Add us to your taskbar by dragging this icon RGI logo to the bottom of your screen.

Close

Management and research tackle rolling contact fatigue

01 Jul 2003

INTRO: Rolling contact fatigue is high up the agenda of Dutch infrastructure manager ProRail. A research policy has been established, and with knowledge shared internationally, ProRail has started to modify rail composition and head profiles, and adopted a regular grinding policy

BYLINE: Jeroen Smulders

Project Manager, Rolling Contact Fatigue, ProRail

INVESTIGATION of a new type of rail defect caused by rolling contact fatigue (RCF) began in the Netherlands in the mid-1990s. Defects can be subdivided into head checks, squats and shelling (Fig 1). Squats are fatigue defects that grow from an indentation in the rail. Shelling defects grow from a material imperfection in the rail, and are very rarely seen in modern, clean steels.

Until the mid-1990s a view prevailed that head checks would seldom lead to a vertical rail breakage, but in 1996 our research found that cracks resulting from head checks can cause a vertical breakage much more often than had generally been assumed, and at the end of the decade it was found that there had been strong growth in the scope of the phenomenon with many more occurrences.

In the past, the RCF problems were not present or were present only to a limited degree. The immediate cause of the problem is excessive tangential stresses overloading the rail surface, but it is clear that changes to the design of infrastructure and rolling stock, in network usage and in maintenance practices over a long period lie behind its appearance (Fig 2).

Investigation of the problem in the Netherlands was intensified and large-scale inventories made following the Hatfield derailment in the UK on October 17 2000, the result of a multiple rail breakage caused by head checks (RG 3.01 p157). ProRail's maintenance management department ruled that combating RCF was a high priority task, and a national project to alleviate the problem was therefore launched.

The objective of the project is to identify the causes of the problem, find methods of keeping it under control and determine a more preventative approach. Policies must be incorporated into ProRail's regulations and should be implemented in operational processes.

Close co-operation between ProRail, its maintenance contractors, Eurailscout, AEA Technology Rail and Nedtrain Consulting resulted in the compilation of a policy plan covering the operational process, which came into effect on October 29 2002. This plan is followed throughout the Netherlands, ensuring a uniform approach to RCF and producing consistent and quantifiable information. Regular discussions with all parties involved, including maintenance contractors, cover all RCF problems, with sharing of knowledge on investigations performed and experience acquired.

Work to investigate the RCF phenomenon began as early as 1996. Initial investigations concerned RCF and crack growth speeds, and at the beginning of 2000 research began to target the influence of rail qualities and railhead geometry.

Close co-operation with other parties is one of the main pillars of the project. To ensure that budgets were spent efficiently, in 2001 we entered into a co-operative association with Railtrack for two-way sharing of experience and knowledge. Bi-annual workshops are held in which the results of the investigations are considered and discussed, increasing the speed at which knowledge is advanced.

In order to retain a sharp focus on the most likely answers to the problems of RCF, in July 2002 ProRail analysed the best short and long-term solutions (box p433). We have since primarily focused on those methods that will bring the quickest and most effective improvements.

Our inventory showed that straight track was only affected to a very limited degree. Switches were much more heavily affected, but the greatest return on investment could be achieved on curved track.

Testing programme

During 2000 various types of rail were installed for testing on a number of curves susceptible to head checks. The aim was to identify materials with low crack growth speeds.

Micro-alloyed head-hardened (MHH) rails showed the best results. Very shallow head checks did occur rapidly, but exhibited a low speed of growth in line with expectations. Grade 260Mn carbon-manganese rail, the standard grade in the Netherlands, was installed as a reference, and this exhibited head checks at a later stage. Once the cracks were established, they grew faster and soon far exceeded the crack depth in the MHH rails.

The MHH rails were approved for use in November 2002, and will now be installed in locations susceptible to head checks. It is expected that the service life of the rail will be increased, and grinding intervals extended.

In October 2003 we plan to test a two-layer rail developed as part of the European Infrastar Project. This consists of a rail web made from standard 260Mn grade steel on to which a 3 to 5mm thick 'Duroc' alloy layer is welded using a laser cladding process (Fig 3). We expect that no head checks at all will occur in this rail.

Another approach has been to devise an anti-head check profile for 54E1 rails, based on European Railway Research Institute D173 reports. Developed in consultation with Speno, this profile relieves the rail shoulder by moving the wheel-rail contact area towards the top of the rail. A test section was installed in 2001, and on the basis of the results the anti-head check profile has been ground in at susceptible sites since mid-2002 (Fig 4).

ProRail has also been working in conjunction with Network Rail of the UK to develop a national rail grinding policy. Over the next few months this 'technical' RCF grinding policy will be converted into an operationally-achievable grinding strategy which will be rolled out across the network in 2004. At specified sites with sharp curve radii, this will see approximately 0·1 to 0·2mm of metal removed for every 15 million gross tonnes passing over the rail (Table I).

Optimising wheel profiles

As RCF occurs at the wheel-rail interface, our research programme has also looked at rolling stock. Earlier investigations had shown that two types of electric multiple-unit frequently operated over sections of track severely affected by RCF, namely the ICM and DD-IRM classes. The wheels of these classes have also been found to exhibit RCF defects (Fig 5) similar to those found on rails, and investigations showed that these defects originated when the trains passed through curves.

Over the last decade the wheelset mileage of the NS fleet decreased by 50%, from around 300000 km to 150000 km between reprofiling. This is because wheelsets now tend to exhibit defects such as flats, squaring and pitting more frequently and at an earlier stage in their wear history. As rolling stock tends to have a newer wheel profile with less wear, the wheel-rail contact is altered, and when vehicles are negotiating a curve, the same piece of rail material is under load more often. Reduced load distribution contributes to the formation of RCF.

In consultation with the operator and maintainer, it was decided to perform a test with a newly-determined wheel profile, known as Modified S1002 (worn). The improved wheel profile would move the wheel-rail contact area to the top of the rail head during the negotiation of a curve (Fig 6), so reducing the contact stresses. This more favourable distribution of the contact area and the resulting reduction in contact stresses would ideally ensure a lower RCF crack growth speed and a longer wheelset service life.

The tests have involved two ICM trainsets, each with the standard S1002 profile on 50% of the axles and a modified profile on the remainder. The results should be available in October this year.

Detection and prevention

The extent of RCF is monitored by a programme of visual inspections that are undertaken twice-yearly by maintenance contractors. These inspections aim to determine the quantity, precise location and severity of head checks and squats, quantifying the length of plain line track affected by these defects or the number of switch and crossing units. Head checks are classified as Light, Moderate, Heavy or Severe (Figs 7a, 7b, 7c, 7d), while squats are subdivided into three classes - A (Light), B (Moderate) and C (Heavy) (Figs 8a, 8b and 8c). Maintenance contractors are provided with reference photographs, enabling them to identify RCF defects.

Data produced in April this year reveals the scale of the RCF problem in the Netherlands. On a network with 6690 km track-km, 399·6 km of rail has head checks with 25 km falling in the Severe category (Fig 9a). Squats have been detected on a total of 1072·6 km, of which 74·8 km are Heavy (Fig 10a). Head checks were also detected on 2836 out of 8700 switches and crossings (Fig 9b), of which 98 were Severe, while squats were present in 1179 units, of which 125 fell in the Heavy category (Fig 10b).

Rails that exhibit heavy or severe head checking are investigated more closely by means of ultrasonic inspection using hand-held equipment. Depending on crack depth, the rail can be replaced immediately, after four weeks or after three months. In some cases a temporary speed restriction is placed on the affected section of track, and in other cases emergency fishplates can be fitted to extend the period of time before the rail must be replaced.

In contrast to the reactive strategy of detection and replacement followed by ProRail until mid-2002, a more preventative approach is now possible thanks to recent research. This comprises the application of the anti-head check profile as standard during preventative grinding at susceptible locations, and since the beginning of 2003 the use of rail meeting specification PVE00090 at such locations on plain track; MHH rail complies with this requirement.

Susceptible sites are defined as curves with a radius between 750m and 3000m and where annual traffic exceeds 5 million gross tonnes. On curves with a radius less than 750m, wear-resistant 350LHT or 350HT rail is used, while standard 260Mn grade rail is used on curves with a radius above 3000m. The introduction of a cyclical grinding programme in the course of 2004 should complete ProRail's RCF prevention programme.

The cost of RCF

Although we have taken major steps towards RCF prevention, investigation is still required in several areas. Under a further programme of research agreed by ProRail and Network Rail in April 2003, topics for investigation include the scope for automatic RCF detection, accurate crack depth determination, time of flight defraction detection methods, and the influence of trackwork stiffness.

In 2002 ProRail spent €22m on tackling RCF, of which €18m was for the replacement of sections of rail over 100m in length. Smaller sections, replacement switch and crossing components and maintaining inventories of replacement parts cost €3m, while central project and research costs amounted to €1m.

With the quantity of head checks remaining reasonably stable, this level of annual investment appears adequate to stabilise the RCF problem. We estimate that the preventative policy of using special rails, anti-head check profile grinding and cyclical grinding should deliver a cost reduction of at least 60%.

Tackling RCF in the Netherlands

Short-term solutions

Special rail qualities

Modified rail head geometry

Modify the rail maintenance regime

Long-term solutions

Modified wheel profiles

Modify track stiffness

Modify rolling stock characteristics

TABLE: Table I. Grinding schedule adopted by ProRail to control RCF

Grinding category Curve radius Grind after

Class 1 Severe, under 2000m 15million gross tonnes

Class 2 Mild, 2000m to 3000m 30million gross tonnes

Class 3 Tangent, over 3000m 45million gross tonnes

The ProRail network

6690 km track-km

8700 points and crossings

4526 bridges, viaducts and tunnels

2964 level crossings

382 stations

CAPTION: Fig 1. Head checks (top left), squats (left) and shelling (bottom left) are three examples of rolling contact fatigue that generated considerable concern in the Netherlands in the late 1990s

CAPTION: Fig 2. A combination of technical changes affecting the wheel-rail interface over recent decades lies behind the RCF phenomenon. Taken together, these changes have led to overloading of the rail surface in the form of excessively high tangential stresses

CAPTION: Fig 3. Trials are planned with a novel rail design featuring an alloy layer welded to the head using a laser cladding process

CAPTION: Fig 4. In consultation with Speno, ProRail has adopted a profile for its 54E1 rail designed specifically to combat head checks. Grinding the gauge corner from profile O to profile N moves the wheel-rail contact patch closer to the centreline of the rail head

CAPTION: Fig 5. RCF damage to the wheel surface of an NS EMU - the phenomenon occurred most frequently on ICM (below) and double-deck IRM electric multiple-units

CAPTION: Fig 6. Distribution of the contact position on a wheel during negotiation of a curve as a function of distance run using the S1002 profile. The adopted modified profile is equivalent to the S1002 profile after 75 000 km. On the horizontal axis 0 represents the centre of the rail

CAPTION: Fig 7. ProRail identifies four different categories of head check: Light (a), Moderate (b), Heavy (c) and Severe (d)

CAPTION: Fig 8. Three categories of squats are used by ProRail: Light (a, top), Moderate (b, centre) and Heavy (c, bottom)

CAPTION: Fig 9. Data produced in April revealed that nearly 400 km of plain line on the 6 690 track-km Dutch network were affected by head checks, which ProRail placed into four categories (a). Head checks were also identified at 2 836 locations on switches and crossings, 98 being in the Severe category (b)

CAPTION: Fig 10. In April squats were found on more than 1 000 track-km, with nearly 75 cases in the Heavy category (a). There were 125 squats in the Heavy category at 1 179 locations on switches and crossings (b)

CAPTION: Fig 11. ProRail is testing eddy-current inspection equipment to detect RCF phenomena

La direction et la recherche prennent les défauts du rail à bras le corps

Management und Forschung nehmen Gleisdefekte in Angriff

Gestión e investigación abordan los defectos de los carriles

ProRail, le gestionnaire d'infrastructure néerlandais, a lancé un programme urgent afin de combattre un accroissement rapide du nombre de défauts des rails causés par la fatigue au niveau du contact du roulement. Travaillant de manière rapprochée avec les entreprises de maintenance et partageant l'expérience avec Network Rail, au Royaume Uni, ProRail est à l'initiative des tests devant déterminer la meilleure façon de se saisir du problème, en insistant sur les méthodes qui ont offerts les améliorations les plus rapides et les plus efficaces. De nouveaux types de rails et des roues au profil modifié vont être essayés, une nouvelle politique de meulage sera mise en place sur le réseau en 2004

Die niederländische Infrastruktur-Organisation ProRail hat ein Dringlichkeitsprogramm zur Bekämpfung des rapiden Anstiegs der Anzahl Gleisdefekte, welche durch Rollkontakt-Ermüdung auftreten. In enger Zusammenarbeit mit den Unterhalts-Unternehmungen und einem Erfahrungsaustausch mit Network Rail in Grossbritannien hat ProRail Versuche angeordnet, welche den besten Weg zum Angriff des Problems aufzeigen sollten, wobei die Methoden, welche eine schnelle und kostengünstige Verbesserung herbeiführen sollten prioritär behandelt werden. Neue Gleistypen und modifizierte Radprofile werden untersucht, und eine neue Schleifpolitik wird 2004 netzweit eingeführt

El gestor de infraestructuras holandés ProRail ha lanzado un programa urgente para combatir el r