Optimization using a physics-based surrogate model for inertial-plastic protective devices in reinforced concrete slabs subjected to blast loads

Abstract

At present, there is no unique methodology for designing protective devices for structural members subjected to blast loads. Moreover, a bad design can increase damage with respect to the unprotected case. Therefore, casting the design problem as a simulations-based optimization problem becomes appealing. The dilemma arises in balancing the complexity and fidelity of the physical model used for the simulations. This paper presents a novel methodology for designing inertial-plastic protective devices based in two models of different fidelities. A high-fidelity model serves for updating a physics-based surrogate model within a two-levels optimization loop. The physics-based surrogate model allows exploiting an established heuristic, that is related to the full densification of the device, for quasi-analytically optimizing its yield force. This approximately optimized yield force is iteratively evaluated in the high-fidelity model. In two test examples, the main loop converged monotonically in just 3 iterations. Besides, both the validity of the heuristic and the actual optimality of the solutions attained were assessed by comprehensive parametric simulations using the high-fidelity model.