by Ravi Ravitharan, Director, Institute of Railway Technology, Monash University
There is a desire amongst many railway transport operators in Australia for rail industry regulators to revise current track maintenance standards to allow for the implementation of new technologies in place of existing requirements.
In the long term, it is expected that such changes would reduce safety risks and costs associated with human inspections and asset failures, whilst providing cost benefit incentives for operators, allowing them to re-invest in new technologies which would benefit the railway transport industry as a whole.
The rail transport industry requires standards that effectively ensure safety by considering all assets and interfaces as a single system and understanding full lifecycle impacts. With specific regard to the implementation of new technology, current regulations regulate safety and asset failure associated safety risks, but do not mandate adoption of new technology.
Whilst it is the operators’ choice not to implement new technology, the long-term value of railway asset lifecycles and associated maintenance costs means that failure mitigation is critically important. Even where failures are not occurring, it should be apparent that new technology can better facilitate efficient operation of their railway network.
The implementation of current railway industry standards such as AS7635:2013 (the current Australian Standard relating to track geometry) have greatly boosted the safety and efficiency of present day railways. Whilst derailment frequency has decreased tremendously in the last decade, a number of industry experts are now advocating for revisiting standards to address gaps where standards are believed to be lacking.
The ever-escalating demands on modern railways (tonnage, speeds and throughput) will inevitably warrant new approaches for managing and assessing track geometry and rolling stock condition, and the leveraging of new and established technologies so that standards can remain effective into the future.
Monitoring track condition
It is vitally important to manage railway track geometry and vehicle condition together as irregularities in either can rapidly lead to elevated vehicle dynamics, increased track loading, asset wear, adverse vehicle/track interactions and eventually the risk of vehicle derailment. Track condition should be regularly monitored as even subtle changes can risk leading to adverse vehicle track interaction and accelerated degradation.
At present, track irregularities are typically monitored using track recording vehicles (TRV) which use a variety of instruments to record track geometry metrics and provide measurements at discrete increments along the track. International railways abide by numerous regulator-mandated track geometry condemning limits with the expectation that compliance will mitigate adverse vehicle dynamics, asset failures and derailment.
Whilst some recommendations are provided in Rail Safety and Standards Board (RISSB) standard AS7635:2013, in Australia, the wide diversity of different railway operations has led to individual railway operators being allowed to develop their own asset condemning limits which best optimise the asset life and safety for their own particular operation in accordance with the guidelines and regulatory requirements of Rail Safety National Law.
Complex vehicle-track interactions
It is important to note though, that individual condemning limits should only be considered as bare minimum requirements as derailments are still being reported at locations that have no condemnable conditions. This suggests that the method of condition assessment based on discrete track geometry measurements is not in itself adequate to prevent derailments and accelerated deterioration of track assets and this should be addressed in revised standards.
The primary issue with the current approach of specifying discrete limits is that it does not adequately account for the myriad of interactions and multiplicative effects that can occur between combinations or clusters of non-condemnable conditions occurring around the same location, together with rolling stock variability. RISSB standard AS7635:2013 has the statement “where combinational defects are detected, these should be investigated and reviewed by the Rail Infrastructure Manager’’. The minutia of vehicle-track interactions is so complicated that humans would not likely be capable of correlating all the conditions they observe and making the logical connections required to foresee possible interactions and adverse outcomes.
This is a very significant issue for railway operators which is complicated further where; multiple cyclic irregularities can occur that lead to elevated vehicle dynamics (track loading, wheel climb, hunting); variability in rolling stock condition occurs (wheel wear, off-centred loads, suspension characteristics); and where trains have deviated from the normal operating conditions (train speeds, axle loads) for which the track was designed.
More frequent inspections key
Another issue associated with current methods of condition assessment is the lack of regularity (frequency) in measurements. The frequency of coverage of TRVs on any section of track is dependent on the number of recording vehicles available to the operator, the length of the railway to be measured, the ability to schedule track access time around nominal operations, standards (which mandate inspection frequency), and of course cost. All this means that TRV coverage frequency is typically in the order of months. RISSB standard AS7635:2013 mandates minimum track geometry recording car inspection frequency of four months for main line interstate corridors and a lot can happen to the track in this period of time.
Track issues can arise very rapidly, in some cases going from new condition to condemnable in a matter of weeks, or more rapidly in the case of extreme weather (track buckling, formation washouts), so at any given time it is unlikely that TRV measurements will accurately reflect the present condition of the track and it is unlikely the track defects will be caught in time to prevent asset failures or derailments.
The irrationality of this approach becomes more apparent when we realise that the inspection and/or maintenance response requirements for detected defects (running every four months) in many cases requires action within a 24-hour period. A typical example would be wheel burn, squat or weld dip on the track. This begs the question, why is the defect allowed to remain in place for the four month period between inspections? Assuming TRV geometry data is even accurate, it still needs to be used in conjunction with a comprehensive range of other inspections in order to adequately manage track condition.
Benefits of continuous monitoring
Using continuous monitoring systems which incorporate autonomous track geometry and rail profile measurement systems, the frequency of track measurements can approach time intervals commensurate with nominal traffic conditions on any individual line. The benefits of continuous monitoring systems to modern railways are abundant. The high frequency of condition measurements allows for:
- Data to be continuously fed into the railway operator’s risk assessment models to greatly improve response times to developing issues
- Facilitates the development of track condition/degradation forecasting models
- Allows operators to rapidly gauge the effectiveness or ineffectiveness of maintenance activities.
Moving towards predictive maintenance
Measuring maintenance effectiveness also allows for operators to build business cases to justify undertaking more substantial maintenance where, for example, records show repeated instances of ineffective tamping.
Continuous monitoring systems have shown significant advantages in data driven maintenance planning for the railway industry by facilitating the shift from reactive (costly, unscheduled or emergency correction of track condition), to predictive (condition based) maintenance approaches.
Predictive maintenance approaches can identify, prioritise and pre-schedule inspections and maintenance to locations based on realistic probabilities of degradation and risk, and rectify them prior to entering a state of rapid or uncontrolled degradation. This approach improves planning and scheduling of maintenance and minimises the risks to infrastructure and rolling stock, whilst also reducing the need for temporary speed restrictions and unscheduled track closures.
It is accepted that track geometry data provides an accurate summary of the spatial position of the track, and whilst peak errors and standard deviations may not necessarily determine the overall condition of the track, track geometry in conjunction with deterministic vehicle response modelling (Universal Mechanism®, Vampire®, NuCars®) and neural networks can be a very powerful research tool. The benefits of vehicle modelling for railway operators are; that they can predict responses to track geometry and/or defects for a range of vehicle types and operational conditions; they can identify particular track conditions or combinations of conditions which can lead to derailments; can assess the interactions between multiple vehicles including longitudinal in-train dynamics, and many others. Vehicle model simulations and machine learning are very powerful research tools, but are still limited to the frequency of track geometry recordings and is not a direct measure of vehicle responses as can be obtained using continuous vehicle response monitoring systems.
Vehicle response measurements
Continuous monitoring systems such as the Instrumented Revenue Vehicles (IRV) have already been shown to be effective at measuring vehicle responses and inferring track condition with a high degree of repeatability, accuracy and regularity. The inclusion of vehicle response measurement requirements in maintenance management plans would result in far more realistic assessments of the actual impact of track condition on rolling stock responses, inclusive of any cumulative and cyclic defects, and would allow all operators to move from basic condition detection methodologies to performance based systems.
Performance based systems actively determine how vehicles perform over sections of track given the specific geometry (including defects), the types of vehicles (age and suspension characteristics) and the specific train operating conditions. Vehicle response measurements also allow for operators to instantly quantify the impacts of changes made in track geometry or operational aspects of the network even justify maintaining sections of track to a higher or lower degree based on the vehicle responses at nominal speeds and loads.
Implementation of new technology
The Office of National Rail Safety Regulator (ONRSR) noted at the workshop Current Track Maintenance Standards and their Relevance to Modern Railways arranged by the Institute of Railway Technology, Monash University in March 2017, that as regulators, they will support and promote the ‘safe’ implementation of new technology. The challenge for industry operators is now to convince regulators, who are inherently risk-averse, that any change to standards, such as relaxing manual inspections requirements in favour of continuous monitoring technologies can be done so safely.
The ONRSR is a risk-based, independent industry-government administrator that audits and reviews compliance with RSNL. The Australian Government dictates the law, industry operators are required to implement the law and the ONRSR is responsible for administrating and encouraging correct implementation of standards in Australia in accordance with Regulation 19 (General engineering and operational systems safety requirements) of RNSL. Any changes to existing standards needs to very carefully weigh up the benefits and opportunities for improvement against any safety, performance and cost consequences.
Changes must recognise that failures to any single asset can cause problems for the rest of the system (disruptions, damage, delays at port). Additionally, it should be noted that the vast scale of the many systems operating throughout the rail transport industry, involving a very large number of assets and personnel, presents a problem when implementing change. The sheer momentum behind such systems means that change can typically only be implemented very gradually and operators will need guidance on how to best implement changes.
Holistic condition monitoring approach
It is recommended that the Rail Safety and Standards Board (RISSB) should consider revising the standards and adopt a more holistic approach that considers the safety and performance of the railway track and rolling stock together as a system. New standards should specify some basic minimum requirements for assessing geometry using modern detection and monitoring technology to continuously update each individual operator’s risk assessment models, which can then be validated and signed off on by regulators. This minimum adoption of modern monitoring technology would begin to force the cultural shift in the railway industry from reactive to proactive maintenance.
As new technologies begin to be adopted, regulations can move towards being based on actual vehicle and track interaction and performance as opposed to track geometry measurement pass or fail conditions. A performance based continuous condition monitoring system would be able to relax discrete constraints in individual metric compliances and give operators more freedom to adapt to improve their network overall condition.
Finally, the use of deterministic vehicle-track simulations to quantify effects of cyclic and/or cumulative track geometry defect categories would still be very beneficial where validated using measured vehicle response data and may be used to optimise the specification of condemning limits on individual and combinations of track geometry defects and would be a great benefit to every modern railway.