Adjusted rail profile makes Dutch rail virtually free from Head Checks

Since 2006, infra-manager ProRail has been keeping the Dutch rail free from Head Checks by grinding the rails in an Anti Head Check profile. Head Checks (HC) are rail defects that are created by wheel-rail contact. These can be serious safety threats. I obtained my doctorate in 2010 with the dissertation titled ‘Design of an Anti Head Check profile based on stress relief 1,2. I designed a rail profile that saves ProRail €50 million of maintenance costs for the rails per year. This rail profile made the volume of HC decrease by over 70% since late-2008. Head Checks are becoming extinct in the Netherlands.

In the Netherlands, approximately 70% of the total annual maintenance budget is spent on rails, including foundation, sleepers, ballast, constructions and switches3. Rails may seem to be simple elements, but they deserve ample attention. The wheel-rail contact is the force that brings the degeneration of both separate systems together. All failing mechanisms can eventually be brought down to this dynamic contact system. This also played a role in the serious and fatal UK rail accident in 2000 at Hatfield, where rails affected by HC broke down.

Shocked by this, infra-manager ProRail took a look at the situation in the Netherlands. Inspections proved that here HC was a serious problem as well: 10% of the curves (rails) appeared to have been affected. The safety, reliability and availability of the rails was in danger. The problem increased and expo – nentially grew each year. In the peak year 2004, ProRail spent €50 million on fighting HC.

In 2001, the infra-manager with partners DeltaRail, Lloyd’s and the Delft Technical University set up a special RCF (Rolling Contact Fatigue) project group responsible for prevention specifically with respect to HC. The assignment was: find a cost-efficient and quick implementation solution for the problem in the area of maintenance at the lowest cost and with a low risk of derailment. As Head of the RCF Research Project Group, I decided to gain more in-depth knowledge of wheel-rail contact – contact forces and contact stress and wear, and eventually I obtained my doctorate at the University of Twente with my research into how to prevent HC.

Difference between Head Checks and Squats

Wear of rails or RCF can be classified into:

• Head Checks: multiple cracks in curves, in particular curves with a radius of 1,500m, in the upper leg in the gauge corner (trans – ition area between the gauge side and the running band)

• Squats: impressions and cracks in the running band on the top side of the rail head. Squats specifically occur in the free track (tangent track) and on yards, but with curves to a lesser extent.

Squats hardly form a threat to operational safety. The largest part is a single rail defect. They require a different approach than HC, so I entirely focused my research on Head Checks. These can grow from risky, serious cracks into a rail bursting open and breaking. My point of departure was the cause mechanism – first the situation in practice and then science – and how Head Checks can be prevented. I did not take the cracks and their development into account. My eventual goal was: a proven Anti Head Checks rail profile. The doctoral research started in 2007 at the University of Twente.

Coefficient of friction

Inspections carried out by ProRail show that Head Checks particularly appear in the upper leg in curves with a maximum radius of 3,000m and occur in a concentrated number between 500-2,000m. Apart from this, they occur at switches and crossings. The contact surface – contact where the wheel rolls down on the rail head – is very small. Often too large forces are transferred on this contact surface. Here too high spinning and slipping forces are created in curves with certain values of a coefficient of friction. The coefficient of friction between the wheel and the rail determines whether maximum shear stress will eventually occur at the surface or even on the inside of the rail material. HC-initiations are easily created with maximum shear stress on the surface. As long as the coefficient of friction remains below 0.4, there is a small chance of this happening. The weather continuously influences the coefficient of friction. For comparison: when the weather is dry and warm, it is 0.7 and in wet weather it is 0.05.

Contact geometry

Apart from the coefficient of friction, the load (axle load and MGT) and certainly the contact geometry as well – both wheel and rail profile separately, as the combined action of these two – are factors in the development of HC. Conformal contact between wheel and rail in curves causes HC-initiation in de-gauge corner. This can be prevented by guaranteeing a twopoint gauge side – running band contact in curves. The cheapest and quickest solution to the problem is optimum adjustment of the contact geometry.

It is possible to optimally adjust the wheel profile to the wheel-rail contact in curves, but this same wheel runs in a curve at one moment and the following moment on the free track. Therefore the profile does not always fit and this leads to instability of the train run, in particular at high speeds. The solution for fighting HC lies in adjusting the rail head geometry.

Anti Head Check rail profile

The HC-growth at the beginning of this century was mostly due to higher speeds, heavier and other types of rolling stock – the newest generation of trains – and higher stiffness in wheelsets for more stability and comfort for the travellers (Primary Yaw Stiffness). Another very important cause is the use of the renewed UIC rail type 54E1 in Europe. This was implemented because of the European unification, in order to allow trains to cross borders without problems. The rail type used in the Netherlands until the 1990s was much less sensitive to HC. Based on this 46E3, I designed in broad lines an Anti Head Check rail profile (AHC 54E1) in 2005. All this was still a best guess back then, but we very soon appeared to be on the right track.

Hertz-model impracticable

When the geometry of the wheel contact is OK, train wheels make contact with the rail under high dynamic load at two places in curves as described above. Literature search shows that 95% of the existing literature about Head Checks uses the Hertz-contact model as a basis for calculating this wheel-rail contact. In order to be able to predict where HC will develop, this model makes use of averages of the two calculated contact areas, as if this involves a wheel-rail contact with one contact point. Field research into HC shows that these theoretical values are way beside the truth. Use of the Hertz-contact model to simply calculate the equivalent of two contact points based on one plus one makes two results in an incorrect presentation of reality. This makes 95% of the literature about HC completely useless. Several countries have based their way of fighting HC on this literature.

For my research, the eventual design of the AHC 54E1 and the later standardised 54E5 profile, I started from a non-Hertzian contact model, based on the theory of Kalker2. The software model of Kalker (Contact) is based on half-space modelling which was later worked out by Dr. Zili Li (Delft University of Technology) into a quasi static quarter space modelling for the rail environment (software: Wear) for his doctoral research. During my research, I further developed this software in order to be able to calculate HC (two-point contacts between wheel and rail) in all cases.

Software validated

Based on the research into wheel-rail con tact and calculations and the behaviour of stiffly dimensioned trains (bogies) in curves, I designed the software programme for calculating the two-point contact surfaces. The programme calculates for each situation the HC initiation zone on rail heads by entering various parameters, such as train type, speed, load (axle load), and wheel and rail profile. With this we can quite accurately predict when and where HC may occur. A small margin remains, because it is never known if the rails are driven on by old or new wheels, but this is negligible in practice.

In tests with a wheel-rail test rig at VoestAlpine, the calculations always turned out to be correct. With this the developed software has been validated. HC already appear between 20,000-50,000 wheel passages. The test in the laboratory of VoestAlpine ended at 100,000 wheel passages, after which the running band of the rails started to show signs of HC. Also, tests on the rail outside confirmed the calculations of my software programme: Head Checks develop in the Netherlands with mixed traffic, in particular with the VIRM train type, almost always at a distance of 7-12mm measured from the gauge side at the gauge corner of the upper leg. Here the shear stress is at a maximum because of the high geometrical spin in the wheel-rail contact. In view of the place where HC occur, actually they should be called Gauge Shoulder Cracks (GSC).

Grinding programme for rails

The solution for the HC problem obviously is in a modification of the rail profile. This modification is calculated using the validated software. The new profile must be correctively ground in rails in the upper leg at curves with a maximum radius of 3,000m and after this it must be maintained every 15 MGT using a cyclic grinding system (reduction of 0.2mm). At the place where train wheels may cause a lot of spin and shear stress, a layer is ground off the rails, creating a gap between the wheel and the rail. Where there is air, there is no contact stress situation, which eliminates the chance of HC.

ProRail started this grinding programme in 2006. In 2009, almost all curves all over the Netherlands had been ground correctively into an Anti Head Check profile. At this moment, all old and new rails in curves have this profile, except in switches on yards where trains ride with speeds of under 40km/h. The AHC-profile is desired here as well, but it is not feasible in view of the costs.

The programme costs ProRail approxi – mately €20 million per year. With an innovation effort using another quality of steel, the required grinding frequency may be reduced. Further savings can be achieved in this area in the future.

Layer of rust as an indication

The software programme calculates at millimetre level how much is to be taken off at various places of the rails (54E1) in order to avoid harmful wheel-rail contact. One cannot just thoughtlessly grind off an equal layer at all places with a slanting corner line, the method applied by SNCF. For this there are too many differences in, for example, train type, load and driving speed. When too much is ground off, the train will ride restlessly. And when material is removed at the wrong place, HC will remain. Just when the grinding based on the results of the Hertz-contact model (see earlier).

Natural wear by riding trains will change the rail again in the course of time. The width of the layer of rust in the gauge corner directly shows the size of the gap, the space between the wheel profile and AHC rail profile during wheelrail contact (system) in a curve. The layer of rust is an indication for new cyclic grinding. There is no wheel-rail contact with a layer of rust in the gauge corner of 7mm wide or more. Therefore no risk of HC. With a reduction of a few millimetres, problems may already arise and the grinding train is to be used. It is important to manage the AHC profile as a maintenance activity through a balanced cyclic grinding system. This makes total absence of HC feasible and cost-effective for infra-managers.

Results

In comparison with the peak year 2004, ProRail managed to reduce HC by 70% already in 20084. Nowadays, the rail track has considerably fewer Head Checks in the rail track and the number is still decreasing every year. When they occur, the cracks grow half as quickly because of the HC-delaying rail qualities MHH (TataSteel) en 370LHT (VoestAlpine) that ProRail only applies in the upper legs in curves with a maximum radius of 3,000m. The result: a substantial reduction of safety risks. The rails have considerably longer life spans, varying from 2-4 years with HC to 23-35 years with the current maintenance using cyclic grinding. Failures occur less often and the emission of CO2 is decreased, because new rails are necessary less often. From a financial pointof- view, the new profile offers a net saving of €50 million per year for maintenance costs.

54E5, the first official standard Anti Head Check profile was introduced in Europe (EN13674) on 1 June 2009. Since that time, steel companies roll rails as a standard in this profile, so ProRail does not have to make extra costs for this type of rail. ProRail has HC under control thanks to the grinding programme. At this moment, the infra-manager focuses on preventing Head Checks in switches. On behalf of ProRail, I do research into squats and how these can be prevented. This is trail-blazing as well and world-leading at this moment. To be continued.

More information about the Anti Head Check Profile (54E5) and the software for calculating the grinding profile dependent on wheel/rail profile combined with the type of material can be obtained from the author and the owner through e-mail: [email protected].

References

1. Dollevoet, R.P.B.J., Li, Z., and Arias-Cuevas, O., A method for prediction of head checking initiation location and orientation under operational loading conditions, Proc. IMechE Part F: Journal of Rail and Rapid transit (JRRT), volume 224 no.5, September 2010, pp. 369 – 374.

2. Dollevoet, R.P.B.J., Design of an Anti Head Check profile based on stress relief, PhD dissertation, University of Twente, the Netherlands, 7th of October 2010, ISBN 978-90-365-3073-6.

3. Zoeteman, A., and Dollevoet, R.P.B.J., Combating rolling contact fatigue: strategies adopted in the Netherlands, Rail Engineering International, number 1, printed in the United Kingdom, ISSN 0141-4615, 2010, pp. 4-7.

4. Zoeteman, A., and Dollevoet, R.P.B.J., Successful examples of co-operation of wheel rail interface management in the Netherlands, World Congress on Railway Research (WCRR) at Lille, France, 25th of May 2011

About the author

Dr. Rolf Dollevoet has always worked on R&D departments as Programme Leader. Since 2003 he has been active at ProRail as System Expert Track and leader of RCF- and wheel/rail contact project research groups. Research was focused on RCF; especially Head Checks (HC, since 2001) and later Squats (2005). HC is already solved in theory (see UTwente PhD dissertation 2010: Design of an Anti Head Check profile based on stress relief3, and this article) and successfully proven in the Dutch network (track). Today, HC renewal costs reduction of €50 million has almost been achieved in Holland. Squats research is still going on in cooperation with TU Delft (Railway engineering department (Dr. Z. Li) and will be finished in 2013. The Wheel/Rail conditioning research project has started up as a pilot. Goal is to manage the friction coefficient independent of weather on the Dutch network. First priority is to reduce the noise and secondly the wear of wheel/rail during operations.