For guardrail designers, it is essential to achieve a crashworthy and optimal system design. One of the most critical parameters for an optimal road restraint system is the post embedment depth or the post-to-soil interaction. This study aims to assess the optimum post embedment depth values of three different guardrail posts embedded in soil with varying density. Posts were subjected to dynamic impact loads in the field while a detailed finite element study was performed to construct accurate models for the post–soil interaction. It is well-known that experimental tests and simulations are costly and time-consuming. Therefore, to reduce the computational cost of optimization, radial basis function–based metamodeling methodology was employed to create surrogate models that were used to replace the expensive three-dimensional finite element models. In order to establish the radial basis function model, samples were derived using the full factorial design. Afterward, radial basis function–based metamodels were generated from the derived data and objective functions performed using finite element analysis. The accuracy of the metamodels were validated by k-fold cross-validation, then optimized using multi-objective genetic algorithm. After optimum embedment depths were obtained, finite element simulations of the results were compared with full-scale crash test results. In comparison with the actual post embedment depths, optimal post embedment depths provided significant economic advantages without compromising safety and crashworthiness. It is concluded that the optimum post embedment depths provide an economic advantage of up to 17.89%, 36.75%, and 43.09% for C, S, and H types of post, respectively, when compared to actual post embedment depths.
Upset Prediction in Friction Welding Using Radial Basis Function Neural Network