Abstract: With the development of high-speed railway, more and more attention has been focused on the rail damage caused by wear and rolling contact fatigue, and it is found that rail service life could be extended by the pre-grinding technology. To determine the optimal pre-grinding profile, a probability model of wheel-rail contact based on the Monte Carlo method is proposed, in which the lateral displacement of wheel set is assumed to follow the Gauss distribution. The probability distributions of contact stress of rail are obtained by a three-dimensional finite element model in which the lateral displacement of wheel set is defined by the probability model, while a train passing. The results show that there is a significant influence of lateral displacement on the maximum contact stress, while the lateral displacement varies with −10 mm, 0 mm and 8 mm, the corresponding maximum contact stress are 1 712.6 MPa, 1 362.1 MPa and 448.7 MPa respectively. Although the value of accumulated or average contact stress could be employed as a index to evaluate the mechanical properties of a rail profile, the prediction results are not consistent with each other, for instance, the average contact stress of CHN60 at inner rail corner(within 30 mm from the rail tread central area) is nearly four times of that at rail tread central area, while the value of accumulated contact stress at rail tread central area is 10 times larger than that at inner rail corner. For avoiding the contradiction, a weighted index which both considering the accumulated contact stress and the average contact stress is proposed and used for evaluating the rail pre-grinding profile. An optimal profile is determined from four types of rail pre-grinding profiles which are designed by the engineering experience according to the weighted index, that is, the minimum value of the index for the rail pre-grinding profile. The present work would be a certain reference value for drawing up the rail profile optimization and rail grinding methods.

Key words: Contact, Monte Carlo, Profile optimization, Rail grinding