A data-driven rate and temperature dependent constitutive model of the compression response of a syntactic foam

Polymeric syntactic foams are used in aerospace and marine applications requiring low density and low moisture absorption together with high specific strength and stiffness. Their mechanical response is highly sensitive to temperature and strain rate and such sensitivity must be modelled accurately. In this study, the uniaxial compressive response of a polymeric syntactic foam is measured at strain rates in the range [10-3, 2.5·103] /s and temperatures varying between -25°C and 100°C. The resulting dataset is used to train a neural network to predict the compressive response of the foam at arbitrary strain rates and temperatures. It is found that the surrogate model is highly effective in predicting the material response at temperature and rates not included in its training set. Finally, a stochastic version of the data-driven model to allow predictions of the variability in the stress versus strain response is proposed.