Data-Driven Modelling of Thermo-Viscoelasticity of Rubber

Viscoelasticity as a unique property of rubber materials, has a significant influence on the product performance, for example, the wet grip and the rolling resistance of tires. Understanding and controlling of this viscoelasticity is essential in compound development and tire performance prediction.

To this end, the viscoelasticity of rubber has been described by mathematical models and characterized by experimental approaches. In mathematical models, much effort has been made to capture the nonlinear stress-strain relationship of rubber observed in the quasi-static stretching, whereas the viscous behavior is usually assumed to be linearly related to the strain rate only. In other words, the viscosity of rubber is assumed to be constant and independent of temperature, which is not consistent with the experiments and can lead to false prediction. From a materials’ science perspective, the viscosity of a rubber compound is determined by multiple factors, including the molecular weight and structure of a polymer, the amount and type of filler as well as the amount of plasticizers. Most of the factors are temperature dependent. The temperature dependence of the viscosity is even more pronounced when new ingredients like resins are introduced in rubber compounds. The objective of this work is to provide a better prediction of rubber viscoelasticity by developing a thermo-viscoelasticity model, where the temperature dependence of viscosity is taken into account.

To develop the thermo-viscoelasticity model, a data-driven approach is used. The data is obtained by conducting the experiments of the Dynamic Mechanical Analysis (DMA) on a typical tire tread compound, where the storage modulus and the loss modulus are measured by varying the frequency and temperature. The measured data is applied to build the thermo-viscoelasticity model. Finally, this developed model has to be validated by comparing the predicted strain-stress behavior with the measured data.