Optimisation with genetic algorithm of the acoustic performance of T-shaped noise barriers with a reactive top surface

In this study, the aim is to optimise the acoustical efficiency of T-shaped noise barriers whose top is covered with a series of wells. This research work uses an optimisation method in order to find the best noise barrier profile considering several variable parameters. Numerical simulations of the acoustical propagation are achieved by use of a 2D boundary element method code. The optimisation part is carried out with a global and direct evolutionary optimisation method: a genetic algorithm. The parameters to optimise are the shape of the protection (the depths of the wells on the crowning) and the flow resistivity of absorbing materials considered. The cost function to maximize is defined through a mean value of the acoustical efficiency of the protection compared to a reference configuration, averaged on several receiver points. Final results show significant optimised values of parameters for efficient protections in order to improve classical noise barriers.