Stochastic calibration of local constitutive models through measurements at the macroscale in heterogeneous media

A methodology is presented suitable to identify properties of single phases, locally at the microscale, within a heterogeneous periodic material, on the basis of measurements at the macroscale and homogenization techniques. Reference is made to periodically perforated steel sheets, regarded as a prototype of a ductile composite material with periodic microstructure. The proposed method might be applied to industrial situations where the fabrication process of a composite material alters the mechanical properties of individual phases or constituents. Experimental data are obtained through monotonic uniaxial tensile tests at the macroscale. An elastic–plastic hardening model is attributed to the matrix material. Weighted least-squares, Monte Carlo simulations and sequential non-linear Kalman filters are employed to solve the parameter identification problem in a stochastic context, and comparative assessments of diverse strategies of inverse analysis are presented, as to accuracy and computational effectiveness.