Computational Technology Resources - CCC

Z. Dai, J. Zuo, Y. Pan, J. Ding, "A Numerical–MLP-Based Framework for Rapid Prediction of EHL Pressure and Film-Thickness Profiles in Wheel–Rail Contact", in J. Pombo, (Editor), "Proceedings of the Seventh International Conference on Railway Technology: Research, Development and Maintenance ", Civil-Comp Press, Edinburgh, UK, Online volume: CCC 15, Paper 9.3, 2026, doi:10.4203/ccc.15.9.3

Keywords: wheel–rail contact, elastohydrodynamic lubrication, surrogate model, multilayer perceptron, machine learning, adhesion.

Abstract

Under wet rail conditions, wheel–rail adhesion is highly dependent on the normal contact behaviour at the interface, which is usually analysed using elastohydrodynamic lubrication (EHL) models. However, high-fidelity EHL calculations are time-consuming, especially when a large number of operating conditions need to be evaluated. This study proposes an MLP-based surrogate framework to predict the pressure and film-thickness distributions along the contact centreline. The training data are generated from a high-fidelity EHL model under different axle loads and train speeds, and two output-specific networks are trained separately for pressure and film thickness due to their different scales and response characteristics. Comparison with numerical results shows that the surrogate models can reproduce important response features, such as pressure peaks and film-thickness variations, with generally small errors across the majority of the parameter space. Relatively larger deviations occur in a small number of cases, mainly near the boundaries of the load–speed parameter space. In addition, the computation time is reduced from seconds to milliseconds, making the proposed framework practical for rapid wheel–rail adhesion analysis.

download the full-text of this paper (PDF, 15 pages, 988 Kb)

go to the previous paper
go to the next paper
return to the table of contents
return to the volume description

	Computational & Technology Resources an online resource for computational, engineering & technology publications
	not logged in - login
Front Page Browse CCC CCP CSETS CTR IJRT Other Authors Search Purchase Guide FAQ Contact us	Civil-Comp Conferences ISSN 2753-3239 CCC: 15 PROCEEDINGS OF THE SEVENTH INTERNATIONAL CONFERENCE ON RAILWAY TECHNOLOGY: RESEARCH, DEVELOPMENT AND MAINTENANCE Edited by: J. Pombo Paper 9.3 A Numerical–MLP-Based Framework for Rapid Prediction of EHL Pressure and Film-Thickness Profiles in Wheel–Rail Contact Z. Dai^1,2, J. Zuo^1,2, Y. Pan¹ and J. Ding¹ ¹College of Transportation, Tongji University, Shanghai, China ²Shanghai Key Laboratory of Rail Infrastructure Durability and System Safety, Tongji University, China doi:10.4203/ccc.15.9.3 Full Bibliographic Reference for this paper Z. Dai, J. Zuo, Y. Pan, J. Ding, "A Numerical–MLP-Based Framework for Rapid Prediction of EHL Pressure and Film-Thickness Profiles in Wheel–Rail Contact", in J. Pombo, (Editor), "Proceedings of the Seventh International Conference on Railway Technology: Research, Development and Maintenance ", Civil-Comp Press, Edinburgh, UK, Online volume: CCC 15, Paper 9.3, 2026, doi:10.4203/ccc.15.9.3 Keywords: wheel–rail contact, elastohydrodynamic lubrication, surrogate model, multilayer perceptron, machine learning, adhesion. Abstract Under wet rail conditions, wheel–rail adhesion is highly dependent on the normal contact behaviour at the interface, which is usually analysed using elastohydrodynamic lubrication (EHL) models. However, high-fidelity EHL calculations are time-consuming, especially when a large number of operating conditions need to be evaluated. This study proposes an MLP-based surrogate framework to predict the pressure and film-thickness distributions along the contact centreline. The training data are generated from a high-fidelity EHL model under different axle loads and train speeds, and two output-specific networks are trained separately for pressure and film thickness due to their different scales and response characteristics. Comparison with numerical results shows that the surrogate models can reproduce important response features, such as pressure peaks and film-thickness variations, with generally small errors across the majority of the parameter space. Relatively larger deviations occur in a small number of cases, mainly near the boundaries of the load–speed parameter space. In addition, the computation time is reduced from seconds to milliseconds, making the proposed framework practical for rapid wheel–rail adhesion analysis. download the full-text of this paper (PDF, 15 pages, 988 Kb) go to the previous paper go to the next paper return to the table of contents return to the volume description
Back to top	©Civil-Comp Limited 2023 - terms & conditions