Consider the scenario below. An elastic-plastic rectangular block with circular cut-outs is fixed at the bottom and extended from the top to achieve 25% global strain. Plane-strain has been used in the out of plane direction, and the specimen is prevented from lateral motion at the upper and lower boundaries. A compressible Neo-Hookean elasticity model was used, with a von Mises yield criteria and linear hardening. Dimensions and material properties are given below. A grid cell size of 0.0025 m was used, with 2 particles/cell in each direction.
Non-physical fluctuations in the von Mises stress and the plastic strain, are shown below. Higher particle density and XPIC improve these results modestly, but the issue remains. How might these fluctuations be eliminated?
It is common in MPM to apply a prescribed displacement boundary condition by creating an object using rigid particles, giving them a prescribed velocity, perhaps time varying, and impose the rigid object's velocity on the neighboring objects using a contact algorithm. This is convenient and easy, but leads to stress fluctuations near the rigid object. Suggestions on how to fix the issue, as illustrated below, are welcome.
Below we see that this simple problem comes with considerable error in the particle stress.