Mixed material models in peridynamics

Good morning all. New user here. About 8 hours total experience. So very much still over-revving the engine and making the clutch smoke.

Does anyone know if it is possible to mix material models when running peridynamics models?

I have successfully used various models seperately (pair_style peri/lps, pair_style peri/pmb etc) but I would like some parts of my model to behave as brittle materials and some parts to behave as elastic-plastic materials (or possbily viscoelastic materials, I haven’t decided what the best material model for my application is yet).

I imagine that it would be possible to make an elastic-plastic model behave very like a brittle material with the right choice of parameters/coefficients (I can do this in finite element models) but I am wondering if there is a more elegant way.

I can see that it is possible to allow multiple styles using pair_style hybrid but I have not yet been able to use this to make multiple peridynamics models work together, so I am curious to know whether it is possible and I should keep trying, or impossible and I should give up.

Thank you for your time.


I doubt that using multiple peridynamics models in a hybrid pair style makes much sense. Remember that those are not particle models, but grid-free continuum models, and for that it seems most logical to me that it would require devising a model that combines multiple behaviors into a single potential function if you have material that can exhibit different behavior. There are all kinds of complications with that though that are far beyond the scope of this forum.

Besides the conceptual argument, there is also a technical problem: with a hybrid pair style you can model two sets of atoms with different potential functions, but you also need a model describing how one set interacts with the other (unless they should be completely separate and non-interacting, but in that case you could model each part independently and don’t need to use a hybrid style) and that brings you back to the conceptual issue I mentioned before.

@akohlmey Thank you for taking time to reply. The application is laminated glass under bullet impact. So I potientially have three materials types - glass (brittle), the polymer interlayer between the layers of glass (viscoelastic or possibly elastic-plastic) and the bullet itself (elastic-plastic). The laminated glass is sort of a continuum (you would expect displacement continuity at the interface between the glass and the interlayer) and we are interested in what happens to the bullet too because the damage it sustains passing through the laminated glass pane, as well as the reduction in its kinetic energy, has a bearing on what it does to the next pane of glass (in a double-glazed unit, for example).

So what I meant was could separate material domains (in the same model) have different peridynamics models. I guess the answer, for now at least, is no.

I will continue to experiment. For a while at least! Thanks again.


Using LAMMPS for this kind of application is certainly a long shot: I don’t see how you could realize it at full detail without writing your own pair style(s) or make additions to existing ones.

Since this is the “beginners” category a few suggestions that may help:

  • if you have not done so, it is probably worth looking into how to realize such a model with a finite element modeling software. While that may be missing some of the advantages of the peridynamics approach, you can at least gain some insight into what kind of expressions can be used to represent the various materials sufficiently well for your needs.
  • at the beginning it is advantageous to limit the complexity of the model so that you can validate the components of a more complex model individually. So it should be possible to model each component (glass, polymer, bullet) individually and also do a crude deformation/breakage model using fix indent (standing in for either a hard sphere or a flat surface in case of the bullet where you can also reverse the frame of reference to keep the bullet static for simplicity).

@akohlmey Very sensible suggestions indeed. I have done quite a lot of FEA but it does have some limitations, I was curious to try the peridynamics approach to see what it offered following a recommendation from a colleague. I have had some success with simpler models - I do try not to run before I can walk but sometimes get carried away! At the moment I am having difficulty stopping my bullet from completely disintegrating along with my glass. I shall keep trying. Thanks again.