Wear simulation using AC-wheel-rail-element

Wear of rail profiles along track and wheel profiles can be simulated during a simulation run with numerical integration. Wear is based on abrasive wear theory according Archard /1/. Wear volume is proportional to frictional work. It depends on mild and heavy wear. Frictional power in wheel/rail contact is computed in parallel to the simulation in time domain. Profiles used in simulation are changing due to wear. The module was validated comparing with measured wear in a turnabout of S-Bahn Stuttgart, Germany.

Wear on wheels can develop quite fast. The wheel shown has done only 20tkm. But fast means still it took some month. 

worn wheel after 25 tkm

The WEAR module runs in parallel to the simulation, changing profiles according defined timesteps, wear amount or distance steps. The advantage of  wear computation during dynamic simulation is shown using a simple example. Two wheelsets having the same profiles.The wheelset is driven by a moment of 6 kNm.

   Simulation was running for 10 s with an output step size of 0.01 s. Profiles are generated for each output step, that means 1000 profiling steps or wear steps have been done. Compared to a simulation of the model without wear computation simulation is slower by about 9 times. Sampling wear and computing wear profiles at the end of each simulation run would mean to get 9 times wear profiles for the same cpu-time. Dividing the 1000 profiling steps of the parallel method by 9 means that parallel method is approximately 111 times faster. In the 10 s a load of 1.8 MT was passing the site and the wheels where doing a mileage of 253 Tkm.

Validation of the module

First verification of the wear module was done in a  project  „Analysis of wear between wheel and rail for inclined rails of 1/40 due to wheels having profiles according DIN 5573”, sponsored by Deutsche Bahn AG. Wear in a turn about of S-Bahn in Stuttgart was measured and simulated using the MBS-tool MEDYNA of ArgeCare.

Best approach of wear distribution, amount of wear and load was reached with the two vehicle model having a 50/50 mix of measured wheel profiles having done 45 tkm and 115 tkm.

  What can not be simulated yet is the flow of the material on low rail due to plasticity effect.   

A result gained with VI-Rail concerns a tram wheel profile. At the User-conference 2014, Fon Azobi from Bombardier Transportation, Bogies, in Siegen presented the comparison of wheel profile measurements and wear simulation of a street car. They modelled the actual track the vehicles where running and made a comparison after 27 Tkm and 39 Tkm the vehicle had done. Beside the fact that flange height of real vehicle and model was different there is nearly no difference between simulation and measurement.  

Black is the new profile, dark blue the measurement and light blue simulation. Wear in treat is slightly smaller and in transition tread to flange slightly bigger in  Some tasks which have been done: Find the wheel profile having least wear for a vehicle running on curved track with comparatively small radii.  Rail profile is UIC60 with inclination 1/30. Wear was summed on wheelset 1 and 3 and on wheelset 2 and 4. Simulation thus includes runs for left and right hand curves but only in one direction.  Diagrams show wear of wheelset 1 and 3 and 2 and 4.

The diagram wear area about mileage shows, wear is developing exponential. For profile 1 wear grow is fastest and simulation was stopped when flange wear got to the limit. Profile 3 is the one which behave best.

The next few diagrams show profile development on wheels on axle 1 and 3 and of axle 2 and 4. For profile 1 there is big flange wear on axle 1 and 3 and nearly only tread wear on wheelset 2 and 4.

Profile development and wear area of wheel profile 1 depending on mileage of the vehicle

 Profile development and wear area of wheel profile 3 depending on mileage of the vehicle

Profile 3 has nearly same wear amount on wheelset 1 and 3 and 2 and 4. Wear develops nearly only on tread. Steering of the vehicle is much better with this profile than for profile 1.

Wear in a tram switch

Question was: will wheel – rail forces exceed the limit depending on the amount of wear.

Forces did not exceed the limiting value, but wheel climbing occurred. Wear simulation was stopped, when wheel climbing exceeded 5 mm..

Modelled profiles in the frog area are given in red.  Distance between profiles is not constant. But wear calculation is done for same distance around profiles with constant wear step size in distance.                                       

When wheel lift exceeded 5 mm, the profiles in the frog looked like the bottom diagrams.