Assessing the fast non-Hertzian methods based on the simulation of wheel–rail rolling contact and wear distribution

Abstract

This paper aims at assessing several fast non-Hertzian methods, coupled with two wear models, based on the wheel–rail rolling contact and wear prediction. Four contact models, namely Kik-Piotrowski's method, Linder's method, Ayasse-Chollet's STRIPES algorithm and Sichani's ANALYN algorithm are employed for comparing the normal contact. For their tangential modelling, two tangential algorithms, i.e. FASTSIM and FaStrip, are used. Two commonly used wear models, namely the Archard (extended at the KTH Royal Institute of Technology) and USFD (developed by the University of Sheffield based on T-gamma approach), are further utilized for wear distribution computation. All results predicted by the fast non-Hertzian methods are evaluated against the results of Kalker's CONTACT code using penetration as the input. Since the two wear models adopt different expressions for calculating the wear performance, the attention of this paper is on assessing which one is more suitable for the fast non-Hertzian methods to utilize. The comparison shows that the combination of the USFD wear model with any of the fast non-Hertzian methods agrees better with CONTACT+USFD. In general, ANALYN+FaStrip is the best solution for the simulation of the wheel–rail rolling contact, while STRIPES+FASTSIM can provide better accuracy for the maximum wear depth prediction using the USFD wear model.

Applying Two Simplified Ellipse‐Based Tangential Models to Wheel‐Rail Contact Using Three Alternative Nonelliptic Adaptation Approaches: A Comparative Study

“…(i) In previous comparative studies [27][28][29][30], the normal modeling used simplified contact models, namely, KP, Linder, STRIPES, and ANALYN. Their normal results were only similar to the exact solution and then the inaccurate traction bound (the product of normal pressure and friction coefficient) could affect the prediction of stick-slip division and tangential stress distribution in tangential modeling.…”

“…where k(x, y) is the wear rate associated with pressure p(x, y), running speed v, relative slip s(x, y), and hardness of wheel-rail material H as shown in Figure 5. For a detailed explanation of this method, the readers are referred to [28,34].…”

“…The core of these three approaches is modifying the flexibility parameters used in FASTSIM by equivalent ellipse [24] or local ellipse [25,26] to replace the original ones derived from elliptic shape. The tangential performance of these three approaches for nonelliptic contact cases was compared and verified by [27][28][29][30].…”

“…However, it is worth mentioning in [27][28][29][30] that the normal contact is solved by different simplified models that cannot provide an exact solution for nonelliptic contact. Since tangential traction boundary depends on normal pressure distribution and flexibility parameters are determined based on contact shape, the estimation errors made by simplified contact models in normal modeling may compensate the estimation errors in the tangential solution.…”

“…In this paper, both FASTSIM and FaStrip are adapted to nonelliptic contact condition using three available approaches proposed in [24][25][26]. The novelty of this paper is the nonelliptic normal contact solution is accurately solved by Kalker's NORM algorithm [12] rather than simplification as done in [27,28]. Thus, it helps making a comparative study not affected by inaccurate normal modeling.…”