Performance of various thin concrete slabs under projectile impact: Sobol’s sensitivity analysis with aid of metamodels

Abstract

The ballistic performance of thin concrete slabs is influenced by impact velocity, slab thickness, steel rebar content, bullet mass and prestress level. We aim to quantitatively identify which of the mentioned input variables play a major role in the ballistic response of concrete slabs, such as ballistic limit, residual velocity, and ejected concrete volume at the front and rear faces of the slab. The Sobol’s Functional Analysis of Variance (FANOVA) is adopted as method of this research, which consists of computing variance-based indexes that rank the importance of the input variables and their interaction. Test and simulation results from a former study are used to feed Blind Kriging models and artificial neural network (ANN) models. After model selection, the ANN model is combined with probabilistic sampling techniques to calculate the first and second order Sobol’s sensitivity indexes. Results are presented and discussed. It is found that the slab thickness and prestress can particularly enhance the ballistic limit of the concrete slabs. Other parameters are discussed as well.