Comparison of numerical and computational aspects between two constraint-based contact methods in the description of wheel/rail contacts

Abstract

The numerical and computation aspects of the Knife-edge Equivalent Contact (KEC) constraint and lookup table (LUT) methods are compared in this paper. The LUT method implementation uses a penetration-based elastic contact model for the flange and a constraint-based formulation at the wheel tread. For the KEC method, where an infinitely narrow rail contacts an equivalent wheel, regularization of the tread-flange transition is adopted to simultaneously account for tread and flange contacts using constraints. A comparison between the two methods is carried out using well-known numerical integrators to show the applicability and limitations of both methods.Two fixed-step-size integrators, the explicit Runge–Kutta (RK4) and the predictor–corrector Adam–Bashforth–Moulton (ABM) methods, and two variable-step-size Matlab built-in function integrators, the explicit $ode45$ o d e 45 and implicit $ode15s$ o d e 15 s , were applied to get the numerical solutions to the dynamic problems and study the relative numerical performance of the two contact description methods. To complete the railway vehicle model, both contact methods were implemented for the multibody model of a benchmark railway vehicle (the Manchester wagon 1). Numerical results were obtained for different railway tracks with and without irregularities. Profiles of the S1002 wheel and LB-140-Area rail, which demonstrate the two-point contact phenomenon, were considered. Both methods were implemented in Matlab and validated against commercial simulation software. The kinematic results for both approaches show good agreement, but the KEC method was up to 20% more efficient than the LUT method regardless of integrator used.

Physics-based dynamic simulation opportunities with digital twins

“…The research related to efficient and accurate simulation techniques is in progress, which is beneficial for the labeled data generation with a computer [4,6]. Therefore it is expected that the simulation is capable of merging tighter with the real world more effectively.…”

Influence of Hertzian and non-Hertzian contact models on two-point wheel/rail contact in different planes

“…In all case studies, the multibody model of the Manchester wagon [8] is implemented in Matlab environment. The fixed-step-size integrator Adam-Bashforth-Moulton (ABM) [9] with time step size ∆𝑡 0.5 ms is used. The wheel and rail profiles are the S1002 wheel and LB140-Area rail profiles, which present a unique two-point contact scenario in the tread-flange transition.…”