1. Complementarity Models for Rolling and Plasticity Mihai Anitescu Argonne National Laboratory At ICCP 2012, Singapore With: A. Tasora (Parma), D. Negrut (Wisconsin), and S. Negrini (Milan)

2. 1. DVI MODELS, STATE OF RESEARCH

3. Nonsmooth contact dynamics—what is it?  Differential problem with variational inequality constraints – DVI  Truly, a Differential Problem with Equilibrium Constraints Friction Model Newton Equations Non-Penetration Constraints Generalized Velocities

4. Granular materials: abstraction: DVI  Differential variational inequalities Mixture of differential equations and variational inequalities.  Target Methodology (only hope for stability): time-stepping schemes.

5. Smoothing/regularization/DEM  Recall, DVI (for C=R+)  Smoothing  Followed by forward Euler. Easy to implement!!  Compare with the complexity of time-stepping  Which is faster?

6. Simulating the PBR nuclear reactor  Generation IV nuclear reactor with continuously moving fuel.  Previous attempts: DEM methods on supercomputers at Sandia Labs regularization)  40 seconds of simulation for 440,000 pebbles needs 1 week on 64 processors dedicated cluster (Rycroft et al.) Simulations with DEM. Bazant et al. (MIT and Sandia laboratories).

7. Simulating the PBR nuclear reactor  160’000 Uranium-Graphite spheres, 600’000 contacts on average  Two millions of primal variables, six millions of dual variables  1 CPU day on a single processor…  We estimate 3CPU days, compare with 150 CPU days for DEM !!!

8. An iterative method  Convexification opens the path to high performance computing.  How to efficiently solve the Cone Complementarity Problem for large-scale systems?  Our method: use a fixed-point iteration (Gauss-Seidel-Jacobi)  with matrices: ..and a non-extensive separable projection operator onto feasible set KT=KT=

9. 2. Plasticity Models

10. Need for plasticity models  This is a very important effect in materials.  It is also true for granular materials where it is about phenomena of cohesion, friction, compliance and plasticization.  Think of the effect of water bridges and cohesion in powders such as appear in medication.

11. Type of plasticity models in 1D.  Plasticity elicits a different constitutive law compared to rigid contact  Force-displacement relationship at a contact a) Rigid b) Compliant c) Nonlinear d) cohesion e) Cohesion + Compliance f) Cohesion + Compliance + plastic

12. The yield surface.  Standard rigid contact 1-a can be turned to cohesive one by the transformation:  In addition, rule in 1-d satisfies (at contact):  In this context is called the yield surface.  The displacement is normal at the Yield surface, such constitutive rules are called associative.

13. Example: Friction and Cohesion in 3D by playing around with Yield surfaces  Shift the Coulomb cone downward and make it an associative yield surface.

14. Modeling plasticity  The yield surface give by Coulomb needs to be modified as no infinite reaction is now allowed (crushing)

15. The model  Separate elastic and plastic displacement:  The elastic part of the force allows to compute the force at the contact if the plastic part of the displacement is known.  Except, of course, the plastic part is NOT known. But now we use the associated plasticity hypothesis to constrain the plastic displacement evolution with a variational inequality.  Mathematically, it is the same idea with normal velocity at the contact; EXCEPT that the containing set is no longer a cone, proper (it can be a shifted cone, or just some convex set. So we can time-step and close it.

16. Time-stepping  After some notation  After this is solved, all we have to do is to solve the plasticity displacement  Use the velocity update  Replace in the plasticity velocity equation; to obtain the variational inequality  Over the cartesian product of cones:

17. Damping and VI  We can similarly accommodate damping  And obtain the VI  Where  And  Once we compute new contact impulse, we compute velocity, and update position, and iterate.

18. How do we solve it:  There is no need to change the PGS algorithm as the matrix N is still symmetric positive definite.  Only the projection over more complex sets needs to be implemented, but the problem and the algorithm have the same abstraction; that is.  We are expending the proof to the general case of convex sets, we do not expect problems.

19. Numerical Experiments  Compaction and shear test of a granular media.  Advanced cases of earth-moving machines (bulldozzers, vehicles) need such things to understand soil reaction.  Configuration: soil sample put in 0.1m x 0.1 mx0.2 m,  Top part is pulled by a “drawer” after a mass is dropped on top; shear force is measured and compared with experiment.

20. Results.  Configuration: Getting parameters is hard; Normal and tangentia compliances are 0:910e-7 m/N and damping coefficient is = 0.1. Sphere distribution.  Evolution of the shear force and vertical force. These will be measured from experiments. Note spike in forces – including shear! When load is dropped (probably some lock-in before it relaxes and pushes on the drawe)

21. The advantage of a simulator is that we can have a deeper understanding of the forces.  Evolution of the contact network. Note the “rolling” behavior.  Calibration and experiments are in progress.

22. Nonassociative plasticity  However, not all plasticity is associative.  How do we do things in the nonassociative case?  Extension  H is positive definite and symmetric, but N is only symmetric.  Do we have existence?

23. Existence for non-associative plasticity  Consider the eigendecomposition of H  We then have that the nonassociative plasticity problem is equivalent to: the following problem over a rotated cone:

24. Existence result  We use results from Kong, Tuncel and Xiu to obtain:  In addition, we can solve the problem by continuation:  P_0 guarantees the path exists; R_0 that it accumulates to a solution.  What to do about stationary iterations: unclear

25. To do  Lots and lots of things.  Extend the nonassociative plasticity to general compact convex sets  Extend projected iteration to the general convex sets and nonassociative plasticity  Simulations for cohesion and plasticity.  Do physical experiments to compare; choose the right parameters.

26. 3. Rolling Friction Models

27. Need for rolling friction models.  Rolling friction models important phenomena in computational mechanics  A very important example is the one of tires: rolling friction is critical important (as forward motion in vehicles is due to rolling friction)  The configuration of stacked granular materials is also critically dependent on rolling friction.  Ball bearing performance is affected by rolling friction,  …  In the DVI space, rolling friction was approached before (Leine and Glocker 2003), here we discuss a related approach but also other issues such as convexification.

28. Phenomenon  A constant force T is needed in the center to keep a ball rolling at constant speed.  That is equivalent to a resistive momentum M.  In a DVI framework:

29. Components of the rolling friction  The contact plane has a normal vector and consists of two tangential vectors  Force between bodies at contact, has a normal component and a tangential component  Torque between bodies at contact, has a normal component and a tangential component.  Dual motion quantities: tangential velocity, normal rotation and tangential rotation:

30. Model: rolling, spinning, sliding  Sliding friction, friction coefficient  Rolling coefficient,  Spinning: Spinning coefficient,

31. Restating the model to expose the maximum dissipation principle  This exposes the conic constraints:

32. Time-stepping with convex relaxation.  The reaction cone:  Define total cone (inclusive of bilateral constraints) and its polar:  Using the virtually identical sequence of time stepping, notations and relaxation we obtain the cone complementarity problem.

33. General: The iterative method  ASSUMPTIONS  Under the above assumptions, we can prove convergence.  The method produces in absence of jamming a bounded sequence with an unique accumulation point  Method is relatively easy to parallelize, even for GPU! Always satisfied in multibody systems Use  overrelaxation factor to adjust this Essentially free choice, we use identity blocks

34. Efficient Computation of the Projection over Intersection of Coupled Quadratic Cones  Structure of the variable  Structure of one cone:  Structure of the projection:  Key observation: For each cone the direction of the projected vector is known except for first component there exists 0 <= t_i <= 1 such that :

35. Efficient Computation of the Projection over Intersection of Coupled Quadratic Cones  In the end, we obtain  We can solve in each t_i easily to obtain:  The function is piecewise quadratic and convex (the graph of each component is a convex parabola union with a flat line with matching derivative.  We can compute the answer efficiently.

36. Validation: linear guideway with recirculating balls  We get close matching to analytical result (we allow a bit of compliance to resolve indeterminacy).

37. Rolling Friction Effects on Granular Materials  Converyor belt scenario: Question can we reproduce rolling friction effects over the repose angle of piles (observed)?  Yes, for same sliding friction the results are different.

38. Angle of repose.  Dependence on rolling friction coefficient – confirms trends in other experiments.

39. Conclusions  We presented extensions of DVI time-stepping schemes to rolling contact and elasto-plastic contact.  These include convex relaxation and algorithmic considerations.  The method is implemented in ChronoEngine.  Plasticity and in particular Non-associative plasticity open the need for more sophisticated complementarity VI models and analysis. (R_0 matrices over convex sets?).  This project is very much in progress.