Specimen design for material parameter identification using topology optimization

Abstract

Constitutive relations close the equations of continuum mechanics, and serve as a surrogate for a material in the design and engineering process. They are often specified in a parameterized form with parameters identified by experiment. In this paper, we propose a framework for identifying experimental configurations that are maximally informative for constitutive model discovery. The framework strongly couples modeling and experimentation: the model leverages high-dimensional data from full-field measurements, while the current uncertainty in the model guides the design of future experiments. We formulate this goal by integrating Bayesian optimal experimental design with topology optimization. The Bayesian design criterion quantifies expected information gain, which drives the topology optimization of the specimen geometry.