Four classical damage models for concrete (three of which are available in commercial codes) have been compared and critiqued, showing that they all share the notions of a “teardrop” yield surface that can harden and (for some models) translate until reaching a three-invariant fracture limit surface that then collapses to account for softening (i.e., permanent loss of strength). Practical engineering models for rock and ceramics are similar. The common drawbacks of these models (primarily severe mesh dependence) can be mitigated, though not eliminated, by seeding their material properties in the simulation with spatial variability (aleatory uncertainty) and by using appropriate scale effects for the strength and failure progression properties.
Seubpong Leelavanichkul (Research Associate, UofU)
Brian Leavy (PhD student, Mech. Engr., UofU)