement
2.4 HEAT TRANSFER
Following section gives steady state/transient heat transfer formulation for the isotropic case
of 2/3D continuum, including dierential equation, boundary and initial conditions.
Window 2-6: Heat transfer formulation in strong form
Fourier equation:
( T,
i
),
i
+
H
t
= c

T
t
on
Boundary conditions:
Temperature BC , with prescribed temperature T:
T = T on
T
Heat ow BC , with prescribed heat ux q:

T
n
= q on
q
Convective BC , with prescribed ambient temperature T
e
:

T
n
= h(T T
e
) on
c
Note: (
q

c
)
T
= ; but (
q

c
)
T
=
Initial condition:
Known temperature eld T
0
at time t = 0: T(x,0) = T
0
(x) on
where :
T temperature, [
o
K]
t time, [day]
heat conductivity, [kN/(
o
K day)]
c

= c heat capacity, [kN/(m


2

o
K)]
c specic heat, [kN m/(kg
o
K)]
mass density, [kg/m
3
]
q external heat ux, [kN/(m day)]
T
e
ambient temperature, [
o
K]
h heat convection coecient, [kN/(m
o
K day)]
H heat source [kN/m
2
]
Window 2-6
Note: (
q

c
)
T
= ; but (
q

c
)
T
= which means that setting temperature B.C.
exclude other boundary conditions at given part
T
, while heat ow B.C. and convective
B.C. may coexist on a common part of the boundary i.e. mixed condition may be set on

c
as:
T
n
+h(T T
e
) +q = 0. Leaving a part of the boundary with no B.C. specied
explicitly, corresponds to setting no heat ow, i.e. adiabatic condition
T
n
= 0 on it.
Steady state problem described by Laplace equation
( T,
i
),
i
= 0
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM2.4:1
Preface Problem Statement
can be formulated. Physically, it describes continuum at thermal equilibrium state, while
mathematically it corresponds to a limit of the transient problem at time t with all
state variables independent from time.
The source term H adopted in the formulation is related to the phenomena of heat emission
during concrete hydration process and is described by the following set of equations:
Window 2-7: Concrete hydration heat source
heat source as a function of maturity M :
H(t, T) = H

aM
1 +aM
maturity M as a function of absolute temperature T and time t:
M(t, T) =
t
_
t
d
exp
_
Q
R
_
1
T
f

1
T
__
dt
where:
H

total value of concrete hydration heat per unit volume [kJ/m


3
],
a heat source parameter [1/day]
Q/R activation energy/universal gas constant [
o
K]
T
f
reference temperature, normally 20
o
C=293
o
K [
o
K]
t
d
dormant period [day]
Window 2-7
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM2.4:2
Preface Problem Statement
2.5 HUMIDITY TRANSFER
Following section gives steady state/transient humidity transfer formulation for the isotropic
case of 2/3D continuum, including dierential equation, boundary and initial conditions.
Window 2-8: Humidity transfer formulation in strong form
Ficks equation:
(D(W)W,
i
),
i
=
W
t
on
Boundary conditions:
Humidity BC, with prescribed relative humidity
W = W on
W
Perfect isolation, with humidity ux q
W
= 0:
q
W
=
(DW)
n
= 0 on
q
Note:
q

W
= ; but
q

W
=
Initial condition:
Known relative humidity eld W
0
at time t = 0: W(x,0) = W
0
(x) on
with :
W moisturre potential, i.e. relative humidity []
t time, [day]
D(W) diusion coecient as a function of W, [m
2
/day]
D(W) = D
1
_
_
_
a +
1 a
1 +
_
1W
1W
1
_
4
_
_
_
where:
D
1
diusion coecient for a moisture potential of W = 1, [m
2
/day]
W
1
moisture potential at which D(W) =
1
2
D
1
(1 +a) []
a factor, to dene diusion at low relative humidity []
Window 2-8
Note:
q

W
= ; but
q

W
= which means that setting humidity B.C. exclude
perfect isolation BC at given part
W
. Leaving a part of the boundary with no B.C. specied
explicitly (default), corresponds to setting no humidity ux, i.e. perfect isolation condition
(DW)
n
= 0 on it.
Steady state problem described by Laplace equation
(D W,
i
),
i
= 0
can be formulated. Physically, it describes continuum at diusion equilibrium state, while
mathematically it corresponds to a limit of the transient problem at time t with all
state variables independent from time.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM2.5:1
Preface Problem Statement
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM2.5:2
Preface
Chapter 3
MATERIAL MODELS
ELASTICITY
CREEP
CONSOLIDATION
PLASTICITY
SWELLING
AGING CONCRETE
APPENDICES
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.0:3
Preface Material Models
3.1 ELASTICITY
Hookes law is used as the basis of the model. In the subsequent windows a review of formulae
from general to some particular cases is presented:
Hooke law
Plane Strain
Axisymmetric Analysis
To simplify the writing, dierent sets of material constant are introduced. The relations
between Lames constant and , the coecient of compressibility K, Youngs modulus E
and Poissons ratio are given in Table 3.1:
Table 3.1: Elastic constants
G = E K
,
(3 + 2)
+

2( +)
3 + 2
3
E,
E
(1 +)(1 2)
E
2(1 +
E
E
3(1 2)
K,
3K 2
3

9K
3K +
3K 2
2(3K +)
K
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.1:1
Preface Material Models
Window 3-1: Hooke law
Generalized constitutive equation

ij
= D
ijkl

kl
(1)
where D
ijkl
is the modulus tensor with 36 independent components
Orthotropic case 9 independent material constants

11
=
1
E
1

11


21
E
2

22


31
E
3

33

22
=

12
E
1

11

1
E
2

22


32
E
3

33

33
=

13
E
1

11


23
E
2

22

1
E
3

33
(2)

12
=
1

12

12
,
13
=
1

13

13
,
23
=
1

23

23
Isotropic case 2 independent material constants
Isotropy conditions:
=
12
=
21
=
31
=
13
=
23
=
32
E = E
1
= E
2
= E
3
(3)
=
12
=
13
=
23
one obtains:

11
=
1
E
(
11

22

33
)

22
=
1
E
(
11
+
22

33
) (4)

33
=
1
E
(
11

22
+
33
)

ij
=
1

ij
, i ,= j
Isotropic case: tensor notation
Strain vs stress

ij
=
1
E
[(1 +)
ij

kk

ij
] (5)
Stress vs strain

ij
=
kk

ij
+ 2
ij
(6)
Isotropic case: volumetricdeviatoric split

kk
= 3K
kk
(7)
s
ij
= 2e
ij
(8)
s
ij
, e
ij
denote the components of stress and strain deviators respectively
Window 3-1
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.1:2
Preface Material Models
Window 3-2: Plane Strain
Plane Strain assumptions:

33
=
13
=
23
0
Under the above assumptions from Eqs (4), one obtains:

33
= (
11
+
22
) (1)
and
_
_

11

22

12
_
_
=
E(1 )
(1 +)(1 2)
_

_
1

1
0

1
1 0
0 0
1 2
2(1 )
_

_
_
_

11

22

12
_
_
(2)
or,
_
_

11

22

12
_
_
=
_
_
1 0
+ 2 0
0 0
_
_
_
_

11

22

12
_
_
Window 3-2
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.1:3
Preface Material Models
Window 3-3: Axisymmetric Analysis
In an axisymmetric analysis, the following notation for coordinates and components of dis-
placements, in the cylindrical coordinates system, are used:
x
1
= r the radial coordinate u
1
= u
r
the radial displacement
x
2
= y the axial coordinate u
2
= u
y
the axial displacement
x
3
= the circumferential coordinates u
3
= u

the circumferential displacement


Analysis of axisymmetric body is assumed as well as that all state variables are independent
of i.e. they are dependent on r and y only. Hence three dimensional problems can be
reduced to 2dimensional ones.
In the axisymmetric torsionless case it is additionally assumed:
u

= 0
which results in

r
=
y
= 0

r
= 0 and
y
= 0
Rewriting the constitutive equations in vector form, one gets:
_

11

22

12

33
_

_
=
1
E
_

_
1 0
1 0
0 0
E

0
0 1
_

_
_

11

22

12

33
_

_
. (1)
First inverting the above stressstrain relations, one obtains:
_

11

22

12

33
_

_
=
_

_
+ 2 0
+ 2 0
0 0 0
0 + 2
_

_
_

11

22

12

33
_

_
.
Remark:
Below matrix consists of the equivalent planestrain matrix plus a fourth row and column.
Hence planestrain conditions can be obtained from the axisymmetric case by ignoring the
fourth row and column, and setting
33
= (
11
+
22
).
Window 3-3
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.1:4
Preface Material Models
3.2 CONSOLIDATION
Primary consolidation is discussed in this section. It results from the coupling of loadinduced
Darcy ow with the motion of a quasisaturated medium.
DARCY LAW
FLUID MOTION
OEDOMETRIC TEST
Related Topics
THEORY: TWO PHASE MEDIA
NUMERICAL IMPLEMENTATION: CONSOLIDATION
GEOTECHNICAL ASPECTS: TWO PHASE MEDIA
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.2:1
Preface Material Models Consolidation
3.2.1 GENERALIZED DARCY LAW
The generalized Darcy law expresses conservation of momentum with the inertia term ne-
glected, it is summarized in Window 3-4
Window 3-4: Darcy ow
Darcys ow velocity
1
:
q =
K

i
p
= k i (1)
where:
q = nv
F
relative uid velocity [m/s]
v
F
average velocity through holes [m/s]
n porosity
K permeability [m
2
] (function of porosity,
independent of uid properties)
k = K

=
F
g
K

permeability coecient [m/s]

F
uid mass density [kg/m
3
]
g earth acceleration [m/s
2
]
uid viscosity [N s/m
2
]
i
p
= i pressure gradient [N/m
3
]

F
=
F
g specic weight [N/m
3
]
i hydraulic gradient [nondim.]
Threedimensional extension (with appropriate sign convention),
q = k : grad = k : grad(p
F
/
F
+z) (2)
and in indicial notation
q
i
= k
ij
_
p
F

F
+z
_
,
j
(3)
Window 3-4
1
See K. Terzaghi (1943) Theoretical Soil Mechanics, Wiley.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.2.1:1
Preface Material Models Consolidation
3.2.2 FLUID MOTION
Conservation of mass in the liquid phase is expressed by Eq. (1). The time variation of
apparent specic mass splits into two terms as shown in Eq. (2). The variation of porosity
can in turn be related to the volumetric strain, Eq. (3). The uid density change is related
to the uids volumetric strain by Eq. (4). Assuming a slightly compressible uid, Eq. (5),
replacing then in the mass conservation equation (Eq. (1), using Eq. (??)) and with Q = 0,
Eq. (6) is obtained.
The convective contribution can be neglected for small strain and Darcy ow, (v grad
w
) is
small, this leads to equation (7) after division by
w
.
Eq. (??) in Window ?? results nally from Darcy ow coupled with (7). If uid compressibility
is negligible, then the term in t can be ignored. If consolidation eects are negligible then
the square parantesis in (6) and (7) is ignored.
Window 3-5: Stress induced uid motion in porous medium
Conservation of mass

F
t
+ div
_

F
v
F
_
= Q (1)
where:

F
= n
F
apparent specic mass

F
specic mass of uid phase
q = nv
F
velocity
Q mass source term (zero if no source)
n porosity
Further,

F
t
=
n
t

F
+

F
t
n (2)
with
n
t
=

kk
t
(3)

F
t
=
F

F
kk
t
(4)
p
t
= K
F

F
kk
t
(5)
results in

F
_

kk
t

n
K
F
p
t
_
+ div (
F
q) = 0. (6)
Finally

kk

n
K
F
p +q
k
,
k
= 0 (7)
with K
F
the uid bulk modulus.
Window 3-5
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.2.2:1
Preface Material Models Consolidation
3.2.3 OEDOMETRIC TEST
An oedometer test can be used for the identication of the consolidation parameters. The
oedometer test consists of a onedimensional drained compression test used to determine the
oedometric compressibility modulus E
oed
and the in situ preconsolidation stress
c
Fig. 3.2.3.
A bilinear approximation of the test in a semi-logarithmic referential leads to the denition
of
c
; the slopes c
s
and c
c
are the unloading and loading moduli, respectively. The test can
be used, given an initial state (e
0
,
0
), the corresponding preconsolidation stress, to dene
the yield point for a plastic cap model (at P in the Fig. 3.2.3).
Oedometer test. is the compressibility index, such that c
c
= log
e
10.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.2.3:1
Preface Material Models
3.3 PLASTICITY
SKETCH OF PLASTICITY APPROACH
MOHRCOULOMB CRITERION
DRUCKERPRAGER CRITERION
CAP MODEL
CAM-CLAY MODEL
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3:1
Preface Material Models Plasticity
3.3.1 SKETCH OF THE PLASTICITY APPROACH
Plasticity is a nonlinear constitutive theory and leads to a nonlinear system of equations, which
is solved iteratively for d, the displacement increment, using a tangent (local) stiness.
Once the displacement increment is known the corresponding strain increment results from
the usual straindisplacement relations. From the strain increment a trial stress can be
deduced which, if it lies outside of the yield criterion, must be returned onto the criterion
using a ow rule. This ow rule denes the direction of the stress return. The amplitude of
the return results from the consistency condition, which requires the new stateofstress to
lie on the yield criterion. The objective, in this section, is to dene precisely the new keywords
introduced for plasticity theory. The basic ingredients of the elastoplasticity theory are as
follows:
Strain decomposition
Assume that the total strain increment (or rate) is the sum of an elastic
e
and a plastic
p
contribution:
=
e
+
p
and that the following constitutive equation holds:
= D
e
(
p
)
with D
e
the elastic constitutive tensor.
Flow rule
The ow rule denes the direction of the plastic ow by:

p
= d r(, q)
where d is a positive scalar which denes the amplitude of the plastic ow and r (in general
a function of the stress state and set of a hardening parameters q) denes the direction in
space. The calculation of d will be described later on.
For associative plasticity the direction of the ow r coincides with that of the normal a to
the yield surface:
a = r =
F

while for nonassociative plasticity, we assume the existence of a plastic potential surface Q
such that:
r =
Q

Hardening law
The most general form of hardening law can be expressed in rate form as follows:
q = d h(, q)
where h (in general a function of the stress state and of a set of hardening parameters q)
is called hardening function.
The amplitude of the plastic ow
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.1:1
Preface Material Models Plasticity
The amplitude of plastic ow can be derived from the consistency condition, which expresses
that the stress point remains on the yield surface during plastic ow:

F =
F

: +
F
q
: q = a : D
e
: (
p
) +
F
q
: q
= a : D
e
: ( d r) +
F
q
: d h = 0.
From the above equation, in which the ow rule and the hardening rule have been applied,
the amplitude d can be evaluated:
d =
(a : D
e
: )
(a : D
e
: r)
F
q
: h
Derivation of the elastoplastic tangent matrix
Applying the amplitude d to the general constitutive equation the tangent elastoplastic
constitutive matrix is dened in the following way:
= D
e
: (
p
) = D
e
: ( d r) = D
e
:
_

_
r
(a : D
e
: )
(a : D
e
: r)
F
q
: h
_

_
=
_

_
D
e

D
e
: r : a : D
e
(a : D
e
: r)
F
q
: h
_

_
= D
ep

In case of perfect elastoplasticity (no hardening is introduced) the denition of the tangent
matrix reduces to:
D
ep
= D
e

D
e
: r : a : D
e
(a : D
e
: r)
.
For a nonassociative plastic ow rule the resulting tangent matrix is nonsymmetric.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.1:2
Preface Material Models Plasticity
3.3.2 MOHRCOULOMB CRITERION
The MohrCoulomb (MC) criterion is more common in soil mechanics. Traditionally, a soil
is described by its cohesion C and its angle of friction . The MC criterion then states
that the shear stress required for yielding depends on the cohesion, the friction angle and the
pressure normal to the slip surface.
Window 3-6: Mohr-Coulomb yield criterion
Mohr-Coulomb yield criterion
[ [= c +
n
tan (1)
where
=

11
+
22
2
(2)
and,
R
MC
= c cos + sin (3)
is the maximum shear stress equal to radius of the Mohr circle at failure, i.e.:
R
MC
=

(
11

22
)
2
4
+
2
12
(4)
Window 3-6
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.2:1
Preface Material Models Plasticity
3.3.3 DRUCKER-PRAGER CRITERION
The DruckerPrager (DP) yield criterion is, mathematically speaking, the most convenient
choice and often numerically the most ecient. The DP criterion is dened in stress space,
by the following equation:
F() =a

I
1
+
_
J
2
k = 0 (3.1)
where the invariants I
1
and J
2
are dened in Section 1.2 and a

and k are positive material


properties. For a

= 0, HuberMises criterion results.


Flow rule
Associative plasticity: the direction of the ow r coincides with that of the normal a to the
yield surface:
a = r =
F

= a

ij
+
1
2

J
2
s
ij
Nonassociative plasticity: the existence of a plastic potential surface Q of DruckerPrager
type is assumed:
r
ij
=
Q

ij
= a

ij
+
1
2

J
2
s
ij
Notice that the corresponding ow is associative in the deviatoric component and non
associative in the volumetric components, as a

= a

.
Cut-o condition
The following tensile cutto plasticity condition can be activated in conjunction with the
DruckerPrager plasticity criterion:
F() =
1

3
I
1
+
_
J
2

3
I

1T
= 0.
It has two basic features, rst that maximum rst stress invariant I
1
is limited to the value
I

1T
for zero deviatoric stress s and the second that the maximum stress ratio dened as:
q
p
=
3

3J
2
I
1
3
is limited to the value which can be reached in the uniaxial compression test.
The ow rule has been assumed as fully associated so the plastic ow vector r is:
r
ij
=
F

ij
=
1

ij
+
1
2

J
2
s
ij
.
Matching of DruckerPrager criterion
DruckerPrager constants could be derived directly from experiments, instead of calculated
from cohesion and friction angle. Assuming that the material is properly identied by a
MohrCoulomb criterion, the matching with the DruckerPrager criterion can be done for
various stress states.
Three dimensional matching
Collapse load matching
Elastic domains matching
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.3:1
Preface Material Models Plasticity
Window 3-7: DP criterion: Three dimensional matching
Both criteria are represented in the deviatoric plane along with threedimensional matching
coecients.
Deviatoric sections of MohrCoulomb (MC) and DruckerPrager (DP) criteria
External apices of the MC criterion yields (axial compression):
a

=
2 sin

3(3 sin )
k =
6c cos

3(3 sin )
(1)
Internal apices:
a

=
2 sin

3(3 + sin )
k =
6c cos

3(3 + sin )
(2)
Window 3-7
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.3:2
Preface Material Models Plasticity
Window 3-8: Matching of collapse load (plane strain conditions)
Matching collapse load of DP and M-C criteria under plane strain conditions is the default
adjustment adopted in the program when plane strain is activated.
Assumptions:
perfect plasticity:
e

p

e
= 0; =
p
(1)
plane strain:
p
33
=
p
13
=
p
23
= 0 (2)
ow rule Eq. (3.3.3) :
p
ij
= d r
ij
= d
_
a

ij
+
1
2

J
2
s
ij
_
, (3)
r
ij
=
Q

ij
From (3)
s
33
= 2a

_
J
2
; s
13
= s
23
= 0
and invariants
I
1
=
3
2
(
11
+
22
) 3a

_
J
2
; J
2
=
_
[(
11

22
) /2]
2
+
2
12
_
(1 3a
2

)
=
(R
MC
)
2
(1 3a
2

)
From DP criterion, Eq. 3.1:
3
2
a

(
11
+
22
) +
R
MC
(1 3a

)
_
(1 3a
2

)
k = 0 (4)
one obtains
R
MC
=
_
(1 3a
2

)
(1 3a

)
_
3a

(
11
+
22
)
2
+k
_
(5)
Identication with MohrCoulomb criterion, Eq. 4
sin = 3a

_
(1 3a
2

)
(1 3a

)
, c cos = k
_
(1 3a
2

)
(1 3a

)
(6)
Associated ow a

= a

=
tan
_
9 + 12 tan
2

, k =
3c
_
9 + 12 tan
2

(7)
Deviatoric ow a

= 0
a

=
sin
3
, k = c cos (8)
a

specied:
a

=
sin
3
_
a

sin +
_
1 3a
2

_
1
, k = c cos
_
a

sin +
_
1 3a
2

_
1
(9)
Window 3-8
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.3:3
Preface Material Models Plasticity
Window 3-9: Matching of elastic domain
Plane strain conditions are assumed.
DP criterion (square form of Eq. (3.3.3):
a
2

I
2
1
2a

I
1
k +k
2
= J
2
(1)
and invariants:
I
1
= (
11
+
22
)(1 +)
J
2
=
1
3
_
(
11

22
)
2
(1 +
2
) +
11

22
(1 2)
2

+
2
12
(DP) :
_

11

22
2
_
2
+
2
12
= k
2
2a

k(1 +)(
11
+
22
)
+(
11
+
22
)
2
_
a
2

(1 +)
2

1
12
(1 2)
2
_
(MC) :
_

11

22
2
_
2
+
2
12
= c
2
cos
2
2

11
+
22
2
sin c cos
+
_

11
+
22
2
_
2
sin
2

Matching the constant, linear and quadratic terms (


11
+
22
) yields:
a

=
sin
2(1 +)
k = c cos
= 0.5
i.e. stressstateindependent matching is possible for arbitrary c and only when = 0.5.
Alternatively, when c = 0, matching is possible for arbitrary for a specied .
Orthotropic Plane Strain:
For such a case one can obtains:
a

=
sin
2(1 +
t
)
k = c cos

t
=
13
=
23
= 0.5
=
12
, arbitrary.
This adjustment corresponds to plane strain with
33
= 0.5(
11
+
22
) i.e. s
33
= 0.
Window 3-9
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.3:4
Preface Material Models Plasticity
3.3.4 CAP MODEL
While the constitutive models described previously could be applied to any kind of material,
the following one is more specic to soils.
Model is describe in the subsequent windows:
Yield surface
It combines the DruckerPrager criterion with an ellipsoidal cap closure analogous to the
CAMCLAY ellipse and the tensile cutto dened in Section 3.3.3 (if needed). Multisur-
face plasticity algorithms require the cap denition to be extended to the zone which is
covered by the DP criterion (that is for p < p
cs
) where it takes the form of a cylinder. p
c
denotes the preconsolidation pressure denes the current cap size.
Flow rule
Associative ow is assumed on the cap; the corresponding ow vector is derived in Sec-
tion 3.3.8.
Hardening law
The hardening law denes the evolution of the size of the cap yield surface. This requires
the evolution law for p
c
as a function of plastic strain. The corresponding derivation
is given in Window 3-11, where Eqs (2) and (3) dene respectively the total and the
elastic contributions in (4); Eq.( 5) results which relates the hardening parameter p
c
to the
volumetric plastic strain
Remark:
Underlined variables are positive in compression.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.4:1
Preface Material Models Plasticity
Window 3-10: Cap model: Yield surface
Yield function criterion
DruckerPrager criterion:
F
DP
= a

I
1
+
_
J
2
k = 0
Cap:
F
C1
= q
2
+
M
2
(R 1)
2
(p p
c
)(p +p
c
2p
cs
) = 0 if p p
cs
F
C2
= q
2
+
M
2
(R 1)
2
(p
cs
p
c
)(p
c
p
cs
) = 0 if p < p
cs
Tensile cuto:
F
CT
=
1

3
I
1
+
_
J
2

3
I

1T
= 0
with
q =

3J
2
, J
2
=
1
2
s
ij
s
ij
p =
I
1
3
, p
cs
=
p
c
+ (1 R)p
T
R
M = 3

3a

, p
T
=
k
3a

June 25, 2003


Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.4:2
Preface Material Models Plasticity
Flow vectors
Dened in reduced stress space = q, p are as follows:
(DP): generally nonassociated ow dened by a

r
DP
=
_
1/

3
3a

_
CAP: associated ow rule is used for both segments of CAP surface
r
C1
=
_
_
2q
_
M
R 1
_
2
(2p 2p
cs
)
_
_
if p p
cs
r
C2
=
_
2q
0
_
if p < p
cs
CutO: associated ow rule is used
r
CT
=
_
1/

3
3/

3
_
Window 3-10
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.4:3
Preface Material Models Plasticity
Window 3-11: Cap hardening
Recall:
relation between volumetric strain rate and void ratio rate
d
kk
=
de
1 +e
0
(1)
void ratio evolution : derived from the logarithmic (e ln p) approximation of the virgin
consolidation path
de =

p
c
dp
c
(2)
void ratio variation for elastic unloading path
de
s
= (1 +e
0
)d
kk
= (1 +e
0
)
dp
c
K
(3)
Evolution law for p
c
elasto-plastic volumetric strain increment

tot
kk
=
e
kk
+
p
kk
(4)
From (13) the hardening law for p
c
is derived
d
p
kk
=
_

1 +e
0
1
p
c

1
K
_
dp
c
Evolution law for cap shape R parameter
R = R
IN
(R
IN
R
0
)
p
c
p
c0
a +p
c
p
c0
(5)
parameters R
IN
and a are set automatically by the numerical procedure to preserve approx-
imately the same dilatancy d =
Q/p
Q/q
for stress paths with stress ratio
q
p
=
M
2
. These
depend on initial preconsolidation pressure p
c0
and on p
T
. R
0
is given by the user. This
parameter enables proper modelling of the K
0
coecient for normally consolidated state
(R
0
> 1).
Window 3-11
Window 3-12: Evaluation of p
co
, R
o
from oedometer test
The initial preconsolidation stress p
co
and cap shape parameter R
o
can be set based on
oedometer test once
VM
(vertical stress at which transition from secondary to primary
consolidation path occurs) and K
NC
o
(K
o
coecient at state of normal consolidation) values
are given (see Window 3-13).
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.4:4
Preface Material Models Plasticity
In the oedometer test the following relation holds (assuming that plastic strains are large
compared to elastic ones):
d
p
V
=
3
2
d
p
D
where: d
p
D
=
_
2
3
de
p
ij
de
p
ij
, d
p
V
= d
p
ii
.
This equation can be rewritten in the form (using plastic ow rule and eect of hardening):
1
H
n
2
p
dp +
1
H
n
p
n
q
dq =
3
2
_
1
H
n
q
n
q
dq +
1
H
n
q
n
p
dp
_
where:
n
p
=
Q
C1
p
(Q
C1
= F
C1
elliptic cap surface)
n
q
=
Q
C1
q
dq/dp =
K
o
(along K
o
path)
H =
F
C1
p
c
p
c

p
V
n
p
(plastic modulus for constant shape ratio parameter R)

K
o
=
3 (1 K
NC
o
)
2 K
NC
o
+ 1
Window 3-12
Window 3-13: Procedure of evaluation of p
co
, R
o
from oedometer test
Given material properties: e
o
, E, , , , DP-size adjustment (a
k
, a

VM
(vertical stress at the transition point from secondary to primary consolidation line),
K
NC
o
(K
o
at state of normal consolidation)
Find: p
co
, R
o
initialize:
i = 0
p
T
=
a
k
3a

M = 3

3a

p
(i=0)
co
=
2 K
NC
o
+ 1
3

VM
next iteration: i = i + 1
nd R
o
(see Window 3-14) (for p
T
= 0)
nd modied shape ratio parameter R
IN
for real value of p
T
and p
(i)
co
(see Window 3-15)
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.4:5
Preface Material Models Plasticity
nd corrected p
(i+1)
co
value (see Window 3-16)
iterate until [ p
(i+1)
co
p
(i)
co
[ >10
8
Window 3-13
Window 3-14: R
o
evaluation
Given: M, p
co
,
K
o
= q/p (at K
o
path) =
3 (1 K
NC
o
)
2 K
NC
o
+ 1
Find: R
o
(using bisection method)
Initialize:
i = 0
R
(i=0)
o
= 1.01;
R = 10
3
i = i + 1
R
(i)
o
= R
(i1)
o
+ R
for given: M, R
(i)
o
, p
co
,
K
o
compute mean stress p at the intersection point of elliptic cap
surface F
C1
and K
o
line
compute corresponding deviatoric stress: q =
K
o
p and n
p
, n
q
,
F
C1
p
c
compute residuum of the governing equation for oedometer test:
f
K
o
= n
p
/n
q

3
2
;
if i > 1 then
if f
last
K
o
f
K
o
0 set: R
(i+1)
o
= ( R
(i)
o
R
o
/2) and EXIT
save: f
last
K
o
= f
K
o
and go to next iteration
Window 3-14
Window 3-15: R
IN
evaluation
Given: p
co
Find: modied shape ratio parameter R
IN
such that dilatancy parameter d = n
p
/n
q
is
the same along trial stress path
M
= M/2 (some arbitrary path) both for elliptic cap with
p
T
= 0 and cap surface with real p
T
value.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.4:6
Preface Material Models Plasticity
Initialize:
Compute dilatancy parameter d
o
= n
p
/n
q
for given: p
co
, p
T
= 0,
M
and shape ratio
parameter R
o
set: R
IN
= 0.01
set: R
last
= R
o
i = 0
i = i + 1
R
(i)
IN
= R
(i1)
IN
+ R
for given M,p
co
, p
T
,
M
and shape ratio parameter R = R
(i)
IN
compute compute mean
stress p at the intersection point of elliptic cap surface F
C1
and stress path line q/p =
M
compute corresponding deviatoric stress: q =
K
o
p and n
p
, n
q
compute dilatancy parameter d = n
p
/n
q
if i > 1 then
if d
o
> d
last
AND d
o
< d OR
d
o
> d AND d
o
< d
last
then
set: R
(i+1)
IN
= R
(i)
IN
+
d
last
d
o
d
last
d
R
IN
and EXIT
set: d
last
= d and go to next iteration
Window 3-15
Window 3-16: Evaluation of corrected p
(i+1)
co
value
Given: M, R
IN
, p
T
, ,
VM
Find: p
(i+1)
co
Set:
K
el
o
=

1

K
o
el
=
3(1 K
el
o
)
1 + 2K
el
o
p =
(1 + 2K
el
o
)
VM
3
q =
K
o
el
p
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.4:7
Preface Material Models Plasticity
For p, q, p
T
, M solve quadratic equation (elliptic cap equation) F
C1
= 0 for
unknown p
(i+1)
co
value
Window 3-16
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.4:8
Preface Material Models Plasticity
3.3.5 MOHR-COULOMB (M-W)
Yield surface
The original MC criterion (see Window 3-6) which leads to a nonsmooth multisurface plas-
ticity problem, is substituted by its smooth, singlesurface approximation, being a particular
case of a general 3parameter criterion developed recently by Mentrey (see Menetrey, Willam:
A triaxial failure criterion for concrete and its generalization. ACI Structural Journal 92(3)
p.311318). This criterion takes the form described in Window 3-17.
Window 3-17: Menetrey criterion
.
F(, , ) = (A
f
)
2
+m
f
[B
f
r
f
(, e) +C
f
] D
f
= 0 (1)
where , , are Haigh-Westergaard stress coordinates equal to:
=
1

3
I
1
(2)
cos 3 =
3

3
2
J
3
J

3
2
2
(3)
=
_
2J
2
(4)
with I
1
, J
2
, J
3
being the usual stress invariants (??eq:WinElast11)
Function r
f
= r
f
(, e), 0.5 < e 1, describes the shape of the surface in deviatoric section
r
f
(, e) =
4 (1 e
2
) cos
2
+ (2e 1)
2
2 (1 e
2
) cos + (2e 1)
_
4 (1 e
2
) cos
2
+ (2e 1)
2
(1 e
2
)

1/2
(5)
Window 3-17
The eccentricity parameter e can be calibrated to t exactly the MohrCoulomb surface
on both extension and compression meridians which leads to the smooth MohrCoulomb
surface. All other parameters of the generalized criterion are also expressed in terms of the
MohrCoulomb friction angle and cohesion c as shown in Window 3-18.
Window 3-18: MohrCoulomb yield criterion by Menetrey criterion
Given:
friction angle, 0

< < 90

c cohesion c 0
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.5:1
Preface Material Models Plasticity
e =
3 sin
3 + sin
(1)
A
f
= 0 (2)
B
f
=
3 sin

24c cos
(3)
C
f
=
1

3c
tan (4)
m
f
= 1 (5)
D
f
= 1 (6)
Deviatoric sections of the smooth Mohr-Coulomb yield surface: a) =10

, e=0.89, b)
=50

, e=0.59
Window 3-18
Flow rule
The ow rule which denes the direction of the plastic ow is given in a standard form:

p
=

The ow potential adopted here takes a form similar to the yield surface, but with assumption
that r is independent of Lodes angle
Q(, ) = (A
q
)
2
+m
q
(B
q
r
q
+C
q
)
radius r
q
is taken as the radius r
f
for the extension meridian.
Other parameters of ow potential A
q
, B
q
, C
q
, m
q
are evaluated in a manner analogous to
yield parameters A
f
, B
f
, C
f
, m
f
by formulae of Window 3-18, but using the dilatancy angle
instead of the friction angle .
The recommended form of the ow potential is identical with the DruckerPrager ow po-
tential. It leads to non associated plasticity even in case when = due to the dierent
forms of r
q
and r
f
. Alternative forms proposed here are tabulated next.
Matching the smooth MohrCoulomb criterion with MohrCoulomb criterion
The proposed smooth MohrCoulomb criterion reduces to a vonMis`es criterion through
external vertices when the excentricity factor e tends towards 1, i.e when tends to zero. A
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.5:2
Preface Material Models Plasticity
Table 3.2: Plastic ow parameters
Flow type A
q
B
q
C
q

0
1

2k

3a

DruckerPrager
0 a

and a

are computed from , c using tensile meridian adjustment


(putting instead of into formula for k one gets k

value)
1
2
_
2 tan
c
f
c
f
t
_

3
_
1

2
tan
c
_
1

2
B
q
+
2

3
f
t
4

<
c
<
35.3

Axisymmetric HoekBrown
(HoekBrown or
concrete models only)

c
- dilatancy angle for uniaxial compression
_
=
f
c

3
, =
_
2
3
f
c
, =

3
_
size adjustment must therefore be activated i.o. to achieve the proper stability results. This
consists in replacing the user dened cohesion c and friction angle by adjusted values which
can be derived following the same reasoning as for the DruckerPrager criterion; this is done
in Window 3.3.3-3.
Window 3-19: Matching of collapse load (planestrain conditions)

p
ij
= d r
ij
= d
_
a

ij
+
1
2

J
2
s
ij
_
(1)
(plane strain,
e
<<)

p
33
=
p
13
=
p
23
= 0 (2)
from (1)
s
33
= 2a

_
J
2
; s
13
= s
23
= 0 (3)
I
1
=
3
2
(
11
+
22
) 3a

_
J
2
=

3 (4)
J
2
=
_
[(
11

22
) /2]
2
+
2
12
_
(1 3a
2

)
(5)
J
2
=
R
2
(1 3a
2

)
=

2
2
(6)
replacing in Eq. (1) one gets, after some transformations
R =

6
A
c
f
cos
f

6
2A
sin
f
(
11
+
22
); (7)
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.5:3
Preface Material Models Plasticity
index f is added on c and sin , cos to avoid confusion in Eq. (10) below:
A =

2
_
3 sin
f
_
r(e, ) 2

6a

sin
f
2
_
1 3a
2

. (8)
Identifying with the MohrCoulomb criterion of Window 3-17
c cos =

6
A
c
f
cos
f
(9)
sin =

6
A
sin
f
(10)
and alternatively
c
c
f
=
tan
tan
f
. (11)
Special cases:
arbitrary ow
sin
f
=
r(e, )(3 sin
f
) 2

3a

sin
f
2

3
_
1 3a
2

sin (12)
results from
cos 3 =
3

3
2
a

J
3
J
3/2
2
(13)
which for plane strain failure and
e
neglected yields:
deviatoric ow yields
a

= 0 yields = 30

(14)
sin
f
=
r(e, )(3 sin
f
)
2

3
sin (15)
if in addition = 0 , then c
f
= 0.866c
Window 3-19
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.5:4
Preface Material Models Plasticity
3.3.6 HOEKBROWN CRITERION (SMOOTH)
The empirical strength criterion proposed by Hoek and Brown for rock masses is:
f(
1
,
2
) =
_

3
f
c
_
2
+
hb

1
f
c
c
hb
= 0
where
1
,
3
the material parameters c
hb
and
hb
are measures of cohesive and frictional
strength, and f
c
designates for uniaxial compressive strength. This yield surface is generated
with the general yield surface presented in Window 3-20 by identication of the adjustment
parameters as presented next.
Window 3-20: Smooth Hoek-Brown criterion
Given f
c
, f
t
and e , the uniaxial compressive, and tensile strength, and the surface eccentricity.
0.5 < e 1 and the following eccentricityvalue is recommended 0.5 < e 0.6.
A
f
=

1.5f
t
f
c
, B
f
=
1

6 f
c
, C
f
=
1

3 f
c
, m+f = 1, c = f
t
Calibration of the generalized criterion to: 1) HuberMises, 2) parabolic Leon, 3) smooth
HoekBrown
Window 3-20
Remark:: The same ow rules as for the Mohr-Coulomb criterion applies.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.6:1
Preface Material Models Plasticity
3.3.7 CUT-OFF CONDITION AND TREATMENT OF THE APEX
As an additional feature of most criteria a tensile Cuto condition of the following form may
be attached if required
I
+
1
max
I
1
t
The presence of the Cuto condition places the problem into the class of multisurface
plasticity problems as the cuto may be formally treated as additional stress constraint.
Treatment of the apex
The ow potential accompanying the Cuto surface has an associative form:
Q() = I
1
.
If the Cuto is disregarded the surface possesses an apex located at the stress point

A
=
_
_

3
3

A
,

3
3

A
, 0,

3
3

A
[, 0, 0
[, 3D
_
_
with

A
=
D
f
m
f
C
f
If the trial stress state is located inside of the apex cone which means it fullls condition:

tr
< (
tr

A
)
GB
f
r
q
KC
q
it would return to the apex.
In the case of an active Cuto condition, limiting value I
1
t
should be compatible with the
position of the surface apex, i.e.

t
< (1 )
A
with
t
=
I
1
t

3
and = 10
2
If the above condition is not met, then the maximum possible cuto position is set auto-
matically as:
I
1
t
=

3 (1 )
A
.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.7:1
Preface Material Models Plasticity
3.3.8 MULTILAMINATE MODEL
In this model the existence of up to ML
max
= 3 weakness planes characterizing anisotropy
of the material behaviour is assumed. On each plane separately, MohrCoulomb plasticity
condition and a tension cuto condition must be fulllled.
Window 3-21: Weakness plane plasticity conditions
Yield function and ow potential isolines
F
(1i)
= +
n
tan
i
c
i


F
(1i)
=
_
tan
i
, 1
_
T
(1)
Q
(1i)
= +
n
tan
i


Q
(1i)
=
_
tan
i
, 1
_
T
(2)
F
(2i)
= +
n
tan
i
c
i


F
(2i)
=
_
tan
i
, 1
_
T
(3)
Q
(2i)
= +
n
tan
i


Q
(2i)
=
_
tan
i
, 1
_
T
(4)
F
(3i)
=
n
f
t


F
(3i)
= 1, 0
T
(5)
Q
(3i)
=
n


Q
(3i)
= 1, 0
T
(6)
Note:

2

= 0.
Window 3-21
This leads to a multisurface plasticity problem which require that a set of up to 3*ML
max
plasticity conditions must be simultaneously fullled by any stress state in the multilaminate
material, i.e
F

_

(i)
_
0; J; J : 1, . . . , 3ML
T
i
, T are linear transformation matrices describing transition between (
xx
,
yy
,
xy
,
zz
),
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.8:1
Preface Material Models Plasticity
global coordinate system stress components and the i-th weakness plane components
n
,
T

(i)
=
_

n

_
(i)
= T
(i)
_
_
_

xy
_
_
_
, T
(i)
=
_
s
2
c
2
2sc
sc sc c
2
s
2
_
where :
s = sin
(i)
c = cos
(i)
.
Plastic strains emerge due to violation of any of those conditions by the trial elastic stress.
The total plastic strain is the sum of each planes contribution.

p
=


p
=


p
NB : A perfectly elastoplastic behaviour (no hardening) is assumed.
The ow rule is governed by a ow potential Q, the form of which is analogous to the form
of corresponding yield function F. As the dilatancy angle
i
specied for each weakness
plane may in general dier from the corresponding friction angle, the ow rule adopted is
nonassociative (Q

,= F

) .
The model requires the data of ML, the number of assumed weakness planes (ML 3).
For each i-th plane , (i = 1, .., 3ML) the following data should be specied:

(i)

inclination angle of ith weakness plane


(90 degrees, positive counterclockwise)

(i)
friction angle

(i)
dilatancy angle
c
(i)
cohesion.
The following constitutive model is derived.
Window 3-22: Physical origin of ow potential Q
During plastic slip ( = tan
n
+c):
dilatancy angle, ,=
if = associativity is preserved.
Window 3-22
Window 3-23: Constitutive equations of multilaminate model
= D : (
p
) (1)

p
=

(2)
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.8:2
Preface Material Models Plasticity
Frictional slide model to explain physical origin of ow potential
with yield and loading/unloading conditions for 1, 2, ...3ML

0 (3)
F

() 0 (4)

() = 0 (5)

() = 0. (6)
In expanded form
if F

() 0 or F

() = 0 and

F

() < 0

= 0
if F

() = 0 and

F

() = 0

= 0
( constraint is active).
Window 3-23
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.8:3
Preface Material Models Plasticity
3.3.9 MODIFIED CAM CLAY MODEL
Window 3-24: Yield and ow potential surface
The elliptical yield surface (and ow potential as well) is described by the equation (see Fig.
below)
F(,p
c
) = q
2
+M
2
c
r
2
() p (p p
c
) = 0
Modied-Cam Clay yield surface
The M
c
parameter is the slope of the critical state line along compression meridian, p
c
is a
preconsolidation pressure adjusted along p axis and r() is a function of Lode parameter
describing shape of the yield surface in the deviatoric plane. The r() function is taken after
van Ekelen.
r () =
_
1 sin(3)
1
_
n
sin(3) =
3

3
2
J
3
J
3
2
2
n = 0.229 0.7925
The relation between k = M
E
/M
C
(M
E
is the slope of critical state line for the tension
meridian) and parameter is as follows
=
k
1
n
1
k
1
n
+ 1
and the default setting for the parameter k can be set as
k =
3
3 +M
c
Remark:
The applied r() function is applicable for friction angles up to 46.55
o
.
Window 3-24
Window 3-25: Nonlinear elastic behavior within yield surface
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.9:1
Preface Material Models Plasticity
The reversible part of the deformation is governed via nonlinear elasticity assuming that
K = K(p), G/K =const. and the well-known formula for bulk modulus is applied
K =
1 +e
o

p
where e
o
is an initial void ratio, is the slope of secondary compression line in eln(p) axes.
Remarks
Shear modulus G is a linear function of p
Poissons coecient is a constant
Mean stress may reach a zero value for innitely large tensile volumetric strains
Any calculation carried out with Cam Clay model requires explicit setting of the initial
stresses
Window 3-25
Window 3-26: Hardening/softening law
The evolution of the hardening parameter is dened through the following equation
dp
c
=
1 +e
o

p
c
(d
p
kk
)
where is a slope of a primary compression line in eln(p) system (see Fig. given below).
Isotropic compression test
Window 3-26
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.9:2
Preface Material Models Plasticity
Window 3-27: Modeling normally consolidated soils with Modied Cam-Clay model
The Modied Cam-Clay model can describe basic macroscopic phenomena observed for nor-
mally consolidated cohesive soils. The model behavior is well represented if we consider a
standard triaxial compression test. In that case (OCR = 1), as shown in Fig. below, the
yield surface follows the current stress state.
0
50
100
150
200
250
300
350
0 50 100 150 200 250 300 350
p
q
1
M
C
Stress path
Initial state
Intermediate state
Limit state
CSL
OCR=1
Drained triaxial compression test (OCR = 1)
During the application of an axial strain to the specimen the compressive volumetric strain
is produced while deviatoric stress grows monotonically up to the value q = M
c
p (see Fig.
below).
0
20
40
60
80
100
120
140
160
0 0.02 0.04 0.06 0.08
e ps -de v
q
-0.006
-0.005
-0.004
-0.003
-0.002
-0.001
0
0 0.02 0.04 0.06 0.08
e ps -de v
e
p
s
-
v
o
l
OCR=1
OCR=1
densification
Stress/strain characteristics (OCR = 1)
Window 3-27
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.9:3
Preface Material Models Plasticity
Window 3-28: Modeling overconsolidated soils with Modied Cam-Clay model
The Modied Cam-Clay model describes also some macroscopic phenomena observed for
overconsolidated cohesive soils. A typical model behavior, analyzed for a drained triaxial
compression test (for OCR = 5) is shown in gures given below. In that case the eective
stress path passes the critical state line until the current yield surface is met and then goes
down until critical state is achieved. This eect corresponds to strain softening.
0
100
200
300
400
500
600
0 100 200 300 400 500 600
p
q
M
C
1
CSL
Initial state
Intermediate state
Limit state
OCR=5
Drained triaxial compression test (OCR = 5)
During the application of the axial strain to the specimen the compressive elastic volumetric
strain is produced rst. Once the yield surface is met a dilatant volumetric strain is grow-
ing up tending to an asymptote at the critical state. The deviatoric stress does not grow
monotonically exhibiting peak and residual values.
0
50
100
150
200
250
0 0.01 0.02 0.03 0.04
e ps -de v
q
-0.0005
0
0.0005
0.001
0.0015
0.002
0.0025
0 0.01 0.02 0.03 0.04
e ps -de v
e
p
s
-
v
o
l
OCR=5
OCR=5
Strain softening
dilation
Stress/strain characteristics (OCR = 5)
Window 3-28
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.3.9:4
Preface Material Models
3.4 CREEP
Creep is a timedependent deformation under maintained stress. It is assumed that the stress
can be split into volumetric and deviatoric components and the corresponding timedependent
strain components are the volumetric and the deviatoric creep. The following formulation is
adopted for onedimensional creep :

cr
=
e
inst
f(t) = C (t)
Creep is considered to be proportional to the instantaneous elastic deformation. C(t) is the
creep law corresponding to a unit stress.
The three-dimensional creep law is then:

cr
= ED
1
C(t) = D
1
o
C(t)
and C(t) is assumed to be e.g. of the form
C(t) = A(t t
0
)
m
where A and m are material parameters.
In most situations the same creep law will be adopted for both the volumetric and the
deviatoric components. Dierent parameters can however be chosen for the volumetric and
deviatoric creep components. Great care has to be taken in that case because this may
generate a Poisson coecient which varies in time.
CREEP UNDER VARIABLE STRESS
CREEP PARAMETER IDENTIFICATION FROM EXPERIMENT
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.4:1
Preface Material Models Creep
3.4.1 CREEP UNDER VARIABLE STRESS
Creep under variable stress requires a principle of superposition. The adopted power law does
not lend itself easily to such in principle. It is therefore replaced, in the implementation, by
a series of Kelvin elements. This is described next in Windows 3-30 and 3-29.
The automatic adjustment of the Kelvin element parameters to the prescribed power law is
derived in Window .3-31
Window 3-29: Kelvin element under constant unit stress
Constitutive equation
q = G +
Creep strain increment, due to
q = H(t t
0
)
where: H Heaviside unit step function

cr
=
1
G
_
1 exp
_

(t t
0
)
__
= A
_
1 exp
_

1
B
(t t
0
)
__
=
= C(t t
0
) with = 1
N.B.:

cr
= 0 at t = t
0

cr
=
q
G
at t =
Kelvin mechanical model
Window 3-29
Window 3-30: Creep under variable stress
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.4.1:1
Preface Material Models Creep
Let the creep strain under variable stress be

cr
(t) = D
1
0
_
_
_

t
_

0
[ (t) (
0
)]
C(t )

d
_
_
_
with D
0
= E
1
D
For unit stress and a single Kelvin element
C(t, ) = A
_
1 exp
_

1
B
(t )
__
C

=
A
B
exp
_

t
B
_
For a chain of N Kelvin elements in series, and introducing the volumetric-deviatoric split
C
v
(t, ) =
N

i=1
A
v
i
_
1 exp
_

1
B
v
i
(t )
__
C
d
(t, ) =
N

i=1
A
d
i
_
1 exp
_

1
B
d
i
(t )
__

cr
m
(t) =
_
D
v
0
v
_
1
_
_
_

t
_

0
[
m
(t)
m
(
0
)]
C
v

d
_
_
_
e
cr
(t) =
_
D
d
0
_
1
_
_
_

t
_

0
[s (t) s (
0
)]
C
d

d
_
_
_
The creep strain increment, required by the general nonlinear incremental scheme, as de-
scribed in section FULL/MODIFIED NEWTON-RAPHSON ALGORITHM (4.6.1) can be
derived by recurrence

cr
=
cr
n+1

cr
n
= D
1
0
N

i=1

cr
i
and
(
i
)
cr
n+1
= [
n
+
n+1
(
0
)] A
i
_
1 exp
_

t
n+1
B
i
__
+
_
exp
_

t
n+1
B
i
_
1
_
(
i
)
cr
n
(
i
)
cr
n+1
= (
i
)
cr
n
+
cr
i
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.4.1:2
Preface Material Models Creep
In the case of nonlinear creep the above recurrence formula is modifed by an additional term
scaling the amplitude of the creep strain A
i
in the following manner
(
i
)
cr
n+1
= [
n
+
n+1
(
0
)] (C
1 n
+C
1 n+1
)
A
i
_
1 exp
_

t
n+1
B
i
__
+
_
exp
_

t
n+1
B
i
_
1
_
(
i
)
cr
n
The nonlinear term C
1
() is expressed by the equation
C
1
() =1+a SL
b
where a (0 a 10) and b (1 b 10) are material parameters and SL is a stress level
expressing the relative distance of the stress state from the yield surface.
Eqs (????) also apply for volumetric and deviatoric creep components when , , D
0
, A
i
, B
i
are specialized appropriately.
(D
v
0
)
1
=
E
3K
= 1 2
and
_
D
d
0
_
1
is a (4 4) matrix with the following diagonal
diag
_
D
d
0
_
1
=
_
(1 +) , (1 +) , 2 (1 +) , (1 +)

which is valid for axisymmetry and for plane strain.


Window 3-30
Window 3-31: Approximation of the creep law
Creep function identication
Parameter identication procedure
Initial approximation
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.4.1:3
Preface Material Models Creep
1. Select
_
B
1
, B
2
, . . . , B
j

=
_
10
4
, 10
3
, . . .

N.B.: Six components [10


4
, . . . , 10] seem to be appropriate, less may be sucient
2. Dene
_
t
1
, t
2
, . . . t
j

such that
A
j
= A
j
_
1 exp
_
t
j
B
j
__
Select : 0.5 e.g., then
t
j
= B
j
ln(0.5)
3. Dene such that for j = 1 to J
C(t
j
) =
j1

k=1
A
k
+A
j
_
1 exp
_

t
j
B
j
__
hence
A
j
=
_
1 exp
_

t
j
B
j
__
1
_
C (t
j
)
j1

k=1
A
k
_
2nd approximation
A
(2)
=
_
1 exp
_

t
j
B
j
__
1
_
C (t
j
)
j1

k=1
F
k
(t
j
)
_
F
k
(t
j
) = A
(1)
k
_
1 exp
_

t
j
B
j
__
.
Iterative scheme
A
(0)
j
=
_
1 exp
_

t
j
B
j
__
1
_
C (t
j
)
j1

k=1
A
(0)
k
_
i-th approximation
A
(i)
j
=
_
1 exp
_

t
j
B
j
__
1
_
C (t
j
)

k=j
F
k
(t
j
)
_
F
k
(t
j
) = A
(i1)
k
_
1 exp
_

t
j
B
k
__
.
t
j
= B
j
ln (0.5)
B
j
= 10
j4
[days] 1 < j < 6.
Convergence test
max
_
A
(i)
j

A
(i1)
j
A
(i)
j
_
< 10
2
Window 3-31
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.4.1:4
Preface Material Models Creep
3.4.2 CREEP PARAMETER IDENTIFICATION FROM EXPERI-
MENTS
The easiest way to adjust creep parameters A and m for C(t, t
0
) = A(t t
0
)
m
from an
experimental curve f(t, t
0
) is to pick two points situated at (t t
0
) = 1 for the adjustment
of A and then at a large value of (t t
0
) for the adjustment of m.
Depending on the available test the procedure diers only slightly. Four typical experiments
are analyzed next; additional situations can easily be extrapoled using D
1
0
with
D
1
0
=
_

_
1 0
1 0
0 0 2(1 +) 0
0 1
_

_
.
If volumetric and deviatoric creep are dierent i.e.
C
v
(t, t
0
) = A
v
(t t
0
)
m
v
,= C
d
(t, t
0
) = A
d
(t t
0
)
m
d
then
(D
v
0
)
1
=
E
3K
= 1 2
and
_
D
d
0
_
1
=
_

_
1 + 0 0 0
0 1 + 0 0
0 0 2 (1 +) 0
0 0 0 1 +
_

_
.
Window 3-32: Uniaxial test

c
= D
1
0
C (t t
0
)
_

c
1

c
2
0

c
3
_

_
=
_

_
1 0
1 0
0 0 2(1 +) 0
0 1
_

_
_

1
0
0
0
_

_
C(t t
0
)
hence with
C(t t
0
) = A(t t
0
)
m

c
1
=
1
A(t t
0
)
m
.
Window 3-32
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.4.2:1
Preface Material Models Creep
Window 3-33: Triaxial test
Assuming
c
v
is measured in the experiment:
_

c
1

c
2

c
12

c
3
_

_
=
_

_
1 0
1 0
0 0 2(1 +) 0
0 1
_

_
_

2
0

3
_

_
C(t t
0
)
hence, with
C(t t
0
) = A(t t
0
)
m

c
v
= 3 (1 2)
m
A(t t
0
)
m
.
Window 3-33
Window 3-34: Triaxial deviatoric test
Assuming
c
2

c
2
=
c
is measured for a unit q = (
1

2
)

c
d
= D
1
0
_

2
0

3
_

_
and
3
=
1
then

c
2
= (
2
2
1
) C (t, t
0
)

c
1
= [
1
(
2
+
1
)] C (t, t
0
)

c
= (1 +) qC (t t
0
) .
Window 3-34
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.4.2:2
Preface Material Models Creep
Window 3-35: Oedometer test

c
= D
1
0
_

_
K
0

2
0
K
0

2
_

_
C (t, t
0
)
then

c
2
= (1 2K
0
)
2
C (t, t
0
) .
Window 3-35
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.4.2:3
Preface Material Models
3.5 SWELLING
A nonlinear creep approach has been adopted here to model swelling phenomenon
2
. Creep
seems to be the simplest phenomenological approach which can reproduce macroscopic be-
haviour of swelling soils/rocks both quantitatively and qualitatively without considering all
the eects in microscale.
Oedometric swelling test
Memorizing of in situ stress
0
Correction of
os
with respect to
o
state
Three-dimensional generalization
An oedometric swelling test is taken as the basis for further three-dimensional generalization
according to the theory given by Wittke and Kiehl
3
.
A typical relation between swelling strain and vertical stress (after Huder and Amberg) ob-
tained from oedometer test is shown in g. below.
Window 3-36: Oedometric swelling test
Five basic parameters are needed to calibrate swelling :
os
,
cs
, , B
o
. and
s
.
Parameter
os
denes a minimum vertical stress value which stops swelling evolution while

cs
bounds the excessive increase of the swelling for small or even tensile stress state.
Parameter denes the slope of the line
s
y

y
for stress range
cs

y

os
in semi-
logarithmic system. Parameter B
o
denes maximum swelling evolution rate while
s
reduces
2
concerns versions: ACADEMIC, PROFESSIONAL, EXPERT only
3
W.Wittke. Stability Analysis of Tunnels.Fundamentals.Verlag Glueckauf Essen, 2000.
J.R.Kiehl.Interpretation der Ergebnisse von Grossquellversuchen in situ durch dreidimensionale numerische
Berechnungen, Proc. 7th.ISRM Congress, Aachen 1991,pp.1534-1538)
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.5:1
Preface Material Models
swelling evolution rate according via formula:
B() =
B
o
1 exp(
s
)
= MIN (1,
|
0
|
|
REF
os

REF
cs
|
)
where the current eective stress is denoted by , in situ stress is denoted by
0
and
appropriate reference stress states
REF
os
,
REF
cs
are dened in the following windows.
Window 3-36
Window 3-37: Memorizing of in situ stress
0
The reference stress
0
is required for evaluation of swelling rate function B() and has to
be memorized during computation. As the computation scenario can be very complex the
following recipe is used to set up
0
reference state:
If the initial state analysis is run then for all active (at time t = 0) materials
0
is taken as
the result of the last step of the initial state.
If the initial state is not specied in list of drivers but swelling is activated for some materials
then
0
is taken as the stress state at rst converged step since material activation. NB. In
such case swelling strain increment will not be generated in the rst time step (since material
activation).
Window 3-37
Window 3-38: Correction of
os
parameter with respect to
0
state
If K
o
axes do not coincide with global x-y-z system axes
transform

0
=T
GL

0
else

0
=
0
IF

0
y
<
os
modify parameter
os
:
os
=

0
y
.
Window 3-38
Window 3-39: Three-dimensional generalization
We assume that the generalized reference stress state
REF
os
and
REF
cs
, in xyz system, can
be dened as follows:

REF
os,xyz
=
_
K
ox

o

o
0 K
oz

o
0 0
_
T

REF
cs,xyz
=
_
K
ox

c

c
0 K
oz

c
0 0
_
T
where: K
ox
= K
oz
=
v
1 v
(if not explicitely dened)
It is possible to set up the in situ K
o
coecients K
ox
, K
oz
in local coordinate system x/y/z/
rotated with respect to xyz system. In such a case the reference stress states are dened as
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.5:2
Preface Material Models

REF
os,x

z
=
_
K
ox

o

o
0 K
oz

o
0 0
_
T

REF
cs,x

z
=
_
K
ox

c

c
0 K
oz

c
0 0
_
T
and the following transformation takes place to set them up in xyz system:

REF
os,xyz
= T
LG

REF
os,x

REF
cs,xyz
= T
LG

REF
cs,x

where T
LG
is a transformation matrix from local to global system.
The increment of creep strain (swelling strain) is computed using following recurrence formula:

CR
=
t
B()

CR(*)
The following approach is used to compute
CR(*)
for given eective stress state :
nd eigenvalues
i
and eigendirections d
i
of
transform
REF
os,xyz
,
REF
cs,xyz
to principal directions of
REF(*)
os,xyz
,
REF(*)
cs,xyz
transform accumulated creep strain
ac-CR
to principal directions of
ac-CR(*)
compute predicted swelling strain components in principal directions of using following
formula:

CR
i
=
_

_
ln

i

REF(*)
o
ii
if
REF(*)
c
ii

i

REF(*)
o
ii
0 if
i

REF(*)
o
ii
ln

REF(*)
c
ii

REF(*)
o
ii
if
i

REF(*)
c
ii
_

_
compute
CR(*)
based on
CR
i
, d
i
and
ac-CR(*)

CR(*)
=
_
MAX(
CR
i
,
ac-CR(*)
ii
)
ac-CR(*)
ii
_
d
i
d
T
i
Remarks:
An explicit integration scheme has been adopted here and thus the maximum time step value
is limited and should satisfy the condition:
0 < t <
2B
MIN
E
oed
(1)
where
MIN
is the principal normal stress which yields the largest swelling strain.
Window 3-39
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.5:3
Preface Material Models
3.6 AGING CONCRETE
4
The aging concrete model represent time dependent mechanical properties as well as rheo-
logical behavior of concrete in early age.
It consists of a set of parallel Maxwell units, as shown in the window 3-40. Each unit is
described by a maturity dependent Young modulus E
k
(M) = W
k
(M)E, and retardation
time
k
=
E
k

k
. The contribution of each unit depends on maturity measure M (expressed
in time units), by the set of weighting factors W
k
, (

k
W
k
= 1), given for specied time
instants t
i
.
Window 3-40: Aging concrete model
Maxwell chain model for aging concrete
Window 3-40
Remarks:
For details of the numerical implementation see:
AGING CONCRETE MODEL - IMPLEMENTATION SCHEME
4
concerns versions: ACADEMIC, PROFESSIONAL, EXPERT only
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.6:1
Preface Material Models
3.7 APPENDICES
SAFETY FACTORS AND STRESS LEVELS
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.7:1
Preface Material Models Appendix
3.7.1 SAFETY FACTORS AND STRESS LEVELS
The following denitions are used in Z Soil

; of course these denitions are not unique.


Window 3-41: Stress level (SL)
I. for axisymmetric criteria
SL =
q
q
failure
with q =
_
3J
2
II. for Lodes angle dependent criteria
SL =
1

with
failure
=
m
+s
Remarks:
1. I is a subset of II
2. obviously will be dierent for each material point.
Window 3-41
Window 3-42: Safety factor (SF)
I. Let SF
1
be the multiplier which can be applied uniformly to each deviatoric stress in the
structure, such that failure occurs when

=
m
+ SF
1
s
II. For twoparameter criteria (C, ) an alternative denition is available which is
SF
2
=

failure
d
_

where is the failure surface and can, as shown elsewhere, be obtained numerically by a
progressive reduction of C and tan using:
c

=
c
SF
2
(tan )

=
tan
SF
2
.
In the general MentreyWillam denition this amounts to replace the parameters of the
criterion by
A

f
= SF
1
A
f
B

f
= SF
1
B
f
Window 3-42
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM3.7.1:1
Preface
Chapter 4
NUMERICAL IMPLEMENTATION
WEAK FORMULATION
ELEMENTS
INCOMPRESSIBLE AND DILATANT MEDIA
FAR FIELD
OVERLAID MESHES
ALGORITHMS
CONSTITUTIVE THEORIES
APPENDICES
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.0.1:2
Preface Numerical Implementation
4.1 WEAK FORM AND MATRIX FORMS OF THE PROBLEM
SINGLE PHASE MEDIUM, TIME INDEPENDENT LOADING
TWOPHASE MEDIUM, RHEOLOGICAL BEHAVIOUR
HEAT TRANSFER
In the subsequent sections the following Hughes notations is used:
w
(i,j)
=
1
2
(w
i,j
+w
j,i
)
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.1:1
Preface Numerical Implementation Weak Form
4.1.1 SINGLE PHASE MEDIUM, TIME INDEPENDENT LOAD-
ING
In the nite element solution procedure one does not solve the strong form of the boundary
value problem but an equivalent weak (variational or virtual work) form which is usually
discretized following a Galerkin technique. The equivalence of the strong or dierential form
of the problem statement and the weak form is discussed in texts concerning nite element
implementation (see [Hughes, 1987]).
Window 4-1: Weak and matrix form of single phase medium
Weak form:
_

w
(i,j)

ij
d =
_

w
i
f
i
d +
_

t
w
i

t
i
d (1)
u
i
= u
i
on
u
where
t
and
u
are boundaries with prescribed tractions and displacements, respectively,
and
= =
u
+
t
.
Approximation functions for w and u
w
i
= N
a
w
ia
(2)
u
i
= N
a
u
ia
Index a identies the element node number.
Matrix form
The matrix form results from the introduction of the appropriate derivatives of these ap-
proximations into the weak form and invoking the arbitrariness of the weighting functions
w
Ku = F (3)
u
ia
= u
ia
on
u
where:
K = /
e=1,N
(K
e
) , / the assembly of elemental stiness contributions
and K
e
is the elemental stiness contribution
K
e
=
_

e
B
T
D B d
e
(4)
will be established when elements are discussed.
Internal force: term Ku
It can alternatively be written as
K u = /
e=1,N
_

e
B
T
d
e
(5)
which will be needed later.
Window 4-1
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.1.1:1
Preface Numerical Implementation Weak Form
4.1.2 TWO-PHASE MEDIUM, RHEOLOGICAL BEHAVIOUR
Twophase medium. Weak form
Integration in time domain
Choice of shape functions
Matrix form for overall equilibrium
Matrix form for uid mass concervation
Coupled system of linearized equations
Explicit treatment of uid state parameters S and k
r
Special cases. Matrix form
Consolidation. Time stepping algorithm
The strong statement of the twophase boundary value problem was given in Section 2.2.
The corresponding weak statement and matrix form are shown in Windows 4-2 and 4-7.
Window 4-2 describes the weak form of the overall equilibrium equation (expressed in terms
of total stresses according to extended Bishops eective stress principle) and weak form
of the uid ow continuity equation with associated boundary conditions. The problem is
formulated in terms of nodal displacements u and pore pressures p
F
.
The actual implementation of the seepage boundary conditions is done using a twosided
surface element (see Window 2-4) in which boundary conditions are imposed on the external
face by user.
Window 4-2: Twophase medium. Weak form
Equilibrium equations
_

w
(i,j)

tot
ij
d
(total internal force)
=
_

w
i
b
i
d
(body load)
+
_

t
w
i

t
i
d
(surface traction)
_

w
F

kk
d
(skeleton def. rate)

w
F
,k
v
F
,k
d
(internal ux)

w
F
c
_
p
F
_

p
F
d
(storage)
+
_

q
w
F
q d
(surface ux)

s
w
F
k
v
(p
F
p
Fext
) d
(seepage penalty term)
= 0
Solid boundary conditions
u
i
= u
i
on
u
T

tot
ij
n
j
=

t
i
on
t
T
Fluid boundary conditions
p
F
= p
F
on
_

p
+
2
s
_
T
v
F
i
n
i
= q on
_

q
+
1
s
_
T
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.1.2:1
Preface Numerical Implementation Weak Form
Initial conditions
u
i
(t
0
) = u
i0
on
p
F
(t
0
) = p
F
0
on
Window 4-2
The matrix form of the problem results from the specic choice of shape functions for inter-
polation of displacements and pore pressure elds within the element. The matrix equations
are shown in Window 4-7. The F
int
(u) is a nonlinear internal force term due to plasticity and
thus a linearized problem statement is required. The application of the implicit integration
scheme in time permits an elimination of time derivatives from set of linearized equations
which is nally solved for u the increment of displacement u and p
F
the increment of pore
pressure p
F
. Iterations are needed if the constitutive law of the solid skeleton is nonlinear.
Window 4-3: Integration in time domain
The following predictor-corrector scheme has been adopted for integration of the displace-
ments and pore pressures in time.
u
i
N+1
= u
i
N
+ (1 ) t u
i
N
+ t u
i
N+1
= u
i
N
+ t u
i
N
. .
u
predictor
i
N+1
+ t u
i
N+1
. .
u
i
N+1
u
i
N
p
N+1
= p
N
+ (1 ) t p
N
+ t p
N+1
= p
N
+ t p
N
. .
p
predictor
N+1
+ t p
N+1
. .
p
N+1
p
N
The integration coecient has been assumed to satisfy the sucient condition for stability

1
2
.
Window 4-3
Window 4-4: Choice of shape functions
Let us introduce the following nite element interpolation functions for approximated eld(s),
i.e. displacements u
h
and pore pressure p
h
, within a nite element
u
h
= N
u
u
e
w
h
= N
u
w
e
p
h
= N
p
p
e
q
h
= N
p
q
e
With these denitions the strain-displacement relation is expressed in the standard form

_
u
h
_
= Bu
e
Window 4-4
Window 4-5: Matrix form for overall equilibrium
Sum of weighted residuals for overall equilibrium is expressed as
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.1.2:2
Preface Numerical Implementation Weak Form
R
1
(w
e
, u
e
, p
e
) =
_

w
e

T
_
B
T
+ B
T
1 S N
p
p
e
_
d
_

w
e

T
N
u

T
b d
_

t
w
e

T
N
u

T
t d = 0
It is satised for any w
e
if the following equation holds
_

_
B
T

N+1
+B
T
1S N
p
p
e
N+1
_
d =
_

(N
u
)
T
b
N+1
d +
_

t
(N
u
)
T
t
N+1
d
. .
F
EXT
N+1
Window 4-5
Window 4-6: Matrix form for uid mass concervation
Sum of weighted residuals of uid mass conservation is expressed as
R
2
(u
e
, q
e
, p
e
) =
_

q
e

T
N
p

T
S 1
T
B u
e
d
_

q
e

T
N
p

T
v
F
d
_

q
e

T
N
p

T
c N
p
p
e
d +
_

q
q
e

T
N
p

T
q d
_

q
q
e

T
N
p

T
k
v
(p p
ext
) d = 0
where generalized storage coecient c(p) is dened as
c(p) = n
_
S
K
F
+
S
p
_
S
p
=
_

F
_
2
(1 S
r
)
_
1 +
_
p

F
_
2
_3
2
p
It is satised for any q
e
if the following equation, written for the current time step N + 1,
holds
C
T
N+1
u
e
N+1

F
H
N+1
p
e
N+1
P p
e
N+1
M
N+1
p
e
N+1
= H
N+1
z
e
N+1
Q
f
N+1
P p
e
ext
N+1
. .
Q

N+1
where
C
T
N+1
=
_

(N
p
)
T
S
N+1
1
T
B d H
N+1
=
_

(N
p
)
T
k
r
(S
N+1
) (N
p
) d
M
N+1
=
_

(N
p
)
T
c(p
N+1
) N
p
d Q
f
N+1
=
_

q
(N
p
)
T
q
N+1
d P =
_

S
(N
p
)
T
k
v
N
p
d
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.1.2:3
Preface Numerical Implementation Weak Form
the elevation z
e
N+1
=
1
_
b
T
N+1
b
N+1
x
e

T
b
N+1
is measured against the direction of
current gravity vector b
N+1
.
As the switch between pressure/ux type of the boundary condition has been made through
penalty approach, one can discretize the boundary
S
by introducing so-called seepage
surface elements [?] which consist of the two faces, one connected to the continuum
element in the domain, and the second, on which a proper pressure p
ext
boundary condition
is set
The approximate value of the penalty factor k
v
, is usually dened as
k
v
= 10
6
trace(k
r
(S
N
) k)

F
h
e
The h
e
is a characteristic size of the continuum nite element e, adjacent to a given
seepage element. This size can be computed as
Ndm

e
. The trace(k
r
(S
N
) k) is a magni-
tude of an equivalent permeability coecient, computed by considering the ow properties
of the continuum nite element adjacent to the seepage one.
This penalty term (due to seepage surface BC) is treated in an explicit way as the actual
permeability depends on saturation ratio obtained at the time step t
N
.
Window 4-6
Window 4-7: Coupled system of linearized equations
Matrix equations:
_

B
T
D
ep(i)
N+1
B d
. .
K
(i)
N+1
u
e(i+1)
+
_
_
_

B
T
1
_
S
(i)
N+1
+
S(p
(i)
N+1
)
p
p
(i)
N+1
_
N
p
d
_

(N
u
)
T
b
S
S(p
(i)
N+1
)
p
d
_
_
. .
G
p
e(i+1)
=
F
(i)
EXT
N+1

_
B
T

(i)
N+1
+B
T
1S
(i)
N+1
N
p
p
e(i)
N+1
_
d
. .
F
(i)
INT
N+1
_
C
(i)
N+1
_
T
u
e(i+1)

_
1

F
H
(i)
N+1
+PC
p(i)
N+1
+
1

F
H
p(i)
N+1
+M
p(i)
N+1
H
pz(i)
N+1
_
p
e(i+1)

M p
e(i+1)
= Q
(i)
N+1

_
C
(i)
N+1
_
T
u
e(i)
N+1
+
1

F
H
(i)
N+1
p
e(i)
N+1
+M
(i)
N+1
p
e(i)
N+1
+P p
e(i)
N+1
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.1.2:4
Preface Numerical Implementation Weak Form
Introducing the iterative increments of the displacement and pressure ratesinto above equation
and multiplying it by t one can obtain
_
C
(i)
N+1
_
T
u
e(i+1)

_
t
_
1

F
H
(i)
N+1
+PC
p(i)
N+1
+
1

F
H
p(i)
N+1
+M
p(i)
N+1
H
pz(i)
N+1
_
+M
_
p
e(i+1)
=
t
_
Q
(i)
N+1

_
C
(i)
N+1
_
T
u
e(i)
N+1
+
1

F
H
(i)
N+1
p
e(i)
N+1
+M
(i)
N+1
p
e(i)
N+1
+P p
e(i)
N+1
_
. .
Finally, the resulting coupled system of linearized equations takes the form
_
K
(i)
N+1
G
(i)
N+1
(C
N+1
)
T

_
t H
(i)
N+1
+M
(i)
N+1
_
_
_
u
e(i+1)
p
e(i+1)
_
=
_
F
(i)
EXT
N+1
F
(i)
INT
N+1
t Q
(i)
N+1
_
where
G
(i)
N+1
=
_

B
T
1
_
S
(i)
N+1
+
S(p
(i)
N+1
)
p
p
(i)
N+1
_
N
p
d +
_

(N
u
)
T
b
S
S(p
(i)
N+1
)
p
d
H
(i)
N+1
=
1

F
H
(i)
N+1
+PC
p(i)
N+1
+
1

F
H
p(i)
N+1
+M
p(i)
N+1
H
pz(i)
N+1
C
p(i)
N+1
=
_

(N
p
)
T
_
S(p
(i)
N+1
)
p
1
T
B u
e(i)
N+1
_
N
p
d
H
p(i)
N+1
=
_

(N
p
)
T
_
k
r
S
S (p
N+1
)
p
k N
p
p
e(i)
N+1
_
N
p
d
H
pz(i)
N+1
=
_

(N
p
)
T
_
k
r
S
S (p
N+1
)
p
k N
p
z
e(i)
N+1
_
N
p
d
M
p(i)
N+1
=
_

(N
p
)
T
_
c(p
(i)
N+1
)
p
N
p
p
e(i)
N+1
_
N
p
d
c(p
(i)
N+1
)
p
= n
_
1
K
F
S (p
N+1
)
p
+

2
S (p
N+1
)
p
2
_
Window 4-7
Window 4-8: Explicit treatment of uid state parameters S and k
r
To simplify the computational scheme all the uid state parameters, that is S and k
r
, are
treated in the explicit manner, based on the pressure values obtained for the time step t
N
.
In that case the resulting nite element system of equations reduces to
_
K
(i)
N+1
C
N
C
T
N
( t H
N
+ t P+M
N
)
_ _
u
e(i+1)
p
e(i+1)
_
=
_
F
EXT
N+1
F
(i)
INT
N+1
t Q
(i)
N+1
_
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.1.2:5
Preface Numerical Implementation Weak Form
where
Q
(i)
N+1
= Q

N+1
C
T
N
u
e(i)
N+1
+
1

F
H
N
p
e(i)
N+1
+M
N
p
e(i)
N+1
+P p
e(i)
N+1
Q

N+1
= H
N
z
e
N+1
Q
f
N+1
P p
e
ext
N+1
Automatically, for the linear constitutive equation of the solid phase the whole problem
becomes linear within the time step. The size of the applied time step is limited, otherwise
pore pressure may begin to oscillate during the integration in time.
Window 4-8
Window 4-9: Special cases. Matrix form
Steady state ow (incremental form)
[H
N+1
+P]
_
p
e(i+1)
_
= Q
(i)
N+1
where
Q
(i)
N+1
= Q

N+1
+
1

F
H
N+1
p
e(i)
N+1
+P p
e(i)
N+1
Q

N+1
= H
N+1
z
e
N+1
Q
f
N+1
P p
e
ext
N+1
Remark:
In that case the only possible computational scheme is the one in which the actual perme-
ability varies with current saturation ratio during the iterative process
Transient ow

_
t H
(i)
N
+M
(i)
N
+ t P
__
p
e(i+1)
_
= Q
(i)
N+1
Q
(i)
N+1
= Q

N+1
+
1

F
H
N
p
e(i)
N+1
+M
N
p
e(i)
N+1
+P p
e(i)
N+1
Q

N+1
= H
N
z
e
N+1
Q
f
N+1
P p
e
ext
N+1
Remark:
In that case we use the scheme with explicit treatment of uid state parameters
Uncoupled two-phase total stress analysis
The uncoupled total stress analysis can be derived from the general formulation for the
problem of consolidation by dropping the transient terms. This way we have a set of two
equations to be solved, that is the steady state uid ow, as given above, and the overall
equilibrium expressed as
_
K
(i)
N+1
_
_
u
e(i+1)
_
=
_
F
EXT
N+1
F
(i)
INT
N+1
_
The internal force vector is integrated using the Bishops extended eective principle via
F
(i)
INT
N+1
=
_

e
B
T
_

(i)
N+1
+S
N+1
1p
N+1
_
d
Remark:
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.1.2:6
Preface Numerical Implementation Weak Form
The steady state uid ow has to be solved rst and then computed pore pressures are
substituted into the internal force vector in the mechanical analysis step.
Window 4-9
Window 4-10: Consolidation. Time stepping algorithm
For each step
Initialize
u
(i=0)
N+1
= u
N+1
= u
N
+ t u
N
u
(i=0)
N+1
= u
N
p
F(i=0)
N+1
= p
F
N
+ tp
F
N
p
F(i=0)
N+1
= p
F
N
For each iteration, assume, c, S constant computed at step n
F
(i)
N+1
= F
ext
N+1
F
int
(u
(i)
N+1
) C
N
p
F(i)
N+1
Q
(i)
= tQ
N+1
C
T
N
u
(i1)
N+1
+H

N
p
F(i1)
N+1
+M
N
p
F(i1)
N+1
+P
(i1)
N
_
p
F(i1)
N+1
p
Fext
N+1
_
with
F
ext
N+1
=
_

N
T
u
bd +
_

N
T
u

t d
Q
N+1
=
_

N
T
p
qd H
N
z
Solve
_
K
T
C
N
C
T
N
H

N
_ _
u
(i+1)
p
F(i+1)
_
=
_
F
(i)
N+1
Q
(i)
N+1
_
Update
u
(i+1)
N+1
= u
(i)
+ u
(i+1)
p
F(i+1)
N+1
= p
F(i)
N+1
+ p
F(i+1)
u
(i+1)
N+1
=
_
u
(i+1)
N+1
u
(i)
N+1
_
t
u
(i+1)
N+1
= u
(i)
N+1
+ u
(i+1)
N+1

p
F(i+1)
N+1
=
_
p
F(i+1)
N+1
p
F(i)
N+1
_
t
p
F(i+1)
N+1
= p
F(i)
N+1
+ p
F(i+1)
Iterate as needed.
Window 4-10
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.1.2:7
Preface Numerical Implementation Weak Form
4.1.2.1 PRESSURE STABILIZED FINITE FORMULATION FOR
PARTIALLY SATURATED TWO-PHASE MEDIA
General stabilization scheme
Matrix form for stabilized formulation
A one-dimensional estimate for
FPL
Setting
FPL
for two- and three-dimensional problems
Denition of element size h

for Q4 and B8 nite elements


Denition of element size h

for T3, W6 nite elements


Denition of directional seepage coecient
Denition of equivalent stiness modulus E

Window 4-11: General stabilization scheme


The general form of discretized variational formulation for the problem of consolidation mod-
ied by some stabilization term R
FPL
takes the form
R(w
e
, u
e
, q
e
, p
e
) = R
1
(w
e
, u
e
, p
e
) t R
2
(u
e
, q
e
, p
e
) +R
FPL
(w
e
, u
e
, q
e
, p
e
) = 0
where
R
FPL
(q
h
, p
h
) =
Nele

i=1
_

e
_
q
h

p
h
d
e
Remark:A stabilization factor factor

depends on mesh size h.


The above equation written in terms of nodal quantities q
e
, p
e
for time step N +1 takes the
form
R
FPL
(q
e
, p
e
N+1
) =
Nele

i=1
_

e
[N
p
q
e
]

N
p
p
e
N+1
d
e
Linearized form of the stabilization term is expressed as
R
FPL
(q
e
, p
e
N+1
) =
Nele

i=1
_

e
[N
p
q
e
]
T

_
N
p
p
e (i)
N+1
+N
p
p
e (i+1)
_
d
e
Replacing the rate of the pore pressure p by
p
e
t
and shifting factor t to the denition
of the stabilization factor the following expression for FPL-stabilization term is obtained
R
FPL
(q
e
, p
e
N+1
) =
Nele

i=1
_

e
[N
p
q
e
]
T

FPL

_
N
p
p
e (i+1)
+ t N
p
p
e(i)
N+1
_
d
e
where newly introduced stabilization factor
FPL
is dened as
FPL
=

t
.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.1.2.1:1
Preface Numerical Implementation Weak Form
Window 4-11
Window 4-12: Matrix form for stabilized formulation
The resulting matrix form is expressed as
_
K
(i)
N+1
C
N
C
T
N

_
t H
N
+P+M
N
+H
FPL
N+1
_
_ _
u
e(i+1)
p
e(i+1)
_
=
_
F
EXT
N+1
F
(i)
INT
N+1
t Q
(i)
N+1
_
where
H

=
t

F
HMH
FPL
Q

= Q+Q
FPL
H
FPL
=
Nele

i=1
_

e
[N
p
]
T

FPL
[N
p
] d
Q
FPL
= H
FPL
p
e(i)
N+1
d
Window 4-12
Window 4-13: A one-dimensional estimate for
FPL
To obtain a one-dimensional estimate for stabilization factor let us consider a simple 1-D
test problem discretized with a single FPL stabilized two-node element as shown in Fig.
below. For the estimation the rst time step has been considered, assuming that the pressure
boundary condition dened at the top node is such that p
2
= p
o
(x = h). The initial stresses
and pressures (at time t = 0) are assumed to satisfy the both overall equilibrium and uid
mass conservation equations.
h
Initial state
x
Loading
q
) x ( p
) x (
o
yo
s
p ) x ( p
) x (
o
yo
D +
s D + s
0 u
1
=
) h x ( p p
o 2
= =
One-dimensional test problem
Applying the condition
p
lim
=
q
E
oed
c
S
+S
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.1.2.1:2
Preface Numerical Implementation Weak Form
to the pressure dierence between its current value and the one assumed at the initial state,
at node 2, derived from the stabilized matrix form, the following one-dimensional formula for
factor
FPL
has been obtained.

FPL
=
1
4
S
2
h
2
E
oed
+
1
6
h
2
c
k k
r
t

F
Window 4-13
Window 4-14: Setting
FPL
for two- and three-dimensional problems
The parameters
F
, , t, c remain unchanged in 1D, 2D or 3D problems, but the element size
h, the permeability coecient k and the modulus of deformation E
oed
have some directional
meaning in the multi-dimensional cases. To handle that let us introduce the equivalent
element size h

, the equivalent permeability coecient k

and the equivalent stiness modulus


E

which are measured along the direction of the current pressure gradient vector v

= p.
This gives the warranty that the GLS formulation will always give the same results for 2D/3D
problems which are reducible to 1D for certain types of boundary conditions.
Hence, the general formula for the stabilization factor
FPL
can be expressed in the form

FPL
=
1
4
(h

E
)
2
S
2
E

. .

E
+
1
6
(h

c
)
2
c
. .

k
r
t

F
. .

k
Window 4-14
Window 4-15: Denition of element size h

for Q4 and B8 nite elements


In the simplied version we can assume that h

E
= h

c
= h

.
The directional element size, for the class of quadrilateral and brick nite elements, can be
dened as follows
h

=
Ndm

i=1
h

i
_
v

i
_
_
v

_
_
_
2
h
()
i
=
_
_
_x
_
A
(+)

i
_
x
_
A
()

i
__
_
_
where
_
A
(+)

i
_
k
=
ki
,
_
A
()

i
_
k
=
ki
, v

= J
1
v

,
1
= ,
2
= ,
3
=
and J is the Jacobian matrix of the isoparametric mapping, v

i
is the i-th component of the
pressure gradient vector in local (, , ) coordinate system,
ki
is the Kroneckers symbol,
A
(+)

i
and A
()

i
are points with Ndm coordinates dened in the local system, and v

is the
pressure gradient vector in the global system. The geometrical interpretation of the denition
is given in Figure below.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.1.2.1:3
Preface Numerical Implementation Weak Form
X
Y
X
Y
h
h
h
x
O
e
x e
h
h
x
Denition of the element size for element Q4
Window 4-15
Window 4-16: Denition of element size h

for T3, W6 nite elements


We assume that h

E
= h

c
= h

.
The directional element size, for any class of nite elements, can be dened as follows (see
Fig. below)
A
1
A
4
A
2
A
3
v
t
h
D1
t
Denition of the element size
h

= max(A
1
A
i

| v

|
) min(

A
1
A
i

| v

|
) i = 2, Nen
where Nen is a number of element nodes.
Window 4-16
Window 4-17: Denition of directional seepage coecient
The directional permeability coecient k

can be taken as the k


loc
11
component of the perme-
ability matrix k transformed to the local coordinate system. If the rst axis of the selected
local system is parallel to the pressure gradient vector v

then the permeability coecient


k

is dened as follows
k

= k
loc
11
=
_
v

| v

|
_
T
k
_
v

| v

|
_
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.1.2.1:4
Preface Numerical Implementation Weak Form
Window 4-17
Window 4-18: Denition of equivalent stiness modulus E

The stiness modulus E

is expressed as
E

=
det (D

i=1,2

j=1,2
(1)
i+j
det (D
(ij)
)
and the auxiliary matrix D

(for the 3D case) takes the form


D

=
_
_
D
11
D
12
D
14
D
21
D
22
D
24
D
41
D
42
D
44
_
_
The matrices D
(ij)
are extracted from D

by deleting the row and column of the entry


D

ij
.
The auxiliary matrix D

is a submatrix of D from which rows and columns corresponding


to shear terms are deleted
In order to give an objective denition of the modulus E

in the general case we can


transform the current stiness matrix D to local system dened through axes of principal
directions of the total strain increment
N+1
. From this transformed matrix we delete
all rows and columns corresponding to shear terms and those rows and columns which
correspond to zero normal strains after transformation to the principal directions. As a
result matrix D

, required for the evaluation of E

, is obtained
For the plane strain case third row and third column is always deleted as well as fourth row
and column (due to
zz
= 0 condition)
The above formula has been derived assuming a quite simple form of the stiness matrix
which, in the case of elasto-plasticity, can be denitely more complicated. In that general
case it is almost impossible to obtain such simple stabilization factor formula. It may also
happen that for certain plastic models the proposed generalized stiness modulus E

can
be negative. To remedy such a situation we could use in the formula for E

the actual
elastic stiness or we could compute it via
E

= E
(e)
1
T
D
ep
1
1
T
D
e
1
where the generalized stiness modulus, computed for the actual elastic stiness matrix
D
e
, is denoted by E
(e)
.
By considering elasto-plastic models which describe liquefaction phenomenon, even the
formula (4-18) may fail and theonly choice is to use purely elastic generalized stiness
modulus E
(e)
.
Window 4-18
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.1.2.1:5
Preface Numerical Implementation Weak Form
4.1.3 HEAT TRANSFER
Following section describe the solution algorithm for transient and steady state heat transfer
problem formulated in section 2.4. In particular hints on
Weak and matrix form of heat transfer analysis
Time marching algorithm for transient analysis
Newton-Raphson schema
Heat source term
are given.
Notation
T -temperature eld t -time
T
E
-external (ambient) temperature

X -time derivative of X
H -heat source -time increment
q
T
-heat ux -gradient
c

-heat capacity X -matrix/vector representation of X


-heat conductivity X
e
element contribution of X
M -maturity
Weak form and matrix formulation
Window 4-19: Weak and matrix form of heat transfer analysis
Approximation for temperature eld T(x,t)
T(x,t) = N(x)T (1)
where N are shape functions and T are nodal temperatures.
Matrix form of governing dierential equation for transient heat transfer analysis
C

T+ (K+K
b
)T =

H+ K
b
T (2)
As Z Soil

system is generally based on incremental approach, equations have to be linearized


which nally leads to following linearized form:
C

T+ (K+K
b
+H

)T =Q (3)
where :
Q = Q+

H+K
b
T
E
C

T(K+K
b
)T (4)
For each matrix X = /
e=1,N
(X
e
) , where symbol /
e=1,N
represents the assembly of element
contributions
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.1.3:1
Preface Numerical Implementation Weak Form
X
e
, as listed below.
K
e
=
_

e
N
T
N d
e
C
e
= c

e
N
T
N d
e
K
e
b
= h
T
_

e
N
T
N d
e
Q
e
=
_

e
N
T
q
T
d
e

H
e
=
_

e
N
T

H
e
H
e
=
_
_

H
M
M
T
+

H
T
_
_
=
_
_

H
M
M
T
+

H
T
_
_
_

e
N
T
N d
e
All those matrices are evaluated using Gauss quadratures.
Window 4-19
The general system can be reduced to steady state conditions which is expressed by the set
of equations:
(K+K
b
)T = Q+K
b
T
E
(K+K
b
)T (4.1)
Window 4-20: Time marching algorithm for transient analysis
Applying nite dierences in time domain , with (0, 1):
T
n+1
= T
n
+ (1 )t

T
n
+ t

T
n+1
=
_
T
n
+ t

T
n
_
+
predictor
t

T
n+1
(1)
thus:

T
n+1
=
T
n+1
t
(2)
Introducing above into the linearized form 3, the following set of incremental equations is
obtained:
[C+t (K+K
b
+H

)] T =t Q (3)
Window 4-20
Implementation scheme ( NR/MNR methods):
The full Newton-Raphson and modied NR schemes are foreseen for solving of equation 3.
In both schemes the eect of heat source matrix is shifted on the R.H.S. The full NR and
modied NR schemes applied to steady state equations 4.1 are both exactly the same and
can be expressed as follows (no iterations are needed):
(K+K
b
)T
(i)
n+1
= K
b
T
E n+1
(K+K
b
)T
(i1)
n+1
(4.2)
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.1.3:2
Preface Numerical Implementation Weak Form
For transient case the eect of source matrix is moved to R.H.S to avoid L.H.S. matrix
recalculation at each iteration. The full NR scheme is summarized by equation ( n-time step
counter, i-iteration counter) :
[C+ t
n+1
(K+K
b
)] T
(i)
n+1
= t
n+1
(Q
(i)
n+1
H
(i)
n+1
T
(i1)
n+1
) (4.3)
Q
(i)
n+1
= Q
n+1
+

H
(i1)
n+1
+ K
b
T
E n+1
C

T
(i1)
n+1
(K+K
b
)T
(i1)
n+1
(4.4)
To avoid recalculation the L.H.S. at each time step the modied NR scheme can be used
which is summarized by equation :
[C+ t

(K+K
b
)] T
(i)
n+1
= (4.5)
t
n+1
_
Q
(i)
n+1

_
H
(i1)
n+1
+
t
n+1
t

t
n+1
(K+K
b
)
_
T
(i1)
n+1
_
(4.6)
The t

is the time increment value for which the L.H.S. was evaluated at last update. The
detailed implementation scheme is given below:
Window 4-21: Newton-Raphson schema for heat transfer analysis
initialize new time step:
t
n+1
= t
n
+ t
n+1

T
(0)
n+1
=

T
n
(only for transient analysis)
M
n+1
= M
n
in each G.P (only for transient analysis)
_
M
T
_
n+1
=
_
M
T
_
n
in each G.P (only for transient analysis)
initialize iteration counter:
i = 0
compute temperature predictor (only for transient analysis):
T
(0)
n+1
= T
n
+ t

T
n
T

n+1
= t

T
n
next iteration:
i = i + 1
evaluate L.H.S. (if required):
for steady state independently on algorithm type:
K

= K+K
b
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.1.3:3
Preface Numerical Implementation Weak Form
for transient case:
search for time increment which would be used during next L.H.S. reevaluation according
to user dened algorithm type:
K

= C+t

(K+K
b
)
evaluate R.H.S.:
for steady state independently on algorithm type :
Q
(i)
n+1
= Q
n+1
+K
b
T
E n+1
(K+K
b
)T
(i1)
n+1
for transient case:
if L.H.S. is/will be reformulated at this time step:
Q
(i)
n+1
= t
n+1
(Q
n+1
+

H
(i1)
n+1
+K
b
T
E n+1
C

T
(i1)
n+1
(K+K
b
)T
(i1)
n+1
)
otherwise:
Q
(i)
n+1
= t
n+1
(Q
n+1
+

H
(i1)
n+1
+K
b
T
E n+1
C

T
(i1)
n+1
(K+K
b
)T
(i1)
n+1

t
n+1
t

t
n+1
(K+K
b
)T
(i1)
n+1
)
solve system of equations:
K

T
(i)
n+1
= Q
(i)
n+1
update total temperatures, temperature total increments and their rates:
T
(i)
n+1
= T
(i1)
n
+ T
(i)
n+1
T

n+1
= T
()
n+1
+ T
(i)
n+1

T
(i)
n+1
=

T
(i1)
n+1
+
T
(i)
n+1
t
n+1
if convergence norm
_
_
_Q
(i)
n+1
_
_
_is not satised go to next iteration.
Window 4-21
Handling of concrete hydration heat source term.
In the case of heat source term related to concrete hydration 1, which appear in denition
of matrices H,

H, integration of both heat source term H and maturity factor derivative


with respect to temperature
M
T
via eqn. 2, in each integration point in an element are
required. Even with assumption that in given time increment the temperature varies linearly,
the integrals 2 can not be evaluated using elementary functions, so it has to be integrated
numerically as follows:
Window 4-22: Heat source term
heat source:
H = H

aM
1 +aM
(1)
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.1.3:4
Preface Numerical Implementation Weak Form
maturity factor:
M =
t
_
t
d
exp
_
Q
R
(
1
T
ref

1
T
)
_
dt (2)
numerical integration shema:
M
n+1
= M
n
+ exp
_
Q
R
(
1
T
ref

1
T

)
_
t
n+1
if t
n+1
> t
BEG
+t
d
_
M
T
_
n+1
=
_
M
T
_
n
+
Q
R
_
1
T

n+1
_
2
exp
_
Q
R
(
1
T
ref

1
T

n+1
)
_
t
n+1
if t
n+1
> t
BEG
+t
d
M
n+1
= 0 if t
n+1
t
BEG
+t
d
_
M
T
_
n+1
= 0 if t
n+1
t
BEG
+t
d
where:
T

n+1
= T
n
(1 ) +T
n+1
Window 4-22
The rst existence time of an element to which the integration point belongs is denoted by
t
BEG
. In the above scheme in case when t
n+1
< t
n+1
t
BEG
t
d
the value t
n+1
should
be replaced by appropriate value.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.1.3:5
Preface Numerical Implementation
4.2 ELEMENTS
FINITE ELEMENTS FOR 2D/3D CONTINUUM PROBLEMS
NUMERICAL INTEGRATION
STRAINS
STIFFNESS
BODY FORCES
INITIAL STRESSES
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.2:1
Preface Numerical Implementation Elements
4.2.1 FINITE ELEMENTS FOR 2D/3D CONTINUUM PROBLEMS
A family of 2/3D isoparametric elements with 1st order interpolation function is implemented
in the program, see Appendix 4.8.1.
In case of single phase analysis the nodal displacements (active for given analysis type) are
the basic unknowns. Derivatives of shape function are necessary for strains analysis and then
for stresses calculation. Basic features are summarized in the Table 4.1.
Table 4.1: Finite elements for 2/3D continuum problems
Analysis type Plane Strain Axisymmetry, r = x Generalized Plane Strain
Ndim 2 2 2
Elementary
volume dV
dx dy 1 2r dr dy dx dy 1
Active
displacement
u(x, y) = [u, v]
T
u(r, y) = [u, v]
T
u(x, y) = [u, v, w]
T
DOF / node 2 2 3
Kinematic
constraints
w = 0,

z
= 0 w

= 0,

= 0

z
= 0
Strain vector
= Bd
_

xx
,
yy
,
xy

T
_

xx
,
yy
,
xy
,

T
_

xx
,
yy
,
xy
,
xz
,
yz

T
Straindisplacement
operator
B
a
B = [B
a
]
a = 1, NEN
_

_
N
a
x
0
0
N
a
y
N
a
y
N
a
x
_

_
_

_
N
a
r
0
0
N
a
y
N
a
y
N
a
r
N
a
r
0
_

_
_

_
N
a
x
0 0
0
N
a
y
0
N
a
y
N
a
x
0
0 0
N
a
x
0 0
N
a
y
_

_
3D
3
dx dy dz
u(x, y) = [u, v, w]
T
3

xx
,
yy
,
xy
,
zz
,
xz
,
yz

T
_

_
N
a
x
0 0
0
N
a
y
0
N
a
y
N
a
x
0
0 0
N
a
z
N
a
z
0
N
a
x
0
N
a
z
N
a
x
_

_
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.2.1:1
Preface Numerical Implementation Elements
4.2.2 NUMERICAL INTEGRATION
Numerical integration is required for the evaluation all integrals resulting from the problem
denition. Gaussian quadrature is used and for a onedimensional integral can be represented
as follows:
_

f(x) dx =

f (
i
) J
x
,

(
i
) w
i
where:
f (
i
) function being integrated, value at
i
J
x
,

jacobian determinant
w
i
Gaussian weighting factor
The extension of the above for multidimensional case is straightforward : the double (triple)
integration is replaced by a double (triple) summation.
In the case of axisymmetry, all integrands include a factor equal to 2r where r = radius of
the integration point under consideration, and 2 can be devided out.
Standard quadrature types used for stiness/forces evaluation for dierent elements are given
in the table below:
Table 4.2: Standard quadrature for typical elements
Element: T3 Q4 TH4 W6 B8
Ngauss 1 2 2 1 2 1 2 2 2
Details on integration point positions, weighting factors, for dierent quadrature types are
given in Appendix 4.8.2
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.2.2:1
Preface Numerical Implementation Elements
4.2.3 STRAINS
Strains at any integration point of a continuum element are evaluated by an unied proce-
dure given in Window 4-23. Beside the standard approach presented here, in the case of
incompressible or dilatant media another approach is recommended, see Section 4.3.
Window 4-23: Evaluation of strains
Quadrilateral element (Q4) outlook
An unied procedure for all 2/3D standard continuum elements include:
isoparametric mapping:
x() =
Nen

a=1
N
a
() x
a
displacement interpolation:
u() =
Nen

a=1
N
a
() u
a
strain:
() = B() u =
Nen

a=1
B
a
() u
a
where the standard form of matrix B is given in the Table 4.1. In the case of strain
projection technique (necessary to simulate incompressible media) resulting form of the
Bmatrix, called BBar, will be described in Window 4-24.
shape function derivatives, present in all concerned B
a
are evaluated as:
N
a
x
=
_
J
1
_
T
N
a

where Jacobian J = x/ for 2D and 3D is calculated as follows


2D: J =
_

_
x

_
; 3D: J =
_

_
x

_
Window 4-23
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.2.3:1
Preface Numerical Implementation Elements
4.2.4 STIFFNESS MATRIX
The element stiness matrix can now be evaluated numerically; recalling that:
K
e
=
_

B
T
D B d
F
e
=
_

B
T
d
Numerical integration yields:
K
e
=
Ngauss

i
B
T
(
i
) D
i
B(
i
) det J(
i
) W
i
F
e
=
Ngauss

i
B
T
(
i
) (
i
) det J(
i
) W
i
where the integrand is evaluated at the integration point i.
The standard integration rules are used for each element, see Table 4.2. If strain projection
method is used, however, in the forming of the Bmatrix included the above stiness terms
(see section Incompressible media). 1 1 gaussian integration of the volume part of the
stiness is incorporated. For such an integration rule, the one integration point is located at
the element centre.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.2.4:1
Preface Numerical Implementation Elements
4.2.5 BODY FORCES AND DISTRIBUTED LOADS
The introduction of the interpolation functions into the weak form yields an elemental body
force term as follows:
F
e
b
=
_

e
N
T
bd
e
For distributed forces, the equivalent nodal force is seen to be:
F
e
t
=
_

e
N
T
t d
e
.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.2.5:1
Preface Numerical Implementation Elements
4.2.6 INITIAL STRESSES, STRAINS
If the initial stress eld is in equilibrium the initial stresses can simply be added to the stresses
resulting from deformations. If not, equivalent nodal forces must be calculated. Nodal forces
equivalent to initial stresses and/or strains can be introduced as follows ; this calls for a
generalized stress-strain relation.
Let
= D(
0
) +
0
where
0
are the initial stresses and
0
are the initial strains. By the denition of the internal
force term:
/
e=1,N
_

B
T
d = /
e=1,N
_

B
T
D(
0
) d + /
e=1,N
_

B
T

0
d
from which the following expression of the nonlinear equilibrium equation results
F
int
(u) = F F

+F

0
with
F

0
= /
e=1,N
_

B
T

0
d
F

= /
e=1,N
_

B
T

0
d.
Note that if initial stresses are in equilibrium F

0
= 0 no deformations will result from the
initial stresses. If not, deformations will result. Similarly if initial strains are compatible with
boundary conditions, no stresses result.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.2.6:1
Preface Numerical Implementation
4.3 INCOMPRESSIBLE AND DILATANT MEDIA
A strain projection technique is necessary in order to simulate incompressible media. The
resulting form of the Bmatrix, called BBar, will be described later on.
Another nite element technique to simulate incompressible and highly dilatant plastic media
is the Enhanced Assumed Strain method (EAS). The general idea and detailed algorithmic
scheme is described in Chapter 4.3.2. This option is activated in ZSOIL when dilatant plastic
ow is active.
INCOMPRESSIBLE MEDIA: B-BAR STRAIN PROJECTION METHOD
DILATANT MEDIA: ENHANCED ASSUMED STRAIN METHOD
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.3:1
Preface Numer. Impl. Incompressible Media
4.3.1 INCOMPRESSIBLE MEDIA : B-BAR STRAIN PROJECTION
METHOD
Material data include the elastic constants i.e.Youngs modulus and Poisson ratio. In fully
incompressible cases, Poissons ratio is 0.5 which within the program is taken as 0.499999.
This leads to mesh locking when a full (2 2) integration scheme is used. Underintegration
(i.e. the use of a single integration point) is therefore required in the volumetric part of the
stiness, while full integration is used for the shear term.
It can be seen from the iterative scheme that the use of seective underintegration would
result in dierent integration schemes for the left and righthand side of the equilibrium
equations. This can, and will, disturb convergence, so strain projection is used to circumvent
this problem.
First, the BBAR matrix is established for the case of torsionless axisymmetry and then the
plane strain case is derived.
Beginning with the standard Bmatrix for node a dened previously in this section, we can
split the matrix into dilatational and deviatoric parts as shown in Window 4-24.
In place of using B
a
the

B
a
is used where the dilatational term has been underintegrated.
The dilatational

B entries are calculated using a reduced 1 1 integration rule whereas the
deviatoric

B entries are calculated using the normal (2 2) integration rule.
Window 4-24: Strain projection B-BAR method
B
a
= B
dev
a
+B
dil
a
where
B
dil
a
=
1
3
_

_
(B
0
+B
1
) B
2
(B
0
+B
1
) B
2
0 0

B
0
+

B
1

B
2
_

_
and B
dev
a
=
_

_
_
2
3
B
1

1
3
B
0
_

1
3
B
2
_

1
3
B
0

1
3
B
1
_
2
3
B
2
B
2
B
1
2
3
B
0

1
3
B
1

1
3
B
2
_

_
In place of use B
a
use

B
a
:

B
a
= B
dev
a
+

B
dil
a
i.e.:

B
a
=
_

_
B
12
B
6
B
10
B
7
B
2
B
1

B
11

B
6
_

_
where
B
0
= N
a
/x
1
B
1
= N
a,x
1
B
2
= N
a,x
2
B
4
=
_

B
1
B
1
_
/3 B
6
=
_

B
2
B
2
_
/3 B
7
= B
2
+B
6
B
10
= B
4
+
_

B
0
B
0
_
/3 B
11
= B
0
+B
10
B
12
= B
1
+B
10
Window 4-24
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.3.1:1
Preface Numer. Impl. Incompressible Media
Remarks:
1. The plane strain case many be obtained from the above formulation by simply setting

B
0
= B
0
= 0 and using Cartesian rather than cylindrical coordinates. The 2r (or r)
factor included in the integrands in the case of axisymmetry must also be ignored.
2. Notice that the D matrix remains 4 4 for plane strain, when

B the formulation is used.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.3.1:2
Preface Numer. Impl. Incompressible Media
4.3.2 DILATANT MEDIA: ENHANCED ASSUMED STRAIN METHOD
INTRODUCTION TO ENHANCED ASSSUMED STRAIN (EAS) APPROACH
EXTENSION OF EAS TO NONLINEAR ELASTOPLASTIC ANALYSIS
REMARKS AND ASSESSMENT OF EAS ELEMENTS
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.3.2:1
Numer. Impl. Incompressible... Dilatant Media
4.3.2.1 INTRODUCTION TO ENHANCED ASSSUMED STRAIN
(EAS) APPROACH
In EAS methods
1
the strain is approximated by two elds, i.e. the compatible and the
enhanced one via equation :
= Bu +M
where B is the standard strain operator, M is the interpolation operator for the additional
strain eld which may be discontinuous across element edges.
The variational basis of the method is the HuWashizu functional in the form:
U =
_

_
1
2

T
D
T
+
T
Bu d
T
N
T
F
_
d.
Substitution of the modied strain eld into functional yields:
U =
_

_
1
2
(Bu +M)
T
D(Bu +M)
T
(Bu +M) +
T
Bu u
T
N
T
F
_
d.
In order to eliminate the statically admissible stress eld from functional the following condi-
tion has to be satised:
_

T
M d = 0.
The system of equations obtained by taking the variation of functional with respect to u and
takes the form:
_
K
uu
K
u
K
u
K

_ _
u

_
=
_
F
ext
0
_
where:
K
uu
=
_

B
T
DB d ; K
u
=
_

B
T
DM d
K
u
=
_

M
T
DB d ; K

=
_

M
T
DM d
Assuming discontinuous approximation for these extra degrees of freedom can be eliminated
at the element level through standard condensation procedure. The interpolation functions
assumed for M

matrix, dened in isoparametric space, for plane strain case and axisymmetry
are given in Windows (4-25) and (4-26). The forms of M

matrix for quasi-plane strain and


3D are given in Windows (??) and (??).
These interpolation functions assumed for M

(here index is used to distinguish between


isoparametric and physical spaces) have to satisfy the orthogonality condition introduced
already in 4-th equation in this section. The transformation of M

matrix from isoparametric


space to physical one, summarized in Window (4-27), is performed with aid of Jacobian of
the isoparametric mapping.
1
Simo J.C., Rifai M.S., A class of mixed assumed strain methods and method of incompatible modes.
International Journal for Numerical Methods in Engineering. Vol.29, pp.1595-1638, 1990
Groen A.E., Improvement of low order elements using assumed strain concepts. TU-Delft report Nr
25.2.94.203, 1994.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.3.2.1:1
Numer. Impl. Incompressible... Dilatant Media
Window 4-25: M matrices for Q4 element in plane strain
EASelement
Number of
extra modes
M

matrix for plane strain


_
=
_

x
,
y
,
xy
,
z

T
_
EAS2 2 M

=
_

_
0
0
0 0
0 0
_

_
EAS4 4 M

=
_

_
0 0 0
0 0 0
0 0
0 0 0 0
_

_
EAS7 7 M

=
_

_
0 0 0 0 0
0 0 0 0 0
0 0 0 0
0 0 0 0 0 0 0
_

_
Window 4-25
Window 4-26: M matrices for Q4 element in axisymmetric case
EASelement
Number of
extra modes
M

matrix for axisymmetry


_
=
_

r
,
y
,
ry
,

T
_
EAS3 3 M

=
_

_
0 0
0 0
0 0 0
0 0
J (, )
r (, ) J
0
_

_
EAS5 5 M

=
_

_
0 0 0 0
0 0 0 0
0 0 0
0 0 0 0
J (, )
r (, ) J
0
_

_
where:
=
1
3
r
T
a
1
r
T
a
0
=
1
3
r
T
a
2
r
T
a
0
r
T
a
0
=
1
4
(r
1
+r
2
+r
3
+r
4
) r
T
a
1
=
1
4
(r
1
+r
2
+r
3
r
4
)
r
T
a
2
=
1
4
(r
1
r
2
+r
3
+r
4
)
r
1
, r
2
, r
3
, r
4
radial coordinates of element nodes.
Window 4-26
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.3.2.1:2
Numer. Impl. Incompressible... Dilatant Media
Window 4-27: Mapping from isoparametric to physical space
The operator M

is mapped onto physical space through the Jacobian of the isoparametric


transformation at the element centroid (which is the origin of the isoparametric space). For
second order tensors the following transformation from the isoparametric to physical space
holds (J is evaluated in current integration point while J
0
is evaluated at the center):

isoparametric
ij
=
det J
det J
0
J
ki
0

physical
kl
J
jl
0
.
With matrix notation this mapping can be rewritten in form :
physical
=
det J
0
det J
T
1
0

isoparametric
.
From the above equation one gets the denition of M matrix: M =
det J
0
det J
T
1
0
M

For plane strain and axisymmetry the transformation matrix T


0
takes the form:
T
0
=
_

_
J
2
11
J
2
21
J
11
J
21
0
J
2
12
J
2
22
J
12
J
22
0
2J
11
J
12
2J
21
J
22
J
11
J
22
+J
12
J
21
0
0 0 0 1
_

_
at ( = 0, = 0)
Window 4-27
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.3.2.1:3
Numer. Impl. Incompressible... Dilatant Media
4.3.2.2 EXTENSION OF THE EAS METHOD TO NONLINEAR
ELASTOPLASTIC ANALYSIS
With the EAS method the resulting nonlinear system of equations to be solved is as follows:
/
e=1,N
_
F
ext
F
int
(u
e
,
e
)

= 0
h
e
(u
e
,
e
) =
_

M
T
d = 0 (e = 1, 2, . . . , Nele)
Linearization of the above yields:
_
K
uu
K
u
K
u
K

_ _
u

_
=
_
F
ext

B
T
d
0
_

M
T
d
_
where:
K
uu
=
_

B
T
D
ep
B d K
u
=
_

B
T
D
ep
M d
K
u
=
_

M
T
D
ep
B d K

=
_

M
T
D
ep
M d
The general nonlinear strategy is summarized in Window 4-28.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.3.2.2:1
Numer. Impl. Incompressible... Dilatant Media
Window 4-28: Computation scheme using EAS elements
1. UPDATE nodal displacements at iteration i + 1 : u
(i+1)
= u
(i)
+ u
(i+1)
2. UPDATE at the element level
(i+1)
e

(i+1)
e
=
(i)
e
+
_
K
(i)

e
_
1
_
h
(i)
e
K
(i)
u
e
u
(i+1)
e
_
3. COMPUTE enhanced strain:
(i+1)
= Bu
(i+1)
e
+M
(i+1)
e
4. COMPUTE stress state
(i+1)
n+1
and corresponding tangent operator D
ep(i+1)
n+1
via standard
strain driven return algorithm
5. INTEGRATE element matrices and internal forces
K
(i+1)
uu
e
=
_

e
B
T
D
ep(i+1)
n+1
B d K
(i+1)
u
e
=
_

e
B
T
D
ep(i+1)
n+1
M d
K
(i+1)
u
e
=
_

e
M
T
D
ep(i+1)
n+1
B d K
(i+1)

e
=
_

e
M
T
D
ep(i+1)
n+1
M d
F
int
(i+1)
e
=
_

e
B
T

(i+1)
n+1
d h
(i+1)
e
=
_

e
M
T

(i+1)
n+1
d
6. PERFORM condensation procedure
F
int(i+1)
e
= F
int(i+1)
e
K
(i+1)
u
e
_
K
(i+1)

e
_
1
h
(i+1)
e
K
(i+1)
e
= K
(i+1)
e
K
(i+1)
u
e
_
K
(i+1)

e
_
1
K
(i+1)
u
e
7. ASSEMBLE and SOLVE for new displacements increment
/
e=1,N
_
K
(i+1)
e
_
u
(i+1)
= F
ext
/
e=1,N
F
int(i+1)
e
8. SET i = i + 1 and perform next iteration if needed.
Window 4-28
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.3.2.2:2
Numer. Impl. Incompressible... Dilatant Media
4.3.2.3 REMARKS AND ASSESSMENT OF EAS ELEMENTS
1. the EAS2 element in plane strain, EAS3 in axisymmetry are strongle recommended for
stability and ultimate state analysis when using isotropic plastic models with dilatant plastic
(for ex. Drucker-Prager if > 0)
2. the EAS4 element in plane strain, EAS5 in axisymmetry are strongle recommended for
stability and ultimate state analysis when using anisotropic plastic models with any type
of plastic ow (only multilaminate model)
3. if one wants to get highly accurate results for beams modelled with aid of Q4 elements
(retaining walls, etc... ) the EAS7 (in plane strain case exclusively) element gives superior
results even for coarse distorted meshes, but its application is limited to elastic type of
materials only (thus this element is not directly supported from the user interface).
4. the incompatible part of deformation M is not taken into account for problem of consol-
idation.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.3.2.3:1
Preface Numerical Implementation
4.4 FAR FIELD
Standard nite elements can be used only to discretize a bounded domain. In the case of
a problem given in an unbounded domain, innite elements, attached to the boundary of
a bounded domain, can be used to describe interaction between bounded domain and its
innite surroundings. The formulation is based on the concept of innite mapped elements
(see
2
). The innite mapped elements concept is summarized in the following windows.
Formulation of the mapped innite element
Mapping for 2D innite element
Evaluation of element matrices for innite mapped elements
Window 4-29: Formulation of the mapped innite element
To summarize the concept of innite mapped elements let us cosider the following one-
dimensional situation given in the Fig. The element which extends from node 1 through node
2 up to node 3, placed at innity, is mapped onto parent element in the local coordinate
system. Once the arbitrary position of the pole of expansion x
0
is choosen one can dene the
position of node 2 using the expression
x
2
= 2 x
1
x
0
The mapping from local coordinate system to the global one can be done in standard way
using the following rule
x() =
2

i=1
N

i
() x
i
where node 3 is disregarded from the summation.
2
O.C. Zienkiewicz, C. Emson, P. Bettes, A Novel Boundary Innite Element, International Journal for
Numerical Methods in Engineering, vol. 19, p.393-404, 1983
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.4:1
Preface Numerical Implementation
The innite mapping functions are dened as follows
N

1
() = 2/(1 )
N

2
() = 1 N

1
()
It can be easily seen that = 1, 0, +1 corresponds respectively to positions of node 1, node
2 and node 3 in global coordinate system.
The basic eld variable is interpolated using standard shape functions and written in the
polynomial form takes the form
u() = a
0
+a
1
+a
2

2
+............
in which coecients a
i
depend on element shape functions. Solving the mapping expression
for we get
= 1
2a
r
where the distance from the pole O to any point within the element is denoted by r and
a = x
2
x
1
. Substituting the expression for into expression for basic eld variable we can
get it in the form expressed in terms of global coordinate r
u() = b
0
+
b
1
r
+
b
2
r
2
+..........
where b
0
= 0 is implied since the variable u vanishes at innity. From the above one can
show that suitable decay functions 1/r
n
for basic variable one can get by choosing proper
shape functions for it.
Window 4-29
Window 4-30: Mapping for 2D innite element
The mapping functions for the 2D element can be easily obtained as a product of one-
dimensional Lagrangian shape functions along local direction and innite shape functions
along direction . If we assume that the poles positions are dened then we can dene
positions of nodes 2 and 3 as
x
2
= 2 x
1
x
O1
x
3
= 2 x
4
x
O4
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.4:2
Preface Numerical Implementation
altough the denition of the innite element geometry is usually done by introduction of
nodes 1, 2, 3 and 4 rather then by poles positions O1 and O2.
The set of the innite shape functions for quadrilateral innite 4-node element INFQ4 is as
follows
M
1
(, ) = N

1
() N
1
()
M
2
(, ) = N

2
() N
1
()
M
3
(, ) = N

2
() N
2
()
M
4
(, ) = N

1
() N
2
()
in which the Lagrangian shape functions N
i
()
N
1
() =
1
2
(1 )
N
2
() =
1
2
(1 +)
The coordinate element mapping is dened as
x(, ) =
4

i=1
M
i
(, ) x
i
The interpolation of the displacement eld within the element is assumed in the form
u(, ) = N
1
(, ) u
1
+N
4
(, ) u
4
Window 4-30
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.4:3
Preface Numerical Implementation
4.5 OVERLAID MESHES
To explain the idea of overlaid meshes let us consider a 2D domain 30m wide and 30m high,
with a square excavation zone of size 6m by 6m placed at the center of the domain. During
the excavation a tunnel lining is installed (see Fig. below).
Tunnel
lining
Kinematic
constraints
q = 1 kN/m/m
In the context of the overlaid meshes approach we can create a relatively coarse mesh, labeled
as layer number 0, which will represent soil behavior far from the excavation zone. In the
neighborhood of the excavation zone, to enhance the quality of the results, we need a denser
mesh and thus we superpose a second mesh labeled as layer number 1. In the 2D case
we could easily generate a mesh in layer 1 in such a way that its external boundary nodes
positions coincide with the nodes of the coarse mesh. However, in 3D applications this is
usually not that easy and it is simpler, from the users point of view, to superpose layer 1
with layer 0 without additional restrictions. This corresponds to overlaid meshes for which
we have to handle the problem of overlapping of few meshes plus the continuity of primary
variables (displacements, pressures etc..) handled via Lagrange multipliers.
In the considered example a mesh consisting of 10 x 10 elements is generated in layer 0 and 20
x 20 elements in layer 1 (layer 1 is 14m wide and 14m high) (see Fig. above). As it is shown in
Fig. below some of the elements in layer 0 are fully overlapped (identied with circles) while
some others are overlapped only partially (identied with triangles). All these elements which
are fully overlapped, and nodes which exclusively belong to these elements, are neglected
during computations. Problems arise in the case of partially overlapped elements as part
of their domain must be deleted from the resulting stiness and right-hand-side vector. To
handle this case an exact geometrical intersection of a given element in layer 0 and elements
in layer 1 must be computed to estimate the overlapped volume. For further application, this
intersection is computed for each integration point subdomain of a given element in layer 0
and set of elements in layer 1.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.5:1
Preface Numerical Implementation
Element fully overlapped Element partially overlapped
If we analyse the situation shown in Fig. below we can notice that for the integration point
number 2 the ratio between nonoverlapped and total volume becomes zero (as the integration
area associated with point 2 is entirely overlapped) while for other integration points it is
smaller than 1 but greater than zero. Hence a simple computation scheme, in which each
integration point volume is weighted by the aforementioned ratio, can be used.
1 2
3 4
Element in layer 0
Elements in layer 1
Non overlapped
area associated
with point 1
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.5:2
Preface Numerical Implementation
4.6 ALGORITHMS
Several solution algorithms can be activated either individually or in a sequence: initial state,
stability, time dependent (driven load/consolidation/creep).
All refer in some way to a NewtonRaphson iterative scheme which is therefore described in
Window 4-31. The specic algorithms are described next.
The simulation of excavation/construction stages is managed through existence functions
associated with elements at the level of the data preparation.
MODIFIED NEWTON-RAPHSON ALGORITHM
CONVERGENCE NORMS
INITIAL STATE ANALYSIS
STABILITY ANALYSIS
ULTIMATE LOAD ANALYSIS
CONSOLIDATION
CREEP
LOAD TIME FUNCTIONS AND TIME STEPPING
EXCAVATION AND CONSTRUCTION STAGES
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.6:1
Preface Numerical Impl. Algorithms
4.6.1 FULL/MODIFIED NEWTON-RAPHSON ALGORITHM
Window 4-31: Full/Modied Newton-Raphson algorithm
1. Step initialization : n = n + 1
set iteration counter i = 0
set u
(i=0)
n+1
= 0
Accumulate total external force vector F
ext
n+1
2. At the element level, at each integration point
if creep is active then compute the creep strain increment
cr
n+1
according to the section
3.4 (MATERIAL MODELS/CREEP)
set the initial stress/strain increments
0n+1
,
0n+1
3. i = i + 1
4. Assemble internal force vector and stiness matrix using computed at the element level
and at each integration point current total stress
(i+1)tot
n+1
and current tangent stiness
matrix D
ep
n+1
( an elastic stress state is computed via formula
(i+1)e
n+1
=
n
+
0n+1
+
D
e
(
(i+1)
n+1

cr
n+1

0n+1
) )
F
int (i+1)
n+1
= /
e=1,N
_

e
B
T

(i+1)tot
n+1
d
K
(i+1)
n+1
= /
e=1,N
_

e
B
T
D
ep(i+1)
n+1
B d
5. Solve system of equations
K
(i+1)
n+1
u
(i+1)
=F
ext
n+1
F
int (i+1)
n+1
(for modied NR technique assume K
(i+1)
n+1
= K
n
(for full NR use curent K
(i+1)
n+1
matrix)
6. Update accumulated increment of displacements: u
(i+1)
n+1
= u
(i)
n+1
+ u
(i+1)
7. If convergence of the iteration process has not been achieved go to step 1
Window 4-31
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.6.1:1
Preface Numerical Impl. Algorithms
4.6.2 CONVERGENCE NORMS
RHS-OUT OF BALANCE EUCLIDEAN NORM
ENERGY INCREMENT NORM
TOTAL ENERGY NORM
CONVERGENCE/DIVERGENCE/CONTINUATION STRATEGY
The right-hand side (RHS) out of balance Euclidean norm, total energy and incremental
energy norms are used in the code to detect convergence or divergence of the Newton-
Raphson iterations. These norms are evaluated separtely for solid and for uid phases in
case of twophase computations. The RHS out of balance norm, incremental energy norm,
and total energy norm are dened in Windows 4-32, 4-33, 4-34 respectively. Detection of
convergence/divergence state and possible continuation strategy is setup inWindow 4-35.
Window 4-32: RHS-out of balance Euclidean norm
The RHS-out of balance Euclidean norm is dened as follows:
|F
ext
n+1
F
int (i+1)
n+1
|
|F
ext
n+1
F
int
n
|
TOL
This norm is evaluated selectively for dierent degree of freedom ( translations u, rotations
, pore pressures p
F
, temperatures T and humidities W). The following table shows active
RHS norms for dierent analysis types and dierent degrees of freedom (+ means active and
- nonactive).
Analysis Type/DOF type u p
F
T W
Deformation + + +
()
- -
Deformation+Flow + + +
()
- -
Flow - + - - -
Heat transfer - - - + -
Humidity transfer - - - - +
The two following tollerances for RHS norm are used i.e. FTOL (set to 0.01 by default)
and QTOL (set to 0.001 by default) related to solid phase DOFs and uid phase DOFs
respectively. The following table shows corresponding tollerances for selected DOFs.
DOF Active tollerance TOL
u FTOL
p
F
QTOL
FTOL
T FTOL
W FTOL
()
The reference value, |F
ext
n+1
F
int
n
|, related to rotational DOFs is set to 0.1 |F
ext
n+1

F
int
n
|
u
L where L is some averaged nite element size, factor 0.1 has been assumed by rule
of thumb and |F
ext
n+1
F
int
n
|
u
is an Euclidean norm of the RHS corresponding to translational
DOFs.
Window 4-32
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.6.2:1
Preface Numerical Impl. Algorithms
Window 4-33: Energy increment norm
Th increment of energy and then the energy increment norm (valid for iteration i 2) are
evaluated as follows (for solid and uid phase separately):
E
S(i+1)
n+1
=
1
2
F
int u(i+1)
n+1
u
(i+1)
E
F(i+1)
n+1
=
1
2
F
int (i+1)
n+1
p
F(i+1)
E
S(i+1)
n+1
E
S(i)
n+1
E
S(i=2)
n+1
ESTOL
_
by default ESTOL = 10
3
_
E
F(i+1)
n+1
E
F(i)
n+1
E
F(i=2)
n+1
EFTOL
_
by default EFTOL = 10
3
_
Window 4-33
Window 4-34: Total energy norm
The total energy and then the total energy norm (valid for step n > 1) are evaluated as
follows (separately for solid and uid phase):
E
S(i+1)
n+1
= E
S
n
+ E
S(i+1)
n+1
E
F(i+1)
n+1
= E
F
n
+ E
F(i+1)
n+1
E
S(i+1)
n+1
E
S(i)
n+1
E
S
n
ESTOL
_
by default ESTOL = 10
3
_
E
F(i+1)
n+1
E
F(i)
n+1
E
F
n
EFTOL
_
by default EFTOL = 10
3
_
Window 4-34
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.6.2:2
Preface Numerical Impl. Algorithms
Window 4-35: Convergence / divergence / continuation
Convergence strategy
The step is converged if all active RHS out of unbalance norms are satsed simutaneously
(see Window 4-32)
|F
ext
n+1
F
int (i+1)
n+1
|
|F
ext
n+1
F
int
n
|
TOL (FTOL/QTOL)
or if the following set of conditions is satised:
|F
ext
n+1
F
int (i+1)
n+1
|
|F
ext
n+1
F
int
n
|
10 TOL (for all active RHS norms)
E
S(i+1)
n+1
E
S(i)
n+1
E
S
n
ESTOL/10
E
F(i+1)
n+1
E
F(i)
n+1
E
F
n
EFTOL/10 (if active)
Detection of divergence
The divergence of the iteration process is detected if the following set of conditions is satised:
|F
ext
n+1
F
int (i+1)
n+1
|
|F
ext
n+1
F
int
n
|
(for all active RHS norms)
E
S(i+1)
n+1
E
S(i)
n+1
E
S
n
and
E
F(i+1)
n+1
E
F(i)
n+1
E
F
n
(if active)
E
S(i+1)
n+1
E
S(i)
n+1
E
(i=2)S
n
and
E
F(i+1)
n+1
E
F(i)
n+1
E
(i=2)F
n
(if active)
For stability driver = 5.0 and for all others = 10.0.
Continuation strategy
In case when the maximum number of iterations is reached the following action is performed
depending on the status of the check box Auto in calculation window called Increase max.
nr of iterations. If it is ON (default setting) and all RHS out of balance norms are below value
1.0 (100%) then the maximum number of iterations is updated automaticaly (increased by 5)
and the process continues unless the new updated maximum number of iterations overreaches
the absolute maximum number of iterations (set in the menu under item Control). If this
check box is OFF, when reaching current maximum number of iterations a dialog box will be
displayed on a screen waiting for new maximum number of iterations to be dened by the
user.
Window 4-35
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.6.2:3
Preface Numerical Impl. Algorithms
4.6.3 INITIAL STATE ANALYSIS
The initial state in soil mechanics results essentially (but not exclusively) from gravity. The
correct implementation of gravity requires simultaneous application (or superposition) of the
gravity itself and the corresponding initial stresses, which may undergo additional constraints
set by the user (initial state K
0
eect ). The corresponding multistep algorithm is described
in Window 4.6.3:
Window 4-36: Initial state multistep algorithm
Initialize:
set time t = 0
assume rst gravity increment and increment size for further steps
set =
0
set step counter n = 0
for each element and for each integration point set
0
= 0
1. Continue
2. Set counter of initial stress reevaluations (for superposition) l = 0
3. Set iteration counter i = 0
4. Continue
5. Evaluate equivalent partial gravity nodal load: F
g
= /
e=1,N
_

e
N
T
bd, where b is the
gravity load
6. Accumulate total external force vector F
ext
n+1
adding to F
g
additional external forces if
needed (also weighted by the factor )
7. Evaluate corresponding stress state
(i)
n+1
and constitutive tangent matrix D
ep(i)
n+1
for given
u
(i)
n+1
,
(i)
n+1
= B u
(i)
n+1
and given
0
, using the following expression:

(i)
n+1
=
n
+
0
+D
e
(
(i)
n+1

p(i)
n+1
)
where the increment of plastic strain
p(i)
n+1
and D
ep(i)
n+1
are obtained from return mapping
algorithm
8. Assemble internal forces (adding the eect of initial pore pressures if needed, here pore
pressures are also weighted by the current gravity multiplier )
F
int (i)
n+1
= /
e=1,N
_

e
B
T
_

(i)
n+1
+
T
p
F(i)
n+1
_
d
9. Assemble stiness matrix: K = /
e=1,N
_

e
B
T
D
ep(i)
n+1
B d
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.6.3:1
Preface Numerical Impl. Algorithms
10. Solve system of equations: K u
(i+1)
= F
ext
n+1
F
int (i)
n+1
11. Update accumulated increment of displacements and strains: u
(i+1)
n+1
= u
(i)
n+1
+u
(i+1)
,

(i+1)
n+1
= Bu
(i+1)
n+1
12. If convergence of the iteration process has not been achieved set i = i + 1 and go to step
3
13. Increase l = l + 1
14. IF l <MAX REPEAT (by default MAX REPEAT=2) then
Update the initial stress increment
0
(via procedure given in Window 4-37 Initial
stress increment correction
Repeat last step introducing corrected initial stresses, GO TO step 2
15. Set : u
n+1
= u
(i+1)
n+1
= 0,
n+1
=
(i+1)
n+1
= 0
16. If n < n
max
set n = n + 1, = + and GO TO step 1
Window 4-36
Window 4-37: Initial stress increment correction
Correction of the assumed initial stresses is done based on the previous stress state
n
,
current stress state
n+1
and additional set of constraints which include the eect of K
0
(the ratio between selected normal stress components can be dened in a local coordinate
system dened by the user). We assume that the y component of the stress state is computed
accurately and we want to full the conditions :
()
x
/
()
y
= K
ox
and
()
z
/
()
y
= K
oz
. The
following steps lead to the evaluation of corrected initial stress increment
0
.
1. Transform
n+1

()
n+1
,
n

()
n
(upper index () means local system, one in which
K
0
has been set up)
2. Compute stress increment between n and n + 1 steps:
()
n+1
=
()
n+1

()
n
3. Assume :
()
0
=
()
n+1
4. IF K
0x
is specied by the user then correct
()
0x
3
:
()
0x
= K
0x

()
0y
5. IF K
0z
is specied by the user then correct
()
0z
:
()
0z
= K
0z

()
0y
6. Transform
()
0
from local K
0
system to global one:
()
0

0
Window 4-37
3
Specication of K
0
can be done exclusively in conned directions
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.6.3:2
Preface Numerical Impl. Algorithms
4.6.4 STABILITY ANALYSIS
The available denitions of safety factors were introduced in Appendix 3.7.1, i.e.
1. , such that global instability corresponds to

global failure
=
m
+s
where is a uniform deviatoric stress multiplier in the domain. From the implementation
point of view the approach amounts to a progressive modication of the yield criteria
parameters until failure occurs.
2. Alternatively for two-parameter (C ) criteria, can be used
SF
2
=

y
d
s
_

s
d
s

where
y
= C +

n
tan

is the yield stress according to the MohrCoulomb criterion, C


the cohesion,

n
the eective normal stress, the friction angle, SF
2
the safety factor and

s
denes the sliding surface. An algorithm which ts the plasticity based approach can
be deduced from the above equation; rewriting the equation as
_

s
d
s
=
_

y
d
s
SF
2
=
_

s
(C +

n
tan

) d
s
SF
2
it is observed that SF
2
can be viewed as the dividing factor of C and tan

for which
instability is reached.
The stability algorithm is summarized in Window 4-38
Window 4-38: Stability algorithm for SF=SF2
1. Initialization
Set SF
n
= SF
0
(start with the prescribed lower bound of the safety factor SF
0
)
2. For each step
SF
n+1
= SF
n
+ SF (increment the safety factor)
C
n+1
=
C
SF
n+1
(scale the cohesion)
tan

n+1
=
tan

SF
n+1
(scale the tangens of friction angle)
3. Solve the boundary-value problem, iterate as needed
4. Go to 2 until divergence occurs
5. At divergence SF
n
SFSF
n+1
; estimate of the safety factor
Window 4-38
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.6.4:1
Preface Numerical Impl. Algorithms
Remarks:
1. Notice that the material characteristics are modied by the algorithm, which is, strictly
speaking, only valid close to SF= 1.
2. Divergence is normally accompanied by a localized strain eld on a slip surface, in which
SF corresponds to the given denition.
3. Let SF
2
be the default denition of the safety factor for two-parameter (C ) criteria.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.6.4:2
Preface Numerical Impl. Algorithms
4.6.5 ULTIMATE LOAD ANALYSIS
An ultimate load analysis follows the classical NewtonRaphson scheme. Computation are
pursued until divergence is reached. Refer to Sect. 4.6.1. The simulation of drained behaviour
should be done using drivers as for 1phase or 2phase medium but using large time step.
The undrained behavior can be done only with 2phase algorithm (Sect. 4.6.6).
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.6.5:1
Preface Numerical Impl. Algorithms
One-dimensional test
4.6.6 CONSOLIDATION ANALYSIS
The consolidation algorithm simulates the transient behavior of the twophase medium. The
time step can vary starting from small values at the beginning up to large steps at the end
of the process. The only limitation is that minimum time step should be greater than the
critical one estimated via following relation:
t t
crit
=

F
h
2
E
oed
k
_
1
4
+
1
6
E
oed
c
_
(4.7)
where:
c compressibility of uid = n/
F
; n porosity,
F
uid bulk modulus
k Darcy coecient

F
uid specic weight
integration coecient (in Z SOIL = 1)
E
oed
=
E (1 )
(1 +) (1 2)
oedometric stiness modulus
h element size.
The one-dimensional test is taken as the basis for the estimation. The condition for nonoscil-
latory pore pressure distribution can be formulated as p
F


tot
1 +E
oed
c
(with a positive sign
for tensile stresses). In the incompressibility limit (c = 0) this condition means that the stress
transferred by the uid cannot be higher then the total stress value. The estimate will be
derived for linear two-node element with an equal interpolation for displacements and pres-
sures. Let us take a mesh which consists of a single nite element, as shown in Fig.(4.6.6),
and assume that the initial pressure is zero. Solving the resulting system of equations with
the set of the boundary conditions (u
1
= 0 and p
2
= 0) we obtain the following value for the
pressure increment in the rst time step:
p
1
=
qh
2E
oed
h
4E
oed
+c
h
3
+
kt

F
1
h
(4.8)
From the consistency condition, which in a general case can be expressed in the form,(see
4
):
p
1

q
1 +E
oed
c
(4.9)
the condition given in eq. 4.7 for the time step t is derived.
4
Vermeer P., Verruijt A. An accuracy condition for consolidation by nite elements.
Int.J.Numer.Anal.Meth.Geomech.5, p.1-14, 1981
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.6.6:1
Preface Numerical Impl. Algorithms
4.6.7 CREEP ANALYSIS
Creep introduces a time dependent deformation at constant or variable stress. The imple-
mentation scheme is summarized in Sect. 4.6.1. An important aspect is that the current
formulation does not produce creeping eect due to initial stresses. The phenomenon starts
due to stress variation starting at time at which the creep is activated. This means that for
example the eect of secondary consolidation understood as a result of creeping should be
modelled running the consolidation and creep simultaneously. Since creep is a time dependent
process it requires therefore a time stepping procedure, which is shown in Window 4-39.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.6.7:1
Preface Numerical Impl. Algorithms
4.6.8 LOAD FUNCTION AND TIME STEPPING PROCEDURE
For time dependent processes like creep, consolidation, transient ow, each item consisting
load or prescribed boundary condition may have attached a function describing its variability
in time (physical or pseudo) during the process. This is called a Load Time Function and
is dened as a list of pairs (t
k
, v
k
), as shown in Window 4-39 assuming linear interpolation
for intermediate points. Dierent loads acting on the model may have attached dierent
load time functions, but its argument must be common for all time function in a job. It
must have the meaning of a physical time (consistently with constitutive data ) for such a
Time Dependent problems as: Transient Fluid, Heat, Humidity Flow, Creep,
Consolidation, while for the remaining (i.e. Driven Load, Deformation, all Steady
State) ones, may be treated as the non-physical one.
Window 4-39: Time stepping procedure
Load time history
Applied loads are characterized by a prescribed load amplitude and a load multiplier (load
factor). At time t
n+1
the applied load is then
F
n+1
= F
0
LF
n+1
.
The load factor LF is dened as a function of time. Each load can be associated with a
dierent load factor.
Load timehistory
Time step
The time dependent processes start always with the initial time step increment dened by the
user t
1
= t
BEG
. Each next time step is predicted through relation t
n+1
= t
n
t
MULT
with time multiplier t
MULT
dened by the user. This next time step increment can be
corrected automatically by the system if in between t
n
t
n
+ t
predicted
n+1
at least one of the
existence functions changes its value (from OFF to ON or vice versa) or some characteristic
point on one of the load time functions is detected. Once such situation is detected the time
step increment is reset to the initial value t
BEG
.
Window 4-39
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.6.8:1
Preface Numerical Impl. Algorithms
4.6.9 SIMULATION OF EXCAVATION AND CONSTRUCTION STAGES
The simulation of excavation/construction stages is normally embedded in every time depen-
dent driver (driven load / consolidation etc..). This type of simulation requires a stiness
update at each excavation stage. The corresponding appropriate algorithmic choice is done
automatically by the system.
On the other hand, the input mesh must correspond to the maximal mesh. This may corre-
spond to the nal stage in a construction simulation or to the initial state for an excavation
simulation.
Management of stiness
The simulation of construction stages leads to a variable size of the stiness matrix and
therefore to a variable number of nodes and elements. This requires additional data manage-
ment. The input mesh must correspond to the maximal mesh. The corresponding node and
element numbering will be kept throughout the analysis.
At each excavation stage an optimization of the nodal numbering will be performed, and
a correspondence table with the input-output node numbering will be established. The
optimized mesh node numbering is used for analysis purposes.
Progressive unloading after excavation
If no LTF (unloading function) is specied for excavated elements, interaction forces from
excavated media will vanish immediately at the moment of excavation. In the case of elasto-
plastic media this may cause diculties in obtaining converged solution. To prevent this,
progressive unloading after excavation can be used which helps the system to redistribute
stresses in the surroundings of the excavated domain, and in consequence to obtain convergent
solution concerning new equilibrium state.
Unloading after excavation can be controlled. The interaction force between the excavated
and the remaining medium will be computed by performing
F
intEXC
=
_

exc
B
T

tot
d
_
LTF(t)
0 if no unloading function is given
_
over the excavated medium. The association of a load timehistory with this set of forces
provides the means for the control of unloading. For more details see section 7.6.2.
Related Topics
GEOTECHNICAL ASPECT. EXCAVATION, CONSTRUCTION ALGORITHM
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.6.9:1
Preface Numerical Implementation
4.7 CONSTITUTIVE THEORIES
DRUCKER-PRAGER MODEL-STRESS RETURN ALGORITHM
MOHR-COULOMB MODEL-STRESS RETURN ALGORITHM
CAP MODEL-STRESS RETURN ALGORITHM
MOHR-COULOMB(SMOOTH)-STRESS RETURN ALGORITHM
MULTILAMINATE MODEL-STRESS RETURN ALGORITHM
MODIFIED-CAM CLAY MODEL-STRESS RETURN ALGORITHM
AGING CONCRETE MODEL-IMPLEMENTATION SCHEME
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7:1
Preface Numer. Impl. Constitutive Equation
4.7.1 DRUCKER-PRAGER PLASTICITY STRESS RETURN PROB-
LEM
The stress return problem is formulated in Window 4-40 and the cutting plane stress return
algorithm is developed in Window 4-41. In Window 4-42 detailed numerical procedures for
stress return for DP plasticity model with/without tensile cuto condition are presented.
Finally in Window 4-45 the denition of consistent elastoplastic tangent stiness matrix is
derived.
Stress return problem
Stress point algorithm for DP plasticity
Stress point algorithm for DP plasticity including tensile cut-o condition
Consistent tangent stines matrix for DP plasticity
Window 4-40: Stress return problem
Stress return problem
Given:
the stress state at last conguration of the equilibrium
n
the increment of the initial stresses
0
the total eective strain increment
(i+1)
n+1
(here the initial strain and creep strain incre-
ments are already subtracted).
Find:

(i+1)
n+1
such that F
_

(i+1)
n+1
_
= 0
Window 4-40
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.1:1
Preface Numer. Impl. Constitutive Equation
Window 4-41: Stress point algorithm for DP plasticity
Let the trial stress be dened as folows:
(i+1)tr
n+1
=
n
+
0
+D
(i+1)
n+1
The nal stress state can be expressed:
(i+1)
n+1
=
(i+1)tr
n+1
D
p(i+1)
n+1
=
(i+1)tr
n+1
Dr
_

(i+1)
n+1
_
Explicit computation of and nal stress state:
Let:
F
_

(i+1)
n+1
_
= F
_

(i+1)
n+1
+
_

= F
_

(i+1)tr
n+1
_
+
_
F

(i+1)tr
n+1
_
T

= F
_

(i+1)tr
n+1
_

_
F

(i+1)tr
n+1
_
T
Dr
_

(i+1)tr
n+1
_
hence:
=
F
_

(i+1)tr
n+1
_
_
F

(i+1)tr
n+1
_
T
Dr
_

(i+1)tr
n+1
_
.
In case of isotropic elasticity the above equation can be expressed in simplied form expressing
all terms in the denominator in reduced stress space (q, p). In (q, p) space the elastic stiness
matrix D, plastic loading vector F/ and ow vector r take the following form:
D =
_
3 0
0 K
_
;
F

=
_

_
F
q
F
p
_

_
=
_
_
1

3
3a

_
_
; r =
_

_
Q
q
Q
p
_

_
=
_
_
1

3
3a

_
_
where p = I
1
/3, q =

3J
2
, so the simplied expression for plastic multiplier is as
follows:
=
F
_

(i+1)tr
n+1
_
9Ka

+
.
Finally the stress state in (q, p) stress space and then in full stress space is dened as:
p
(i+1)
n+1
= p
(i+1)tr
n+1
K(3a

)
q
(i+1)
n+1
= q
(i+1)tr
n+1
3
_
1

3
_

(i+1)
n+1
=
q
(i+1)
n+1
q
(i+1)tr
n+1
s
(i+1)tr
n+1
p
(i+1)
n+1
The above derivation holds for a general nonassociated ow rule.
Window 4-41
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.1:2
Preface Numer. Impl. Constitutive Equation
Window 4-42: Stress point algorithm: Numerical implementation
Calculate
trial stress:
(i+1)tr
n+1
=
n
+
0
+D
(i+1)
n+1
,
resulting invariants for trial stress: I
tr
1
n+1
, p
tr
n+1
,
_
J
tr
2
n+1
, q
tr
n+1
,
deviatoric stress at trial state: s
(i+1)tr
n+1
=
(i+1)tr
n+1
+p
(i+1)tr
n+1
IF tensile cuto is not active THEN
compute the value of DP surface plasticity condition: F
DP
= a

I
tr
1
n+1
+
_
J
tr
2
n+1
k
IF F
DP
< 0 THEN
set
(i+1)
n+1
=
(i+1)tr
n+1
EXIT
ELSE
Perform a return procedure to the DP surface (Window 4-43 ) with given data:
A

= a

, B

= k, A

= a

, s
(i+1)tr
n+1
, p
(i+1)tr
n+1
, q
(i+1)tr
n+1
EXIT
ENDIF
ELSE (tensile CutO is active)
compute the value of DP surface plasticity condition: F
DP
= a

I
tr
1
n+1
+
_
J
tr
2
n+1
k
compute the value of CUTOFF surface plasticity condition
F
CT
=
1

3
I
tr
1
n+1
+
_
J
tr
2
n+1

1

3
I

1T
IF F
DP
< 0 THEN
IF F
CT
< 0 THEN

(i+1)
n+1
=
(i+1)tr
n+1
EXIT
ELSE
Perform a return to the CUTOFF surface (Window 4-43 ) with given data:
A

=
1

3
, B

=
1

3
I

1T
, A
c
=
1

3
, s
(i+1)tr
n+1
, p
(i+1)tr
n+1
, q
(i+1)tr
n+1
EXIT
ENDIF
ELSE (F
DP
0)
IF F
CT
< 0 THEN
Perform a return to the DP surface (Window 4-43 ) with given data:
A

= a

, B

= k, A

= a

, s
(i+1)tr
n+1
, p
(i+1)tr
n+1
, q
(i+1)tr
n+1
EXIT
ELSE
Perform a return to the intersection of a DP and CUTOFF surfaces (Window 4-44 )
EXIT
ENDIF
ENDIF
ENDIF
Window 4-42
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.1:3
Preface Numer. Impl. Constitutive Equation
Window 4-43: Stress return algorithm for DP type surface
We assume that plastic surface and plastic potential are given in the following form:
F = A

I
1
+
_
J
2
B

= A

p +
1

3
q B

Q = A

I
1
+
_
J
2
= A

p +
1

3
q
The following algorithm performs return mapping for given trial stress data s
(i+1)tr
n+1
, p
(i+1)tr
n+1
, q
(i+1)tr
n+1
,
given plastic parameters A

, B

, A

and elastic constants: , K (extracted from D


e
matrix).
1. Reevaluate plastic surface/potential coecients in p q stress invariants space
A

= 3A

, A

= 3A

2. Compute plasticity condition value: F = A

p
(i+1)tr
n+1
+
1

3
q
(i+1)tr
n+1
B

3. Compute plastic multiplier value: =


F
KA

+
4. Compute nal value of p
(i+1)
n+1
: p
(i+1)
n+1
= p
(i)tr
n+1
K A

5. Check appex condition


IF A

,= 0 p
(i+1)
n+1
< (p
appex
=
B

) THEN
correct the volumetric plastic ow to allow exact return to the apex
and recompute the plastic multiplier value
A

= A

+
=
_
p
(i+1)
n+1
+p
appex
_
_
KA

+
_
KFA

KF
_
p
(i+1)
n+1
+p
appex
_
KA

=
F
KA

_
A

+
_
set the nal stress at the apex:

(i+1)
n+1
= p
(i+1)appex
n+1

EXIT
ELSE
compute q
(i+1)
n+1
: q
(i+1)
n+1
= q
(i+1)tr
n+1
3
_
1

3
_
set the nal stress in full space:
(i+1)
n+1
=
q
(i+1)
n+1
q
(i+1)tr
n+1
s
(i+1) tr
n+1
p
(i+1)
n+1
EXIT
ENDIF
Window 4-43
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.1:4
Preface Numer. Impl. Constitutive Equation
Window 4-44: Stress return: intersection of CUTOFF and DP surfaces
1. Form system of 2 equations expressing the consistency condition for both surfaces simul-
taneously (index
CT
is for cut-o condition):
_

_
_
F
DP

_
T
D
_
Q
DP

_ _
F
DP

_
T
D
_
Q
CT

_
_
F
CT

_
T
D
_
Q
DP

_ _
F
CT

_
T
D
_
Q
CT

_
_

_
_

DP

CT
_
=
_
F
DP
F
CT
_
It can be simplied for the case of isotropic elasticity:
_

_
9Ka

+ 9Ka

3
+
9Ka

3
+ 3K +
_

_
_

DP

CT
_
=
_
F
DP
F
CT
_
2. Solve the above system for
DP
,
CT
3. IF
DP
0 THEN
Perform a return to the cut-o surface (Window 4-43 ) with given data:
A

=
1

3
, B

=
1

3
I

1T
, A

=
1

3
, s
(i+1)tr
n+1
, p
(i+1)tr
n+1
, q
tr
n+1
EXIT
ENDIF
4. IF
CT
0 THEN
Perform a return to the DP surface (Window 4-43 ) with given data:
A

= a

, B

= k, A

= a

, s
(i+1)tr
n+1
, p
(i+1)tr
n+1
, q
(i+1)tr
n+1
EXIT
ENDIF
5. Compute nal values of p
(i+1)
n+1
, q
(i+1)
n+1
q
(i+1)
n+1
= q
(i+1)tr
n+1
3
_

DP
1

3
+
CT
1

3
_
p
(i+1)
n+1
= p
(i+1)tr
n+1
K
_

DP
(3a

) +
CT
_
3
1

3
__
6. Set the nal stress

(i+1)
n+1
=
q
(i+1)
n+1
q
(i+1)tr
n+1
s
(i+1) tr
n+1
p
(i+1)
n+1
7. EXIT
Window 4-44
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.1:5
Preface Numer. Impl. Constitutive Equation
Window 4-45: Consistent tangent stines matrix for DP plasticity
The general denition of the consistent stiness matrix for perfect (no hardening) single
plasticity surface is as follows:
D
ep
= D

_
D

__
D

_
T
_
F

_
T
D

where:
D

=
_
C+

2
Q

_
1
.
The inversion of the above modied elastic compliance matrix can be done with the aid of
the ShermannMorisson formula which is dened as follows:
_
A+uv
T
_
1
= A
1

A
1
vu
T
A
1
1 +u
T
A
1
v
.
This reduces signicantly the computational eort with respect to the direct inversion. The
denition of D

is valid for both DP and CUTOFF surfaces since the second derivative
applies only to the term related to J
2
invariant. In order to make use of ShermannMorisson
formula the following decomposition is made:
(D

)
-1
= C+

2
Q
J
2
J
2
_
J
2

__
J
2

_
T
+
Q
J
2

2
J
2

1
2
T+
_

E
_

T
. .
C
+ A
1
..

2
Q
J
2
J
2
_
J
2

__
J
2

_
T
+ A
2
..

Q
J
2
_
T
1
3

T
_
_
1
2
+A
2
_
T
. .
N
+
_
A
3

A
2
3
_
. .
A
4

T
+A
1
_
J
2

__
J
2

_
T
where:
A
3
=

E
T =
_

_
1 0 0 0 0 0
0 1 0 0 0 0
0 0 2 0 0 0
0 0 0 1 0 0
0 0 0 0 2 0
0 0 0 0 0 2
_

_
(generally for 3D).
Lets rewrite this expression in a form ready to apply ShermannMorisson formula:
(D

)
1
= N+A
4

T
+A
1
_
J
2

__
J
2

_
T
=
_
N+A
4

T
_
. .
K
+A
1
_
J
2

__
J
2

_
T
.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.1:6
Preface Numer. Impl. Constitutive Equation
Finally:
D

=
_
K+A
1
_
J
2

__
J
2

_
T
_
1
= K
1

K
1
_
J
2

_
A
1
_
J
2

_
T
K
1
1 +A
1
_
J
2

_
T
K
1
_
J
2

_
and
K
1
= N
1

N
1
A
4

T
N
1
1 +A
4

T
N
1

.
The inversion of matrix N is trivial since it is a diagonal matrix.
Remark:
In case when both DP and CUTOFF conditions are both active the denition of the
elastoplastic tangent/consistent tangent matrix is more complicated (as it is a multisurface
plasticity case) and is derived in Appendix 4.8.3.
Window 4-45
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.1:7
Preface Numer. Impl. Constitutive Equation
4.7.2 MOHR-COULOMB PLASTICITY STRESS RETURN AL-
GORITHM
The Mohr-Coulomb model supplied with Rankine type cut-o condition belongs to class
of multisurface plastic models. This model is dened by friction angle , dilatancy angle
, cohesion C and uniaxial tensile strength f
t
. The implementation of the stress return
algorithm and consistent tangent elasto-plastic operator D
ep
is based on paper of Larson and
Runneson
5
.
Set of plastic yield functions
Set of plastic ow potentials
Generalized stress return algorithm
Multisurface return procedure
Consistent tangent operator
Window 4-46: Set of plastic yield functions
F
1
=
1

3
K

k = 0
F
2
=
1

2
= 0
F
3
=
2

3
= 0
F
4
=
1
f
t
= 0
where:
K

=
1 sin
1 + sin
k =
2C cos
1 + sin
Derivatives of F
i
with respect to principal stress components
1
,
2
,
3
are as follows:
F
1

=
_
1 K

0
_
T
F
2

=
_
1 1 0
_
T
F
3

=
_
0 1 1
_
T
F
4

=
_
1 0 0
_
T
Window 4-46
5
Larson, Runneson:Implicit integration and consistent linearization for yield criteria of the Mohr-Coulomb
type. Mechanics of cohesive-frictional materials, vol.1, 367-383 (1996).
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.2:1
Preface Numer. Impl. Constitutive Equation
Window 4-47: Set of plastic ow potentials
Q
1
=
1

3
K

Q
2
=
1

2
Q
3
=
2

3
Q
4
=
1
where:
K

=
1 sin
1 + sin
and dilatancy angle is denoted by . A nonassociated ow rule, governed by the value of
dilatancy angle , is assumed only for M-C surface while for all others the associated one is
used.
Derivatives of Q
i
with respect to principal stress components
1
,
2
,
3
are as follows:
Q
1

=
_
1 K

0
_
T
Q
2

=
_
1 1 0
_
T
Q
3

=
_
0 1 1
_
T
Q
4

=
_
1 0 0
_
T
Window 4-47
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.2:2
Preface Numer. Impl. Constitutive Equation
Window 4-48: Generalized stress return algorithm
The upper index (tns) is used to express some quantities in tensorial notation and upper index
(prc) is used to distinguish between full stress space and space of principal stresses
Problem statement:
Given:

n
- stress state at the last converged step n

(i+1)
n+1
- strain increment
Find:

(i+1)
n+1
and elasto-plastic tangent/consistent tangent operator D
ep(i+1)
n+1
Algorithm
1. compute trial stress state:
tr
=
n
+D
e

(i+1)
n+1
2. compute principal stresses :
1
,
2
,
3
for
tr
3. check M-C and Rankine yield criteria:
IF F
1
< 0 and F
4
< 0 THEN

(i+1)
n+1
=
tr
D
ep
= D
e
EXIT
ENDIF
4. check appex condition in case of deviatoric ow
IF K

< 1 and K

= 1 THEN
IF
1
+
2
+
3
> 3C THEN
perturb K

by fraction of K

: K

=
+K

1 +K

( = 0.001)
ENDIF
ENDIF
5. nd eigenbases using Serrins theorem for all three principal stresses:
m
(tns)
i
[33]
=

tr
i
d
e
i
_

tr(tns)
(I
tr
1

tr
i
)
(tns)
[33]
+
I
tr
3

tr
i
_

tr(tns)
_
1
_
6. perform multisurface return procedure to obtain
(i+1)
n+1
and D
ep (i+1)
n+1
described in Win-
dow 4-49
Window 4-48
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.2:3
Preface Numer. Impl. Constitutive Equation
Window 4-49: Multisurface return procedure
1. Dene set of active surfaces Jact = i : F
i
> 0 and number of active surfaces M
2. Initialize: vector of plastic multipliers :
i
= 0
i4
vector of principal plastic strain increments
p(prc)
i
= 0
i3
3. Set vector F
[41]
: F
i
= F
i
(
tr
1
,
tr
2
,
tr
3
)
4. Iterate : k = k + 1
- for given set of active surfaces compute matrix H
[MM]
H
ij
= 2G
_
F
Jact(i)

(prc)
_
T
_
Q
Jact(i)

(prc)
_
+ trace
_
F
Jact(i)

(prc)
_
trace
_
Q
Jact(i)

(prc)
_
- set vector F

[M1]
: F

i
= F
Jact(i)

i=1,M
- solve H
[MM]

[M1]
= F

[M1]
- verify current set of active surfaces
FOR i = 1, M
IF
Jact(i)
+
i
< 0 reduce list Jact by deletion of Jact(i) surface
END DO
- if list of active surfaces has been reduced go to next iteration
- update plastic multipliers and plastic strains
FORi = 1, M

Jact(i)
=
Jact(i)
+
i

(prc)p
k
=
(prc)p
k
+
Jact(i)
Q
Jact(i)

(prc)
k

k=1..3
END DO
- update principal stresses (here is Lame constant)

(prc)
i
=
tr(prc)
i
2G
p(prc)
i
(
p(prc)
1
+
p(prc)
2
+
p(prc)
3
)
i=1,3
- evaluate yield condition values F
i
(
1
,
2
,
3
)
i=1,4
- add to list Jact those surfaces which were nonactive before but now become active due
to violation of the condition F
i
< 0 for i / Jact
- if i : [F
i
[ > TOL goto next iteration
5. Set stress state in full space (see window transition from tensorial to matrix notation)

(tns)
kl
=
3

i=1

(prc)
i
m
(tns)
i
kl
6. Set tangent or algorithmic stiness (ATS) matrix D
ep
(Window 4-50)
Window 4-49
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.2:4
Preface Numer. Impl. Constitutive Equation
Window 4-50: Consistent tangent operator
1. Given : D
e
[NSTRENSTRE]
, H
[MM]
,
F
JACT(i)

(prc)
,
Q
JACT(j)

(prc)
,
[M1]
2. Set yield condition and plastic potential derivatives in full stress space in matrix form
a
[NSTREM]
=
_
_
F
Jact(1)

_
T
_
F
Jact(2)

_
T
....
_
F
Jact(M)

_
T
_
b
[NSTREM]
=
_
_
Q
Jact(1)

_
T
_
Q
Jact(2)

_
T
....
_
Q
Jact(M)

_
T
_
this can be done using F
i
and Q
i
derivatives over principal stress components
1
,
2
,
3
and eigenbases m
(tns)
i
through following formula:
F
Jact(n)

kl
=
3

i=1
F
Jact(n)

i
m
(tns)
i
kl
Q
Jact(n)

kl
=
3

i=1
Q
Jact(n)

i
m
(tns)
i
kl
3. Perform transition from tensorial to matrix notation (taking doubled values for shear terms)
via standard procedure given in Window ( transition from tensorial to matrix)
4. Set tangent stiness matrix is dened:
D
ep
= D
e
D
e
b H
1
a
T
D
e
5. Set algorithmic (consistent) tangent matrix (ATS):
D
ep(ATS)
= D
ep
4G
2
M

n=1,3

Jact(n)
B
n
where matrix B
n
in tensorial notation is dened as
B
n
ijkl
=
3

m=1
Q
Jact(n)

m
d m
m
ij
d
kl
d m
m
ij
d
kl
=

tr
m
d
tr
m
[I
ijkl

ij

kl
I
tr
3
/
tr
m
_
A
ijkl
(
tr(tns)
)
1
ij
(
tr(tns)
)
1
kl
_

ij
m
m
kl
+m
m
ij

kl
I
tr
3
/ (
tr
m
)
2
_
m
m
ij
(
tr(tns)
)
1
kl
+ (
tr(tns)
)
1
ij
m
m
kl
_
+
2
_
I
tr
3
/ (
tr
m
)
3
1
_
m
m
ij
m
m
kl
]
A
ijkl
=
1
2
_
_

tr(tns)
_
1
ik
_

tr(tns)
_
1
jl
+
_

tr(tns)
_
1
il
_

tr(tns)
_
1
jk
_
I
ijkl
=
1
2
(
ik

jl
+
il

jk
)

ij
=
1 if i = j
0 if i ,= j
and conversion of matrices B
n
from tensorial to matrix notation is done in standard way
Window 4-50
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.2:5
Preface Numer. Impl. Constitutive Equation
4.7.3 CAP PLASTICITY STRESS RETURN ALGORITHM
TWO SURFACE STRESS RETURN FOR DP AND CAP SURFACES
The stress return algorithm is presented in Window 4-51. It uses the stress return algorithm
developed for the DP and CUTTOFF surfaces (see Section 4.7.1) and in most complex
situation, when CAP surface and/or DP one are activated, the stress return algorithm de-
veloped for multisurface plasticity (see Appendix 4.8.3). In the following implementation the
plastic volumetric strain was selected as a hardening parameter since the relation between
volumetric plastic strain and preconsolidation pressure can be integrated in exact form. In all
cases the stress return is run in reduced stress space (q p).
Window 4-51: Stress return algorithm for CAP model
Given:
the stress state
n
at last converged step n
the total eective strain increment
(i+1)
n+1
(here the initial strain and creep strain incre-
ments are already subtracted)
Algorithm
Run the trial return algorithm to DP and CUTOFF surfaces exclusively (Window 4-42)
IF DP condition or CUTOFF condition were activated THEN
compute the volumetric plastic strain increment
p(i+1)
v
n+1
using
trial mean stress p
tr(i+1)
n+1
and nal value of the mean stress p
(i+1)
n+1

p(i+1)
v
n+1
=
_
p
tr(i+1)
n+1
p
(i+1)
n+1
_
K
compute the trial preconsolidation pressure p
tr(i+1)
c
n+1
(via procedure described
in Window 4-54) for given initial preconsolidation pressure p
c
n
(at previous converged step) and
p(i+1)
v
n+1
compute current value of shape surface parameter R = R
_
p
tr
c
_
(i+1)
n+1
compute pressure p
CS
at the intersection of the CAP and DP surfaces
p
CS
= p
CS
_
R, p
tr(i+1)
c
n+1
_
IF p
(i+1)
c
n+1
p
CS
THEN
p
(i+1)
c
n+1
= p
tr(i+1)
c
n+1
accept the nal stress state obtained from the trial return to the DP and CUTTOFF
EXIT
ELSE
Run two-surface stress return for D-P and CAP surfaces as described in Window ??
ENDIF
ENDIF
Window 4-51
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.3:1
Preface Numer. Impl. Constitutive Equation
4.7.3.1 TWO SURFACE STRESS RETURN FOR DP AND CAP
SURFACES
The two surface return algorithm for DP and CAP surfaces is run through Multisurface
Closest Point Projection procedure described in Appendix 4.8.3. The procedure is run in
reduced stress space
()
=(p,q) and it requires an explicit derivation of yield condition, ow
vector, hardening function, and some auxiliary derivatives of these quantities with respect to
hardening parameters and stress components for all. These essential procedure ingredients
are set up for D-P surface in Window 4-52 and for CAP surface in Window 4-53.
Window 4-52: Multisurface Closest Point Projection Plasticity: DP surface
Surface index = 1
Plasticity condition
F
DP
= a

I
1
+
_
J
2
k = 0
Plasticity condition derivatives (over stress components):
F
DP

()
=
_

_
F
DP
q
F
DP
p
_

_
=
_
_
1

3
3a

_
_
Flow vector r and its derivatives with respect to stress
()
components:
r
DP
=
_
_
1

3
3a

_
_
r
DP

=
_
0 0
0 0
_
Hardening parameter: (does not exist)
Hardening function: (does not exist)
Window 4-52
Window 4-53: Multisurface Closest Point Projection Plasticity: CAP surface
Surface index = 2
Plasticity condition
F
C
=
_
F
C1
if p p
CS
F
C2
if p < p
CS
_
where:
F
C1
= q
2
+
M
2
(R 1)
2
_
p p
C
__
p +p
C
2p
CS
_
= 0
F
C2
= q
2
+
M
2
(R 1)
2
_
p
CS
p
C
__
p
C
p
CS
_
= 0
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.3.1:1
Preface Numer. Impl. Constitutive Equation
Plasticity condition derivatives (over stress components)
F
C

=
_

_
F
C1

()
if p p
CS
F
C2

()
if p < p
CS
_

_
where:
F
C1

()
=
_
_
2q
_
M
R 1
_
2 _
2p 2p
C
s
_
_
_
F
C2

()
=
_
2q
0
_
Flow vector r
C
:
r
C
=
_
r
C1
if p p
CS
r
C2
if p < p
CS
_
where:
r
C1
=
_
_
2q
_
M
R 1
_
2 _
2p 2p
CS
_
_
_
r
C2
=
_
2q
0
_
Hardening parameter
q =
p
V
Hardening function
h =
r
C1
p
=
_
M
R 1
_
2 _
2p 2p
CS
_
if p p
CS
0 if p < p
CS
Flow vector derivatives
_
r

()
_
:
r
C

()
=
_
_
_
r
C1

()
if p p
CS
if p < p
CS
_
_
_
where:
r
C1

()
=
_
_
2 0
0 2
_
M
R 1
_
2
_
_
r
C2

()
=
_
2 0
0 0
_
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.3.1:2
Preface Numer. Impl. Constitutive Equation
Flow vector derivatives
r
q
:
r
C

()
=
_

_
r
C1

p
v
=
r
C1
p
C
p
C

p
v
if p p
CS
r
C2

p
v
=
r
C2
p
C
p
C

p
v
if p < p
CS
_

_
where:
r
C1
p
C
=
_

_
0
r
C1
p
R
dR
dp
C
+
r
C1
p
p
CS
_
p
CS
p
C
+
p
CS
R
dR
dp
C
_
_

_
r
C2
p
C
=
_
0
0
_
p
C

p
v
=
1
_

1 +e
0
1
p
C

1
K
_
r
C1
p
R
= 2
M
2
(R 1)
3
_
2p 2p
CS
_
(lower index p is for pcomponent)
r
C1
p
p
CS
= 2
M
2
(R 1)
2
Plasticity condition derivative (over hardening parameters
_
F
q
_
)
F
C
q
=
_

_
F
C1
q
if p p
CS
F
C2
q
if p < p
CS
_

_
where:
dF
C1
dp
C
=
F
C1
p
C
+
F
C1
R
dR
dp
C
+
F
C1
p
CS
_
p
CS
R
dR
dp
C
+
p
CS
p
C
_
dF
C2
dp
C
=
F
C2
p
C
+
F
C2
R
dR
dp
C
+
F
C2
p
CS
_
p
CS
R
dR
dp
C
+
p
CS
p
C
_
p
CS
p
C
=
1
R
p
CS
R
=
p
C
+p
T
R
2
dR
dp
C
= (R
IN
R
0
)
a
_
a +p
C
p
C0
_
2
F
C1
R
= 2
M
2
(R 1)
3
_
p p
C
__
p +p
C
2p
CS
_
F
C2
R
= 2
M
2
(R 1)
3
_
p
CS
p
C
__
p
CS
+p
C
2p
CS
_
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.3.1:3
Preface Numer. Impl. Constitutive Equation
Window 4-53
Window 4-54: Iterative procedure to compute cap hardening parameter p
c
Integration of the hardening equation (Window 3-11) yields
= Aln
_
p
c
n+1
p
c
n
_
B
_
p
c
n+1
p
c
n
_

p
v
n+1
= 0
with
A =

1 +e
0
, B =
1
K
Let

_
p
c
n+1
+ p
c
n+1
_

=
_
p
c
n+1
_
+
d
dp
c
n+1
p
c
n+1
= 0
then,
p
c
n+1
=

_
p
c
n+1
_
d
dp
c
n+1
=

_
p
c
n+1
_
_
A
p
c
n+1
B
_
Iterative procedure:
Initialize, p
(i=0)
c
n+1
= p
c
n
i = i + 1
=
Aln p
(i)
c
n+1
p
c
n
B
_
p
(i)
c
n+1
p
c
n
_

p
v
n+1
= 0
p
(i+1)
c
=

_
A
p
(i)
c
n+1
B
_
p
(i+1)
c
n+1
= p
(i)
c
n+1
+ p
(i+1)
c
Iterate until
p
(i+1)
c
p
(i+1)
c
n+1
>TOL
Window 4-54
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.3.1:4
Preface Numer. Impl. Constitutive Equation
4.7.4 MOHR-COULOMB (SMOOTH) PLASTICITY STRESS POINT
ALGORITHM
The stress return algorithm is governed by general Multisurface Plasticity Closest Point Pro-
jection (MSPCPP) described in Appendix 4.8.3. If the trial stress state is inside the vertex
cone the stress point returns directly to the appex as described in 3.3.7.
Both MSPCPP and SSPCPP algorithms require evaluation of the following values for a given
stress point
yield function value F ()
yield function gradients

F
ow potential gradients

Q
ow potential hessian
2

Q
These quantities are given in Window 4-55, for the generalized 3parameter criterion including
the MC criterion as a special case.
Window 4-55: Three parameter general plasticity criterion
Evaluation of the yield function, yield and ow function,gradients and hessian of the ow
potential
Yield function
F (, , ) (A
f
)
2
+m
f
[B
f
r
f
(, e) +C
f
] c = 0
=
1

3
I
1
, =
_
2J
2
, cos 3 =
3

3
2
J
3
J

3
2
2
.
An expression for r
f
is given in 3.3.5.
u = cos
Using the substitution u = cos and a trigonometric identity the expression for cos 3 can
be written in the form:
(u, ) = 4u
3
3u
3

3
2
J
3
() (J
2
())

3
2
= 0
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.4:1
Preface Numer. Impl. Constitutive Equation
Yield function gradients

F =
F

+
F

+
F
u
u

=
1

3
I
1

_
J
2
_

=
1

J
2

3
2
J

3
2
2
J
3

3
4
J
3
J

5
2
2
J
2

3 (4u
2
1)
F

= m
f
C
f
F

= 2Af
2
f
+m
f
B
f
r
f
(u, e)
F
u
= m
f
B
f

dr
f
(u, e)
du
After rearranging

F = t
1F
I
1

+t
2F
J
2

+t
3F
J
3

with
t
1F
=
1

3
m
f
C
f
t
2F
= 2A
2
f
+m
f
B
f
_

_
r


dr (u, e)
du
(4u
2
1)
J

5
2
2
3

3
4
J
3
_

_
t
3F
= m
f
B
f

3
6
dr (u, e)
du
(4u
2
1)
J

3
2
2
.
For u tending to 0.5 the term
dr(u,e)
du
(4u
2
1)
cannot be evaluated directly but it posseses a limit :
lim
u0.5
_
_
_
dr (u, e)
du
(4u
2
1)
_
_
_
=
1 e
2
(2e 1)
2
Flow potential gradients

Q = t
1Q
I
1

+t
1Q
J
2

t
1Q
=
mC
q

3
t
2Q
= 2A
2
q
+m
q
B
q
r
q
1

June 25, 2003


Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.4:2
Preface Numer. Impl. Constitutive Equation
Flow potential Hessian

2
Q

_
t
2Q
()
J
2

_
=
mB
q
r
q
2

2
J

3
2
2
_
J
2

__
J
2

_
T
+t
2Q

2
J
2

Window 4-55
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.4:3
Preface Numer. Impl. Constitutive Equation
4.7.5 MULTILAMINATE MODEL INVISCID PLASTIC APPROACH
The stress point level requires the following to be evaluated:
1. current stress state resulting from prescribed previous stress state and strain increment.
2. tangent constitutive matrix D
ep
.
Both are performend by means of multisurface plasticity algorithm described in Appendix 4.8.3.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.5:1
Preface Numer. Impl. Constitutive Equation
4.7.6 MODIFIED-CAM CLAY MODEL STRESS RETURN AL-
GORITHM
Derivation of the formula for secant bulk modulus K
sec
Integration of hardening law
Residuals of the governing set of update equations
Auxiliary derivatives
Consistent tangent elastic operator
Consistent tangent elasto-plastic operator
The stress return algorithm requires setting the governing update equations concerning stress
state, current value of the preconsolidation pressure, current value of the secant bulk modulus
and plastic criterion. To collect all these governing equations the formula for secant bulk
modulus will be derived rst and then the integration of the hardening law will be done in an
analytical manner.
Window 4-56: Derivation of the formula for secant bulk modulus K
sec
By integrating the incremental relation between the mean stress p and the volumetric elastic
strain
e
kk
dp =
1 +e
o

p d
e
kk
and by applying the following boundary condition
for :
e
kk
=
e
kk
N
p = p
N
we can compute the current mean stress p
N+1
for given elastic strain increment
e
kk
via
formula
p = p
N
exp(
1

e
kk
)
where

=

1 +e
o
Remark:All the quantities with lower index N are taken from the last converged step.
To derive the expression for the secant bulk modulus K
sec
we can write the relation between
the increment of the mean stress p and the increment of the elastic volumetric strain
e
kk
,
all measured with respect to the last converged step
p p
N
= K
sec

e
kk
Substitution of the expression for p into the above equation yields the following relation for
the current secant bulk modulus K
sec
K
sec
= p
N
_
1 exp(
1

e
kk
)
_

e
kk
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.6:1
Preface Numer. Impl. Constitutive Equation
which after replacing the elastic volumetric strain
e
kk
by a dierence between the total and
plastic one gives
K
sec
= p
N
_
1 exp
_

kk
+
F
p
___

kk
+
F
p
Window 4-56
Window 4-57: Integration of hardening law
To perform the analytical integration of the hardening law let us rewrite it in the form
dp
c
p
c
=
1

(d
p
kk
)
where

=

1 +e
o
The integration of the above dierential equation with assumed boundary condition
for :
p
kk
=
p
kk
N
p
c
= p
c
N
yields the following solution
p
c
= p
c
N
exp
_
1

(
p
kk
)
_
It can be reformulated introducing the increment of plastic multiplier and the appropriate
ow vector components to replace the increment of volumetric plastic strain which nally
gives
p
c
= p
c
N
exp
_
1

F
p
__
Window 4-57
Window 4-58: Residuals of the governing set of update equations
The stress return algorithm formulated in the full stress space is governed by the following
set of 4 residuals of the update equations
r

=
N+1
s
N
2GR
1
_
dev(
N+1
)
N+1
_
F
q
q
J
2
J
2

+
F
r
r
sin(3)
sin 3

__
+
1p
N
exp
_

kk
N+1
+
N+1
F
p
__
r
p
c
= p
c
N+1
p
c
N
exp
_
1

N+1
F
p
_
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.6:2
Preface Numer. Impl. Constitutive Equation
r
K
= K
sec
N+1

p
N
_
1 exp
_

kk
N+1
+
N+1
F
p
_
__

kk
N+1
+
N+1
F
p
r

= q
2
+M
2
C
r
2
() p (p p
c
)
where
R =
_

_
1 0 0 0 0 0
0 1 0 0 0 0
0 0 2 0 0 0
0 0 0 1 0 0
0 0 0 0 2 0
0 0 0 0 0 2
_

_
M = I
1
3
1 1
T
and 1 represents Kroneckers symbol.
The generalized vector of residuals r
N+1
takes the form
r
N+1
=
_
r

r
p
c
r
K
r

_
T
= 0
and the vector of primary variables v
N+1
is dened as
v
N+1
=
_

N+1
p
c
N+1
K
sec
N+1

N+1
_
T
To form a Newtons scheme we need to linearize r
N+1
which yields
r
(i)
N+1
= r
(i1)
N+1
+
r
(i)
N+1
v
(i)
N+1
v
(i)
= 0
Index i is introduced here for local iterations performed at the material point level.
The standard update procedure for the vector v
(i)
N+1
is such that
v
(i)
N+1
= v
(i1)
N+1
+ v
(i)
The iterations are continued until certain convergence criteria are fullled, and it should be
emphasized here that each residual term has its own reference value for objective convergence
assessment.
Window 4-58
Window 4-59: Auxiliary derivatives
r
v
=
_

_
r

p
c
r

K
r

r
p
c

r
p
c
p
c
r
p
c
K
r
p
c

r
K

r
K
p
c
r
K
K
r
K

p
c
r

K
r

_
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.6:3
Preface Numer. Impl. Constitutive Equation
r

= I+2G R
1

N+1

_
J
2

__
q
J
2
q
J
2

2
F
qq
+
F
q

2
q
J
2
J
2
__
J
2

_
T
+
F
q
q
J
2

2
J
2

+
sin(3)

_

2
F
sin(3)p
p
I
1
I
1

+

2
F
sin(3) sin(3)
sin(3)

_
T
F
sin(3)

2
sin(3)

+
1p
N
exp(
1

(
kk
N+1
+
N+1
F
p
) (

N+1

)
_

2
F
pp
p
I
1
I
1

+

2
F
p sin(3)
sin(3)

_
T
r

p
c
= 2G R
1

N+1
_

2
F
sin(3) p
c
sin(3)

_
+
1p
N
exp(
1

(
kk
N+1
+
N+1
F
p
) (

N+1

)

2
F
p p
c
r

K
= 2
G
K
R
1
_
dev(
N+1
)
N+1
_
F
q
q
J
2
J
2

+
F
sin(3)
sin 3

__
r

= 2G
sec
R
1
_
F
q
q
J
2
J
2

+
F
sin(3)
sin 3

_
+
1p
N
exp(
1

(
kk
N+1
+
N+1
F
p
) (
1

)
F
p
r
p
c

= p
c
N
exp
_
1

N+1
F
p
_

N+1

2
F
p p
p
I
1
I
1

r
p
c
p
c
= 1 p
c
N
exp
_
1

N+1
F
p
_

N+1

2
F
p p
c
r
p
c
K
= 0
r
p
c

= p
c
N
exp
_
1

N+1
F
p
_
1

F
p
r
K

_
F
p
_ = p
N
exp
_

kk
N+1
+
N+1
F
p
__
_

N+1

kk
N+1
+
N+1
F
p
_

_
1

+
1

kk
N+1
+
N+1
F
p
_
+
p
N

N+1
_

kk
N+1
+
N+1
F
p
_
2
lim
(

kk
N+1
+
N+1
F
p
)
0
_
_
r
K

_
F
p
_
_
_
=
1
2
p
N

N+1
(

)
2
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.6:4
Preface Numer. Impl. Constitutive Equation
r
K

=
r
K

_
F
p
_
_

2
F
p p
p
I
1
I
1

+

2
F
p sin(3)
sin(3)

_
r
K
K
= 1
r
K
p
c
=
r
K

_
F
p
_

2
F
p p
c
r
K

= p
N
exp
_

kk
N+1
+
N+1
F
p
__
F
p
_
1

kk
N+1
+
N+1
F
p
_

_
1

+
1

kk
N+1
+
N+1
F
p
_
+
p
N
_

kk
N+1
+
N+1
F
p
_
2
F
p
lim
(

kk
N+1
+
N+1
F
p
)
0
_
r
K

_
=
1
2
p
N
(

)
2
F
p
r

=
F
p
p
I
1
I
1

+
F
q
q
J
1
J
2

+
F
sin(3)
sin(3)

p
c
= M
2
c
r
2
() p
r

K
= 0
r

= 0
F
r
= 2 M
2
c
r() p (p p
c
)
F
p
= M
2
c
r
2
() (2p p
c
)
F
q
= 2q
F
sin(3)
=
F
r
r
sin(3)
r
sin(3)
=
r n
(1 sin(3))
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.6:5
Preface Numer. Impl. Constitutive Equation

2
r
sin(3) sin(3)
=
r n
2
(n 1)
(1 sin(3))
2

2
F
qq
= 2

2
F
qp
= 0

2
F
q sin(3)
= 0

2
F
pp
= 2 M
2
c
r
2

2
F
p sin(3)
=

2
F
p r
r
sin(3)
= 2 M
2
c
r (2p p
c
)
r
sin(3)

2
F
sin(3) sin(3)
=

2
F
r r
r
sin(3)
r
sin(3)
+
F
r

2
r
sin(3) sin(3)

2
F
sin(3) p
c
=

2
F
r p
c
r
sin(3)
Window 4-59
Window 4-60: Consistent tangent elastic operator
For stress paths inside the yield surface the stress state at step N + 1 is dened as

N+1
=
N
+ 2G
sec
MR
1

N+1
1p
N
exp(

kk

)
Hence the consistent tangent matrix is dened as follows
D
ATS-E
=
d
N+1
d
= 2G
sec
MR
1
+
p
N

exp(

kk

) 11
T
+
2 MR
1

N+1
G
sec
K
sec
K
sec

kk
_

kk

_
T
where
K
sec

kk
=
p
N+1

kk

p
N
p
N+1
(
kk
)
2
lim

kk
0
K
sec

kk
=
1
2
p
N
(

)
2
G
sec
K
sec
=
3
2
1 2
1 +
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.6:6
Preface Numer. Impl. Constitutive Equation
Remark:The elastic consistent tangent matrix is nonsymmetric
Window 4-60
Window 4-61: Consistent tangent elasto-plastic operator
The elasto-plastic tangent operator, which is consistent with the update equations is dened
as follows
D
ATS
=
d
d
=

+

v
v

Due to the fact that the algorithm is driven by a given strain increment
N+1
, which remains
unchanged during local stress return iterations, we can write
dr
d
=
r

+
r
v
v

= 0
from which the term
v

can be computed as
v

=
_
r
v
_
1
r

Substituting the above expression into the general formula for the consistent tangent operator
D
ATS
yields
D
ATS
=



v
_
r
v
_
1
r

r
N+1

=
_

_
r

r
p
c

r
K

_
r

=
2 G
sec
R
1
_
I
1
3
11
T
_

p
N

exp
_

kk
N+1
+
N+1
F
p
__
11
T
r
p
c

= 0
r
K

=
_

_
p
N
_

kk
N+1
+
N+1
F
p
_
2

p
N
exp
_

(
kk
N+1
+
N+1
F
p
)

__
1

+
1

kk
N+1
+
N+1
F
p
_

kk
N+1
+
N+1
F
p
_

_
1
T
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.6:7
Preface Numer. Impl. Constitutive Equation
r
K

=
1
2
p
N

2
1
T
for
kk
N+1
+
F
p
0
r

= 0

N+1
v
=
_

N+1

,

N+1
p
c
,

N+1
K
sec

N+1

N+1

= I
r

N+1
p
c
=
r

p
c

N+1
K
sec
=
r

N+1

=
r


Window 4-61
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.6:8
Preface Numer. Impl. Constitutive Equation
4.7.7 AGING CONCRETE MODEL - IMPLEMENTATION SCHEME
Window 4-62: Aging concrete model-stress point algorithm
For given:
time: t
n+1
,
time increment t
n+1
,
strain increment
n+1
,
maturity (from previous step) M
n
,
stress in each unit
(k)
n
, k = 1, .., NrOfMaxwellUnits
Compute:
imposed strain increment
0n+1
,
maturity(for current step) M
n+1
,
stress in each unit
(k)
n+1
, k = 1, ..NrOfMaxwellUnits,
total stress
n+1
=

(k)
n+1
Young module in each unit E
k
generalized Young module E

Note: i - iteration counter, k - Maxwell unit counter.


Window 4-62
Algorithm details
Evaluate maturity and imposed strain increment
IF (i = 1 ) THEN
IF ( maturity has to be integrated on the basis of heat transfer solution ) THEN
Compute maturity M
n+1
at current time t
n+1
by formula (see also heat transfer for-
mulation/implementation):
M
n+1
= M
n+1
+
N

l=1
e
Q
R

1
T
f

1
T

n+1

t
l
Note: because in mechanical analysis much larger time steps are used than these in
heat transfer or humidity diusion thus integrate the maturity with accuracy as good
as obtained from heat transfer analysis.To nd N and t
l
time subincrements and
then to integrate M
n+1
use time grid from heat transfer analysis and corresponding
set of temperatures. The integrated expression is evaluated for each subincrement
with midpoint rule.
Compute: t

= M
n+1
M
n
ELSE
t

= t
n+1
t
n
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.7:1
Preface Numer. Impl. Constitutive Equation
M
n+1
= M
n
+ t

END IF
Compute the increment of the initial mean strain on the basis of heat transfer and humidity
diusion solutions in time period [t
n
, t
n+1
].
mean
0,n+1
=
T
T
n+1
+
W
W
n+1
, where:
-
T
thermal dilatancy,
W
-hygral dilatancy.
END IF
Compute the full vector of the initial strain increment
0,n+1
:
plane strain:
0,n+1
= (1 +)
mean
0,n+1
, (1 +)
mean
0,n+1
, 0, 0
T
axisymmetry:
0,n+1
=
mean
0,n+1
,
mean
0,n+1
, 0,
mean
0,n+1

T
3D:
0,n+1
=
mean
0,n+1
,
mean
0,n+1
, 0,
mean
0,n+1
, 0, 0
T
Compute the eective strain increment
eff
n+1
:

eff
n+1
=
n+1

0,n+1
.
For the averaged maturity M

= 0.5(M
n+1
+M
n
) interpolate the value of Young module
E(M

) with the given set of material data,


Set matrix D
0
, see the section: CREEP
Generalized Young module and stress state
Initialize: E

= 0,
n+1
= 0,
LOOP k = 1, NrOfMaxwellUnits,
(here :
k
is the retardation time for unit k),
Set Young module for unit k: E
k
= W
k
(M

)E(M

) using linear interpolation for given


set of coecients W
k
,
Compute stress state for each unit:

(k)
n+1
=

k
t

_
1 e

k
_
E
k
D
0

eff
n+1
+
(k)
n
e

k
Accumulate the total stress state:
n+1
=
n+1
+
(k)
n+1
Accumulate the generalized Young module:
E

= E

+E
k
_
1 e

k
_
END
Set the current stiness constitutive matrix for global stiness matrix evaluation purpose:
D =E

D
0
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.7.7:2
Preface Numerical Implementation
4.8 APPENDICES
SHAPE FUNCTIONS AND REFERENCE ELEMENTS FOR 2D/3D
NUMERICAL INTEGRATION DATA
MULTISURFACE PLASTICITY CLOSEST POINT PROJECT ALGORITHM
SINGLE SURFACE PLASTICITY CLOSEST POINT PROJECTION ALGORITHM
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.8:1
Preface Numerical Implementation Appendices
4.8.1 SHAPE FUNCTION DEFINITION AND REFERENCE ELE-
MENTS FOR 2/3 D
Reference element Shape functions
N
i
=
1
4
(1 +
i
) (1 +
i
)
_

_
N
1
=
1
4
(1 ) (1 )
N
2
=
1
4
(1 +) (1 )
N
3
=
1
4
(1 +) (1 +)
N
4
=
1
4
(1 ) (1 +)
N
1
= 1
N
2
=
N
3
= .
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.8.1:1
Preface Numerical Implementation Appendices
4.8.2 NUMERICAL INTEGRATION DATA FOR DIFFERENT EL-
EMENTS IN 1/2/3D

3
3
= 0.5773502691896
Integration points
Element shape number Positions Weigthing
N
gauss
i :
i

i
factors W
i
1 1 0.0 2.0
2
1 :
2 :

1.0
1.0
1 1: 0.33333 0.33333 0.5
in 2
1 :
2 :
3 :
0.16666
0.16666
0.66666
0.16666
0.66666
0.16666
0.16666
0.16666
0.16666
1 1: 0.0 0.0 4.0
22=4
1 :
2 :
3 :
4 :

1.0
1.0
1.0
1.0
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.8.2:1
Preface Numerical Implementation Appendices
4.8.3 MULTISURFACE PLASTICITY CLOSEST POINT PROJEC-
TION ALGORITHM
A generalized, unconditionally stable, algorithm designed for the integration of constitutive
equations for class of multisurface elastoplastic models is presented here. This subject has
been worked out rst by Simo & Hughes but procedures described therein were not as general
as needed. Three crucial problems e.a.: selection of the reduced set of active mechanisms
solving the plastic corrector problem in case when the number of active mechanisms is greater
than the stress space dimension , activation of initially nonactive plastic mechanisms due to
eects of hardening/softening and nally denition of the general form of a consistent tangent
matrix for multimechanism models with generally nonassociated ow rule, and including hard-
ening/softening, were introduced later by Szarli nski and Truty. Slightly modied algorithm
including an additional substeping scheme is described here. The complete set of information
on Multisurface Closest Point Projection Stress Return Algorithm ( MCPPSS) can be found
in the following Windows:
Notation
Basic set of equations
Consistent tangent operator D
ep
Scheme of the generalized stress return algorithm
Window 4-63: Notation
M number of all plastic mechanisms,
Jact actual set of active plastic mechanisms,
mechanism index,
a
= t plastic multiplier value,

r (, q) ow vector,

h(, q) column matrix of hardening/softening functions,

q
column matrix of plastic (hardening/softening)
parameters

p
column matrix of total plastic strains,

F
the value of plasticity condition for given stress
state and plastic parameters
D elastic stiness matrix
C elastic compliance matrix
k index of previous closest point projection algorithm iteration
k + 1 index of actual iteration
n index of last conguration of equilibrium
n + 1 index of actual conguration of equilibrium
Window 4-63
Window 4-64: Set of basic equations
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.8.3:1
Preface Numerical Implementation Appendices
The integration of constitutive equations for any elastoplastic model consists of two stages
e.g. elastic predictor and plastic corrector one. Assuming the fully implicit integration scheme
for plastic strains and hardening/softening parameters this second stage is expressed by the
following set of equations which is the basis of the closest point projection algorithm (in the
following windows this will be called the basic set of equations):

p(k+1)
n+1
=
p
n
+

=1,M

(k+1)
n+1

r
(k+1)
n+1
q
(k+1)
n+1
= q
n
+

=1,M

(k+1)
n+1

h
(k+1)
n+1

F
(k+1)
n+1
=

F
(k)
n+1
+

F
(k)
n+1
= 0
Jact

(k+1)
n+1
= D
p(k)
n+1
_

_
In most cases this system is nonlinear and NewtonRaphson procedure is needed for its
solution. To apply this procedure the consistent linerization of the above set of equations
has to be done. Denoting by R

and R
q
the residuals of the rst two equations which are
dened by the following expressions:
R

=
p
n

p(k)
n+1
+

=1,M

(k)
n+1

r
(k)
n+1
R
q
= q
n
q
(k)
n+1
+

=1,M

(k)
n+1

h
(k)
n+1
the linearized system of rst two equations of the basic set using expressions for R

and R
q
and 4th equation from basic set is as follows (summation for
Jact
):
_
A
(k)
n+1
_
1
=
_
D 0
0 I
_
_

p(k)
n+1
q
(k)
n+1
_
=
_

_
R

=1,M

(k)
n+1

r
(k)
n+1
R
q
+

=1,M

(k)
n+1

h
(k)
n+1
_

_
where:
A
(k)
n+1
=
_

_
D
1
+

=1,M


r
(k)

=1,M


r
(k)
q

=1,M


h
(k)

=1,M


h
(k)
q
I
_

_
1
.
In order to nd increments of plastic multipliers the third equation from basic set written for
each mechanism

F : Jact has to be solved. After linearization it takes the form:

F
(k+1)
n+1
=

F
(k+1)
n+1
+

F
(k)
n+1
=

F
(k+1)
n+1
+


F
(k)
n+1


(k)
n+1
+


F
(k)
n+1
q
q
(k)
n+1
(1)
It can be written in matrix form using 4-th equation from basic set which leads nally to the
following:

F
(k)
n+1
=
_


F
(k)
n+1


F
(k)
n+1
q
_
(2)
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.8.3:2
Preface Numerical Implementation Appendices
In the above equation vector of unknowns which consists of plastic strain and plastic variables
increments can be eliminated via expression derived for
_
A
(k)
n+1
_
1
and nally after some
matrix manipulations this expression can be rewritten in the form:
U
(k)
n+1
A
(k)
n+1
P
(k)
n+1

(k)
= f
(k)
n+1
U
(k)
n+1
A
(k)
n+1
R
(k)
n+1
where:
R
(k)
n+1
=
_
R

R
q
_
P
(k)
n+1
=
_
r

1
. . . r

N
h

1
. . . h

n
_
U
(k)
n+1
=
_

_
_
f

_
T
_
f

1
q
_
T
.
.
.
.
.
.
_
f
N

_
T
_
f

N
q
_
T
_

_
Window 4-64
Window 4-65: Consistent tangent operator D
ep
The general denition of the tangent matrix is as follows:
D
ep
=
d
d
.
This denition after introducing of the applied integration scheme leads to the so called
consistent tangent matrix. The starting point for its derivation is the basic set of equations
given in Window 4-64 which expresses the fully implicit scheme applied for the integration
of plastic strains and plastic variables. All next steps will lead to derivation of the relation
between dd and encapsulating from it the algorithmic constitutive operator. To do that
lets dierentiate this basic set of equations which yields:
d
p
=

=1,M

d+

=1,M


r
q
dq+

=1,M

r
dq =

=1,M

d+

=1,M


h
q
dq+

=1,M

h
d

F =


F

d+


F
q
dq = 0;
Jact
d = D( d d
p
)
Considering only rst two equations from the above set and eliminating the increment of
plastic strains using fourth equation we can write the following matrix equation:
_

_
C+

=1,M

=1,M


r
q

=1,M

=1,M


h
q
I
_

_
. .
A
1
_
d
dq
_
=
_
d
0
_

_
_
_
_
_
_

=1,M
d

=1,M
d

h
_
_
_
_
_
_
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.8.3:3
Preface Numerical Implementation Appendices
Let us evaluate the stress increment considering once again the fourth equation from linearized
basic set
d = D( d d
p
) = D
_

=1,M

=1,M


r
q
dq

=1,M
d


r
_
= D
_
_
_
_
_
d
_

=1,M

=1,M


r
q
_
. .
a
_
d
dq
_

=1,M
d


r
_
_
_
_
_
.
By elimination of the plastic strain dierential from the above equation we can rewrite as:
d = D
_
_
_
_
d aA
_
d
0
_
+aA
_

=1,M
d

=1,M
d


h
_

_
_
_
_
_
.
Let us introduce some auxiliary matrices to simplify further derivations:
matrix r which consists of column stored ow vectors for each active mechanism r =
[r
1
, . . . , r
N
]
vector of plastic multiplier increments: d =
_

_
d
1
.
.
.
d
N
_

_
matrix K
[(NSTRE+NHARD)NSTRE]
: K =
_
I
0
_
.
With these defnitions one can reform the equation for d:
d = D
_
d aA
_
d
0
_
+ (aAPr) d
_
but still it is necessary to dene vector d. It can be done writing appropriate consistency
condition for each active mechanism. Thus for all (NJACT) active mechanisms one gets the
following system
_


F
q
.
.
.
.
.
.

N
F

N
F
q
_

_
_
d
dq
_
=
_

_
0
.
.
.
0
_

_
.
Eliminating vector of unknowns the expression for increments of plastic multipliers will take
nally the form:
d = (UAP)
1
UA
_
d
0
_
.
Introducing the above formula into the equation for d, after some matrix manipulations,
we get the most general form of the multimechanism consistent elasto-plastic matrix:
d = D
_
I aAK+ (aAPr) (UAP)
1
UAK

. .
D
epconsistent
d.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.8.3:4
Preface Numerical Implementation Appendices
Window 4-65
Window 4-66: Multisurface stress return algorithm
Given:

n
: stress at previous converged step

n+1
: the imposed strain increment
q
n
: hardening parameters at previous converged step

p
n
: plastic strain at step n
nd:

n+1
= D
_

n+1

p
n+1
_
satisfying:

0
F

(
n+1
, q
n+1
) 0

(
n+1
, q
n+1
) = 0
Jact
N.B.:
all symbols with tilde are introduced for temporary copies of some objects,
all symbols with overhead bar - are introduced for accumulated values of plastic strain
increment, hardening parameters etc...
1. Initialization: q
(0)
n+1
= q
(0)
n
,
(0)
n+1
=
(0)
n
STEP TOL=1.0e-4
DSTEP MIN=0.05
STEP SUM=0.0 (substep accumulation)
DSTEP=1.0 (initial substep)
2. Predictor phase:

tr
=
n
+
tr
,
tr
= D
n+1

(0)
n+1
=
n
, =
n
3. Next substep:

(k)
n+1
=
(k1)
n+1
+DSTEP
tr
IF STEP SUM=0 T THEN
q
(0)
n+1
= q
(0)
n
, q
(0)
n+1
= 0
q = q
n
, q = 0

p(0)
n+1
= 0,
p
= 0
Jact () = TRUE

Jact () = TRUE

(0)
= 0,
TRIAL
= 0, = 0
END IF
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.8.3:5
Preface Numerical Implementation Appendices
4. Check values of plasticity conditions and singular cases (like return to the appex for
DP)
F

(k)
n+1
, q
(k)
n+1
_

IF singular case is detected then EXIT and RUN separate procedure


5. Set initial activity of surfaces: Jact() =
_
F

(0)
n+1
, q
(0)
n+1
_
> 0
_
IF nr of active surfaces =0 THEN
STEP SUM = STEP SUM + DSTEP
DSTEP = min(DSTEP, 1 - STEP SUM)
IF [DSTEP[>STEP TOL THEN
=
(0)
n+1
GOTO Next Step
ELSE
D
ep
= D
END IF
END IF
6. Iterate: k = k + 1 (iteration counter)
7. Evaluate (for each mechanism from set Jact)
plasticity conditions gradients :
F

ow vectors : r

ow vectors gradients :
r

;
r

q
hardening functions : h

(, q)
hardening functions gradients :
h

;
h

q
8. Evaluate residuals R
(k)
n+1

Jact
(see Window 4-64)
9. Check convergence via procedure given in Window 4-67
10. Evaluate matrices A
(k)
n+1
, P
(k)
n+1
, U
(k)
n+1
(summation)

(see Window 4-64) and com-


pute increments of plastic multipliers:
U
(k)
n+1
A
(k)
n+1
P
(k)
n+1

(k)
= F
(k)
n+1
U
(k)
n+1
A
(k)
n+1
R
(k)
n+1
IF matrix U
(k)
n+1
A
(k)
n+1
P
(k)
n+1
is ingular or negative denite THEN
perturb the matrix U
(k)
n+1
A
(k)
n+1
P
(k)
n+1
by its diagonal to remove eect of singularity
or negative denitness
reevaluate once again increments of plastic multipliers using perturbed matrix
END IF
11. Update total plastic multipliers
Jact
IF
(k1)
+
(k)
< 0 THEN

(k)
= 0
Jact() =FALSE
after checking of each mechanism GOTO Next Iteration
END IF
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.8.3:6
Preface Numerical Implementation Appendices
12. Evaluate the increment of plastic strains and hardening parameters
_

p(k)
n+1
q
p(k)
n+1
_
=
_
D 0
0 I
_
_
A
(k)
n+1
_
_

_
R

=1,M

(k)
r
(k)
n+1
R
q
+

=1,M

(k)
h
(k)
n+1
_

_
13. Update plastic state variables and stresses

p(k)
n+1
=
p(k1)
n+1
+
p(k)
n+1
, q
p(k)
n+1
= q
p(k1)
n+1
+ q
p(k)
n+1

(k)
n+1
=STEP SUM
tr
n+1
D
p(k)
n+1

Jact

(k)
=
(k1)
+
(k)
14. Final check for iteration
Check plasticity conditions: F

(k)
n+1
, q
(k)
n+1
_

Jact
IF
/ Jact
: f

(k)
n+1
, q
(k)
n+1
_
THEN
Jact()=FALSE
END IF
15. GO TO next iteration
Window 4-66
Window 4-67: Convergence test
1. IF [R
(k)
n+1
[ < TOL1 AND F

(..) < TOL2


Jact
IF k >MAXITER THEN
IF DSTEP >DSTEP MIN
DSTEP=DSTEP/2

(0)
= ,
p(k)
n+1
=
p
,
(0)
n+1
=
Jact () =

Jact ()

q
(0)
n+1
= q, q
(0)
n+1
= q
GO TO next substep
ENDIF
END IF
STEP SUM = STEP SUM + DSTEP
DSTEP=min(DSTEP, 1-STEP SUM)
IF [DSTEP[>STEP TOL THEN
GOTO Next Substep
ELSE (algorithm failure)

n+1
=
n
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.8.3:7
Preface Numerical Implementation Appendices
D
ep
= D
EXIT
END IF
STEP SUM = STEP SUM + DSTEP
DSTEP=2 DSTEP
DSTEP=min(DSTEP, 1-STEP SUM)
=
(k)

p
=
p(0)
n+1
=
(0)
n+1

Jact () =Jact()

q = q
(k)
n+1
, q = q
(k)
n+1
IF [DSTEP[>STEP TOL GO TO next substep
Evaluate the consistent tangent stiness matrix D
ep
(see Window 4-65)
EXIT
ENDIF
Window 4-67
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.8.3:8
Preface Numerical Implementation Appendices
4.8.4 SINGLE SURFACE PLASTICITY CLOSEST POINT PRO-
JECTION ALGORITHM
The case of a single surface plasticity with/without hardening is handled via multisurface
plasticity theory and algorithms assuming M=1 (nr of plastic surfaces), see Appendix 4.8.3
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM4.8.4:1
Preface
Chapter 5
STRUCTURES
TRUSSES
BEAMS
MEMBRANES
APPENDICES
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.0.4:2
Preface Structures
5.1 TRUSSES
TRUSS ELEMENT
RING ELEMENT
ANCHORING OF TRUSS AND RING ELEMENTS
PRE-STRESSING OF TRUSS AND RING ELEMENTS
UNI-AXIAL ELASTO-PLASTIC MODEL
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.1:1
Preface Structures Trusses
5.1.1 TRUSS ELEMENT
GENERAL IDEA OF TRUSS ELEMENT
GEOMETRY AND DOF OF TRUSS ELEMENTS
INTERPOLATION OF THE DISPLACEMENTS AND STRAINS
WEAK FORMULATION OF THE EQUILIBRIUM
STIFFNESS MATRIX AND ELEMENT FORCE VECTOR
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.1.1:1
Preface Structures Trusses
5.1.1.1 GENERAL IDEA OF TRUSS ELEMENT
Elements are designed to model the presence of dierent kind of anchors, reinforcements ,
or separate bar (3D case). Note the dierence between truss/anchor and 2D membrane
elements , see point 5.3 .
Window 5-1: Truss elements in Plane Strain
Plane Strain case
Window 5-1
Window 5-2: Truss elements in axisymmetry
Axisymmetric case
Window 5-2
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.1.1.1:1
Preface Structures Trusses
5.1.1.2 GEOMETRY AND DOF OF TRUSS ELEMENTS
Window 5-3: Truss element geometry
The reference element, shape functions, construction of isoparametric mapping, DOF
setting for 2/3 node truss element
Window 5-3
Geometry of the truss element is identied by the geometry of the centroid line:
o
x() =

i=1,2
N
i
()x
i
Local cross-sectional coordinate system x
L
, y
L
, z
L
directions are set as:
e
x
L
=
o
x,

|
o
x,

|
; e
z
L
=
_
0, 0, 1
_
T
; e
y
L
= e
z
L
e
x
L
,
which for 2D cases correspond to the following rules:
1. x
L
axis is tangent to element with the direction pointing from 1st to 2nd node of the
element,
2. z
L
is perpendicular to global XY plane (meridian plane for Axisymmetric) and equal to
global Z axis,
3. y
L
is set from x
L
, z
L
by righthand screwdriver rule .
For an arbitrary point of an element identied by its reference coordinate and local cross
sectional positions ry
L
, (z
L
), the mapping to the global coordinate may be put as:
x(, y
L
, z
L
) =
o
x() +y
L
e
y
L
+z
L
e
z
L
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.1.1.2:1
Preface Structures Trusses
The Jacobi matrix of the above mapping may be put as:
J =
_
o
x,

; e
y
L
; e
z
L

.
The localglobal x
L
x
G
transformation matrix may be put as:
T =
_
e
x
L
; e
y
L
; e
z
L

, T
T
=
_
_
e
x
L
)
e
y
L
)
e
z
L
)
_
_
For for any vector:
v
G
= Tv
L
, v
L
= T
T
v
G
Evaluation of strains requires the derivatives of displacement components versus local x
L
axis
:
v,
x
L
= v,

x
L
= v,

D with D = J
1
)
1
e
x
L
For 2D cases a relation between local and global DOF as well as forces may be established
in a form :
v
G
= Tv
L
, v
L
= T
T
v
G
, with T =
_
c s
s c
_
,
where:
e
x
L
=
_
c, s
_
T
, e
y
L
=
_
s, c
_
T
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.1.1.2:2
Preface Structures Trusses
5.1.1.3 INTERPOLATION OF THE DISPLACEMENTS AND STRAINS
For the point with given reference coordinate translation displacement components u
G
(referred to global coordinate system x
G
) are interpolated from its values at nodal points
u
G
() =

i=1,2
N
i
()u
G
i
The only strain component taken into account while evaluating element stress is strain along
the element, and, as no bending is taken into account, this strain is assumed to remain
constant in the whole cross-section of the element:

xx
L
(x
L
, y
L
) =
u
L
x
L
= e
x
L
T
u
G
x
L
,
The formulae relating the only strains component
xx
L
any point within the element, with its
DOF vector u may put in general for:

xx
L
() = B

x
L
()u =
Nen

i=1
B
i
x
L
()u
i
,
For each analysis type in takes form:
Plane Strain, Axisymmetry:
u
i
=
_
u
i
v
i

T
B
i
x
L
=
_
e
x
L
x
DN
i
,

e
x
L
y
DN
i
,

Generalized Plane Strain:


u
i
=
_
u
i
, v
i
, w
i

T
,
B
i
x
L
=
_
c
2
e
x
L
x
DN
i
,

c
2
e
x
L
y
DN
i
,

scDN
i
,

,
s = sin(), c = cos()
angle of inclination from xy plane (default = 0).
3D:
u
i
=
_
u
i
, v
i
, w
i

T
,
B
i
x
L
=
_
e
x
L
x
DN
i
,

e
x
L
y
DN
i
,

e
x
L
z
DN
i
,

.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.1.1.3:1
Preface Structures Trusses
5.1.1.4 WEAK FORMULATION OF THE EQUILIBRIUM
The virtual work principle expressing equilibrium of a system may be put in the general form:
nd such that:
for any , u
_
V

T
dV
_
V
u
T
pdV = 0
The contribution of the truss elements in the above may be easily derived, with integrals taken
along the element length. A is an area attributed to the assumed computational domain i.e.
to the unit slice for the Plane strain, Generalized Plane Strain , to the whole circumference
(independently from current radius) for the Axisymmetric case or to one distinct bar for the
3D case.
for any , u
_
L

xx
L

xx
L
AdL
_
L
u
T
pdL = 0
For the description of 1D uniaxial elastoplastic material model used for truss elements see
the Window 5-13.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.1.1.4:1
Preface Structures Trusses
5.1.1.5 STIFFNESS MATRIX AND ELEMENT FORCE VECTOR
Stiness matrix K
e
and force vector f
e
of the truss element are derived in a standard way
from the weak formulation of a problem. It would require constitutive module D
xxxx
as well
as
xx
L
stress evaluation for given strain increment. This will be performed by 1D uniaxial
elastoplastic model.
Numerical integration technique is used to evaluate integrals over the length of the element.
In the case of a 2node linear element, 1 integration point is used (Ngaus = 1, W
1
= 2.0,

1
= 0.0). Integration over the cross section of the element is hidden in given values of
integral characteristics of the cross section area A.
K
e
=
1
_
1
B

x
L
T
()D
xxxx
B

x
L
()|
o
x()|d =
Ngaus

igaus=1
B

x
L
T
(
igaus
)D
xxxx
B

x
L
(
igaus
)|
o
x()|W
igaus
f
e
=
1
_
1
B

x
L
T
()
xx
()|
o
x()|d =
Ngaus

igaus=1
B

x
L
T
(
igaus
)
xx
(
igaus
)|
o
x(
igaus
)|W
igaus
The element p load is assumed to act along the centroid line leading to load induced forces
evaluated as:
f
p
=
1
_
1
N
T
()p|
o
x()|d =
Ngaus

igaus=1
N
T
(
igaus
)p|
o
x(
igaus
)|W
igaus
where:
N() =
_
N
i
()I
(dim)

, i = 1, Nen
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.1.1.5:1
Preface Structures Trusses
5.1.2 RING ELEMENT
GEOMETRY AND KINEMATICS OF A RING ELEMENT
WEAK FORMULATION OF THE EQUILIBRIUM
STIFFNESS MATRIX AND ELEMENT FORCES
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.1.2:1
Structures Trusses Ring Element
5.1.2.1 GEOMETRY AND KINEMATICS OF A RING ELEMENT
The geometry of the ring element (available exclusively for Axisymmetric analysis type) is
represented by the coordinates of its 1st node and attached area.
Window 5-4: Geometry and DOF of a ring elements
Ring element. Geometry and DOF
Window 5-4
Although only u contribute to element strain both u, v displacement components are kept
as element DOF, to allow force resulting from vertical load to be transmitted to the system
via ring element.
The only strain component taken into account while evaluating element stress is strain in
circumferential direction, constant in the whole crosssection of the element, related only to
the radial component of the displacement and current radius:

zz
=
u
r
The formulae relating
zz
, with elements DOF vector u:

zz
= B

zz
u, u =
_
u, v

T
B

zz
=
_
1
r
, 0
_
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.1.2.1:1
Structures Trusses Ring Element
5.1.2.2 WEAK FORMULATION OF THE EQUILIBRIUM
From the general form of the V.W.P it may be stated as:
for any , u :

zz

zz
(2rA) u
T
p(2r) = 0
For the description of 1D uniaxial elastoplastic material model used for ring element see
the Window 5-13
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.1.2.2:1
Structures Trusses Ring Element
5.1.2.3 STIFFNESS MATRIX AND ELEMENT FORCES
Stiness matrix K
e
and force vector f
e
of the truss element is derived in a standard way
from the weak formulation of a problem. It would require constitutive module D
zzzz
as well
as
zz
stress evaluation for given strain increment. This will be performed by 1D uniaxial
elastoplastic model, see the Window 5-13
K
e
= B
T

zz
D
zzzz
B

zz
2rA
f
e
= B
T

zz

zz
2rA
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.1.2.3:1
Preface Structures Trusses
5.1.3 ANCHORING OF TRUSS AND RING ELEMENTS
For the sake of generality the nodal points of truss elements may be placed in arbitrary
position within the domain occupied by continuum or structural element. Despite modeling
convenience, the option also possesses some physical meaning as then the truss element
force are distributed over surrounding nodes of continuum element and the eect of force
concentration is diminished.
Window 5-5: Anchoring of truss and ring elements
Conditions to be fullled:
displacement compatibility:
u
Truss
= u
Cont
force equivalence (weak form):
for any u :
u
T
Truss
f
Truss
=
Nen

i=1
u
(i)T
Cont
f
(i)
Cont
Anchoring of a truss/ring node within continuum element
Window 5-5
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.1.3:1
Structures Trusses Anchoring ...
5.1.3.1 NUMERICAL REALIZATION OF ANCHORING
Let N
i
are shape functions of the element in which kth node of a truss is anchored, and

are local coordinates of the anchorage point within its reference element. Displacement of a
truss node u
Truss
are to be evaluated from displacement of surrounding element u
Cont
via
compatibility condition :
u
Truss
= u
Cont
(

) =
NenCont

i=1
N
i
(

)u
iCont
= N
T
u
Cont
where:
N = [N
i
(

)I
Ndm
] , i = 1, NenCont
In the case of anchoring within structural (i.e. beam or shell element), shape function matrix
Nmust be understood in more general sense i.e. as a matrix relating translation displacement
at arbitrary point within the element with all its DOF and takes the form given in point 1.2,
1.3.
In turn, forces evaluated at the node of a truss f
Truss
are transferred to the nodes of the
surrounding element as f
Cont
basing on (weakly formulated) equivalency
for any u :
u
T
Truss
f
Truss
=
Nen

i=1
u
(i)T
Cont
f
(i)
Cont
= u
Cont
f
Cont

u
T
Truss
f
Truss
= u
T
Cont
Nf
Cont
= u
Cont
f
Cont
f
Cont
= Nf
Truss
Finally, the stiness of the truss element is assembled on the DOF of surrounding element,
and takes a form as bellow, where K
jk
are jth and kth nodal submatrix of the truss
stiness:

K =
_

_
K
jj
.
.
. K
jk
N
T
s s s
NK
kj
.
.
. NK
kk
N
T
_

_
In the case of the ring element, its stiness assembled on the DOF of surrounding elements
takes form:

K = NK
kk
N
T
.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.1.3.1:1
Preface Structures Trusses
5.1.4 PRESTRESSING OF TRUSS AND RING ELEMENTS
Pre-stress can be applied only to anchor/truss or ring elements. Prestress diers from a
standard initial stress situation by the fact, that the internal force is assumed to be known a
priori. According to the denition of current tangent stiness matrix it follows that:
K =
f(u)
u
= 0
since internal force f(u) is constant. As a consequence, if pre-stress is being applied then
the stiness of the anchor is always set to zero while it is evaluated via standard procedures
if pre- stress is nonactive.
Strains in the anchor / ring element are monitored during the pre-stress as:

i
=
i1
+
i

i
=

prestres
(t
i
)
i1
E
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.1.4:1
Preface Structures
5.2 BEAMS
GEOMETRY OF BEAM ELEMENT
KINEMATICS OF BEAM THEORY
WEAK FORMULATION OF THE EQUILIBRIUM
INTERPOLATION OF THE DISPLACEMENT FIELD
STRAIN REPRESENTATION
STIFFNESS MATRIX AND ELEMENT FORCES
MASTER-CENTROID (OFFSET) TRANSFORMATION
RELAXATION OF INTERNAL DOF
BEAM ELEMENT RESULTS
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.2:1
Preface Structures Beams
5.2.1 GEOMETRY OF BEAM ELEMENT
Beams in 2D analysis types (Plane Strain Axisymmetry, Plane Stress)
Window 5-6: Beams in 2D analysis type modeling situations
In the case of Plane Strain analysis, beam elements can be used for two - modeling situations:
Modeling situations for beam elements in Plane Strain and 2.5D analysis
In the case of Axisymmetric analysis, only continuum shell option can be used:
Beam elements in axisymmetric analysis
Window 5-6
System x
G
, y
G
, z
G
is the global one. Local element system x
L
, y
L
, z
L
is created in the
element in a following way:
1. x
L
- axis is tangent to the elements centroid axis at a given point, with the direction
pointing from 1-st to 2-nd centroid node of the element
2. z
L
- is perpendicular to global XY plane (meridian plane for Axisymmetric case) and
corresponds to global Z axis
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.2.1:1
Preface Structures Beams
2D beam element: nodal DOF and coordinates systems
3. y
L
-is set from x
L
, z
L
by right-hand screw driver rule
e
x
L
=
o
x
,
|
o
x
,
|
, e
z
L
=
_
0, 0, 1
_
T
, e
y
L
= e
z
L
e
x
L
Relation between local and global DOF (degrees of freedom) as well as forces may be estab-
lished in a form :
v
G
= Tv
L
, v
L
= T
T
v
G
, T =
_
_
c s 0
s c 0
0 0 1
_
_
are local coordinate unit vectors.
Local cross sectional axis are these to which the geometry of the crosssection is referred.
In the case of elastic beam model they must strictly correspond to centroidal and principal
crosssectional inertia axis . In the case of layered, nonlinear beam model this demand may
be satised only approximately, as most meaningful eects, related to axial strain and stresses
resulting from axial force and bidirectional bending action are taken into account by cross
sectional numerical integration. Then cross sectional axis setting will be used to input layer
centers positions. Moreover, cross sectional axis setting will be used to refer bidirectional
shear and torsion elastic characteristics.
For local cross sectional positions ry
L
, (z
L
), the mapping to the global coordinate may be
put as:
x(, y
L
, z
L
) =
o
x() +y
L
e
y
L
+z
L
e
z
L
The Jacobi matrix of the above mapping may be put as:
J(, y
L
, z
L
) =
_
o
x
,
+y
L
e
y
L
,

+z
L
e
z
L
,

; e
y
L
; e
z
L

The local-global x
L
x
G
transformation matrix may be put as:
T =
_
e
x
L
; e
y
L
; e
z
L

, T
T
=
_
_
e
x
L
)
e
y
L
)
e
z
L
)
_
_
thus for any vector :
_
v
G
= Tv
L
,
v
L
= T
T
v
G
.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.2.1:2
Preface Structures Beams
Evaluation of strains requires the derivatives of any displacement components versus local xL
axis :
v
,x
L
= v
,

xL
= v
,
D with D = J
1
)
1
e
x
L
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.2.1:3
Preface Structures Beams
5.2.2 KINEMATICS OF BEAM THEORY
The adopted kinematics of a beam is based on Timoshenkos hypothesis, i.e. that originally
plane cross- section perpendicular to beam centroid line remains plane but not necessarily
perpendicular to deformed beam axis . Moreover small displacements & strains are assumed
Window 5-7: Kinematic assumption of beam theory
Kinematic assumption of beam theory
Window 5-7
Displacements at any point r with given local coordinate x
L
, y
L
, z
L
are then given by
displacement
o
u of centroid line and independent rotation of a cross section plane (ber)
as :
u(x) =
o
u(x
L
) +(x
L
) r
For 2D case this leads to:
u
L
=
o
u
L
(x
L
) (x
L
)y
L
v
L
=
o
v
L
(x
L
)
Two formulations of beam element are used:
layered approach (Nonlinear Beam option, analysis type Plane Strain, Axisymmetry, )
composite sections allowed, nonlinear or elastic material models, setting of the centroid of
the crosssection may be done in approximate manner as M N coupling is taken into
account in the model);
integral approach (Elastic Beam option, Plane strain, Axisymmetry, ), uniform section
described by its integral characteristics, elastic material model only, setting of the centroid
of the crosssection must be done in rigorously exact way, as MN coupling is disregarded
by the model)
For the layered approach strains are evaluated at each layer. These (related to local coordinate
system) are:
(2D case)
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.2.2:1
Preface Structures Beams
normal strain

xx
L
(x
L
, y
L
) =
u
L
x
L
= e
T
x
L

o
u
G
(x
L
)
x
L
y
L

z
L
(x
L
)
x
L
,
shear strain

xy
L
=
v
L
x
L
+
u
L
y
L
= e
T
y
L

o
u
G
(x
L
)
x
L

z
L
(x
L
)
additionally, for Axisymmetry , circumferential strain is set as:

zz
L
(x
L
, y
L
) =
u
G
r
=
1
r
_
o
u
G
(x
L
) cy
L

z
L
(x
L
)
_
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.2.2:2
Preface Structures Beams
5.2.3 WEAK FORMULATION OF THE EQUILIBRIUM
In both layered and integral approach, the treatment of a transversal shear (and torsion
for the 3D case) is decoupled from bending and extension state analysis and limited to
linear stress- strain relation. This kind of procedure is exact in linear cases, while for the
nonlinear one, consists a commonly accepted simplication. Coupled treatment of both,
shear & bending / extension state would require additional signicant numerical eort and
much more complex formulation. In both formulations, the shear behavior is treated in
integral way, with usage of averaged shear angle, stress-resultant tangent force and a shear
correction factor.
The virtual work principle expressing the equilibrium of a beam may be put in the general
form:
nd such that:
for any , u
_
V

T
dV
_
V
u
T
pdV = 0
From the above, taking into account kinematic assumptions both layered and integral
approach, detailed formulation for each analysis type may be derived, i.e.:
layered formulation:
(2D, Plane Strain case)
for any , , u :

i
_
V
(i)

(i)
xx
L

xx
L
dV
(i)
+
_
L

xy
L
(k
y
GA)
xy
L
dL
_
V
u
T
pdV = 0
(2D, Axisymmetric case, dV is conical volume element)
for any , , u :

i
_
V
(i)
_

(i)
xx
L

xx
L
+
(i)
zz
L

(i)
zz
L
_
dV
(i)
+ 2
_
L

xy
L
(k
y
GA)
xy
L
r dL
_
V
u
T
pdV = 0
integral formulation for elastic beam model
(2D, Plane Strain case, case)
for any , ,
z
, u :
_
L
[
o

xx
L
N +
z
M +(k
y
GA)] dL
_
V
u
T
pdV = 0
For material model description see the Appendix, Window 5-13.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.2.3:1
Preface Structures Beams
5.2.4 INTERPOLATION OF THE DISPLACEMENT FIELD
For the point lying on the element centroid line with given reference coordinate , both
translations u and rotational , displacement components (referred to global coordinate
system ) are interpolated from its values at the centroid nodes Ci. Note, that these may
be obtained from element DOF by master to centroid oset transformation ,see the
Appendix 5-12.
o
u
G
() =

i=1,2
N
i
()u
G
Ci
o

G
() =

i=1,2
N
i
()
G
Ci
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.2.4:1
Preface Structures Beams
5.2.5 STRAIN REPRESENTATION
The formulae relating strains any point within the element with its DOF vector q, based on
kinematic assumptions is given in general form:
(, y
L
, z
L
) = B(, y
L
, z
L
)q =
Nen

i=1
B
i
(, y
L
, z
L
)q
i
, q
i
=
_
u
C
i

C
i
_
The displacement vector q consist of:
2DPlane Strain, Axisymmetry:
q
i
=
_
u, v,
z

T
i-th node submatrix of B-matrix relating shear strain with nodal DOF has a form in the case
of:
2DPlane Strain, Axisymmetry:
B

i
=
_
e
y
L
x
DN
i
,

; e
y
L
y
DN
i
,

; N
i

Remaining strain component onenode submatrix B


i
takes form (for straight geometry
element):
layered approach:
2DPlane Strain:
axial strain
B
x
L
i
=
_
e
x
L
x
DN
i
,

; e
x
L
y
DN
i
,

; DN
i
,

y
L

2DAxisymmetry:
axial strain:
B
x
Li
=
_
e
x
L
x
DN
i
,

; e
x
L
y
DN
i
,

; DN
i
,

y
L

circumferential strain:
B
x
Li
=
_
N
i
r
; 0;
N
i
r
y
L
e
y
L
y
_
integral approach:
2D-Plane Strain:
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.2.5:1
Preface Structures Beams
axial strain at centroid line:
B

i
=
_
e
x
L
x
DN
i
,

; e
x
L
y
DN
i
,

; 0

curvature:
B

i
=
_
0; 0; DN
i
,

June 25, 2003


Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.2.5:2
Preface Structures Beams
5.2.6 STIFFNESS MATRIX AND ELEMENT FORCES
Stiness matrix K
e
and force vector f
e
of a beam element are derived in a standard way from
the weak formulation of a problem. It would require constitutive module matrix D for each
stressstrain component. Numerical integration technique is used to evaluate integrals over
the length of the element. In the case of a 2node linear element selected reduced integration
technique (SRI) with 1 integration point is used ( Ngaus = 1, W
1
= 2.0,
1
= 0.0).
Integration over the cross section of the element is performed numerically in the case of
layered approach, or it is hidden in given values of integral characteristics of the cross section
( shear area for uncoupled treatment of shear, area and rotational inertia for elastic beam).
Integrals in layered approach case use longitudinal jacobians
l
J = det(J(, y
L
l , z
L
l )) eval-
uated at each layer l separately, in order to enhance accuracy in case of curved elements.,
while other integrals over the element length use jacobians referred to centroid line
o
J =
det(J(, 0, 0)). For all 2D cases, contribution of shear part is evaluated in a common way
as:
K

=
1
_
1
B
T

()(kGA)B

()
o
J d
=
Ngaus

igaus=1
B
T

(
igaus
)(kGA)B

(
igaus
)
o
J
igaus
W
igaus
f

=
1
_
1
B
T

()(kGA())
o
J d
=
Ngaus

igaus=1
B
T

(
igaus
)(kGA(
igaus
))
o
J
igaus
W
igaus
2DPlane Strain, layered approach (Nonlinear beam):
K
e
=
1
_
1

l
B
T
x
L
(, y
L
l )D
l
xxxx
B
x
L
(, y
L
l )
l
J d +K

=
Ngaus

igaus=1

l
B
T
x
L
(
igaus
, y
L
l )D
l
xxxx
B
x
L
(
igaus
, y
L
l )
l
J
igaus
W
igaus
+K

f
e
=
1
_
1

l
B
T
x
L
(, y
L
l )
xx
(, y
L
l )
l
J d +f

=
Ngaus

igaus=1

l
B
T
x
L
(
igaus
, y
L
l )
xx
L
(
igaus
, y
L
l )
l
J
igaus
W
igaus
+f

June 25, 2003


Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.2.6:1
Preface Structures Beams
Plane Strain , integral approach (Elastic beam model ):
K
e
=
1
_
1
_
B
T

(); B
T

()

_
EA 0
0 EL
z
_ _
B

()
B

()
_
o
J d +K

=
Ngaus

igaus=1
_
B
T

(
igaus
); B
T

(
igaus
)

_
EA 0
0 EL
z
_ _
B

(
igaus
)
B

(
igaus
)
_
o
J
igaus
W
igaus
+K

f
e
=
1
_
1
_
B
T

(); B
T

()

_
EA()
EI
z
()
_
o
J d +f

=
Ngaus

igaus=1
_
B
T

(
igaus
); B
T

(
igaus
)

_
EA(
igaus
)
EI
z
(
igaus
)
_
o
J
igaus
W
igaus
+f

2DAxisymmetry, layered approach (Nonlinear beam):


K
e
= 2
_
_
1
_
1

l
_
B
T
x
L
(, y
L
l ); B
T
z
L
(, y
L
l )
_
D
l

_
B
x
L
(, y
L
l )
B
z
L
(, y
L
l )
_
r()
l
Jd +rK

_
_
= 2
_
Ngaus

igaus=1

l
_
B
T
x
L
(
igaus
, y
L
l ); B
T
z
L
(
igaus
, y
L
l )
_
D
l

_
B
x
L
(
igaus
, y
L
l )
B
z
L
(
igaus
, y
L
l )
_
_
r(
igaus
)
l
J
igaus
W
igaus
+ 2rK

f
e
= 2
_
_
1
_
1

l
_
B
T
x
L
(, y
L
l ); B
T
z
L
(, y
L
l )

_

xx
(, y
L
l )

xz
(, y
L
l )
_
r()
l
J d +rf

_
_
=
2
_
Ngaus

igaus=1

l
_
B
T
x
L
(
igaus
, y
L
l ); B
T
z
L
(
igaus
, y
L
l )

_

xx
(
igaus
, y
L
l )

xz
(
igaus
, y
L
l )
_
_
r(
igaus
)
l
J
igaus
W
igaus
+ 2rf

where constitutive module matrix: D


l
(22)
=
_
D
xxxx
D
xxzz
D
zzxx
D
zzzz
_
.
The element load p is assumed to act along the centroid line leading to load induced forces
evaluated as:
f
p
=
1
_
1
N
T
()p
o
J d =
Ngaus

igaus=1
N
T
(
igaus
)p
o
J
igaus
W
igaus
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.2.6:2
Preface Structures Beams
N() =
_
N
i
()I
(dim)

, i = 1, Nen
All element forces and stiness, as given above, are referred to centroidal nodes of the
element. Prior to the agregation to global ones, they are submitted to Mastercentroid
(oset) transformation, see Appendix 5.4.1 .
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.2.6:3
Preface Structures Beams
5.2.7 MASTER-CENTROID (OFFSET) TRANSFORMATION
In order to deal with the frequently encountered situation where beam elements are connected
to other elements of the model by nodes, which are not the centroids of a beam cross section,
oset transformation of element displacement, forces and stiness is introduced. The DOF
of the element are placed on its master nodes dening connectivity. Based on rigid body
movement of the master, translation and rotation displacements of the centroid are
evaluated.
For the details of master-slave oset transformation see the Appendix 5.4.1 .
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.2.7:1
Preface Structures Beams
5.2.8 RELAXATION OF INTERNAL DOF
Stiness matrix and force vector derived above concerns element with all DOF active. Situa-
tion when user demanded group of DOF is relaxed (what means that no forces are transmitted
by the element in selected directions).The directions of relaxed DOF might be related to local
element directions at both ends of the element.
If given directions at beam node are relaxed, the node is implicitly duplicated and DSC
(Node-to-node interface ) elements are introduced to the model. The stiness of that in-
terface remains 0 at relaxed direction, while it is assumed to take penalty values, estimated
automaticaly from beam stiness on xed DOF.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.2.8:1
Preface Structures Beams
5.2.9 BEAM ELEMENT RESULTS
Element stress resultants are referred to cross section local coordinate system at the integra-
tion point. The sign convention is as follows:
N normal force positive in tension
M
i
bending moment positive, are such which leads to positive (tensile) stresses in the
points with positive local coordinate on the complementary axis
Q
i
shear forces positive is such that produce positive shear stresses acting in the
ith local direction
M
x
torsion moment positive moment is represented by the vector directed towards
outer normal
Note, that in case of all analysis types, signs of some stress resultants are related to node
order C1 C2
Window 5-8: 2D beam element result setting
2D beam element results sign convention
Window 5-8
Note, that for Plane Strain and Generalized plane strain Analysis case of discrete beam system
in Z direction stress resultants are referred to single beam but not a unit slice of a model.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.2.9:1
Preface Structures Beams
Window 5-9: Axisymetric shell element result setting
Axisymmetric shell element results sign convention
Window 5-9
Note, that in the case of Axisymmetry, evaluated stress resultants are referred to the unit
length of the axisymmetric shell both in circumference and meridian direction.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.2.9:2
Preface Structures
5.3 MEMBRANES
ELEMENT GEOMETRY MAPPING AND COORDINATE SYSTEM
DISPLACEMENTS AND STRAINS FIELD
CONSTITUTIVE MODELS
WEAK FORMULATION OF THE EQUILIBRIUM
STIFFNESS MATRIX AND ELEMENT FORCES
Elements are designed to model dierent kind of soil reinforcement (solid phase) such as
geotextiles, geo-grids. In modern geo-technical practice there is a variety of dierent types
of such a means. Membrane elements are available in all 2D and 3D analysis types. Although
membrane and truss elements use the same DOF (at least for statics), note the dierence
between membrane and truss elements for all 2D analysis. The dierence concerns the
possibility of using a wider list of constitutive models in the case of membrane elements.
Window 5-10: Membrane elements
The family of membrane elements
Window 5-10
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.3:1
Preface Structures
Note the signicant dierence between membrane and truss elements for the Axisymmetry
analysis. For the membrane elementary volume dV changes with the current point radius r,
while for the truss elements it remains constant, see the Window 5-1. The geometry of the
element is set as iso-parametric surface element (all T3/Q4) in 3D space or as 2 node linear
element in 2D space. It posses one layer of nodes - as it is required in statics. As element
has only 1 layer of nodes it can not be used to model a ow in the normal direction - (2 node
layer necessary) -thus membrane elements can not be used to model impermeable surface.
When necessary to simulate the drainage or impermeable surface in the normal direction, one
node layer membrane element should be placed together with 2-node layer ow interface:
Window 5-11: Flow trough membrane elements
Modelling of ow trough membrane elements
Window 5-11
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.3:2
Preface Structures Membranes
5.3.1 ELEMENT GEOMETRY MAPPING AND COORDINATE SYS-
TEM
In all 2D cases geometry setting for membrane element is identical with these for truss
element, see Window 5-3.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.3.1:1
Preface Structures Membranes
5.3.2 DISPLACEMENTS AND STRAINS FIELD
The displacements within the element are interpolated from nodal values as:
u() =

a=1,Nen
N
a
()u
a
The strains within the element (only in-plane are taken into account)
2D analysis types:
Plane Strain:
=
_
_

xxL

yyL

xyL
_
_
=
_

_
u
xL
x
L
0
0
_

_
=
_

_
e
T
xL

u
x
L
0
0
_

_
Axisymmetry:
=
_
_

xxL

yyL

xyL
_
_
=
_

_
u
xL
x
L
u
xG
r
0
_

_
=
_

_
e
T
xL

u
x
L
u
xG
r
0
_

_
The formulae relating strains any point within the element, with its DOF vector u may be
put in the general form:
= Bu =
Nen

a=1
B
a
u
a
,
Plane Strain:
u
a
= [u
a
, v
a
]
B
a
() = [e
xLx
DN
a

, e
xLy
DN
a

]
Axisymmetry:
B
a
() =
_

_
e
xLx
DN
a

e
xLx
DN
a

N
a
r
0
0 0
_

_
,
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.3.2:1
Preface Structures Membranes
5.3.3 CONSTITUTIVE MODELS
Constitutive models used in membrane elements (solid phase) for all analysis types re-
late in- Plane Strains = [
xxL
,
yyL
,
xyL
, ]
T
with membrane stress components =
[
xxL
,
yyL
,
xyL
, ]
T
in local directions of membrane element. For some models, constitu-
tive data concerning stiness and strength must be given in reference to the whole thickness
of the element, in units [force/length]. The models are:
Elasto-plastic membrane (isotropic)
Data: stiness K [force/length], Poisson ratio
, tensile and compressive strength f
t
, f
c
[force/length]
Applications: geotextile
Stress criterion:
1
f
t
,
2
> f
c
where:
1
,
2
principal stresses
Elasticity matrix: D
e
=
_

_
K K 0
K K 0
0 0
K
2(1 +)
_

_
Elasto-plastic membrane (anisotropic)
Data: stifness K
11
, K
22
, K
12
[force/length] tensile
strength f
t1
, f
t2
[force/length] compressive
strength f
c1
, f
c2
[force/length] angle be-
tween local element axis x
L
and an-isotropy
1-st axis x
1
, evaluated from projection of
a direction vector onto element surface, see
Appendix ?? .
Applications: geo-grids
Stress criterion:
_

11

22
_
= T
Elasticity matrix: D
e
= T
T

_
_
K
11
K
12
K
12
K
22
_
_
T
where: T =
_
c
2
s
2
sc
s
2
c
2
sc
_
;
s = sin , c = cos
Elasto-plastic bre
Data: elasticity moduleE, area per unit length
A, tensile and compressive strengthf
t
, f
c
,
angle between local element axis x
L
and
bre direction x
1
, evaluated from projection
of a direction vector onto element surface,
see Appendix ?? .
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.3.3:1
Preface Structures Membranes
Applications: reinforcement layer
Note: In case of Axisymmetry, the model is suitable for modelling circumferential reinforce-
ment ( = 90

), while for longitudinal one, usage of truss elements is recommended.


Stress criterion f
t
, > f
c
uniaxial stress in the bre direction
Elasticity matrix: D
e
= Et t
T
where: t =
_
c
2
s
2
sc

T
.
Elasto plastic-plane stress member
Data: elasticity moduleE, Poisson ratio , area
per unit lengthA, MenetreyWillam crite-
rion data
Applications: thin lining (steel, concrete, etc.)
Stress criterion MenetreyWillam (plane stress)
Elasticity matrix: standard plane stress
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.3.3:2
Preface Structures Membranes
5.3.4 WEAK FORMULATION OF EQUILIBRIUM
The contribution of the membrane elements to the virtual work principle expressing equilibrium
of a system may be put as:
nd such that:
for any , u
_
S

T
AdS
_
S
u
T
p dS = 0
with
=
_

xxL
,
yyL
,
xyL
,

T
;
= [
xxL
,
yyL
,
xyL
, ]
T
;
u =
_
[u v]
T
for Plane Strain, Axisymmetry
[u v w]
T
for Gen. plane strain, 3D
A thickness of the element (= 1 for models using membrane forces instead of stresses).
Integrals are taken over the surface of the element. S is an area attributed to the assumed
computational domain i.e. to the unit slice for the Plane Strain,Generalized Plane Strain or
to the whole circumference for the Axisymmetric case.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.3.4:1
Preface Structures Membranes
5.3.5 STIFFNESS MATRIX AND ELEMENT FORCES
Stiness matrix K
e
and force vector f
e
of the membrane element are derived in a standard
way from weak formulation of a problem. It would require constitutive module matrix D as
well as stress evaluation for given strain increment. Numerical integration technique is used
to evaluate integrals over the element with inegration point number as shown in the table.
Element M L2 M T3 M Q4
Ngaus 1 1 22
2D cases:
Plane Strain, Generalized Plane Strain:
K
e
=
1
_
1
1
_
1
B
T
DB | x,

| Ad =
Ngaus

igaus=1
B
T
DB | x,

| A W
igaus
f
e
=
1
_
1
1
_
1
B
T
| x,

| Add =
Ngaus

igaus=1
B
T
| x,

| A W
igaus
Axisymmetry
K
e
=
1
_
1
1
_
1
B
T
DB | x,

| 2r Ad =
Ngaus

igaus=1
B
T
DB | x,

| 2rA W
igaus
f
e
=
1
_
1
1
_
1
B
T
| x,

| 2rAdd =
Ngaus

igaus=1
B
T
| x,

| 2rA W
igaus
.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.3.5:1
Preface Structures Appendices
5.4 APPENDICES
MASTER-SLAVE (OFFSET) TRANSFORMATION
UNI-AXIAL ELASTO-PLASTIC MODEL
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.4:1
Preface Structures Appendices
5.4.1 MASTER-SLAVE (OFFSET) TRANSFORMATION
In order to deal with the frequently encountered situation where elements are connected
to other elements of the model by nodes, which are not the centroids of a cross section,
oset transformation of element displacement, forces and stiness is introduced. The DOF
of the element are placed on its master nodes dening connectivity. Based on rigid body
movement of the master, translation and rotation displacements of the centroid being the
slave node are evaluated. The above concerns beams as well as shell elements.
Window 5-12: Master-slave (oset) transformation
Displacements at the slave:
u
S
= u
M
+
M
o

S
=
M
where oset vector is used:
o = x
Slave
x
Master
o = xi, yi, (zi)
T
.
In turn, forces and moments evaluated initially at slave are moved to masters in a way
preserving static equivalency:
t
M
= t
S
m
M
= m
S
+o t
S
Expressing the above in a matrix form one can get:
displacement transformation :
q
S
= Oq
M
force transformation:
f
M
= O
T
f
S
stiness transformation:
as : f
M
= O
T
f
S
= O
T
K
S
q
S
= O
T
K
S
Oq
M
= K
M
q
M
K
M
= O
T
K
S
O
Oset transformation matrix takes the form:
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.4.1:1
Preface Structures Appendices
(all 2D cases)
O =
_
_
1 0 yi
0 1 xi
0 0 1
_
_
Window 5-12
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.4.1:2
Preface Structures Appendices
5.4.2 UNI-AXIAL ELASTO-PLASTIC MATERIAL MODEL
The uni-axial stress-strain relationship to be used commonly for truss, ring, beam, bers is
given as follows:
Window 5-13: Uniaxial elastoplastic material model
Uniaxial elasto-plastic material model
Window 5-13
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.4.2:1
Preface Structures Appendices
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM5.4.2:2
Preface
Chapter 6
INTERFACE
CONTACT
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM6.0.2:3
Preface Interface
6.1 CONTACT OF SOLIDS AND FLUID INTERFACE
GENERAL OUTLOOK
DISPLACEMENT & STRAINS
CONSTITUTIVE MODEL
STIFFNESS MATRIX AND FORCE VECTOR
AUGMENTED LAGRANGIAN APPROACH
CONTRIBUTION TO CONTINUITY EQUATION
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM6.1:1
Preface Interface Contact
6.1.1 GENERAL OUTLOOK
Window 6-1: Interface elements: General remarks
Mechanical contact as well as ux (of uid, heat, humidity) through the surface between
two bodies is modelled by nite element discretization of the interface between them. The
interface elements use nodes belonging to the FEmesh of both adjacent solids with assumed
invariable topology (small displacement theory). Moreover the compatibility of the initial
positions of nodes is required (this is assured by preprocessing tools, see Interface option).
Interface (contact) elements between 2 adjacent bodies
Interface element geometry is based on the iso-parametric mapping from the reference ele-
ment. As nodes of both layers of element are assumed to occupy the same position :
x() =

i=1,..,Nen
N
i
()x
i
.
Family of interface elements
Window 6-1
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM6.1.1:1
Preface Interface Contact
Window 6-2: Interface: Mechanical contact
As an additional feature, for mechanical contact, the displacements continuity option is
introduced. This is related to interface element status.
Status CONTACT: The interface elements reproduce the force action between the two
bodies based on the relative displacements of the interface nodes. The elastoplastic friction
model is used, allowing for sliding and separation, while the elastic properties of the interface
impose penalty constraints multipliers excluding penetration, see Window 6-4 for the details.
Status CONTINUITY(u,p,T) or CONTINUITY(u,T): Displacements continuity across
the interface is enforced. Nodes on both sides of the interface share the same kinematical
DOFs.
Window 6-2
Window 6-3: Fluid Phase
In case of Flow or Deformation+Flow analysis mode, interface elements may posses
pressure DOF at nodes of both layers. Interface elements can be used to model following
situations:
Fully permeable interface
Pressures continuity across the interface is enforced. Nodes on both sides of the interface
share the same pressure DOFs.
Impermeable interface
No ow takes place in the direction normal to the interface (no ux q
n
= 0 condition
on both faces is imposed). Resulting pressures on both interface faces will be (in general)
discontinuos. In this case the interface element does not contribute to the equation system.
Permeable interface
Both isotropic and anisotropic ow conditions can be handled.
Window 6-3
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM6.1.1:2
Preface Interface Contact
6.1.2 DISPLACEMENTS AND STRAINS
Displacements at each layer of the interface L1 and L2 are interpolated using standard
approach, from given nodal displacements u:
uL1
L2
() =

i=1,..,Nen
i=Nen+1,..,2Nen
N
i
()u
i
. The generalised strains at the integration point of the interface element are mutual displace-
ments of both layers transformed to the local basis t, n of the element. In 3D case local
base on the element surface is created according to unied procedure given in Appendix ??.
Generalized strains in contact element
The relation between the above generalised strain and nodal displacements may be put in
unied form,
() = B()u =
2Nen

i=1
B
i
()u
i
with B matrix given as:
B()= T() [N
i
()I
NDOF
, N
i
()I
NDOF
]
T
, i = 1, Nen
where:
T()transormation matrix such that U
TN
= T()u
I
NDOF
unit matrix, NDOF is displacement component number per node
2D cases
Plane Strain, Axisymmetry:
u
i
= [u
i
, v
i
]
T
() = [u
t
, u
n
]
T
= [u
t
L2
u
t
L1
, u
n
L2
u
n
L1
]
T
T() =
_
c, s
s, c
_
, c =
x,

_
x,
2

+y,
2

, s =
y,

_
x,
2

+y,
2

Moreover, initial gap may be accounted for while evaluating element strains.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM6.1.2:1
Preface Interface Contact
6.1.3 CONSTITUTIVE MODEL
Constitutive behaviour of the interface is described in the terms of:
generalised strains , evaluated from nodal displacements of the interface
eective stresses

, submitted to appropriate stress criterions resulting from


1. cohesive Mohr - Coulomb condition
2. no-tension condition.
For Plane Strain and Axisymmetry number of stress components Nstre = 2, while for
3D, Generalized Plane Strain analysis Nstre = 3. In the case of analysis mode Defor-
mation+Flow, the concept of eective stress is used taking into account pressures p and
saturation ratio S:
Nstre = 2 Nstre = 3
=

+S
_
0
p
_
=

+S
_
_
0
0
p
_
_
ow rule with a ow potential in the form analogous to Mohr-Coulomb yield function
allowing for non-associative ow rule in the case when ,= ,
constitutive matrix D.
Both

and D are evaluated within the frame of perfect multisurface elastoplasticity


theory, with components related to the plane of the interface and its normal.
Formulation of both cases of contact constitutive law is given in the Window 6-4.
The trial stresses

are evaluated as:

=
n
+D
el

using the previous stress


n
, strain increment , the elastic (penalty) interface stiness
D
el
.
The elastic stiness K
n
should be large enough to prevent signicant penetration in the case
of compression, but can not undertake arbitrarily large values as it might spoil conditioning
of the resulting FE equation system and lead to diculties in obtaining convergence of the
solution. Estimation of penalty stiness is done as follows in the Window 6-5.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM6.1.3:1
Preface Interface Contact
Window 6-4: Constitutive law of frictional contact
Analysis type: Plane Strain, Axisymmetry 3D,Generalized Plane Strain
Nstre 2 3
Data:
c cohesion
friction angle
dilatancy angle (0 )
Stresses:
= [,
n
]
T
with:
the shear interface stress

n
the stress normal to the
interface
= [
1
,
2
,
n
]
T
with:

1
,
2
the shear interface stresses
in local directions e
1
, e
2

n
the stress normal to the
interface
Stress conditions Slip activated when:
F
1
(

) = + tan()

n
c > 0 if 0
F
2
(

) = + tan()

n
c > 0 if < 0
F
1
(

) =
_

2
1
+
2
2
+
tan()

n
c > 0
No tension (cuto) activated when:
F
3
(

) =

n
> 0 F
2
(

) =

n
> 0
Graphic presentation:
Flow potential
Q
1
(

) = + tan()

n
c > 0 if 0
F
2
(

) = + tan()

n
c > 0 if < 0
Q
1
(

) =
_

2
1
+
2
2
+
tan()

n
c > 0
Gradients
a =
F

,
b =
Q

:
a
1/2
=
_
1, tan

T
,
b
1/2
=
_
1, tan

T
.
a
1
=
_

1

,

2

tan
_
T
b
1
=
_

1

,

2

tan
_
T
=
_

2
1
+
2
2
Elasticity matrix: D
el
=
_
K
t
0
0 K
n
_
D
el
=
_
_
K
t
0 0
0 K
t
0
0 0 K
n
_
_
Window 6-4
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM6.1.3:2
Preface Interface Contact
Window 6-5: Estimation of the penalty stiness and permeability
This is done automatically (default) based on the following algorithm:
1. Find neighbouring elements ( active at the current time t
n
+ 1)
2. For neighbouring elements nd the maximum size in the direction normal to the interface
Penalty stiness estimation
1. Estimate normal stiness as:
K
n
= min
_
E
1
h
1
,
E
2
h
2
_
A

Neq
2. Set tangent stiness K
t
of the interface as:
K
t
= 0.01K
n
3. In the case of Flow or Deformation+Flow analysis mode estimate penalty permeability:
k
f
=
B

Neq
min
_
k
1
h
1
,
k
2
h
2
_
where k
m
=


i=1,N dim
k
2
ii
k
r
(S(p))
taking into account permeability multiplier k
r
dependent on current saturation S = S(p)
In the above:
A, B arbitrary factors(default A = 10
4
, B = 10
3
)set by numerical experience
Neq total equation number in the system
precision (machine dependent small number)
Parameters adopted under points 3, 4, 5 may be multiplied by user-dened factors.
Window 6-5
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM6.1.3:3
Preface Interface Contact
Window 6-6: Stress point algorithm
Depending on the trial stress

, 3 dierent types of the behaviour will be modelled:


sticking,
sliding,
separation.
The algorithm of stress and constitutive matrix evaluation is as follows:
Stress-point algorithm
if

n
> 0 then
separation:

n+1
= 0
D = 0
else
if F
1
(

) 0
if Nstre=2
..
F
2
(

) 0then
sticking

n+1
=

D = D
el
else
sliding (F
1
(

) > 0) :

n+1
=
n
+D
el
( b)
=
F
i
(

)
a
T
D
el
b
D = D
el

(D
el
b) : (D
el
a)
T
a
T
D
el
b
where : a =
F
i

, b =
Q
i

end if
end if
Note that stress return for the sliding case is performed with a one step cutting-plane
procedure. This is due to the linear form of the yield function and ow potential in 2D case
(Nstre = 2) as well as possible radial return for the 3D case (Nstre = 3).
Window 6-6
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM6.1.3:4
Preface Interface Contact
6.1.4 STIFFNESS MATRIX AND ELEMENT FORCE VECTOR
These are evaluated as: 2D cases:
Plane Strain, Generalized Plane Strain
K
e
=
1
_
1
B
T
DB | x,

| d =
Ngaus

igaus=1
B
T
DB | x,

| W
igaus
f
e
=
1
_
1
B
T
| x,

| d =
Ngaus

igaus=1
B
T
| x,

| W
igaus
Axisymmetry
K
e
=
1
_
1
B
T
DB | x,

| 2rd =
Ngaus

igaus=1
B
T
DB | x,

| 2rW
igaus
f
e
=
1
_
1
B
T
| x,

| 2rd =
Ngaus

igaus=1
B
T
| x,

| 2rW
igaus
with constitutive matrix D and stresses being the result of the point level algorithm, see
Window 6.5.4 . Note that in the case of Analysis mode Deformation +Flow, the total
stresses including pressure is used to evaluate element forces. In order to avoid oscillatory
patterns of normal stress, B matrix is evaluated using following averaging technique, in which
the part related to normal strains B
n
is evaluated in central integration point exclusively:
B(
igaus
) =
_
_
B
t
(
igaus
)
s
B
n
(
igaus
)
_
_
The numerical integration is used with integration point numbers given in the table:
Element C L2 C T3 C Q4
Ngaus 2 3 22
Described above concept of averaging normal strain-related part of B matrix is equivalent to
selective integration technique, with 1-point integration rule for normal strain-related part of
stiness and force vector.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM6.1.4:1
Preface Interface Contact
6.1.5 AUGMENTED LAGRANGIAN APPROACH
1
A common problem of the standard penalty approach used for problems of contacting
elasto-plastic media is that the resulting overpenetration is too large and contact stresses
may be underestimated. Usage of high values of penalty stifnesses usually results in loss of
convergence and oscillatory contact stress distribution. In the case of soil-structure contact
interaction the stress resultants may be underestimated as well. To handle this deciency an
Augmented Lagrangian Approach can be used. In the current contact formulation (segment
to segment approach) each time the state of the global static equlibrium is achieved the
contact resulting normal stresses are memorized, penalty stiness is increased (by default
through factor of 2). The Augmented Lagrangian Approach is summarized in window given
below.
Window 6-7: Augmented Lagrangian Approach
1. initialize: = 0,
(=0)
f = 1
2. at each integration point at contact element set:
(=0)

n
N+1
=
n
N
3. solve: F
ext
N+1
F
int
(u
N+1
) = 0 assuming that that the trial normal stress at each inte-
gration point of contact element is computed as:
trial
n
N+1
=
()

n
N+1
+
()
f k
n

n
N+1
4. at each integration point of contact element check overpenetration: [
n
N+1
[ >TOL (?)
5. if overpenetration is too large (at any integration point) perform augmentation procedure:
6. set: = + 1
7. increase penalty parameter:
()
f =
(1)
f g (g = 2 by default)
8. if
()
f > f
max
set:
()
f = f
max
9. if <MAX-AUGMENTATIONS go to step (2)
Window 6-7
1
concerns versions: ACADEMIC, PROFESSIONAL, EXPERT only
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM6.1.5:1
Preface Interface Contact
6.1.6 CONTRIBUTION TO CONTINUITY EQUATION
In the case of active uid phase interface elements contribute to matrix H and ux vector Q
(see point 4.1.2) with the following:
H
Interface
=
_

_
N
N
_
k
f
_
N
T
, N
T

d
Q
Interface
= H
Interface
t
w
where:
N
T
=
_
N
1
() . . . N
Nen
()

shape function vector


t
T
w
=
_
t
w1
, . . . t
wNen

nodal pressure vector.


The integration technique analogous to the one used for Stiness matrix and element force
vector evaluation is used.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM6.1.6:1
Preface
Chapter 7
GEOTECHNICAL ASPECTS
TWOPHASE MEDIUM
EFFECTIVE STRESSES
SOIL PLASTICITY
INITIAL STATE
SOIL RHEOLOGY
ALGORITHMIC STRATEGIES
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.0.6:2
Preface Geotechnical aspects
7.1 TWO-PHASE MEDIUM
In this section, an attempt is made to relate modelling parameters to geotechnical aspects.
The soil is modeled as a two-phase medium, this means that equilibrium of the medium
requires the solution of a coupled system of dierential equations where one set of equations
represents the equilibrium of the solid and the second set of equations represents the continuity
of the uid ow. Both sets include coupling terms. Drained and undrained conditions are
limiting cases of particular interest. The corresponding boundary conditions are shown in
Window 7-1
Drained conditions
Boundary conditions are such that, in the long term, the local stress is carried by the
skeleton ( =

).
Undrained conditions
When boundary conditions and material properties are such that no uid motion relative to
the solid is possible, the condition is undrained and the medium behaves in an essentially
incompressible manner.
Window 7-1: Drained and undrained condition
Drained condition: p
F
= p
F
on
p
(left); Undrained condition q = 0 on
v
(right)
Window 7-1
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.1:1
Preface Geotechnical aspects
7.2 EFFECTIVE STRESSES
Eective stresses allow a unied approach to the analysis of the drained and undrained
conditions, this concept is extended here to account for partially saturated media. Let:
=

+p
F
S
where

is the eective (grain to grain stress) , p


F
the interstitial pressure, and S the
saturation ratio.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.2:1
Preface Geotechnical aspects
7.3 SOIL PLASTICITY
DRUCKER-PRAGER VERSUS MOHR-COULOMB CRITERION
CAP MODEL
DILATANCY
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.3:1
Preface Geotechnical aspects Soil plasticity
7.3.1 DRUCKER-PRAGER VERSUS MOHR-COULOMB CRITE-
RION
It is common to describe soils as elasticperfectly plastic MohrCoulomb materials. A smooth
MohrCoulomb criterion described earlier is available in this program. If a DruckerPrager
criterion with cap closure is preferred, the size of the DruckerPrager criterion can be adjusted
to match the MohrCoulomb criterion. This is illustrated in Window 7-2. The various
matching options are derived in the theoretical section; for plane strain, the most meaningful
matching for an ultimate load analysis is the matching of collapse loads. The most important
sizeadjustments are summarized in Window 7-2.
Window 7-2: Matching Mohr-Coulomb and Drucker-Prager criteria
Deviatoric sections of MohrCoulomb (MC) and DruckerPrager
Matching a

k
External apices
2 sin

3(3 sin )
6c cos

3(3 sin )
Internal apices
2 sin

3(3 + sin )
6c sin

3(3 + sin )
Planestrain collapse
sin
3D
; D =
_
a

sin +
_
1 3a
2

_
c cos D
1
Elastic domain
sin
3
c cos
(plane strain
t
= 0.5)
1
a

, k : parameters of the DruckerPrager criterion


: friction angle
C : cohesion
: angle of nonassociativity (in doubt, use default value)
Window 7-2
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.3.1:1
Preface Geotechnical aspects Soil plasticity
Related Topics
THEORY: M-C VERSUS D-P
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.3.1:2
Preface Geotechnical aspects Soil plasticity
7.3.2 CAP MODEL
The cap model accounts for nonlinear behavior of soil under dominant volumetric (pressure)
stress. The initial size of the cap is derived from the oedometric test, it is determined by
the preconsolidation pressure. Once the yield point (i.e. the cap) is reached by the stress,
hardening takes place.
Hardening results from the reduction of the void ratio under increasing pressure, it is again
controlled by the oedometric test.
Related Topics
CAP MODEL
CAP MODEL - STRESS POINT ALGORITHM
OEDOMETRIC TEST
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.3.2:1
Preface Geotechnical aspects Soil plasticity
7.3.3 DILATANCY
Two alternative dilatancy parameters can be accommodated by the material models proposed
earlier. The rst parameter, d, relies on the availability of experimental results. The second
one, , assumes the empirical knowledge of , by analogy with the friction angle .
Dilatancy parameter d
Volumes changes in soils, in the plastic regime, are conveniently described as follows; let
d be the dilatancy
d =
d
p
V
d
p
D
=
Q/p
Q/q
=
r
p
r
q
where d
p
V
is the volumetric plastic strain increment and d
p
D
is the plastic deviatoric
strain increment. Q is the plastic potential, r
p
is the norm of the volumetric plastic ow
component and r
q
, the deviatoric one. d
p
V
, d
p
D
will usually be retrieved from a tri-axial
test and r
p
/r
q
will be derived from the plastic model.
Dilatancy extraction from triaxial test
Experimental results from a triaxial test can be plotted as follows:

V
=
1
+ 2
3

D
= (2/3)(
1

3
)
In the plastic regime d
V

= d
P
V
and d

= d
P
hence:
d
V
d
D
= d
Dilatancy with Drucker-Prager plasticity
The plastic potential is in this case:
Q = a

I
1
+
_
J
2
then,
r
p
=
Q
p
=
Q
I
1
I
1
p
= 3a

r
q
=
Q
q
=
Q

J
2

J
2
p
= 1/

3
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.3.3:1
Preface Geotechnical aspects Soil plasticity
and nally,
d = 3

3a

Given d from the experiment, a

can be found for the material model.


Dilatancy with Mohr-Coulomb plasticity
Plastic ow is again governed by a Drucker-Prager type surface in the program, even when
yield is governed by a Mohr-Coulomb criterion. The rate of non-associativity generated by
a given (experimental) d is therefore dependent on the size adjustment.
In 2D, for size adjustment option matching plane strain collapse load , given , c and d,
one gets:
a

= d/3

3
a

=
1
3
sin [a

sin +
_
1 3a
2

]
1
k = C cos [a

sin +
_
1 3a
2

]
1
For a 3D situation in which denes the size adjustment with respect to external and
internal matching, we get
a

= d/3

3
a

= (1 )[2 sin /(

3(3 sin ))] +[2 sin /(

3(3 + sin ))]


k = (1 )[6C cos /(

3(3 sin ))] +[6C cos /(

3(3 + sin ))]


NB: a

should be such that (0 < a

< a

).
Alternative experimental results can also be used in order to dene d, as illustrated next.
Alternative experimental representations
d =
3
2
_
1 + 2d

1 d

_
d =
d

1
d

3
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.3.3:2
Preface Geotechnical aspects Soil plasticity
Dilatancy angle
Assuming a plastic potential given by Q = [[
n
tan . The value of can result
from empirical knowledge or be retrieved from experiments as before, using the following
formula,
tan =

p
n

p
t
where
p
n
and
p
t
are normal and tangential plastic strain increments or alternatively:
sin =

p
V

p
max
=

p
1
+
p
3

p
1

p
3
Application with Drucker-Prager plasticity
If is specied, then the same size adjustment as for will be assumed in order to retrieve
a

. For example, assuming a 2D situation for which , C, and size adjustment plane
strain collapse are specied, then introduced into the associated ow option yields:
a

= tan /
_
9 + 12 tan
2

can then in turn be introduced into formulas in order to retrieve k and a

.
In a 3D situation with f, k, and specied, a

can be retrieved from formula,


replacing ; a

and k are, in this case independent of

. These operations are, of course,


done automatically and hidden to the user.
Dilatancy with smooth Hoek-Brown criterion
Flow options available on Hoek-Brown criterion include:
Deviatoric, corresponding to incompressible ow.
Tensile meridian. The ow will be radial in the deviatoric plane and follow the normal
to the tensile meridian in the meridian plane.

c
prescribed (Hoek-Brown ow).

c
is dened as the angle of dilatancy at failure under uniaxial compression
c
=
arctan(d/d) and
_
arctan(
f
t

2f
c
) <
c
< arctan(
1

2
)
_
.
Alternatively
c
= arctan[
1

6
dI
1
d

J
2
; the resulting dilatancy will vary with loading paths:
from coincidence with the loading path under uniaxial tension to
c
under uniaxial
compression.
Following table contains indicative values of dilatancy characteristics
/ d
c
, []
o
Clay 0 0 -
Sand - - -
Gravel - - -
Rock 0.67 1 - -
Concrete - - 4
o
35
o
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.3.3:3
Preface Geotechnical aspects
7.4 INITIAL STATE
COEFFICIENT OF EARTH PRESSURE AT REST, K0
STATES OF PLASTIC EQUILIBRIUM
INFLUENCE OF POISSONS RATIO
COMPUTATION OF THE INITIAL STATE
INFLUENCE OF WATER
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.4:1
Preface Geotechnical aspects Initial state
7.4.1 COEFFICIENT OF EARTH PRESSURE AT REST, K0
K
0
is by denition the ratio of horizontal eective stresses to vertical eective stresses
K
0
=

2
.
The coecient can be determined from triaxial experiments, measured with a pressuremeter
or evaluated using approximate formula. Commonly used formula are given in Window 7-3.
Window 7-3: Coecient of earth pressure at rest, K
0
Normally consolidated soil (NC) (K
0
)
NC
= 1 sin [JAK48]
Overconsolidated soil (OC) (K
0
)
OC
= (K
0
)
NC
OCR

= sin
Conned elastic medium K
0
=

1
Window 7-3
Overconsolidation ratio (OCR)
The overconsolidation ratio OCR is obtained from an oedometric test. It is maximum close
to soil surface and tends to 1 a depth. The identication of OCR is illustrated in
Window 7-4.
Given the void ratio, the water content, a preconsolidation pressure can be associated with
each vertical stress and the corresponding overconsolidation ratio can be computed as illus-
trated in Figs 7-4, 7-4.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.4.1:1
Preface Geotechnical aspects Initial state
Window 7-4: Overconsolidation ratio OCR
Eective stress at depth z :

z
=

z,
initial void ratio:
e
0
=
_

s
1 +w

_
1
where:

s
: unit weight of solid particles
w : water content ratio
Oedometric test
Vertical stresse and preconsolidation pressure as a function of depth
(right).Overconsolidation ratio as a function of stress (left)
Window 7-4
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.4.1:2
Preface Geotechnical aspects Initial state
7.4.2 STATES OF PLASTIC EQUILIBRIUM
The section deals with the case of semi-innite soil mass with horizontal or inclined surface
subjected to gravity load. Coecients of horizontal pressure K fullling dierent stress
criterion are investigated.
MOHR-COULOMB CRITERION
DRUCKER-PRAGER CRITERION
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.4.2:1
Geotech. asp. Initial ... States of plastic ...
7.4.2.1 MOHR-COULOMB MATERIAL
The plastic equilibrium at depth h of a semiinnite soil mass with horizontal surface, con-
sidered as a MohrCoulomb material, subjected to gravity loading is characterized by two
circles in a Mohr diagram (Window 7-5).
The two Mohr circles correspond to Rankine states. The small circle corresponds to the
active Rankine state, the large circle to the passive state.
A cohesionless material is considered rst. Note that the principal stress orientation
coincides with axes 1 (horizontal) and 2 (vertical). The plastic stress state at depth h is

2
= h (vertical)

1
=
3
(assumption)

1
=
A
= K
A
h (horizontal, active state)

1
=
p
= K
A
h (horizontal, passive state)
Stress states such that,

1
= K
0
h with K
A
< K
0
< K
P
are elastic. K
0
is the earth pressure coecient at rest.
From geometrical considerations equations 1 and 2 (Window 7-5) can be derived for K
A
and K
P
are plotted and provide a useful way to dene possible horizontal stress states, given
the friction angle. For an elastic perfectly plastic material, no stress state outside of these
limits is tolerable.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.4.2.1:1
Geotech. asp. Initial ... States of plastic ...
Window 7-5: Rankine state
Mohr diagram of the plastic state of a semi-innite medium: Cohesionless material (left),
Cohesive material (right)
Earth pressure coecient at rest K
0
, for a cohesionless semi-innite soil mass with
horizontal surface
Active state:
K
A
=
1 sin
1 + sin
= tan
2
_
45


2
_
=
1
N

(1)
Passive state:
K
P
=
1 + sin
1 sin
= tan
2
_
45

+

2
_
= N

(2)
Window 7-5
Similar expressions for the horizontal stress are derived in Window 7-5 for a cohesive ma-
terial, from Window 7-6 and geometrical considerations.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.4.2.1:2
Geotech. asp. Initial ... States of plastic ...
Window 7-6: Rankine states for a Cohesive MohrCoulomb material
Active state

1
= K
A
_

2
+
C
tan
_

C
tan
or

1
= K
A

2
(1 K
A
)
C
tan
.
Passive state

1
= K
P
_

2
+
C
tan
_

C
tan
or

1
= K
P

2
(1 K
P
)
C
tan
.
Window 7-6
Given the friction angle, K
A
and K
P
can be read from Window 7-5. Introducing the cohesion
C, f and K
A
or K
P
into the expressions for
1
yields limiting values of the horizontal stress

1
. These values dene the elastic range. The same discussion holds when the seminite
soil mass is loaded on its surface by a uniform load q. In this case
2
= h is replaced by

2
= (h +q).
The presence of a water table can be accounted for similarly. Plastic states are dened in
terms of eective stresses. The limiting values K
A
and K
P
of K
0
are therefore the same for
a saturated medium as for a dry medium.
If the water table is located at a depth d, the upper layer can be viewed as a surface load on
a saturated medium, and K
A
and K
P
are again the same.
Particular situations
If the semiinnite soil mass is limited by an upper surface inclined at an angle the
Mohr diagram is given in Window 7-7.
The stress state at depth h on a plane inclined at angle can be calculated as:
= hcos
2

= hsin cos .
The corresponding point in the Mohr diagram is Z and the Mohr circles corresponding
to active and passive states can be constructed. The values obtained for K

A
and K

P
(K

=
11
/h) are reported in Window 7-7 as functions of the friction angle and the
angle of the slope .
The case of a semiinnite soil mass limited by an upper surface inclined at an angle ,
submerged by water can be solved similarly and the same values of , K

A
, K

P
apply, associated
with

.
The case of an innite slope under conditions of seepage ow leads to the following stress
state on a plane inclined at angle :
= hcos
2

=
sat
hsin cos .
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.4.2.1:3
Geotech. asp. Initial ... States of plastic ...
A Mohr circle can be drawn as in the previous case and the corresponding limiting states can
be calculated. The values of K

A
and K

P
can again be read from Window 7-5 , with tan
replaced by
tan

=

sat

b
tan .
Similar derivations can be performed for the case of a cohesive material leading to the same
expressions for K

A
and K

P
if is replaced by
0
such that:
tan
0
= tan
_

+C/ tan
_
.
For seepage ow

replaces .
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.4.2.1:4
Geotech. asp. Initial ... States of plastic ...
Window 7-7: Rankine states, inclined surface
Mohr diagram for a semiinnite soil mass with inclined surface. Cohesionless soil
Earth pressure coecient at rest K

0
, for a cohesionless semiinnite soil mass with surface
inclined at angle
K

A
=
_
cos
2
cos
_
cos
2
cos
2

_
_
cos
2

_
1
(1 sin cos ) (1)
K

P
=
_
cos
2
cos
_
cos
2
cos
2

_
_
cos
2

_
1
(1 sin cos ) (2)
with
= arcsin
_
sin
sin
_
;
= arcsin
_
sin
sin
_
+
The horizontal stresses corresponding to active and passive states are expressed as:
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.4.2.1:5
Geotech. asp. Initial ... States of plastic ...
active state:
(
11
)
A
= K

A
h
C
tan
passive state:
(
11
)
P
= K

P
h
C
tan
where K

A
and K

P
are read from Window 7-7.
Window 7-7
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.4.2.1:6
Geotech. asp. Initial ... States of plastic ...
7.4.2.2 DRUCKER-PRAGER MATERIAL
Rankine states can be derived similarly for a DruckerPrager cohesionless material, with the
stress state dened previously. With the assumption
3
=
1
, equations (DP1) Window 7-8
are derived. With the assumption
3
= 0.5 (
1
+
2
), relations (DP2) result.
The results obtained for K
A
and K
P
as function of are reported in Window 7-8. It is
observed that the elastic ranges of MohrCoulomb and DruckerPrager materials coincide,
for the assumption that
3
= 0.5 (
1
+
2
).
As before, the presence of a surface load or of a water table can be accounted for, as for the
MohrCoulomb case.
Window 7-8: Rankine states for Drucker-Prager material
Earth pressure coecient at rest K
0
for cohesionless material, MohrCoulomb (MC),
DruckerPrager
3
=
1
(DP1) and Drucker-Prager with
3
= 0.5 (
1
+
2
) (DP2)
K
A
=

3 sin
2 sin +

3
, K
P
=

_
3 sin
_
2 sin

3
(1)
K
A
=
1 sin
1 + sin
, K
P
=
1 + sin
1 sin
(2)
Window 7-8
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.4.2.2:1
Geotech. asp. Initial ... States of plastic ...
If the semiinnite soil mass, always considered as a cohesionless material, is limited by an
upper surface inclined at an angle , values of K

A
and K

P
can again be derived for
both case, i.e.:
K

A
= cos
2

1 Dtan
1 +Dtan
, K

P
= cos
2

1 +Dtan
1 Dtan
with
D = tan
_
arccos
_
sin
6

3 sin
3

3 sin
__
if
3
=
1
D = tan
_
arccos
_
sin
sin
__
if
3
= 0.5 (
1
+
2
) .
The obtained values of K

A
and K

P
(K

=
11
/h) are plotted in Window 7-9 as functions
of the friction angle and the angle of slope , the same assumptions for
3
as before are
made.
It is noted again that the elastic ranges of MohrCoulomb and DruckerPrager materials,
with the assumption of
3
= 0.5 (
1
+
2
), coincide (see Window 7-8 and Window 7-9).
Similar derivation can be performed for the case of a cohesive material. The case of a
horizontal surface can be treated using K
A
and K
P
from Window 7-7. The horizontal
stresses corresponding to active and passive states are obtained from:
Active state:
(
11
)
A
= K
A
h
3C cos

3 + 2 sin
if
3
=
1
(
11
)
A
= K
A
h
2C cos
1 + sin
if
3
= 0.5 (
1
+
2
)
Passive state:
(
11
)
P
= K
P
h
3C cos

3 + 2 sin
if
3
=
1
(
11
)
P
= K
P
h
2C cos
1 sin
if
3
= 0.5 (
1
+
2
) .
Under condition of seepage ow the value of K

A
and K

P
cannot be read directly from
Window 7-9, they need to be derived explicitly. The same remark holds for the case of a
cohesive material with surface inclined at angle .
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.4.2.2:2
Geotech. asp. Initial ... States of plastic ...
Window 7-9: Rankine states. Inclined surface
Earth pressure coecient at rest K

for a cohesionless semiinnite soil mass with surface


inclined at angle . DruckerPrager with
1
=
3
.
Earth pressure coecient at rest K

for a cohesionless semiinnite soil mass with surface


inclined at angle . DruckerPrager with
3
=0.5(
1
+
2
).
Window 7-9
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.4.2.2:3
Preface Geotechnical aspects Initial state
7.4.3 INFLUENCE OF POISSONS RATIO
Let
12
be the plane containing axes 1 and 2 in which the plane strain problem is dened. If
failure is to occur in the
12
plane,
3
must be the intermediate stress; hence the following
condition must apply:

1

3

2
.
Simultaneously, plane strain holds, i.e. for the elastic case:

3
= (
1
+
2
) and 0.5.
The following limiting conditions results:
1.

3
=
1
=
_

1
_

2
2.

3
= 0.5 (
1
+
2
) .
The second one result from the limits of Poissons ratio. If no tectonic stresses are present,
all elastic states lies within these limits.
These conditions are sometimes met a priori by the boundaryvalue problem (e.g. for the box
shaped medium) or by the adopted matching of DruckerPrager/MohrCoulomb criteria,
as e.g. for the elastic matching proposed earlier. Violating these conditions can have a
signicant eect on the solution.
In addition, some choices of material data lead to plastic behavior already under gravity load-
ing. Since this is the most common loading in soil mechanics, it is interesting to investigate
the corresponding limits of elastic behavior. Some situations of special interest are analyzed
next for a cohesionless material.
Case 1: Box-shaped medium under gravity load, dry, matching collapse load.
The box-shaped medium is the default conguration adopted in the program. Combining
plane strain and the boundary conditions associated with the boxshaped medium leads for
the isotropic medium to:

1
=
3
=
_

1
_

2
;
2
= h
Notice that this coincides with a limiting condition established previously for failure to occur
in the plane
12
and to the active Rankine state. This condition is therefore satised a priori.
The invariants corresponding to this stress state are:
I
1
=
_
1 +
1
_

2
= 3 J
2
=
1
3
_
1 2
1
_
2

2
.
In stress space, the corresponding stress point is located on a cone with its vertex at the
origin, characterized by:
a

I
1

_
J
2
= 0
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.4.3:1
Preface Geotechnical aspects Initial state
from which,
3a

J
2

3
_
1 2
1 +
_
.
For a cohesionless soil, a

characterizes the position of the stress point with respect to the


yield surface. Elastic and plastic stress states can be identied as follows:
a

< a

: elastic state
a

= a

: plastic limit
a

> a

: outofbalance state
Using the matching rules discussed earlier with a

replacing a

, relations are established


which dene the elastic limit as a function of and .
Matching the collapse loads corresponding to MohrCoulomb and DruckerPrager criteria,
under plane strain conditions and deviatoric ow, yields the following result:
a

=
sin
3
or sin
y
=

3
12
1+
.
The lower index y denotes yielding. This curve is shown in Window 7-10.
Note that a similar curve can be derived using directly the MohrCoulomb criterion without
any consideration of matching with the DruckerPrager criterion, leading to:
sin
y
= 1 2.
Material data corresponding to a point located below the curve will automatically generate
a plastic state, and a point located above will generate elastic behavior. Dierent situations
can be analyzed in a similar way.
Window 7-10: Inuence of Poissons ratio
Inuence of Poissons ratio (box-shaped medium, cohesionless soil) (left); Inuence of
Poissons ratio (innite slope, cohesionless soil, dry medium) (right)
Window 7-10
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.4.3:2
Preface Geotechnical aspects Initial state
Case 2: Box-shaped medium under gravity load, dry, matching elastic domains.
Matching of orthotropic elastic domains of DruckerPrager and MohrCoulomb criteria yields:

1
=
_
4 + 1
3
_

2
;
2
= h

3
= 0.5 (
1
+
2
) .
The corresponding invariants were computed earlier and (see matching of elastic domains)
ratio a

is such that yielding occurs if:


3a

J
2

=
1 2
2 (1 +)
sin
This curve is plotted in Window 7-10. Material data corresponding to a point located below
the curve will generate a plastic state.
Case 3: Saturated medium.
The same results as before apply to eective stresses.
Case 4: Innite slope at angle , dry, matching collapse loads.
Principal stresses are:

1
= h(1 sin )

2
= h(1 + sin )

3
= (
1
+
2
) = 2h
Corresponding invariants are:
I
1
= 2 (1 +) h
J
2
=
1
3
(h)
2
_
3 sin
2
+ 1 4 + 4
2
_
then
3a

J
2

=
_
3
_
3 sin
2
+ (1 2
2
)
_
2 (1 +)
.
Note that, for failure to occur in plane
12
,
3
must be the intermediate stress. This yields:

1 sin
2
(see Window 7-10 ) .
Similarly, the adjustment of yield criteria for place strain collapse load yields:
3a

=
_
3
_
3 sin
2
+ (1 2
2
)
_
2 (1 +)
> sin .
For each given slope in a cohesionless soil, a curve = f() can be drawn, which char-
acterizes the limit of elastic behavior and, for the given boundary-value-problem, the limit of
stability (Window 7-5).
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.4.3:3
Preface Geotechnical aspects Initial state
Case 5: Innite slope at angle , dry medium, matching elastic domains (Win-
dow 7-10).
Adjustment of yield criteria for coincidence of elastic domains, with
t
= 0.5. The state of
principal stresses is:

1
= h(1 sin )

2
= h(1 + sin )

3
= 0.5(
1
+
2
) = h
The corresponding invariants are:
I
1
= 3h
J
2
=
2
h
2
sin
2

with elastic matching of failure criteria, this corresponds to yield if:


sin > sin .
The stress state in a cohesionless soil, for the adopted adjustment with the MohrCoulomb
criterion, will be elastic if < and plastic if > . Poissons ratio has no inuence in this
particular case.
Case 6: Innite slope at angle , saturated.
The same results as before apply, corresponding to eective stresses.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.4.3:4
Preface Geotechnical aspects Initial state
7.4.4 COMPUTATION OF THE INITIAL STATE
Box-shaped medium in plane strain
For most static plane strain problems the soil halfplane can be conveniently approximated
by a boxshaped medium with smooth lateral boundaries (Window 5.4.9). The particular
stress-strain state which results can easily be derived.
From plane strain and lateral boundary conditions :

3
= 0
3
= (
1
+
2
)

1
= 0
1
= (
2
+
3
)
therefore:

1
=

1

2
The stress-strain elds which result for some typical loading cases are summarized in Win-
dow 7-11.
Gravity eld
As can be seen in Window 7-11 the correct implementation of gravity requires simultaneous
application of and corresponding initial stresses. This combination is characterized by the
capital in this text.
An initial state corresponding to an urban environment can be established using the same
procedure.
Axisymmetric medium
The default boundary conditions for the axisymmetric case are the same as for planestrain.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.4.4:1
Preface Geotechnical aspects Initial state
Window 7-11: Box-shaped medium with smooth lateral boundaries
Box-shaped medium (
1
=
3
= 0)
No APPLICATION OF: YIELDS: WHERE:
1
Deadweight
downwards

2
= h

1
=
3
=

1
h

2
=
h
E
(1
2
2
1
)
hdepth
unit weight
(BOXD1)
2
Initial stress

02

2
= 0

1
=
3
=

1

02

2
=

02
E
(1
2
2
1
)
(BOXD2)
3
Initial stress

01

2
= 0,
1
=
01

3
= 0,
2
= 0
(BOXD3)
4
gravity eld

02
= h

01
= K
0

02

03
= K
0

02
NB: K
0
=

1
by default

2
= h

1
= K
0

02

3
= K
0

02

2
= 0
(BOXD4)
Window 7-11
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.4.4:2
Preface Geotechnical aspects Initial state
7.4.5 INFLUENCE OF WATER
A steady state Darcy ow model only is included, although the pressure b.c. can vary in time
and for each step steady state solution can be obtained.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.4.5:1
Preface Geotechnical aspects
7.5 SOIL RHEOLOGY
Soil is subjected to long term deformations which cannot be avoided. This phenomenon is
called consolidation.
Modern consolidation theories split the deformations into several mechanisms and two time
periods associated with primary and secondary consolidation.
Primary consolidation is dominated by a mechanism of stressinduced seepage ow which
transfers progressively the part of load carried by the interstitial water to the soil skeleton.
During secondary consolidation, after stabilization of primary consolidation, the deformation
is dominated by creep mechanisms.
Creep can be split into volumetric and deviatoric components.
A carreful choice of boundary to avoid meaningless results.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.5:1
Preface Geotechnical aspects
7.6 ALGORITHMIC STRATEGIES
SEQUENCES OF ANALYSES
EXCAVATION, CONSTRUCTION ALGORITHM
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.6:1
Preface Geotech. aspects Alg. strategies
7.6.1 SEQUENCES OF ANALYSES
Most combinations of drivers are possible provided they are meaningful: initial state, stability,
ultimate load, prestress, consolidation, creep, ow. Obviously, an initial state analysis should
come rst; a stability analysis should not be followed by any other type of analysis unless
provisions are taken to restart before the stability analysis; recall that the stability analysis
goes through the change of the material properties.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.6.1:1
Preface Geotech. aspects Alg. strategies
7.6.2 EXCAVATION, CONSTRUCTION ALGORITHM
The excavationconstruction process shows an analogy with loads and associated loadtime
histories. Each element is associated with an existence timehistory which takes values (0 or
1) depending if the element exists at a given time.
It is possible to simulate excavation and construction processes; corresponding restrictions
are specied in the following remarks.
Remarks
1. The initial mesh numbering will be referred to throughout the analysis, for display of results.
It must therefore include all elements appearing during analysis; some may however, be
inactive at the beginning of the analysis.
2. When performing an excavationconstruction analysis, stiness update must obviously be
required at the beginning of each step. The corresponding algorithmic choice must be
done in the input denition.
3. Excavation stages can not be associated with some types of stability e.g. algorithms; this
options would not be meaningful.
4. If an excavation is followed by a time dependent analysis, progressive unloading will occur.
5. Unloading can be controlled using load time functions attached to the elements, named
as unloading functions. The interaction forces of the excavated medium on the sur-
rounding medium can be computed as follows:
F
intEXC
=
_

exc
B
T

tot
d
_
LTF(t)
0, if no unloading function is given
_
where the integration is carried out over the excavated domain. Each excavated element
is associated with a load function which can be used to control progressive unloading.
If no LTF (unloading function) is specied for excavated elements, interaction forces from
excavated media will vanish immediately at the moment of excavation (situation 2a in
the Figure 5). Forces F
intREM
will act as a load in a rst step after the excavation.
If compressive stresses dominated in the area before the excavation then F
intREM
will
generate tensile load around newly created boundary. In the case of elasto-plastic media
this may cause diculties in obtaining converged solution. In that case progressive
unloading after excavation (situation 2b in the Figure 5) will deminish destabilizing
eect of F
intREM
and helps to redistribute stresses in the surroundings of the excavated
domain, and in consequence to obtain convergent solution of the new equilibrium state,
(situation 3 in the Figure 5). The excavation data should include existence function and,
if needed, unloading function. Both are shown in the gure 5
Above numerical procedure, despite being useful in obtaining solution, corresponds to
technical measures normally undertaken at the construction site in order to prevent failure
of a soil mass during the excavation, like temporary supports or spacers.
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.6.2:1
Preface Geotech. aspects Alg. strategies
Simulation of excavation. Events sequence
Functions controlling excavation process
June 25, 2003
Z Soil

-2D-2PHASE v.6
QuickHelp DataPrep Benchmarks Tutorials
TM7.6.2:2

Potrebbero piacerti anche