A numerical optimization technique for design of wheel profiles

Abstract A method for determination of an optimum railway vehicle wheel profile for a given geometric contact characteristics of wheel/rail using a numerical optimization approach is presented in this paper. Such characteristics are defined through a given rail profile, and the target rolling radii difference (Δ r ) as a function of lateral displacements ( Y ). Cubic spline functions are utilized to connect curve segments of wheel profile to produce smooth variation trends desired. The coefficients of low degree polynomials could represent the curve's generalized coordinates. In this manner the number of necessary coordinates is greatly reduced. It is shown that the total number of design variables is only five, which computationally makes the optimization process very efficient. The method of Complex algorithm within the non-gradient numerical optimization routines is selected to generate a sequence of improving profiles leading to the optimum one. Dynamic performances of the generated profiles are then determined by obtaining their Y− Δ r curves using ADAMS/Rail software package. The objective is to minimize the deviation of the Y− Δ r performances of the generated profiles from the target one, under the given vehicle and track characteristics. Subsequently, the proposed method is used to design a wheel profile for passenger vehicles, compatible with the Iranian north-eastern railway characteristics. The dynamic performances of the designed profile, and those of S1002 and P8 standard profiles, are presented for comparison. It is shown that the wear index of the designed profile is significantly lower than the other two. In addition, other dynamic characteristics such as lateral displacement, angle of attack and derailment quotient are shown to be improved considerably. The procedure worked efficiently and the results proved to be quite satisfactory.