Material parameters identification and experimental validation of damage models for rubberlike materials

Abstract

We test the predictability of our (DPS) microstructure inspired constitutive approach for damage and hysteresis in rubberlike materials by comparing it with the widely adopted Ogden and Roxburgh (OR) and Beatty and Krishnaswamy (BK) models. The experimental validation is analyzed by uniaxial cyclic tests with increasing maximum strains of high damping EPDM rubber specimens adopted in seismic isolation. The numerical identification of the material parameters is obtained by a careful population-based stochastic optimization approach. After testing the stability of our numerical approach, based on a large number of numerical experiments, we deduce that our constitutive model and the OR model are the most accurate. However, differently from our model, we find that the OR model exhibits a strong numerical instability, with a very high variance of the optimal material parameters for the same experimental data sets. We believe that the obtained results are important both for rubber constitutive models and for material parameters optimization approaches.