Improvement of aerodynamic properties of high-speed trains by shape optimization and flow control
Paper in proceedings, 2008
Increase in speed of new high-speed trains has led to new requirements for improvement of their aerodynamic properties. Aerodynamic properties such as drag, crosswind stability, aero acoustics noise and upraise of ballast due to flow have to be treated simultaneously in a multi-objective optimization procedure. This paper demonstrates an efficient optimization procedure that uses metamodels in form of polynomial response surfaces as a basis for search for optimal designs. Such simple models of objective functions makes it possible to use genetic algorithms to explore the design space. As a result of the suggested optimization procedure a set of so called Pareto-optimal solutions was obtained that helps exploration of extreme designs and finding tradeoffs between design objectives. Two examples are demonstrated for the purpose of the validation of the optimization procedure: the optimization of the front of the train for the cross-wind stability and the optimization of passive flow devices (so called vortex generators) for drag reduction. Influence of turbulence models used in computer experiments on optimization procedure is explored. It was found that the choice of turbulence model influences the shape of the train with minimal lift force. Usage of Reynolds Stress (RSM) and RNG k-epsilon model was found to produce noisy data that prevented construction of response surface models.
Pareto optimal front