Maintenance is a vital part of the system approach: appropriate and timely maintenance forms an integral part of the system approach that needs to be adopted by railways if they are to meet the demanding challenges being put upon them

Jay Jaiswal

THE primary mechanisms that deteriorate both the rail and wheel surfaces are wear, plastic deformation, and metal fatigue. They result in rolling contact fatigue (RCF) cracks, loss of profile, and the development of corrugations. The effects are exacerbated by increases in dynamic forces resulting either from incompatible rail-wheel profiles or track irregularity.

Regular rail profile grinding is a proven means of control for these problems. However, the growing demand for train paths is reducing the time available for track maintenance and particularly the operation of grinding trains. Clearly, the technological challenge is to increase grinding productivity. It is also necessary to examine the functionality provided by the current grinding systems and establish whether some of them could be better fulfilled through material selection.

There are two factors that need to be considered: control of rail profile and control of RCF crack initiation and growth.

Considerable work has been done to develop an anti-RCF rail profile. One such profile is the 60E2 being adopted by a number of European railways. Figure 1 shows a comparison of this profile with the standard 60E1 profile while Figure 2 emphasises the magnitude of this difference. Ground profiles have also been optimised to provide gauge corner relief although the magnitude of this relief is often larger than that shown for the 60E2 profile.

[FIGURES 1-2 OMITTED]

The crown profile of the rail is achieved most efficiently and accurately through the hot rolling and roller straightening process using precisely shaped rolls. More importantly, increasing the longevity of the desired profile requires the resistance to wear and plastic deformation to be designed into the material properties of the rail steel rather than repeated correction through in-track rail grinding.

RCF cracks are initiated through the response of the material to the imposed stresses. Therefore it should be possible to increase the period to crack initiation by using more RCF resistant materials. But this needs to go hand-in-hand with the management of the rail-wheel profiles and dynamic interaction to reduce stress.

Although a range of steels with much higher resistance to RCF initiation is available (Figure 3), freedom from RCF cracks in the most susceptible stretches of track requires grinding to remove damaged material and the incipient cracks. This is equivalent to redressing the wear balance to achieve a "Magic wear rate" (this refers to the combined influence of both natural wear and that enforced through grinding). The synergy of optimum rail metallurgy, good track engineering, and non-intrusive high-speed grinding is illustrated in Figure 4.

[FIGURES 3-4 OMITTED]

When restoring the worn rail profile, it is important to consider the profile with reference to RCF crack initiation. The small gauge corner relief apparent in Figure 2 is designed to avoid contact with the wheel and thereby prevent any existing cracks from growing.

On a new rail, the profile reduces the conicity and thereby the forces. Based on very extensive failure investigations of RCF-affected rails from a wide range of track conditions in Britain and regular monitoring of several sites, Corus has established that RCF cracks initiate at a location between ~25 to 30mm from the active side of head of the rail or ~6 to 11mm from the centre of the rail head. As Figure 2 shows, the magnitude of relief available at this position is so tiny that it is likely to be removed by only small amounts of vertical wear or by almost negligible rail roll under dynamic loading. Hence, provided the forces generated are high enough, RCF cracks will eventually initiate at these locations necessitating grinding to remove fatigued material and restore the crown profile and slight gauge corner relief.

A high-speed grinding technique being developed by Stahlberg Roensch, Germany, has significant potential to increase rail life and track availability. Commercial operation of the unit on German track is planned for the last quarter of this year. The unpowered cylindrical grinding stones are arranged at an angle to the longitudinal axis of the rail and rotate at high speed as the vehicle moves (see photo above). The speed of operation makes the process almost non-intrusive to the normal maintenance operation and the small metal removal is sufficient to remove incipient cracks and damaged layers. Frequent removal of such a small amount also helps to remove minor surface irregularities. Corus has also developed a model to predict the frequency of grinding based on track and traffic characteristics.

[ILLUSTRATION OMITTED]

Following the Hatfield derailment in Britain in 2000, Corus helped to establish the original relationship between the surface length and vertical depths of RCF cracks (Figure 5) which was later populated with more data (Figure 6) from sample examinations. This relationship was employed as a criterion for rail renewal based on crack lengths determined by visual inspection. There appears to be considerably more spread than in the original data indicating the influence of other factors.

[FIGURES 5-6 OMITTED]

The need to assess accurately the internal propagation of RCF cracks led to the development of variants of eddy-current-based techniques that measure the penehated length of crack. However, the complexity of the crack network makes accurate measurement difficult.

Although train-mounted eddy current systems are being evaluated in Germany, their accuracy for detection and measurement of RCF cracks has yet to be demonstrated. Furthermore, such techniques do not necessarily capture all the characteristics of RCF cracks such as their linear density and angle, and the position of initiation on the head in relationship to primary running bands.

Train-borne track inspection has until recently been restricted to ultrasonic testing and measurement of track geometry. The recent introduction of the New Measurement Train (NMT) by Network Rail (NR), Britain, and Archimede by Italian Rail Network (RFI) represent a significant technical advance as they incorporate laser measurement of rail profile and video imaging of the rail and track components. Research is being undertaken by Corus, Oxford Lasers, and University of Exeter (with guidance from NR, Heathrow Express, and London Underground and part financial support from the British government), to enable an objective and consistent assessment of the rail-wheel contact band and visible rail head defects such as RCF cracks through innovative image acquisition and Intelligent Image Analysis (IIA).

A reliable inspection technique is needed to monitor parameters related to the rail-wheel interface to enable informed decisions for renewal and maintenance. This would help to reduce the very high cost of managing RCF.

Vast quantities of video data are now being collected on several European networks, but resolution is not sufficient to resolve RCF cracks and none of the systems address motion blur adequately. Furthermore, analysis of the data is restricted to either visual assessment or the use of basic image analysis techniques based on comparison with standard images. Although such techniques will identify the overall maintenance work requirements, the real benefits will come from a deeper analysis of the data from the whole network to extract trends and correlations between asset condition and track and traffic characteristics.

The project should create a step change in the cost-effective inspection of the rail head condition with increased safety. It aims to deliver:

* an enhanced image acquisition system using pulsed laser illumination technology to eliminate motion blur and thereby resolve features such as RCF cracks

* IIA techniques using standard software capable of rapid off-line processing will provide a cost-effective way of conducting reliable and consistent analysis of vast quantities of inspection data

* measurements of key features of rail-wheel contact, such as variation in the width and location of the running bands and the size, angle, distribution and frequency of RCF cracks, which will improve the understanding of track degradation and provide earlier warnings for appropriate planned maintenance intervention, and a significant cost saving by minimising manual inspection and eliminating erroneous placement of emergency speed restrictions or rail renewal.

Railways revolve around the circle of "inspection" followed by "reaction". Now that inspections can be made at higher speeds, earlier intervention should be possible. However, the philosophy is still to "inspect quality in". In other words, the safe and efficient running of the railways cannot be guaranteed without frequent inspections to confirm the condition of the track.

In some industries, the approach is to design out failure mechanisms and regulate inspection on the ability to predict degradation. For railways, the design and operational variables and the response of track components to them are extremely complex and make true predictive condition monitoring difficult. However, some recent advances in sensor technologies, such as the use of fibre optic Bragg gratings, provide a way forward. Harsh conditions in some manufacturing industries have led to the monitoring of deviations from healthy "signature tunes" instead of debating the accuracy of absolute measurements. Such an approach should be applicable to track monitoring.

There is also a need to separate the following types of condition monitoring systems. The most relevant example of operational condition monitoring relates to the operation of switches and the associated mechanisms. Various systems exist. Recent developments of such systems have been in the provision of failure predictive capability based on detection of changes to signals such as motor current. Alarms can then be sent to the maintenance crew. Operational monitoring is also relevant to the operation of level crossing barriers but some recent accidents suggest the need to incorporate further sensors and operational logic.

A key example of policing of compliance to specifications is a wheel impact load detector to identify wheel fiats. Even if a lot of such devices were installed (at great cost), the rogue wheels can still damage the track before they have been detected. However, such devices can be installed at the depots to prevent vehicles with defective wheels from leaving.

Proactive and preventive maintenance enables prediction of residual life and the need for timely intervention to prolong track life. This remains a key challenge and needs further research that is driven by the shrinking time for maintenance (and conversely by the need to increase track availability) and the need to increase worker safety by reducing manual track inspection.

The Intelligent Image Analysis project provides an objective measure of various RCF and running surface characteristics for the entire network so that correlations with track design and traffic characteristics can be established. Equally, the development of a "smart" track capable of monitoring track forces from identified passing vehicles, and undertaking analysis to predict remaining life or the need for maintenance intervention, would be a true application of proactive and predictive condition monitoring. The crunch question that remains is: who should pay for research into the development of such a system?

Dr Jay Jaiswal

Director, Engineering and Technologies, Corus Rail Sector, Britain

COPYRIGHT 2005 Simmons-Boardman Publishing Corporation
COPYRIGHT 2005 Gale Group