Uncertainty quantification for advanced progressive damage models for composites by means of efficient emulators and bootstrapping

Abstract

A methodology was developed and validated to quantify the uncertainty for advanced progressive damage models for composites. It relies on a pragmatic approach entailing the definition of efficient emulators, the use of state-of-the-art computational models, and the employment of bootstrapping statistic techniques. The proposed methodology was calibrated on numerical results obtained running a limited amount of virtual experiments (five for each configuration) of unnotched and open-hole specimens in tension and compression. The structural strength was taken as the quantity of interest, and a methodology was proposed and validated to determine its distribution and associated statistics.