Multi-disciplinary optimization of railway wheels

Abstract

A numerical procedure for multi-disciplinary optimization of railway wheels, based on Design of Experiments (DOE) methodology and automated design, is presented. The target is a wheel design that meets the requirements for fatigue strength, while minimizing the unsprung mass and rolling noise. A 3-level full factorial (3LFF) DOE is used to collect data points required to set up Response Surface Models (RSM) relating design and response variables in the design space. Computationally efficient simulations are thereafter performed using the RSM to identify the solution that best fits the design target. A demonstration example, including four geometric design variables in a parametric finite element (FE) model, is presented. The design variables are wheel radius, web thickness, lateral offset between rim and hub, and radii at the transitions rim/web and hub/web, but more variables (including material properties) can be added if needed. To improve further the performance of the wheel design, a constrained layer damping (CLD) treatment is applied on the web. For a given load case, compared to a reference wheel design without CLD, a combination of wheel shape and damping optimization leads to the conclusion that a reduction in the wheel component of A-weighted rolling noise of 11 dB can be achieved if a simultaneous increase in wheel mass of 14 kg is accepted.