Identification of material parameters for aluminum foam at high strain rate

This paper concerns on the aluminum foam material modeling and identification of constitutive parameters at high strain rates. Lured by its excellent energy absorption capacity under the impact conditions, aluminum foam has been widely used in automotive and aerospace industry as a lightweight filler material. Nevertheless, aluminum foam shows different mechanical properties at low and high plastic deformation rates, moreover, in the engineering practice, the occurrence of high strain rate appears more often than low strain rate. Generally speaking, to obtain the constitutive model parameters of aluminum foam material, a large number of expensive experiments need to be conducted. In addition, the plastic deformation behavior of aluminum foam at high strain rate follows a highly non-linear dynamic process, and its parameter identification requires much more complex numerical procedure. For these reasons, this paper proposes a new procedure to predict Deshpande and Fleck (DC) model parameters for aluminum foam based on successive artificial neural network (SANN) technique and particle swarm optimization (PSO) algorithm. Finite element analyses are performed by using Design of Experiment (DoE) method to establish the SANN model for each of the five constitutive parameters of the DC model. The constitutive model parameters with the minimum discrepancy between experimental and SANN curves are obtained by using the surrogate modeling and PSO methods. Finally, this approach is validated by comparing the FE modeling results against the experimental results. This study demonstrates the effectiveness of such a three phase identification procedure comprising experimentation, optimization, and verification. It provides a useful means for parametric identification of other similar lightweight foam or porous materials.