Estimation of wheel-rail friction at vehicle certication measurements

Sammanfattning: In certification of new rail vehicles with respect to running characteristics, a wide variety of operating conditions needs to be considered. Behaviour on straight and curved tracks, including twisted tracks at various speeds, is of great importance. However, the wheel-rail friction should always be high corresponding to dry conditions. It means that the certification tests have to be carried out during dry weather conditions and unlubricated rails. But measuring the friction at test conditions is a great challenge.Therefore, in a recent work (Petrov et al.), an algorithm was proposed for the continu-ous estimation of wheel-rail friction along both rails. The algorithm is based on wheel-rail forces in all three directions (Y, Q, X ) for both wheels in a wheelset, lateral contact po-sition on its wheels and wheelset angle of attack. The algorithm was evaluated with a fictional vehicle with vehicle-track dynamics simulations on various tracks (straight and curves) and track irregularities.In cooperation with SNC Lavalin (formerly Interfleet Technology), an opportunity arose to get the required data from on-track tests. In this way, all nine quantities above were measured during test runs of a new vehicle, so the algorithm could be evaluated under realistic conditions. The tests in tight curves of radius 150 m are used in the present work for this purpose. The measured data and the algorithm were processed in a Matlab program to get the friction estimate. Apart from the friction, the creepages and spin are also estimated with the aim to serve as quality indicators of the estimated friction.As the vehicle measurements include noise, errors and uncertainties, a statistical tool was introduced. Moreover, a sensitivity analysis was performed. It was observed that in these tight curves, the friction estimation on the outer wheel is poor, but a phenomenon arose that with increasing spin corresponding to even higher contact angle the friction might be estimated again. However, small spin gives a good friction estimation provided the total creep is high enough. Therefore, sharp curves, traction/braking or large track irregularities are necessary to estimate the friction well.In order to continue the work deeper into the area, other tests with the above scenarios would be useful