Quantitative constitutive model selection for inertial microcavitation rheometry using Bayesian inference
Please login to view abstract download link
Understanding and predicting the nonlinear behavior and damage potential of soft materials undergoing finite and high-strain rates (i.e., 103 1/s and higher) is critical to biomedical engineering applications. One of the challenges lies in determining the constitutive model that best describes the material deformations in the high-strain rate regime. The Inertial Microcavitation Rheometry (IMR) technique finds the constitutive model and parameters using laser-induced cavitation (LIC). IMR compares the bubble radius history from LIC experiments to Rayleigh-Plesset-type (forward calculations) results to identify the best constitutive model [1, 2]. However, the fitting procedure is limited to considering the most spherical bubble collapse experimental result of one material. Additionally, the material library of forward-calculations is limited to a subset of available viscoelastic models. We present a Bayesian inference approach for high-fidelity constitutive model selection within the IMR framework. The approach follows Occam’s razor principle and penalizes models with more parameters compared to models with one to two parameters. The approach quantifies the plausibility of different of constitutive models using multiple experimental data sets. The plausibility of the constitutive models for different assumed measurement and model errors is also presented.
