Effects of two testing protocols on the material model parameter identification for rubber-like materials

Abstract

This paper analyzes the polyurethane material that can be used for the construction of seismic isolation devices based on rolling of elastomeric spheres. Such isolators could be used in low-income countries. Uniaxial tensile tests were performed on dumbbell-shaped polyurethane 95 ShA specimens under two different loading protocols. Protocol 1 consisted of applying a cyclic saw-tooth loading centered on a pre-imposed initial deformation, while Protocol 2 consisted of consecutive loadings followed by relaxation at three different deformation levels. Then, a material model comprising three-chains in parallel was calibrated against the tests. The model combined the Yeoh hyperelastic and Bergstrom-Boyce models. The results of the parameter calibration showed that different testing protocols could lead to different model parameter values. In terms of fitting errors, it is observed that fitting to Protocol 1 generates a good prediction on Protocol 2 with an error of 0.26%; however, when fitting to Protocol 2, the behavior observed on Protocol 1 could not be accurately predicted, resulting in an error of 9.10%. Moreover, when comparing fitting to match only Protocol 1 with considering both protocols simultaneously, the total error is only reduced from 0.32% to 0.20%, suggesting that Protocol 2 adds redundant information already contained in Protocol 1. Additional tests with different deformation rates and ranges need to be conducted to define an optimal protocol for material calibration.