Wheel shape optimization approaches to reduce railway rolling noise

Abstract

A wheel shape optimization of a railway wheel cross section by means of Genetic Algorithms (GAs) is presented with the aim of minimizing rolling noise radiation. Two different approaches have been implemented with this purpose, one centred on direct Sound poWer Level (SWL) minimization, calculated using TWINS methodology, and another one emphasizing computational efficiency, focused on natural frequencies maximization. Numerical simulations are carried out with a Finite Element Method (FEM) model using general axisymmetric elements. The design space is defined by a geometric parametrization of the wheel cross section with four parameters: wheel radius, a web thickness factor, fillet radius and web offset. For all wheel candidates, a high-cycle fatigue analysis has been performed according to actual standards, in order to assure structural feasibility. Rolling noise reductions have been achieved, with a decrease of up to 5 dB(A) when considering the wheel component. Response surfaces have been also computed to study the dependency of the objective functions on the geometric parameters and to test the adequacy of the optimization algorithm applied.