Modelling Failure Of Polymers: An Optimization Strategy Based on Genetic Algorithms and Instrumented Impact Tests

Modelling the failure of engineering polymers used in critical structural applications is still a challenging task that is increasingly demanded by the industry to optimize part design and estimate component service life. The difficulties include not only developing constitutive models capable of reproducing the complex polymer response at a reasonable computational cost, but also calibrating related parameters. That is to say, a way to find specific parameters which best represent the actual behavior of a material within the scope and limitations of a given constitutive or failure model. The aim of this study is to contribute in developing a robust inverse method calibration strategy. To address this issue, a novel approach based on genetic algorithms optimization (GA) together with finite element analysis (FEA) is proposed to blindly extract key constitutive and failure parameters from instrumented impact tests on single edge notched bending (SENB) specimens. The method was implemented to infer eight constitutive and failure parameters of a polyamide 12 (PA12) with an elasto-plastic ductile damage model. Triaxiality induced stable-unstable transition was successfully achieved by varying the notch depth of SENB specimens. Accordingly, three optimization schemes were conducted: (i) using only unstable experimental data; (ii) using only stable experimental data and (iii) using both simultaneously (multi-objective). The set of parameters obtained from each scheme were used to perform predictive FEA simulations, which were verified with experimental data. It was proven that both propagation regimes provide substantial information to obtain the mechanical response of the material. Simulation results evidenced the capability of the proposed strategy to predict the PA12 impact response and furthermore to fairly reproduce a completely different load case: a dynamic tensile test.