PhysX for Discrete Element Method

Discrete element method (DEM) is used to simulate granular material at high precision or fracturing in continuous materials. I am interested in using PhysX as a DEM solver.

Some questions about PhysX’s capability as a DEM solver have been addressed in Longshaw et al 2010.

Longshaw lists several shortcomings inherent in PhysX as a DEM solver; however, this paper is nearly a decade old, which leads me to wonder if some have been addressed. I also have a few additional questions myself about PhysX’s potential as DEM solver:

  1. Is PhysX still limited to under 65,536 elements?
  2. Is physics able to set material properties of elements (i.e. restitution coefficient, Youngs modulus, Poisson ratio, friction coefficients)?
  3. Does PhysX have any kind of element-to-element bond model for simulating continuous materials?

I suspect PhysX will not work out-of-the-box for most DEM applications, but its GPU capability and long heritage make it appealing for this kind of work.

Interested to hear what people think!