##### Modeling & Simulation, Testing & Validation (MSTV)

##### 2010

# A STATISTICS-BASED DAMAGE MODEL THAT ACCOUNTS FOR SCALE EFFECTS

#### Abstract

In this context, a damage model is a mathematical algorithm that is used to predict if and when in a
given loading history a structure will fail by ductile fracture. Increments in a damage parameter are related to
strain increments and state of stress. The damage model would operate as part of a numerical simulation, or
separately on an output file. A scale effect in ductile fracture is widely recognized from test data, where a large
structure tends to fail at lower strain than a smaller structure that is geometrically similar and of the same
material. Most damage models are not scale sensitive, and when they are calibrated to data from small
laboratory specimens, they will tend to over-predict the performance (i.e., energy absorbing capability) of a
larger structure. Another factor is scatter in test results even when specimens are made with care to be as
identical as possible. Both of these factors are addressed in the proposed statistics-based damage model. Scale
effects and scatter become part of the actual material behavior, rather than perceived effects of variability in
specimen preparation or test procedures. In addition, it is found that the statistical approach leads to improved
calibration of the model for the effects of triaxiality and Lode parameter.