Optimization workflow to parameterize elastomer material models based on arbitrary time‐domain measurement data

Abstract

Predicting the behavior of elastomer components always requires experimental characterization of their nonlinear mechanical properties. This involves measuring the ratio of stresses and strains and the dissipated energy per load cycle in harmonic deformations of test specimens. Characteristic values, such as the elastomer's stiffness and damping, and their dependence on amplitude and frequency are then calculated and used to identify material model parameters.However, determining these characteristic values imposes significant requirements on test setup and control, as they are only meaningful when calculated from ideally sinusoidal and steady-state load sequences. This can result in unnecessary methodological errors for any real-world measurement. This contribution presents a method for identifying material model parameters that omits determining characteristic values, greatly reducing testing times and prediction errors. The method involves an optimization-based variation of model parameters to directly match the stress response of a material model with measured stresses. Its independence from ideal test sequences is shown by applying the method to numerically generated measurement data with artificial control errors. Application to real measurement data reveals a possible reduction of testing times from about 2 h to below 35 s while preserving the model's accuracy.