A unified non-linear system model view of hyperelasticity, viscoelasticity and hysteresis exhibited by rubber

Abstract

Rubber modeling is an old subject and so many models exist that it is difficult to have a clear vision of what exists and is more appropriate. Rather than attempting a standard review, this paper proposes classification using the traditional system modeling strategy, where raw measurements are either processed to obtain non-parametric models, or used to identify parametric models, whose accuracy can be controlled by order selection or by numerical implementation considerations. A full test campaign, including multi-step relaxation, low speed triangular and sine tests, on a large deformation compression sample is used to illustrate the need to model and combine the base behaviors known as hyperelasticity, viscoelasticity, and rate independent hysteresis. The equivalence between linear viscoelasticity and linear time invariant systems is used to clarify the link between order selection and accuracy of a generalized Maxwell model. Rate independent hysteresis is analyzed using a convolution product like the one used for viscoelastic transients by introducing a relaxation modulus. Measurements of the hysteretic relaxation modulus are used to propose strategies to measure the asymptotic hyperelastic modulus and discriminate between different hysteretic model forms. A parallel between Iwan and Maxwell models is detailed, and non-parametric models are used to show that the two overlap in the low frequency small deformation regime. Regularized rate independent hysteresis and non-linear viscoelasticity are finally shown to lead to a similar view allowing a transition between the rate independent and linear relaxation models. The instantaneous ratio analytic force and displacement signals, or instant complex modulus, is introduced as novel non-parametric estimation of sine measurements and shown to be a powerful tool to analyze and validate the fact that a force rate relaxation with non-linear relaxation frequencies is most appropriate to represent the non-linear coupling of all three effects.