Automatic calibration of the SANISAND parameters for a granular material using multi-objective optimization strategies

The parameter calibration of a constitutive model is a requisite to counter the uncertainty in the parameters and
to approximate the simulation results effectively. Yielding a robust set of parameters for various test conditions is complicated
as innumerable parameter combinations have to be investigated. In previous works, this calibration has been performed manually
by trial and error without checking the robustness of the chosen parameters. Therefore, the present study introduces an automated
calibration procedure using multi-objective optimization techniques. This assists in searching the parameter domain space
extensively for better combinations that simulate the experiment results precisely. Though this approach is quite popular in
various other engineering aspects, proposing the concept of calibrating the soil parameters and validating their efficiency has
been always a challenge and interesting in this framework. In this research, SANISAND model parameters have been calibrated
for crushed glass material under different triaxial conditions considering the barotropy, and pycnotropy effects. The results
demonstrated that the optimized SANISAND parameters approximated the experiment results far better than manually calibrated
results. This calibration approach facilitates in conserving the robust parameters besides dealing with time constraints and
motivates the idea of adapting this automation platform to any constitutive model for significant approximations.