Rep LargeDeformation
Rep LargeDeformation
Rep LargeDeformation
by
A. Truty
A. Urbański
Th. Zimmermann
1 Introduction 3
2 Co-rotational approach 5
2.1 Co-rotational approach outline . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Activation of large deformations option in Z SOIL code . . . . . . . . . . . 7
2.3 Benchmarks for beam elements . . . . . . . . . . . . . . . . . . . . . . . . 7
2.3.1 Euler problem (beam 2D). Eccentric compression in post-buckling range 7
2.3.2 Curved 3D beam . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.4 Benchmarks for shell elements . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.4.1 Cylindrical shell under point forces. . . . . . . . . . . . . . . . . . . 10
2.4.2 Spherical shell under point force. Displacement control (post-buckling) 11
2.5 Benchmarks for continuum elements . . . . . . . . . . . . . . . . . . . . . 12
2.5.1 Euler problem in 2D . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.5.2 Euler problem in 3D . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Introduction
ZSoilr 2020code is designed to handle large rotations and displacements for all structural
elements like shells, beams, membranes, anchors and continuum and large deformations as
far as contact interface is concerned. The corotational approach is exploited to manage
large rotations and its main benefit is that all the stresses and strains at the integrations
points remain the engineering ones although are given in the rotated local frame. The new
contact formulation, developed to manage really large relative motions of bodies, makes use
of so-called slave-master approach in which contacting node (slave) cannot penetrate the
corresponding master element face (master) by means of penalty formulation enhanced by
Augmented Lagrangian approach (if needed).
CHAPTER 1. INTRODUCTION
Co-rotational approach
3 Deformed configuration
Corotated configuration
4
2
True deformation d
{E n }
1
1 {E 0 } {X }
Initial configuration
Goal:
to perform large displacement/rotation analysis re-utilizing standard geometrically linear
element for:
• beams in 2D and 3D
• shells
• membranes
• truss/anchors
Assumption:
Displacements and rotations attributed to rigid body motion could be arbitrarily large, but
”true deformation” remains within small strain limit
CHAPTER 2. CO-ROTATIONAL APPROACH
Performance:
• deduct rigid body motion from total deformation of an element, then evaluate element
forces and stiffness emerging from -”true” -deformation
• introduce of element frame E rigidly attached to the element. Element processing is
performed with respect to these frame
• deduction of rigid body motion is equivalent to a projection:
δ d̄ = Pδd
f = PT f̄
∂ d̄
P=
∂d
∂PT
K = PT K̄P + f̄
∂d
• consistent treatment of arbitrarily large rotation:
F representation of a rotation by the tensor, use exponential mapping (Rodriguez for-
mula):
Q=eΩ Ω=spin(w)
orthogonal ←− skew ←−
vector w
tensor Q −→ symmetric Ω −→
Ω=log Q w=axial(Ω)
with:
sinw 1 − cos2 w 2
Q =eΩ = I+ Ω+ Ω; w = kwk
w w2
F no additive update, use products of rotation tensors:
Rn+1 = ∆RRn
The geometrical nonlinearity option is activated by the check box Large displacement /
rotations in the bottom part of the dialog box Analysis and drivers under menu Control /
Analysis & Drivers. The check box is active only if the version type is set as Advanced during
Control /Project preselection or in the Analysis and drivers dialog.
Remarks:
• once this option is activated the whole analysis will be run as geometrically nonlinear
• standard contact elements (segment to segment) cannot be used, only large deformation
contact is allowed
• switching ON/OFF this option during restarts will yield computation failure
the moment. The comparizon of the reference solution by Życzkowski) and the numerical
one is shown in the second figure.
(e=0.1 L)
0.1m
N
L=1.0
L=1.0 m
0.8
0.7
0.6
0.5
Uy/L
0.4
0.3
Exact
0.2 Numerical
0.1
0
0 1 2 3 4 5
N / Nc rit
Figure 4:Figure
Euler2.4: Euler problem. Force-displacement diagram
problem: Force-displacement diagram
6
8 ZSoilr .PC 070303 report
2.3. BENCHMARKS FOR BEAM ELEMENTS
4.2 Curved 3D beam
2.3.2 Curved 3D beam
File: Litewka-Wriggers.inp
1.0 in
Cross FZ
1.0 in
section
45
R=100 in
Ux=Uy=Uz=0
Φx=Φy=Φy=0
h = 0.094
R = 4.953 free
free
R = 4.953 Symmetry
Y
z
y
x
E =E 10 .5.5××10
= 10 106
6
ν =ν 0=.03125
.3125
X L / 2 = 5.175
L / 2 = 5.175
imposed displacement
Imposed displacement
Figure
Figure 2.6: Free
6: Free edgecylinder.
edge cylinder. TheThe
data data
70000
P [kN]
60000
50000
40000 Z_SOIL
10000
U [m]
0
0 0.5 1 1.5 2 2.5 3
Figure Figure
7: Free2.7: edge cylinder.
Free edge Displacement
cylinder. Displacement historyhistory
E = 68950
E = 68950 N/m2
ν = 0.3
ν = 0.3
L / 2 = 0.7849
L / 2 = 0.7849 m
w
= 2.54
Rw h = 0.09945
R = 2.54 m h = 0.09945 m
70
P [N]
60
50
40
30
20 Z_S OIL
10 w [m]
CAM8x8x16, ref
0 Chróś cielews ki
0 0.1 0.2 0.3 0.4
Figure Figure
9: Clamped spherical
2.9: Clamped sphericalshell. Load-displacement
shell. Load-displacement graph graph
Material imperfection
L/H=10
ML
MR
B
A
Nonsymmetric contact: only Interface1 defined
Slaves-L ={A,B}
or Interface2
Slaves-R={C,D}
Symmetric contact: both Interface1 and
Master-L={ML}
Interface2 exist
Master-R={MR}
Interface1 = { Slaves-L,Master-R}
Interface2 = { Slaves-R,Master-L}
ML
MR
B
A
Nonsymmetric contact: only Interface1 defined
Slaves-L ={A,B}
or Interface2
Slaves-R={C,D}
Symmetric contact: both Interface1 and
Master-L={ML}
Interface2 exist
Master-R={MR}
Interface1 = { Slaves-L,Master-R}
Interface2 = { Slaves-R,Master-L}
MR
B
A
Nonsymmetric contact: only Interface2 is
Slaves-R={C,D}
possible
Master-L={ML}
Interface2 = { Slaves-R,Master-L}
Continuum 3D vs Continuum 3D
A E
B F
H
D
C G
ML MR
Nonsymmetric contact: only Interface1 defined
Slaves-L ={A,B,C,D}
or Interface2
Slaves-R={E,F,G,H}
Symmetric contact: both Interface1 and
Master-L={ML}
Interface2 exist
Master-R={MR}
Interface1 = { Slaves-L,Master-R}
Interface2 = { Slaves-R,Master-L}
slaves
Master faces
Continuum-3D vs Shell
A
B Here symmetric contact
is possible
D
C +
A
Here symmetric contact
B
is not possible
The second approach is more general but it yields some limitations on slave-master setting.
To explain the general idea lets us consider an example of excavation followed by lining
construction and then filling. The aim of the simulation is to to perform an excavation and
then construction of a lining followed by seven stages of filling (see Fig. 3.7). After an
excavation we get some deformations which we will neglect when a tunnel lining is built.
To do that the program memorizes total deformation Ue . This is done automatically (see
Fig.3.8). In the next step the tunnel lining is added and its initial configuration is assumed
to be undeformed (just before construction) regardless nonzero deformation at nodes at the
bottom slab of the lining. This can be managed thanks to the memorized deformation Ue .
The major problem appears when we begin to add fill material. This is so because newly
added fill should satisfy contact kinematics if it touches already deformed lining (due to its
own dead weight and/or loading imposed during previous fill steps)and existing deformation
on the remaining boundaries (see Fig.3.9). If we consider the situation shown in Fig.3.10 we
can notice that the initial undeformed mesh in the zone of fill (stage I) must be mapped onto
deformed configuration caused by a construction of a lining. Hence the boundary nodes along
section A-B must satisfy the contact kinematics (cannot penetrate the lining and cannot be
separated from it), nodes along the boundary A-D and D-E must fit current deformation
equal to U − Ue (the one corresponding to the settlement caused by lining construction),
and nodes along section E-B must remain at the initial elevation.
I
VII
H
VI
G
V
C
F
II IV
E B
I III
D
A
This mapping is made in the code by a finite element solution of a sub-domain subject to
the imposed boundary displacements. As the result we get a shift to the nodal coordinates
of all newly added continuum elements (NB. in this finite element sub-problem we assume
artificial elastic constants E = 1.0 and ν = 0.0). However, to make this mapping, all nodes
along the section A-B, being part of the contact interface, must be slave nodes (!). It
should be emphasized here that incremental deformations during single fill step should be
small otherwise strain incompatibilities along the section A-D can cause stress oscillations
(although total deformation caused by filling can be large).
Lining
Fill layer
Deformation increment
after construction of a lining
D A
boundary nodes
Mapping on Deformed lining
deformed
To summarize the following general rules are used in the second approach:
• structures (beams/truss/membranes/shells) are always added in undeformed configuration;
hence, whenever new structural element is added its initial total deformation must be
memorized as U os (at the element level); the current structure deformation is always equal
to U − Uos
• continuum is added after mapping to the deformed configuration
• at the end of the excavation time step the current total deformation corresponding to that
state is memorized as Ue for all nodal points
NB. Activation of the approach II has be done in the dialog Control /Analysis and drivers
by switching the Large displacement /rotations ON. Then under Settings, in the appearing
dialog checkbox Update coordinates during costruction has to be set ON, otherwise algorithm
I will be performed
Once we have disconnected the mesh along the contour we can generate contact interface in
the following way
1. highlight contact element contour (edges which will play a role of so-called masters) (see
Fig.(3.16), and create masters with a label ”masters”
2. select part of the lining (beam elements) (see Fig.(3.17)and then select nodes from selected
beam elements
3. create group of slave nodes (contactors) with a label ”slave nodes on beam”
4. using an option Create\Update contact elements define the interface by merging the pair
”slave nodes on beam”-”masters” (see Fig.(3.18, Fig.(3.19)
EXF=5
EXF=4
EXF=3
EXF=1
EXF=2
Orientation of beams
Figure 3.24: Setting continuity interface along initial position of the lining contour
Data preparation in this case is somewhat different compared to the standard contact interface
setting known for small deformation applications. In the large deformations regime slave nodes
may interact with different master faces during the analysis hence explicit setting of so-called
contact elements is not possible. In general we may associate the attribute ”contactor node”
with any node in the mesh and another attribute which is the ”master face” with any face
of the finite element (continuum, shell, beam (only 2D) or membrane).
q=F/2R
R=8m
E=500 kPa, ν=0.3
E=`
As the foundation is to be rigid we can generate the mesh as shown in figure 3.26. However,
contact stress recovery is not that easy matter in the node-segment contact implementations.
Hence we can generate different mesh for the foundation with which, using standard tools
available in the postprocessor (cross sections through the mesh), we can easily recover the
interesting values. This second mesh is shown in figure 3.27.
The comparison of normal stresses in the interface for two levels of the force F = 50 kN/m
F
and F = 100 kN/m (q = ) is presented in figure 3.28.
2R
Figure 3.27: Hertz problem in plane strain format - dense mesh in the foundation
Figure 3.29: In menu INTERFACE (LARGE DEFORMATIONS) create group of slave nodes
(contactors)
Figure 3.30: In menu INTERFACE (LARGE DEFORMATIONS) create group of master faces
Figure 3.31: Definition of the interface for Hertz problem. In menu INTERFACE (LARGE
DEFORMATIONS) select one group of slaves and one group of master faces and add to list
of interfaces
Figure 3.32: Finite element meshes for two wheels and foundation
Figure 3.44: Deformation at stage of separation of the wheels from the foundation (at central
point)
Figure 3.45: Deformation at stage of advanced separation of the wheels from the foundation
REFERENCES
1. C.C.Rankin and B.Nour-Omid: The use of projectors to improve finite element perfor-
mance. (Computers & Structures ,1988)
2. C.C.Rankin and B.Nour-Omid: Finite rotation analysis and consistent linearization using
projectors. (Computer Methods in Applied Mechanics and Engineering,1990)
3. M.A.Chrisfield and G.F.Moita: A unified co-rotational framework for solid shells and beams.
(International Journal for Solids and Structures, 1996)
4. B.Skallerud and B.Haugen: Collapse of thin shell structures. Stress resultant plasticity
modeling and finite element formulation. (International Journal for Numerical Methods in
Engineering, 1999)
5. H. Parisch and Ch. Lu̇bbing. A formulation of arbitrarily shaped surface elements for three-
dimensional large deformation contact with friction. (International Journal for Numerical
Methods in Engineering, vol. 40, p.3359-3383, 1997)
6. P. Papadopoulos and R. L. Taylor. A mixed formulation for the finite element solution of
contact problems. Computer Methods in Applied Mechanics and Engineering 94 (1992) p.
373-389.