Vous êtes sur la page 1sur 34

Domain Decomposition Techniques to Couple

Elasticity and Plasticity in Geomechanics

By Horacio Florez, M.Sc., M.S.E.

Research work carried out under the supervision of


Dr. Mary F. Wheeler

The Center for Subsurface Modeling


The University of Texas at Austin

CSM Industrial Affiliates Meeting, Austin, October 14, 2009


Outline

 Summary and motivation

 Model problems
 Elasticity and loose coupling with flow (*)
 Computational plasticity

 State-of-the-art Domain Decomposition schemes in


geomechanics:
 Dirichlet-Neumann (*)
 Mortar FEM

 Numerical experiments in geomechanics on multi-


core processors (*)

 Concluding remarks, future work and references


Part of summer internship work at The ConocoPhillips Company (*)
Summary

We’re going to discuss 3


1   2   3
popular Domain Decomposition
Von Mises
schemes to couple elasticity
and plasticity in geomechanics.
The algorithm of
2
computational plasticity with Druker-Prager

popular failure criteria will be 1

discussed. Top: The Druker-Prager and Von


Mises yield surfaces are depicted.
We present the
Left: A plasticity front starts to
computer implementation with propagate due to the fact that the

FEM and preliminary 2-D material is yielding right there.

results.
Motivation: Domain Decomposition
Elasticity and plasticity in geomechanics: iterative coupling
Reservoir level Near borehole
Outline

 Summary and motivation

 Model problems
 Elasticity and loose coupling with flow (*)
 Computational plasticity

 State-of-the-art Domain Decomposition schemes in


geomechanics:
 Dirichlet-Neumann (*)
 Mortar FEM

 Numerical experiments in geomechanics on multi-


core processors (*)

 Concluding remarks, future work and references


Part of summer internship work at The ConocoPhillips Company (*)
Isotropic Elasticity

 FEM formulation:
     b , in  ; u  0, on D ;   nˆ  t , on N
au, v   l v 
au, v     v  : C :  u d ; l v    t  v ds   b  v d
T

  N 

K u  f
ext

K   B C B  d ; f   t   ds    b  d
T ext T T

  N 

Fig. Applications include borehole


stability and sand production,
reservoir compaction and subsidence
(loose coupling), among others
Loose Coupling: Reservoir Compaction

 Small perturbation assumption :

      b       0 
 b    g ; u  0, on D ;   nˆ  t , on N
      0 , in  ;       C :  -  p f
0

K   B C B  d ; f   t   ds   B   p f  d
T ext T T

  N 

 1st problem: at initial equilibrium, conditions are often unknown


 2nd problem: the stress changes due to pressure drop and changes in
tractions; no changes in body forces
 Superposition principle: solve 2nd problem without knowing the 1st one

(*) Part of these ideas come from a geomechanics course under Dr. Yves Leroy
Computational Plasticity Algorithm

 Rate-independent plasticity (small deformation):

   e   p
  
 p

  C :    ; C      2 
 p   N  , q  ; q   H  , q 

 Path-dependent materials :

f u  f ext  0
int

f u    B   d   t   ds    b  d
int T ext T T
; f
  N 

 n 1  ˆ  n ,  n1  ;  n1  ˆ  n ,  n1 


Computational Plasticity Cont.

 The incremental boundary value problem:

r u n 1   fun1   f extn1  0
int

f u n 1    B ˆ  n ,  u n1   d ; f n1   t n 1   ds    b n 1  d
int T ext T T

  N 

t n 1  n 1 t ; b n 1  n 1 b  f n 1  n 1 f
ext ext

 Non-linear solution: The Newton-Raphson scheme

K T  u
k 1
 
 r u n 1
k

k 1 k 1
u n 1  u n 1   u
k

r ˆ
KT    B D B  d ; D 
T ep ep

 u n 1 u kn1 
 n 1  kn1
Plasticity Return Mapping Algorithm

 Fully implicit elastic predictor/return-mapping:

1. Elastic predictor:
trial trial
 
 en trial   n 1   ;  n 1   n ;  n 1   e
e trial trial
; q n 1  
trial

 
1
n 1 n 1

2. Check plasticity admissibility (yield criterion)


 trial trial

IF f  n 1 , q n 1  0 THEN set n 1  n 1 and then EXIT
trial

3. Return- mapping. Solve the non- linear system


 en 1   en trial
1   N n 1
 0
   
 n 1   n 1   H n 1   0
trial


 f  n 1 , q n 1

 0
  
for  n 1 ,  n 1 , and 
e

4. EXIT
Outline

 Summary and motivation

 Model problems
 Elasticity and loose coupling with flow (*)
 Computational plasticity

 State-of-the-art Domain Decomposition schemes in


geomechanics:
 Dirichlet-Neumann (*)
 Mortar FEM

 Numerical experiments in geomechanics on multi-


core processors (*)

 Concluding remarks, future work and references


Part of summer internship work at The ConocoPhillips Company (*)
Domain Decomposition
Dirichlet-Neumann (DN)

 Lu1  f in Ω1 1
 k 1
u1  0 on Ω1  Ω
 u k 1  k on 
 1
 Lu2  f in Ω2
 k 1
u2  0 on Ω2  Ω
  u k 1   k 1 on 
 n 2
  
k 1   D  u2 k 1  1   D  u1 k 1 


  
 nu2 k 1   N   nu1 k 1  1   N   nu2 k 

 Matching grids
2
 Coloring algorithm (3 colors tool)
 Over-relaxation is important
 Iterative coupling by the BC’s
DN: Coloring and Algorithm

1. White guys (D-guys) go first

2. Hybrid grey guys proceed


after D-guys

3. Black ones can now go

4. Feedback displacements to
white and grey ones and go to
step 1 if there is a residual in
the tractions, stop if not

mn

i j k Fig. General partitioning will require a three-color tool, hybrid


sub-domains show up for touching both D- and N- guys
Speedup for Isotropic Elasticity
Kirsch’s Benchmark Problem
 Quadrilatheral mesh, nn = 13564, ne =
FEM P1 FEM P2
13280, Intel® Xeon® Processor E5440,
2.83 GHz, Quad Core (Harpertown)

Dirichlet-Neumann

8.00

6.00  xx
S 4.00 5.48 6.12 6.84
4.64
2.00 3.69
2.80
1.99
1.00
0.00
1 2 3 4 5 6 7 8

IPFA
Number of Processors
Linear

(*) Part of internship work at The ConocoPhillips Company


Exact FEM Q1
Mortar FEM Method: Elasticity

 Nonconforming discretizations:

a u, v     v  : C :  u d ; l v    t  v ds   b  v d
T

  N 

bu, v    u  v d ; u  u  u  1 2 



2
a u h , v h   bv h ,  h   l v h 


b u , 
 h h  0

1
 
 k 1 B   
0 1 T
  u 1   l 1 

 0 k    B   
2 2 T   2    2  
  u   l 
   B    0
Fig. Non-matching interfaces and
 B 1 2     
      hanging-nodes are treated properly
Outline

 Summary and motivation

 Model problems
 Elasticity and loose coupling with flow (*)
 Computational plasticity

 State-of-the-art Domain Decomposition schemes in


geomechanics:
 Dirichlet-Neumann (*)
 Mortar FEM

 Numerical experiments in geomechanics on multi-


core processors (*)

 Concluding remarks, future work and references


Part of summer internship work at The ConocoPhillips Company (*)
Reservoir Cross-Section: Plane Strain
Boundary conditions and conforming mesh


  nˆ  0

ux  0 ux  0
Fy  0 Fy  0

u y  0, Fx  0

Fig. The pressure field comes from a 100 x 20 black-oil model, the tensor product mesh propagated in the
surroundings is quite inefficient and requires a non-matching treatment
FEM Solution: Conforming Mesh Case
Vertical displacement contour

Fig. The FEM solution shows compaction (in blue) and build-up (in red)
Reservoir Cross-Section: Mortar Case
Boundary conditions and non-conforming mesh


  nˆ  0

ux  0 ux  0
Fy  0 Fy  0

u y  0, Fx  0

Fig. The same tensor-product mesh is used in the pay-zone while the surroundings are meshed with
Delaunay triangulations. The goal is to reduce the computational cost
FEM Solution with 4 Mortars
Vertical displacement contour

Fig. The mortar solution reproduces the same features in the displacement field but the
computational cost was reduced by 50% because of the efficient meshing
Strip-Footing: Plasticity Example

Von Mises yield criterion


P  f  3 J 2   y  p 
  nˆ  0
 y  848.7 KPa
E  10 7 KPa ;   0.48
   x ,  y , xy 
ux  0 ux  0    x ,  y ,  xy 
Fy  0 Fy  0

u y  0, Fx  0

This problem allows determining the bearing capacity


(limit load) of a strip footing before collapsing
Strip-Footing: FEM Solution
Vertical displacement contour and plasticity front

Fig. 1 The elastic trial (top) and the plastic converged


(bottom) solutions are shown for a given load Fig. 2 The plasticity front propagates during
increment the incremental loading process
Outline

 Summary and motivation

 Model problems
 Elasticity and loose coupling with flow (*)
 Computational plasticity

 State-of-the-art Domain Decomposition schemes in


geomechanics:
 Dirichlet-Neumann (*)
 Mortar FEM

 Numerical experiments in geomechanics on multi-


core processors (*)

 Concluding remarks, future work and references


Part of summer internship work at The ConocoPhillips Company (*)
Concluding Remarks

We have presented:
1. Parallel Finite Element CG-Code was developed and tested
on benchmark problems
2. Domain Decomposition techniques for coupling elasticity
and plasticity with DN and mortars
 Scalable speedup obtained for elasticity on 8
processors with DN
 Scalable speedup achieved for plasticity up to 4 cores
(multi-threaded ensemble of tangent matrix)
Future Work

1. Further testing on Linux cluster machines like Bevo, Lonestar,


and Ranger

2. Implement other popular failure criteria such as Druker-Prager


and Cam-Clay

3. Benchmarking with both research and commercials codes


such as HYPLAS, FEAP, Abaqus, etc.

4. Incorporate more physics into the FEM-code: thermal stresses


and coupling with the energy equation

5. We have to try with both Discontinuous Galerkin (DG)


References: Domain Decomp.

1) Toselli, A. and Widlund, O., 2005, “Domain Decomposition Methods – Algorithms and Theory”, Springer
Series in computational Mathematics, New York, USA.
2) Quarteroni, A. and Valli A., 1999, “Domain Decomposition Methods for Partial Differential Equations”,
Numerical Mathematics and Scientific Computation , Oxford University Press, New York, USA.
3) Girault, V., Pencheva, G., Wheeler, M. and, Wildey, T., 2009, “Domain decomposition for linear elasticity with
DG jumps and mortars”, Comput. Methods Appl. Mech. Engrg., 198 (2009) 1751-1765.
4) Girault, V., Pencheva, G., Wheeler, M. and, Wildey, T., 2009, “Domain decomposition for poro-elasticity with
DG jumps and mortars”, in preparation.
5) Badia S. et al, 2009, “Robin-Robin preconditioned Krylov methods for fluid-structure interaction problems”,
Comput. Methods Appl. Mech. Engrg., 198 (2009) 2768-2784.
6) Discacciati M., et al., 2001, “ROBIN-ROBIN DOMAIN DECOMPOSITION FOR THE STOKES-DARCY COUPLING”,
SIAM J. NUMER. ANAL., Vol. 45, No. 3, pp. 1246-1268.
7) Hauret, P. and Le Tallec, P., 2007, “A discontinuous stabilized mortar method for general 3D elastic problems”,
Comput. Methods Appl. Mech. Engrg., 196 (2007) 4881-4900.
8) Flemisch B., Wohlmuth, B. I., et al., 2005, “A new dual mortar method for curved interfaces: 2D elasticity”, Int.
J. Numer. Meth. Engng. 2005, 68:813-832.
9) Hauret, P. and Ortiz, M., 2005, “BV estimates for mortar methods in linear elasticity”, Comput. Methods Appl.
Mech. Engrg., 195 (2006) 4783-4793.
References: Plasticity

1) Neto, E. A. et al, 2008, “Computational methods for plasticity : theory and applications”, Wiley, UK.
2) Simo, J. C. and Hughes T.J.R., 1998, “Computational Inelasticity”, Springer, Interdisciplinary Applied
Mathematics.
3) Lubliner, J., 1990, “Plasticity Theory ”, Dover Publications, Inc., New York.
4) Zienkiewicz, O. C. and Cormeau, I.C., 1974, “VISCO-PLASTICITY AND CREEP IN ELASTIC SOLIDS- UNIFIED
NUMERICAL SOLUTION APPROACH”, International Journal of Numerical Methods in Engineering , Vol. 8, pp.
821-845.
5) Cormeau, I.C., 1975, “NUMERICAL STABILITY IN QUASI-STATIC ELASTO/ VISCO-PLASTICITY”, International
Journal of Numerical Methods in Engineering , Vol. 9, pp. 109-127.
6) Hughes, T.J.R. and Taylor, R. L., 1978, “UNCONDITIONALLY STABLE ALGORITHMS FOR QUASI-STATIC ELASTO/
VISCO-PLASTIC FINITE ELEMENT ANALYSIS”, Computers & Structures, Vol. 8, pp. 169-173.
7) Simo, J. C. and Taylor, R. L., 1985, “CONSISTENT TANGENT OPERATORS FOR RATE INDEPENDENT
ELASTOPLASTICITY”, Computer Methods in Applied Mechanics and Engineering, Vol. 48, pp. 101-118.
8) Simo, J. C. and Taylor, R. L., 1986, “A RETURN MAPPING ALGORITHM FOR PLANE STRESS ELASTOPLASTICITY”,
International Journal of Numerical Methods in Engineering, Vol. 22, pp. 649-670.
9) Wilkins, M.L., 1964, “Calculation of Elasto-Plastic Flow”, In Methods of Computational Physics 3, eds. , B.
Alder et. al., Academic Press, New York.
10) Clausen, J., et al., 2007, “An efficient return mapping algorithm for non-associated plasticity with linear yield
criteria in principal stress plane”, Computers & Structures, Vol. 85, pp. 1975-1807.
References: Poroelasticity

1) Kim, J. et al., 2009, “Stability, Accuracy and Efficiency of Sequential Methods for Coupled Flow and
Geomechanics”, SPE Paper 119084.
2) Liu R., 2004, “Discontinuous Galerkin Finite Element Solution for Poromechanics”, PhD thesis, The University
of Texas at Austin .
3) Gai X., 2004, “A Coupled Geomechanics and Reservoir Flow Model on Parallel Computers”, PhD thesis, The
University of Texas at Austin .
4) Han G. et al., 2002, “Semi-Analytical Solutions for the Effect of Well Shut Down on Rock Stability”, Canadian
International Petroleum Conference, Calgary, Alberta .
5) Chen Z, et al, 2006, “Computational Methods for Multiphase Flows in Porous Media, SIAM, pp. 57; 247-258 .
6) Du J, and Olson J., 2001, “A poroelastic reservoir model for predicting subsidence and mapping subsurface
pressure fronts”, Journal of Petroleum Technology & Science, Vol. 30, pp. 181-197.
7) Grandi, S. and Nafi M., 2001, “Geomechanical Modeling of In-situ Stresses around a Borehole”, MIT,
Cambridge, MA.
8) Charlez A., 1999, “The concept of Mud Window Applied to Complex Drilling”, SPE Paper 56758 .
Contact Us:

Visit us:
http://www.ices.utexas.edu/subsurface/

e-Mail: florezg@gmail.com

Any Questions?

Thanks for your attention End of presentation


Rate Independent Plasticity

We just follow the approach by Simo and Hughes (1998) and Lubliner (1990):

   e   p
 
  C :    p ; C      2 
  E 1  ;   E 1  1  2 
Elastic domain and yield criterion:

 
e :  , q  S x m | f  , q  0   
Flow rule and hardening law:
 p   N  , q 
 
q   H  , q
Kuhn-Tucker complementary conditions:

  0, f  , q   0, and   f  , q   0
Rate Independent Plasticity

Interpretation of the Kuhn-Tucker complementary conditions:

  
f  0   , q  int E     0 (Elastic)
  f  0    0 (Elastic unloading)

 f  0    E   f  0,   0 (Neutral unloading)
  
  f  0,   0 (Plastic loading)
 
f  0
Consistency condition and elastoplastic tangent moduli:

f f
f  :    q
 q

f 
f

 
: C    
p f
q
 q

 
f  f : C :     f : C : N  f H   0
    q 

Rate Independent Plasticity

Assumption for the flow rule, hardening law, and yield condition satisfy:

  
f : C : N  q f  H  0 
 f : C :  x x
f  0    ; x 
 f : C : N   q f  H 2

Finally the so called tensor of tangent elastoplastic moduli becomes:

 
  C :    p  C :    N   C ep : 
 C if   0
 C : N   f : C
C 
ep
C if   0
  f : C : N   q f  H

For the special case of associative flow rule we have:

 
N  , q   f  , q  
Failure Criteria

We just follow the approach by Zienkiewicz and Cormeau (1974) and Hughes (1978):

f  ,    f  ,    y k   0
 f  Q
 p     
 f 0  
Q  f  associativ e plasticity
Q  f  non  associativ e plasticity
 x   x n
The visco-plastic strain rate law:
f  f  m , J 2 , J 3 
 f  Q Q
 p      
 0
f  
Q Q  m Q J 2 Q J 3
  
  m  J 2  J 3 
Q  Q
 
   m
M0   
Q
J 2
MI 
Q
 
J 3
M II  

Druker-Prager Yield Surface

Common expressions are given by:

6sin   6  c  cos 
f   m  3 J2 
3  sin   3  sin  
  0  Von Mises criterion
f  3 J2  2c  J2  y ; 3
1   2   3
y  Uniaxial yield stress

Von Mises

2
Druker-Prager

1

Vous aimerez peut-être aussi