Louis Girardi is Head of Rail Maintenance Standards & Policy Division, SNCF; Dipl-Ing René Heyder is Engineer Materials, Damage analysis at DB Systemtechnik; Didier Lévy is a track expert with RATP in Paris; and Daniel Boulanger is Technical Manager, Corus Rail Products. The authors would like to acknowledge support for the IDR2-Novum research programme received from RFF and Philip Montier
THE INITIATIVE for Development & Research on Rail (IDR2) began with an agreement between SNCF, RATP, rail maker Corus and four research institutions - the French National Institute for Transport and Safety Research, Paris (Inrets), Institut National des Sciences Appliquées de Lyon (Insa), plus the Laboratoire de Mécanique des Solides (LMS) and Mecamix at the École Polytechnique in Palaiseau.
Signed in 2002, the agreement aims to make rail maintenance more cost-effective by co-ordinating research programmes, integrating the results of theoretical work and benefiting from the latest developments in rail products. Financial support is provided by Réseau Ferré de France.
In 2003 three German partners working on the closely-related Novum research programme joined IDR2. DB, the Federal Institute for Materials & Research (BAM) and the GKSS Research Centre in Hamburg are financially supported by the Federal Ministry of Education & Research.
IDR2 pays particular attention to the behaviour of track under different conditions, covering theoretical computation of wheel-rail forces under various vehicles, surface forces between wheel and rail, internal stresses and crack propagation. The programme seeks to determine whether particular forms of defect are caused by the properties of different types of rail or by wheel-rail interaction. Throughout the programme, the research teams went to considerable trouble to ensure that a realistic environment was represented.
Driving the programme is the need for a better understanding of maintenance requirements as train frequencies, speeds and axleloads increase. These all contribute to higher maintenance requirements at a time when there is strong pressure to make cost savings. Current maintenance methods including grinding and arc welding repair have been studied to understand their effectiveness, improve their performance and reduce costs.
One of the basic philosophies of the programme is that rail behaviour cannot be explained without considering the various interaction phenomena between track and vehicles, including the influence of rail surface and track geometry upon forces generated by the passage of vehicles. Geometry is affected by deformation under load and by local irregularities, some of which can arise from routine rail maintenance operations such as welds or arc welding repairs.
Another basic idea is that, in many cases where rail cracks occur, there is a 'stable' crack growing phase which can be monitored before the rail breaks. This expansion phase can be modelled using classical crack expansion principles with stresses governed by vehicle-track interaction. The research also considers crack dimensions and location to derive a strong correlation between theoretical and actual behaviour.
An important part of the programme is devoted to investigating the mutual influence of all these factors. This requires extensive use of modern numerical models in vehicle and track dynamics, as well as engineering observations about rail maintenance operations. High-level theoretical tools are necessary, but these must rest on engineering results and maintenance case studies.
The programme is divided among four task groups. The first is charged with determining the forces between wheel and rail and between the rail and its supports. The second and third task groups are looking at material properties and internal stresses in the rail section, while laboratory and track validations are being undertaken by the fourth task group; all this work is due to be completed by December this year.
Track observations are being carried out by railway and industrial partners. These will allow a clearer insight into the main physical and mechanical factors affecting rail behaviour, helping to rank their mutual influence.
It is essential that realistic loading scenarios are used in the laboratory experiments if the results are to be transferable to actual operating conditions, and so the programme managers decided to use the measurements of dynamic vertical wheel-rail forces taken at three DafuR test sites in Germany.
DafuR is a diagnostic system for testing out-of-round wheels, and 28 measuring sites are installed across the DB network. The core of the DafuR system is the 'long Q-force measuring station' which records dynamic vertical forces arising as a train rolls over the rail. Fig 1 shows a Q-force measurement where the mean value of the Q-force profile corresponds to the static wheel load.
Data collected and evaluated over three months will be used to calculate frequency distributions of the wheel-rail contact forces as a function of running speed for freight and passenger trains. This will identify certain wheel defects such as out-of-roundness and flats, as well as establish time profiles for selected train movements.
Control of wheel-rail contact
Modification of the wheel-rail contact is commonly achieved by grinding the rail gauge corner to eliminate or reduce head checking. Tests to determine the effectiveness of grinding have been performed on conventional and high speed lines in France at sites which suffered from head checking on the outer rail.
Both preventive and curative grinding has been used in the tests, with different amounts of metal ground from the gauge corner of the rail. A small amount was removed for preventive grinding, and 0·7mm was ground off when curative grinding was required (Fig 2).
Periodic hardness monitoring has been conducted at different points over the same rail sections (Fig 3). These points are located at the centre of the rail (blue dots), near the rail centre in a very hard zone (green dots), on the rail corner (pale blue dots) and on a bright zone (red dots) which appears when gauge corner contact between wheel and rail resumes. This bright zone forms a parameter that is easy to check for monitoring the end of a preventive or curative cycle.
Practical observations on different railways reveal a wide range of defects over the rail surface. These include phenomena such as squats and the Belgrospis (crack clusters) observed by DB, which seem to be caused by non-uniform surface forces. Observations also confirm that head checking appears most frequently on the gauge corner of the outer rail in curves.
Different types of internal crack exist, and these may be caused by surface cracks or existing initial inclusions, located at different positions inside the rail head or elsewhere in the rail section. In curves, they are often located near the gauge side, whereas in tangent track they are typically found near the centre of the railhead (below left).
From these examples, we can see whether a defect is initiated internally or externally.
Contact patterns and forces
Hertzian or semi-Hertzian computations give a pseudo-elliptical print for the wheel-rail contact patch, with different patterns for high and low rails (Fig 4).
Differences exist between the two rails because of lateral forces which tend to move the axle towards the external rail, leading to a different contact pattern for the two wheels. Computed pressure patterns have allowed a better understanding of the wheel-rail contact, and improved control of the contact area through grinding.
For a locomotive, pressure patterns are very different after shallow grinding of the gauge corner of the rail (Fig 5a) and after deep grinding (Fig 5b). With a wagon, pressures are more centred on the rail tread (Fig 5c). This suggests that only traction vehicles cause head checking, but more information is needed to support this theory.
If the pressure pattern for shallow grinding of the gauge corner in the case of a single rail and a single wheel is compared with that of a single rail and different wheels, it becomes clear that because of wheel and rail tolerances, strong profile modification is needed to alter the wheel-rail contact pattern significantly.
The track and rail are submitted to external constraints in the vertical and transverse directions, which can cause dynamic effects. These depend on the track geometry, but because of the inertial effects, they exist even in perfect track.
Under the IDR2 programme, multibody load computations will be undertaken, coupled with 3D finite element modelling which takes track and rail dynamics into account. In the case of a moving constant load, the moving force represents the wheel contact, allowing derivation of the dynamic vertical translation at the location of the force. These computed results show that even in the case of perfect track, combined track and rail dynamics cause different behaviour according to the speed of the train. This can also be stated on the basis of experimental data.
Arc welding is commonly used to eliminate rail surface defects (below). Successive heating phases before welding help to avoid rapid cooling which would lead to the formation of martensite. Different thermal deformations occur after heating, and a small residual vertical deformation usually remains after the repair has been completed.
As these residual geometric defects may cause harmful dynamic effects, especially on high speed lines, a transient elasto-plastic thermo-mechanical model has been tested to analyse the effects of boundary and thermal conditions. The purpose of this is to check if the simulation agrees with real-world data, and to fine tune the heating and weld deposit processes to obtain better residual geometry of the overall longitudinal rail shape.
After the preliminary pre-heating phase, the increasing heat during the successive weld layings causes the rail to lift. After completion of the repair, the rail cools, but a final residual permanent deformation remains (Fig 6).
Material properties and mechanical behaviour
IDR2 includes a series of field tests to ascertain if head checking cracks are eliminated naturally where there is considerable wear on the gauge side of the outer rail. The chosen test site was a 985m radius curve at Saint-Benoît, near Poitiers on the Paris - Bordeaux main line, where trains run at 120 km/h. The external rail had been replaced in 1997 by a combination of perlitic 800 (220), 900 (260) and MHH (micro-alloyed, heat-treated) grades.
After an initial grinding operation and 134 million tonnes of accumulated tonnage, the following conclusions were drawn:
- all grades suffered head checking;
- the head checking zone is generally more work-hardened, except for the MHH rail;
- wear is higher for the 800 and 900 grades;
- head checking is less developed for MHH than for other grades (p442). A further conclusion is that for both wear and head checking criteria, MHH appears to be outperforming all other grades.
Bainitic rail was tested at Dieupentale, near Toulouse on the Bordeaux - Toulouse line, where traffic is a mix of TER regional trains and TGVs running at up to 160 km/h. A bainitic rail with hardness of 320HB was laid to replace an external 900 A rail in a 1140m radius curve which was strongly affected by head checking. No grinding was carried out until accumulated tonnage had reached 80million.
The two grades display very different head check development. While there are no head checking cracks on the bainitic rail, cracks are clearly visible on the 900 A rail.
Both the test zones revealed how important material properties can be in making an assessment of rail behaviour; they also served as useful references for validation of the numerical models.
Mechanical behaviour of rail
Much attention has been given to improving knowledge of the material properties of rail steels and to feeding these properties into numerical simulations. The mechanical properties of rail differ at every location, but it is clearly not possible to take test samples that would allow material characteristics over the entire rail section to be known.
IDR2 therefore uses the results of successive indentation tests, in which pressure is applied by a small indenter onto various parts of the rail section, and the mechanical characteristics deduced from the measured force and displacement. Material properties can be identified using a finite element computation, fitting calculated force/displacement responses to the measured data.
Crack initiation and propagation
There are three main phases in crack development: initiation; slow growth; and rapid growth before the break occurs. The slow crack growth phase is really what matters for rail maintenance, because inspection work needs to be minimised without risking a rail break.
The German partners in the project are responsible for carrying out investigations into the first phase of crack development, using a wheel-rail test rig operated by DB Systemtechnik. The second phase of slow growth is being examined by means of fatigue testing on a 1MN servo-hydraulic resonance test machine at the Federal Institute of Materials Research & Testing in Berlin.
The IDR2-Novum research programme covers a wide field of theoretical as well as applied rail research. One of the primary benefits expected is better prediction of the crack growth rate inside the rail in order to establish the opportunities for improvements in track monitoring, component design and service conditions.
Thanks to a system approach it is possible to vary a range of key parameters to improve current maintenance processes such as lubrication and rail profile modification using grinding or other methods. Finally, the programme methodology may lead to a need for significantly fewer tests, and which will be of shorter duration with better understanding of their results.
Rail manufacturers as well as railway infrastructure managers stand to benefit because of the strong links between rail product specifications, maintenance needs and research results.
In the near future the project team hopes that use of a similarly comprehensive approach will help reduce maintenance costs and deliver products that meet the increasingly heavy demands placed on rail networks.
- CAPTION: TOP: One of DafuR's test sites on the DB network
- Fig 1. Q-force profile recorded during passage of a wheel over a rail at a DafuR test site on the Halle - Erfurt line in Germany
- Fig 2. Preventive grinding (left) and curative grinding (right) of the gauge corner on rail affected by head checking in France
- Fig 3. Periodical hardness monitoring of the rail surface
- CAPTION: Progressive transverse rail cracking in curves (top) and in tangent track (above)
- CAPTION: Fig 4. Wheel-rail contact pattern for high rails (left) and low rails (right)
- CAPTION: Fig 5. Different pressure patterns generated by a locomotive on the rail head with shallow (top) and deep (centre) grinding on the gauge face; the pressure pattern for a wagon is also shown (bottom)
- CAPTION: Arc welding to repair a rail surface defect
- CAPTION: ABOVE: Head checking is less developed for MHH rail than for other grades
- CAPTION: Rail surface showing a mixed flash-butt weld between bainitic and 900 A rail. There are no head checking cracks on the bainitic rail, but cracks are clearly visible on the 900 A rail
- Fig 6. The effect of successive weld layings used to repair a surface defect is to lift the rail, but after cooling there is a residual permanent deformation
- CAPTION: DBSystemtechnik's test rig at Kirschm?€?ser is being used to investigate the initiation of cracking
IDR2-Novum programme promises lower rail maintenance costs
A wide-ranging programme of theoretical and field research suggests that a system approach to track-train interaction can lead to more cost-effective rail maintenance. The programme includes investigation of the mutual influence of the rail surface and track geometry on wheel-rail forces, requiring the use of high-level theoretical tools and modern numerical models in vehicle and track dynamics. Other work embraces computation of wheel-rail forces under different vehicles and the influence of variations in the properties of rail steels. Four groups of experts drawn from France and Germany are charged with completing the programme.
Le programme IDR2-Novum promet des coûts de maintenance ferroviaire réduits
Un large programme de recherches, théoriques et d'application, suggère qu'une approche globale de l'interaction entre train et voie peut conduire à une maintenance ferroviaire plus économique. Le programme comprend l'investigation de l'influence mutuelle de la surface du rail et de la géométrie de la voie sur les forces roue-rail, nécessitant des outils théoriques de haut niveau et des modèles numérisés modernes concernant le véhicule et la dynamique de la voie. Les autres travaux comprennent l'évaluation des forces roue-rail sous différents véhicules et l'influence des variations de propriétés des aciers pour les rails. Quatre groupes d'experts appelés de France et d'Allemagne sont en charge de l'achèvement du programme.
IDR2-Novum-Programm verspricht geringere Gleisunterhaltskosten
Ein breit angelegtes Programm an theoretischer und praktischer Forschung lässt folgern, dass eine System-Betrachtung der Interaktion zwischen Zug und Gleis zu kostengünstigerem Gleisunterhalt führen kann. Das Programm umfasst die Untersuchung der gegenseitigen Beeinflussung von Schienenoberfläche und Gleisgeometrie auf die Rad-Schiene-Kräfte, welche hochwertige theoretische Werkzeuge und moderne numerische Modelle von Fahrzeug- und Gleisdynamik ben?€?tigt. Andere Arbeiten betreffen die Berechnung der Rad-Schiene-Kräfte unter verschiedenen Fahrzeugen und den Einfluss von unterschiedlichen Eigenschaften der Schienenstähle. Vier aus deutschen und franz?€?sischen Mitgliedern zusammengesetzte Expertengruppen sind für die Fertigführung des Programms verantwortlich.
El programa IDR2-Novum promete menores costes de mantenimiento de raíles
Un amplio programa de investigación teórica y de campo sugiere que darle un enfoque sistem? tico a la interacción entre la vía y el tren puede proporcionar un mantenimiento m? s rentable del carril. El programa incluye una investigación de la influencia que ejercen la superficie del carril y la geometría de la vía sobre las fuerzas rueda-carril, lo que requiere la utilización de herramientas teóricas de alto nivel y de modernos modelos numéricos en la din? mica del vehículo y de la vía. Otros estudios engloban el c? lculo de las fuerzas rueda-carril con distintos vehículos y la influencia que ejercen los cambios en las propiedades del acero del carril. Cuatro grupos de expertos venidos de Francia y Alemania se encargan de completar el programa.