Design optimization of a MASH TL-3 concrete barrier using RBF-based metamodels and nonlinear finite element simulations

•First design optimization of a roadside barrier based on current safety standard.•Full-scale crash simulations using validated finite element models.•RBF-based design optimization methodology incorporated with FE simulations and a genetic algorithm.

Concrete barriers are one of the most widely used safety features for preventing errant vehicles from entering opposing travel lanes on highways. As a safety device, a concrete barrier is also required to safely redirect a striking vehicle such that it is not bounced into travel lanes and collide with other vehicles. Given its rigidity compared to most vehicular structures, the performance of a concrete barrier is mainly influenced by its shape or the cross-sectional geometry. Over the years, concrete barriers have been continuously improved using roadway crash data and crash tests. Although the current in-service concrete barriers satisfy the requirements of safety standards, the empirical approach is not cost-effective for designing new barrier systems. In this study, a simulation-based optimization approach was adopted to obtain the optimum design of a concrete barrier by combining nonlinear finite element (FE) simulations, metamodeling with radial basis functions (RBFs), and a genetic algorithm (GA). The performance of the concrete barrier was determined by evaluating vehicular responses specified in the current safety standard, Manual for Assessing Safety Hardware (MASH). Nonlinear FE simulations were first carried out on sample designs to obtain the vehicular responses for creating the RBF metamodels, which were then used in the optimization process to replace the expensive FE simulations. An optimal design of the concrete barrier was obtained by the GA and was shown to have improved safety performance over the original design.