Brittle material modeling

[lingsma]

Hello everyone,

I'm currently modelling refractory linings in Ansys workbench 11. These linings consist of brittle materials. (concrete, firebrick) I'm using the Mohr-Coulomb criterion to analyse the stresses, and I'm wondering if anyone can help me with te following questions:

First, is the Mohr-Coulomb criterion right for these materials? Is it normally also being used for this kind of problems?

Second, I've heard about the Drucker-Prager criterion. This is not in the Workbench, but it's also being used for this kind of simulations. However, according to what I've read, it doesn't differ from the Mohr-Coulomb criterion that much. What is the difference? Which problems is the Drucker-Prager criterion used for?

thanks!

 

[omidomidi]

the mode of failure in concrete is highly dependent on the propagation of the cracks.you can use both Willim-Warnke or Drucker-Prager criterions for testing concrete failure in ansys classic using Solid65 element but beacause concrete doesn't act like metals you can't use Mohr's criterion.

 

[TERIO]

Mohr-Coulomb is very similar to Drucker-Prager. If you are more familiar with metals then Mohr-Coulomb is like the Tresca yield surface (hexagonal cross sections) while Drucker-Prager is like von Mises (circular cross sections). Mohr-Coulomb and Drucker-Prager depend on the mean stress (hydrostatic pressure) while Tresca and von Mises do not.

 

[lingsma]

Thanks for the answers!
Terio: that's what I allready suspected; that they are quite similar. I'm currently using the Mohr-Coulomb criterion because it's the easiest option when working with Ansys Workbench. I'm doing rather basic stuff, so my guess is that the MC criterion is sufficient.
Do you have an opinion on the difference between the two criteria, wheter one is better than the other and if so, on which points?
thanks!

 

[TERIO]

I haven't worked with either, so I a just passing along my understanding of what they are. Here are some descriptions from Wikipedia:

http://en.wikipedia.org/wiki/Yield_surface

I assume you are using this as a "failure" criterion in post-processing tool and not as as a yield surface for actually modeling inelastic deformation. As a post-processing failure criterion my guess would be that it does not make much difference but Mohr-Coulomb might be a little more conservative. In the latter case I would be somewhat surprised if Mohr-Coulomb was available because yield surfaces with corners cause computational difficulties (you can't define the direction normal to a corner).