Behavior characterization of visco-hyperelastic models for rubber-like materials using genetic algorithms

• Published in: Applied Mathematical Modelling Vol. 66; pp. 241 - 255
• Main Authors: A. López-Campos, J., Segade, A., R. Fernández, J., Casarejos, E., A. Vilán, J.
• Format: Journal Article
• Language: English
• Published: New York Elsevier Inc 01.02.2019; Elsevier BV
• Subjects: Axial stress; Elasticity; Energy dissipation; Genetic algorithms; Hyperelasticity; Materials elasticity; Mechanical analysis; Multiple objective analysis; Nonlinear viscosity; Optimization; Rubber; Rubber-like materials; Search algorithms; Strain; Time dependence; Viscoelasticity; Viscosity; Nonlinear viscosity; Rubber-like materials; Hyperelasticity; Genetic algorithms
• ISSN: 0307-904X; 1088-8691; 0307-904X
• DOI: 10.1016/j.apm.2018.08.031

•A nonlinear visco-hyperelastic model is numerically studied.•Genetic algorithms are employed to fully characterize the material.•Single and multi objectives versions are considered.•A benchmarking analysis is performed using a known solution.•Uniaxial, biaxial and shear tests are designed to compare the proposed candidates. Rubber-like materials are widely deployed in industry because of their outstanding properties of elasticity and energy dissipation capacity. In order to analyze and improve the behavior of components made of rubber, suitable models for the material behavior must be formulated before their usage in complex evaluations. The aim of such models is to predict the mechanical response of these materials under external loads, using visco-hyperelastic descriptions. In this paper we are interested in reproducing the nonlinear elastic behavior of rubbers as well as the strain-dependent part (viscoelasticity). Typical viscoelastic models are based on linear assumptions, being the viscous part time-dependent but no strain-dependent. Therefore, nonlinear elasticity and linear viscoelasticity are mixed up in the available models. In our work, we propose a model with both parts described as nonlinear. We use a classical hyperelastic model (Mooney–Rivlin) combined with a nonlinear viscous part including both dependencies in strain and time. We also develop a genetic algorithm for searching (optimizing) the model parameters that describe the behavior of this type of materials. We describe simple and multiple objective genetic algorithms for the optimization based on uniaxial and biaxial stresses. We present benchmarking tests for the algorithms, which reproduce different tests with high accuracy and we also discuss the reliability of the model for different stress sates, strain ranges and strain rates. Finally, a real specimen is tested and studied with the algorithm to define its properties according to the material model developed. The study highlights the capabilities of the model to describe complex stress states based on limited information, and it also shows the possible limitations. The procedure we develop can be used as a design tool that allows implementing any simply analyzed material into a realistic material model for further mechanical evaluations, as it is finite elements or similar.