The integrity of the track is a given on railways today. Operators and their passengers rely on trains running at high speed along smooth and continuous rails. That should be Continuous Welded Rail (CWR), and even that acronym shows how important is the integrity of rail welding.
So it should be no surprise that there is an Institute of Rail Welding. An offshoot of The Welding Institute, it holds a more-or-less annual technical seminar so that members can keep up with the latest developments. The 24th such event took place recently at the Riverside Centre in Derby, chaired by senior Network Rail asset manager Ian Davison.
Improved understanding of RCF
The first paper of the day, and one of the most interesting, wasn’t about welding at all. Instead, Brian Whitney, Network Rail principal engineer for track and civils, spoke of the defects that commonly occur in rails and how to detect them.
Rolling Contact Fatigue (RCF) is responsible for the majority of rail surface defects and is caused by the high stresses that occur in the tiny contact patch between wheel and rail. These arise from the vertical loading of the wheel and from traction, braking and steering forces which subject the surface of the rail to vertical, longitudinal and lateral stresses, all of which can be extremely intense.
RCF damage may manifest itself in many forms, typically squats (an indentation in the running surface caused by a sub-surface fatigue crack) and the surface cracks commonly known as gauge corner cracking (GCC), although in fact the latter often occurs elsewhere than the gauge corner.
The rail failure at Hatfield was probably a result of severe surface cracking. In this form, RCF begins as fine cracks that appear on the rail surface at right angles to the direction of wheel travel and continue beneath the rail surface, initially horizontally.
At this stage the cracks may join up and result in small pieces of the rail surface spalling away, leaving surface defects.
However, under continued traffic and without intervention to prevent it, the cracks may eventually begin to turn downwards into the rail. Once these downward cracks grow deep enough, they can propagate suddenly under load, right through the depth of the rail, leading to a rail break. Worse still, where there are many such cracks, more than one break may grow in this way within the same length of rail, causing it to collapse into small pieces. This appears to have been what occurred at Hatfield.
Post-Hatfield, Network Rail adopted a simple method to classify RCF cracks into three categories based upon the visible length of the surface cracks. This was not ideal as the critical measure is not the visible surface crack, nor even the underlying horizontal one, but whether, and to what extent, a crack has turned down towards the rail foot.
Brian described the work undertaken by Network Rail to detect and measure RCF cracks so that potentially dangerous examples could be removed. In fact, it is important to find cracks before they reach a dangerous state so they can be removed in a planned way while they were still safe.
In the meantime, Network Rail applied an intensive rail-grinding programme as a crack control measure. If the initial surface cracks are removed in this way sufficiently early, they cannot reach the point where they start to turn down into the potentially dangerous form.
Network Rail’s existing ultrasonic test units (UTUs), fitted with Sperry technology, were able to detect down turned cracks but only when they had already reached potentially risky depths. ‘Surface noise’ from the top few millimetres of the rail effectively hides defects still in the early stages. So, whilst the UTUs could be used to detect and remove the potentially dangerous deeper cracks, they could not find RCF in time to plan and execute removal before it became dangerous. So removal had to be undertaken as emergencies in very short timescales, incurring train delays and financial penalties for the infrastructure owner.
Network Rail has seen a dramatic reduction in the rail break numbers. Between 1998 and 2015 these have dropped from 952 to 98 per annum with RCF-related breaks coming down from around 120 or 130 to only one or two each year. Network Rail has achieved much of this by re-railing or rail grinding, and the rest by the detection and repair or removal of defects.
This all sits in a context of rapidly- increasing rail traffic. Network Rail is finding that the incidence of heavy and severe RCF is growing so the company’s strategy is to move away from manual track inspections to train-borne systems. Last year, the company removed 25,000 rail defects at an average cost of about £5,000 each. Without better asset management in the future, the major reductions in rail-related delays and costs achieved in recent years are in danger of being reversed.
To reduce train delays and to keep ahead of the expected growth of the RCF problem, the company needed a new RCF detection method that would find down-turned cracks early. An invitation to tender against detailed requirements for a detection and measurement system received five or six responses. These were evaluated, and the chosen one was taken forward.
The selected proposal uses eddy currents induced in the rail head by electromagnets. The currents vary significantly when cracks are present, and this variation can be used both to measure the cracks and determine their position.
Network Rail worked with Sperry to turn this basic theoretical concept into a practical tool for use in a train- borne application. Sperry developed a rubber-tyred wheel probe similar to its ultrasonic ones which is able to maintain a consistent distance between the eddy current probes and the rail surface, a critical requirement.
The 10 probes are distributed across the width of the rail head to cover it from gauge corner to field corner. The wheel has the advantage that it fits directly into the same carrier mechanism as the ultrasonic wheel probes, removing the need to develop a new system for this purpose.
The new detection system has proven to be very effective in detecting RCF cracks early. However, the challenge was the sheer quantity of data produced – 100,000 lines of data per second. To cope, it was necessary to filter the data and digitise it into management information. The results are reports that indicate the largest detected crack depth in each metre of rail by each of the 10 probes.
Welds show up as cracks right across the rail head and 5mm deep. These are now automatically filtered out of reports of RCF. Surface weld repairs are also detectable and can now be identified and ignored.
The project team has been proving the validity of the results by removing and sectioning pieces of rail to determine the actual crack dimensions for comparison with the eddy current results. This has been done for the whole range of crack depths, not just for the deeper ones, it being important to be sure that the system is accurate across the whole range. The results have been very good, showing that the detection system is extremely accurate and reliable.
The system is useful for far more than just finding and removing defects in good time. It is also being used to learn more about how RCF cracks develop, where they occur and why. This will allow the linkage of RCF growth to track geometry resulting in designs and maintenance regimes which prevent RCF initiation.
Rail replacement planning will be improved using the understanding of RCF occurrence and growth that it will provide. The application of premium rail steels can be targeted to sites where it will give the optimum benefit. Similarly, a more-complete understanding of RCF will result in improved planning of rail grinding and better modelling of wheel/rail interactions.
Further development is now planned. The next intended stage is the development of a ‘walking stick’ eddy current detection apparatus, and a trolley-mounted version is also anticipated. These will allow crack detection in S&C and detailed checks on reports from the train-borne systems when these are required.
Removal and replacement
Having detected a defect, it needs to be repaired. Traditionally, that meant cutting out the affected length of rail and replacing it. However, Frédéric Delcroix of RailTech International described a new head wash repair (HWR) process for repairing surface defects in the rail head using alumino- thermic welding (ATW).
Simply explained, instead of cutting out a complete section of rail, only the affected part of the head is removed. The missing material is then replaced using a similar ATW process to that for joining rails, but using a mould which contains the material and allows it to ‘refill’ the missing volume.
The technical advantages of this method are twofold. It avoids two RailTech estimates that some 80% of surface defects can be satisfactorily repaired by the HWR process. The main category of defect that cannot is the wheel burn, which is typically too long. RailTech is developing and seeking approval for a variant of the HWR process, a triple HWR repair, which will greatly increase the applicability of the process by allowing drop of metal already referred to. Everything is thus controlled so that the ignition of the portion occurs at a defined point at its top, and premature tapping is avoided.
Further advantages accrue. The welder is not exposed to combustion fumes as there is no need to bend over a crucible with its lid removed in order to ignite it. In addition, because conventional welds that would fasten the short length of replacement rail into the railway, and it can be done without having to de-stress and then re-stress the rail. As most of the original rail section remains, the process can be undertaken with the rail still stressed. This process saves cost as well as time – RailTech estimates an 80% cost saving.
Minimal new materials or equipment are required, as the moulds used are the same as used for wide gap welds, but with a smaller weld portion (8.5kg). Three possible methods may be employed to cut out the defect in the rail head – flame cutting (as proposed in the UK), grinding with a hydraulically-powered grinder (the preferred option in the USA) or the use of an electrically powered grinder (as preferred by French railways). The flame cut surfaces created by the UK method are ground off before the weld repair is made.
The process was initially tested at TTCI in Pueblo, Colorado in 2008, followed by UK testing in 2010 that led to UK approval in 2012. This led into testing in France in 2013 and approval there the following year.
RailTech estimates that some 80% of surface defects can be satisfactorily repaired by the HWR process. The main category of defect that cannot is the wheel burn, which is typically too long. RailTech is developing and seeking approval for a variant of the HWR process, a triple HWR repair, which will greatly increase the applicability of the process by allowing repair of longer wheel burns.
A further proposed development will, if successful, enable the repair of a squat that sits on top of a flash-butt (FB) weld. A key challenge has been ensuring a reliable seal between the rail and the moulds, around the FB weld collar. RailTech has developed a felt material that is sandwiched between mould and rail to achieve this seal.
Frédéric next introduced the STARTWELLTM system. This is an innovative weld ignition system that has a number of benefits. The normal ignition system risks the early tapping of the weld portion should the welder insert the igniter too deep into it. The new process eliminates this risk as it is ignited by a drop of molten metal that falls from the special igniter into the centre of the surface of the weld portion. The igniter sits in a fixed position in the lid of the crucible, which need not be removed to ignite the portion. Ignition is triggered by applying the special electrical igniter gun to contacts on the top of the igniter. A reaction in the igniter is started and generates the molten drop of metal already referred to. Everything is thus controlled so that the ignition of the portion occurs at a defined point at its top, and premature tapping is avoided.
Further advantages accrue. The welder is not exposed to combustion fumes as there is no need to bend over a crucible with its lid removed in of the nature of the ignition system, the materials are not subject to the strict international transit restrictions applicable to traditional ATW materials.
Documentation made easy
Following Frédéric’s presentation, his colleagues Richard Kyte and Richard Vontak described the new RailTech AE MMS (Mobility Maintenance Software) welding software which has been in use in France for the past three years.
The AE MMS system is designed to ensure that welding managers, project managers and other relevant parties can easily and reliably obtain full records of the welds carried out by the welding teams they employ. It is a web portal system that operates between tablet computers carried by the welders and an internet server and can be used with Windows, Android or Apple software.
The welder’s tablet will download from the server each night the full details of the work the team is to undertake on the next day.
Upon executing a weld, the welder completes a form on the tablet which collects all required information about the weld and the welding process. The form cannot be filed if any of the details are omitted. Typical details include which contractor has employed the team, which team made the weld, which processes and materials were used and the portion number(s), and what grinding was done after the weld was stripped.
Other information may be captured according to the requirements of the project or client, for example welder competencies.
Easy welding of harder rail
Track performance can be improved by using better qualities of steel, and Sean Gleeson of Tata Steel and Ian Davison of Network Rail were on hand to outline the properties of HP335 rail and discuss how it can be welded.
HP335 is a metallurgically-engineered pearlitic steel (hyper-eutectoid). It is used ‘as made’, requiring no heat treatment or other post-rolling attention. As Sean explained, it is made at Scunthorpe in the UK but some specialised sections are rolled in Tata Steel’s factory at Hayange in northern France from the UK steel. It has a finer pearlitic structure than Grade 260 steel and, as the designation 335, suggests, is considerably harder (335 is the Brinell hardness of the steel as-rolled).
Ian took over the presentation at this point in order to go through the catalogue of Network Rail approvals for the welding of HP335 rail. These cover the straightforward ATW and flash butt (FB) welding of rail into CWR and the welding of HP335 leg extensions onto cast manganese crossings. Explosively-hardened cast manganese crossings are specified for this application because this ensures compatibility of hardness between crossing and leg extensions.
Various repair techniques are also under development covering Manual Metal Arc (MMA) repair, use of aluminothermic portions to replace an excavated portion of the head, and flux cored arc welding is also planned. Extension of repair techniques to in-situ repair of switchblades is also under development.
During the day, delegates also heard from Chris Eady of The Welding Institute who gave an update of the European Rail Safe project. The current phase of Rail Safe is called Rail Safe-TR, since it consists of a project in Turkey. Its objective is to develop and deliver a training plan for Turkish rail welders, to meet the need for competent welding staff deriving from the government’s plan for Turkish rail transport.
Rail Safe-TR is about to begin delivery of training courses for welding trainers in a pilot phase intended to prove the methodology. The project is due to finish in November this year.
There were also papers on the current Control Period 5 (CP5) from Network Rail as well as the adoption of new tubular stretcher bars for S&C, as described in issue 120 (October 2014).
This all led to a full and interesting day which delegates were still discussing as they made their way back from the Riverside Centre at the end of another successful IoRW Technical Seminar.