Optimisation of railway crossing geometry considering a representative set of wheel profiles

Abstract

A numerical method for robust geometry optimisation of railway crossings is presented. The robustness is achieved by optimising the crossing geometry for a representative set of wheel profiles. As a basis for the optimisation, a crossing geometry is created where rail cross-sectional profiles and longitudinal height profiles of both wing rails and crossing nose are parameterised. Based on the approximation that the two problems are decoupled, separate optimisations are performed for the cross-sectional rail profiles and the longitudinal height profiles. The rail cross sections are optimised to minimise the maximum Hertzian wheel–rail contact pressure. The longitudinal height profiles are optimised to minimise the accumulated damage in the wing rail to crossing nose transition zone. The accumulated damage is approximated using an objective criterion that accounts for the angle of the wheel trajectory reversal during the transition from the wing rail to the crossing nose as well as the distribution of transition points for the utilised wheel profile set. It is found that small nonlinear height deviations from a linear longitudinal wing rail profile in the transition zone can reduce the objective compared to the nominal design. It is further demonstrated that the variation in wheel profile shapes, lateral wheel displacements and the feasible transition zone length of the crossing will determine the longitudinal height profiles of the wing rail and crossing nose if all wheel profiles are to make their transition within the transition zone.

Keywords:

• railway
• crossing
• geometry
• optimisation
• wheel profile

Acknowledgements

This work has been performed within the Centre of Excellence CHARMEC (CHAlmers Railway MEChanics) (see www.charmec.chalmers.se). The author would like to thank Prof. Jens C. O. Nielsen for fruitful discussions and proof reading. Correspondence with Mr Heinz Ossberger and Mr Erich Scheschy of VAE GmbH and Mr Ingemar Persson of DEsolver AB are also gratefully acknowledged.

Funding

This work was partly funded by Trafikverket (the Swedish Transport Administration), VAE GmbH and SL (Stockholm's Public Transportation).