Maintaining tracks for long-term future service

Posted: 26 November 2007 | | No comments yet

A railway track represents a large investment that is not only meant to enable safe, fast and comfortable passenger and freight traffic, but is also expected to be permanently available. The track should allow decades of intense utilisation with no major interruptions. Track possessions for maintenance work or premature failure of its components imply major costs for the infrastructure operator.

A railway track represents a large investment that is not only meant to enable safe, fast and comfortable passenger and freight traffic, but is also expected to be permanently available. The track should allow decades of intense utilisation with no major interruptions. Track possessions for maintenance work or premature failure of its components imply major costs for the infrastructure operator.

A railway track represents a large investment that is not only meant to enable safe, fast and comfortable passenger and freight traffic, but is also expected to be permanently available. The track should allow decades of intense utilisation with no major interruptions. Track possessions for maintenance work or premature failure of its components imply major costs for the infrastructure operator.

Technical progress has allowed consistent improvement of travel speed and also of the durability of certain track components. However, the increasingly fast and dense traffic puts particular strain on the rails. The following hypothetical, but not unrealistic life-cycle description of a rail on a high-speed track wherever in the world, will illustrate this:

  • A new rail is rolled and checked with state-of-the-art testing equipment to guarantee a crack-free rail head and proper geometry (longitudinal and transverse profile). After welding it to the customised length in a stationary welding plant, the long welded rail (LWR) is installed properly in the track. The track parameters are verified with a measuring train and the new track finally approved for traffic up to 330kph.
  • After 6-9 months, some initial headchecks can be made. Also, a certain corrugation is recognised, which causes increasing noise and which multiplies dynamic forces for rails and vehicles.
  • Over the months, the headchecks develop further – a random measurement after one year indicates a damage depth of up to 1mm.
  • Due to excessive noise emission the track receives anti-corrugation grinding and this requires months to complete due to the possession time for the maintenance work. It is being assumed – but not verified – that the headchecks have been removed in the process.
  • A year later, headchecks are once again clearly visible. Because of high utilisation of the track, the respective maintenance works are postponed to the following year.
  • Some years later, some eddy current tests conducted locally reveal headchecks of significant depth. The track requires urgent attention; however, the appropriate equipment to remove sufficient material from the damaged rail head is often not available at short notice. Until the start of the maintenance works further months pass. Eddy current tests show a crack depth which is the maximum resolution of this measuring method. Due to the large degree of material removal, work progress is subsequently slow. With the possession time each night still being short, the work lasts several months. On top of all that comes major traffic disruption.
  • Before the wear reaches a critical limit it is decided to exchange the rails in long track sections. Even though the railhead still appears to have sufficient material reserve, the effort to remove the necessary amount of material is not considered economical. The old rails are scrapped, having lasted only a fraction of their designated life span. The scrap value is again only a small fraction of their original price, the cost for exchanging and transportation not being included. On top of all that, comes major traffic disruption.

Five critical aspects are responsible for this short and rather costly rail life cycle:

Rapid RCF-related crack growth

For decades, wear on the corner gauge was the significant cause of a limited rail life time, until rolling contact fatigue (RCF) related damage (Headchecks, Squats) has taken over in the final years of the last century. These rail flaws are subject to extensive research and their appearance, their possible cause and the consequences have been broadly published. In the early 90s, approximately 30% of the rail flaws on the DB-infrastructure in Germany were credited to RCF1. Ten years later this figure is raised to 60% for Japan2. High-speed trains have shown to be particularly ‘aggressive’. The average growth rate for headchecks on tracks with traffic up to 200kph is reported to be 0.4mm per year (equiv. to 20 MGT)3. After just a few years, the corner gauge cracks will develop to a degree which requires substantial machining in order to remove the cracked and damaged material layers. On high-speed lines, with trains using powered axles, the local growth rate of cracks can be significantly higher still.

Another feature of RCF-related flaws is the non-linear growth. Initially there are no cracks at all, but the material is being strained and fatigued by every passing train. Once the cracks have been initiated, they propagate more and more rapidly. Environmental influences like rain penetrating into the crack can accelerate crack growth further. After a phase of propagating under a constant angle into the rail head, the headcheck typically changes direction at a certain critical depth and either returns to the surface causing severe shelling at the corner gauge or turns to grow deep into the rail with possible fracture as the ultimate consequence. For safety reasons, but also to protect the investment of the infrastructure, effective maintenance measures are necessary to prolong the rail life span.

Lack of time for maintenance

Due to the high track utilisation and also due to a necessary administrative effort, track possession for maintenance work requires long lead times and offer only limited time for track access, frequently only a handful of hours. The net time to actually conduct the respective work is typically not more than 60% of that, because of long transfer distance between the station and the site or other operational reasons.

Suboptimal use of measuring technology

Measuring methods are the ‘eyes’ of rail maintenance. Without a proper diagnosis of the as-is-condition the correct countermeasure can hardly be selected and deployed. This also applies to the verification that this countermeasure was effective. Considering the alarming growth rate of RCF related rail flaws, it is surprising that analysis tools based on eddy current for detection and characterisation of small cracks only have a rather modest utilisation in the field.

Reactive maintenance

Conventional maintenance measures are being applied as soon as rail flaws exceed a certain critical threshold. This process requires that during cyclical inspection of the track the flaw is not only detected but also characterised regarding its severity. The threshold is typically selected such that when exceeded functional restraints with regards to safety, noise or riding comfort would apply. Only as long as the flaw can be readily found and categorised, provided that maintenance measures can be initiated and conducted in a timely manner, and the further flaw development remains sufficiently slow, this is a proper method to keep the availability of the track high. This does not imply that it is also the most cost-effective strategy.

RCF is an established failure mechanism that does not occur at random. Rather than waiting until the flaws become critical, it should be anticipated and treated in a proactive manner with an effective maintenance strategy.

Evaluation of maintenance measures

The responsible personnel for the infrastructure is always confronted with the question whether a scheduled maintenance measure is economical. The answer to this is simple; if the availability of the track depends on it. The real challenge comes, when having to quantify the value of such a measure in the future, giving particular consideration to the maintenance effort to come, the life span and availability of the track, as well as cost for track renewal and recycling or liquidation of old material. Without such evaluation and a strategy derived from it, the question is frequently reduced to whether a maintenance budget is available or not.

Preventative rail maintenance

Since all RCF-related flaws are initiated at the surface and then progressed into the rail head, the rail can be restored by removing this impaired or damaged material. This reduces the material reserve of the head, so that such a measure cannot be repeated all too often. In addition, there are substantial costs involved and the time effort to remove enough material to eliminate all cracks. State-of-the-art technology is rail milling which removes up to 2mm per pass.

The term ‘preventative grinding’ stands for a modern and sustainable maintenance strategy, which is derived from the perception that rail flaws grow over time in a non-linear, progressive way. Hence, one grinding (or milling) operation after a given time requires more material removal than the sum of two or more milling operations within the same time span. For this reason, some infrastructure operators systematically conduct regular and frequent grinding operations in a phase where the cracks are still short. With this additional effort (more possession time, more grinding operations) the material reserve of the rail head is preserved, so that with continued grinding the life span of the rail can be extended. Hempe predicted in a theoretical analysis, that with grinding operation in three instead of six year intervals, the life span of a rail can be prolonged from 27 to 40 years3. It has been reported from Canada and the USA that with preventative grinding after every 50 Mio t, the life span of the rail could be doubled to 1000 Mio t.

The economical aspect of a maintenance strategy addresses the question of how often and how deep the rails need to be machined, so that the extra service will pay off in the future with regard to a reduced renewal cost. A proper answer to this can only be provided with a Life-Cycle-Cost analysis (LCC), which sums up the net present value of all service and maintenance works during the life time of the rail and compares it to the price for a track renewal3. Based on current shift prices for grinding operations, Hempe calculated the optimal cycle for periodic grinding operations to be between 40 and 80 Mio t, depending on the crack growth rate (=damage function). This translates roughly to a 2-4 year cycle.

The minimum material removal and hence the highest possible life span for the rail could be achieved, if cracks were to be avoided altogether by removing the hardened and brittle surface layer on a regular basis before cracks appear. Such a procedure implies an artificial wear in the same order of magnitude as the damage rate in the material. Kalousek and Magel have introduced the term ‘magic wear rate’4. According to a UIC report, a material removal of 0.1mm for every 50 Mio. t would already reduce RCF related flaws by 50%5. Such a procedure may only be considered if the rail machining could be conduced cost-effectively and fast, preferably without the need for possession time altogether. This implies grinding during train runs. For this purpose Stahlberg Roensch (SR) has developed the unique technology of High Speed Grinding (HSG), which is now available and approved in Germany.

High Speed Grinding features a series of free rotating, non-powered grinding stones, which are offset by a certain angle relative to the rail. When dragged along the rail this offset angle produces an autorotation and a relative motion which grinds the rail. The rail is not being reprofiled but is being artificially worn. Because of the respective contours of rail and grinding stone, the contact line between the two is shaped like a stretched ‘S’ of roughly 60mm in length, which helps to smoothen short corrugations. The most outstanding feature is the grinding speed of approximately 80kph. Based on this technology the grinding machine ‘RC01’ was consistently developed to operate without track possession. Grinding is conducted during scheduled train traffic with 80kph with no need to de-install electronic track sensors. The vehicle is fully within the G-1 gauge, however, switches are excluded from the grinding process. RC01 is not powered but requires the traction of a locomotive with at least 1000 kW for operation on a level track. In cooperation with the DB research departments, the designated material removal is > 0.1mm, which can be achieved with three passes. The process has recently been approved for operation on DB’s High Speed network. Furthermore it is the first maintenance being approved by EBA for operation within regular traffic schedule.

The success of preventative rail grinding is highly dependant on how consistent the regular operations are being conducted. The challenge is to find the ideal operating sequence. For HSG this is currently determined in the course of a long term validation project together with DB-AG, starting with measured crack growth data and subsequently derived damage functions.

Hence the life cycle of rails, and in particular rails on high-speed tracks, depends largely on the infrastructure operator’s respective maintenance strategy. However, the rail does not have to be ground until the material reserve is fully exhausted. Depending on the state of wear the rail may also be de-installed from the track, reconditioned offline and then re-installed in another track. Such a process has been conducted for decades in the German railway network (“Kreislaufwirtschaft”).

During reconditioning, all damaged rail sections and welds are cut out and the rail is straightened. The rails edge opposite the former gauge corner is reprofiled by milling and the rail sections are welded back together to the specified length. Finally the rail is tested with ultra-sound and eddy current to verify it is free of any flaws. The cost for a reconditioned rail is approximately 50% of a new rail and it may well serve another life cycle in a secondary track.

With these developments in mind, the hypothetical life cycle of the rail in the previously mentioned example could have possibly taken quite a different course:

  • Immediately after installation in the track, the rail receives its initial grinding, which removes the decarb layer and optimises the head profile. Directly after the track is cleared for traffic up to 260kph.
  • By strictly following a specified maintenance cycle, the RCF-hardened surface layer of the rails is regularly removed and crack growth avoided.
  • Measuring trains regularly control the status of the entire track using ultrasound, eddy current testing and/or image processing.
  • After 8-10 years the rail head is reprofiled by means of conventional grinding of milling, to optimise the wheel-rail contact, which has been affected by wear.
  • After two decades, corner gauge wear in certain curves has increased to a degree that the rail needs to be replaced. The de-installed old rail is still free of cracks and still has sufficient material reserve on the opposite side, so that the major part can be reconditioned and re-utilised in a communal line.
  • Due to lower rail loading and significantly slower train speeds the rail is subjected to less RCF. Measuring trains continue to verify that it is free of cracks. The danger of accumulating corrugation is counteracted with occasional preventative grinding.
  • Finally after four decades, the material reserve of the rail is fully exhausted. With the ballast and the sleepers still being in good condition, only the rail is continuously exchanged and scrapped.

In the future, the strain on rails is not likely to be less; trains will be faster and traffic will tighten even further. Keeping a track serviceable over a long period of time will increasingly require a sustainable and long-term service policy and an awareness to fully exploit the available maintenance and characterization technologies.


  1. Schultheiß, H.: „Schienen – ein Teil der Fahrbahn“, Eisenbahningenieur 41 (1990) 6
  2. Cannon, D. F. et al: “Rail defects – An Overview”, Fracture of Engineering Materials & Structures (2003) 26
  3. Hempe, T. et al.: “Schienenschleifen als Bestandteil einer technisch-wirtschaftlichen Gleisinstandhaltung”, ZEVrail Glasers Annalen 131 (2007) 3
  4. Kalousek / Magel: The “Magic” Wear Rate, RT&S March 1997
  5. UIC World Executive Council: „Rail Defect Management. Final Report – Part B”, Paris 2003