OPTIMIZATION OF A FEM RAILWAY WHEEL MODEL IN TERMS OF ITS VIBRATION/NOISE CHARACTERISTICS

Recent research in the field of railway noise locates the main source of noise emissions in the wheel. In the BRITE-EURAM Project HEMBOT, LFME developed an optimization method that targeted the major parameters that affects the noise emission of a train wheel: the wheel's design characteristics. This paper presents the optimization of the design of a railway wheel in terms of the wheel's sound power levels emission, with respect to its geometrical properties. To this end, a simplified FEM model of the wheel was employed, that did not include the interaction of the wheel and rail or the influence of the braking system that is assembled on the wheel. The objective of the optimization method was to find a design of the selected railway wheel, which without the use of damping or tuning devices, emits less vibration/noise compared to the original design. The optimization method used, was based on Genetic Algorithms (GAs). GAs are a robust optimization method that performs regardless of the optimization problem. The GA-based optimization method that is presented in this paper, utilized ANSYS running in batch mode for the calculation of the objective function values of the population of each generation. The results of the application of the GA optimization method that was developed, were successful. The GA optimization method was able to find not only a global optimum design which represents a reduction of 48.3% in the sound power levels emitted by the wheel, but also to identify numerous "near-optimal" wheel designs with reductions ranging from 36.5% up to 47%. This result, which is a feature that characterizes GAs, offers more freedom to the wheel manufacturer in the selection of the most appropriate/applicable design. Finally, experimental determination of the emitted sound power of the original wheel design and of the GA-optimized wheel design, showed a reduction of 3 dB for the latter.