The subject of Railway Safety has been ever more prescient in the time since the call for papers on this SaRS Special Issue of the journal has been out. The Rail Accident Investigation Branch (RAIB) has published its report on the passenger train derailment at Carmont which tragically resulted in three fatalities [RAIB Report 20/2022], and two incidents of wrong-side failures of the signalling system have resulted in a derailment at Dalwhinnie, Badenoch and Strathspey [RAIB Report 10/2022] and two trains being in the same signal section at Wingfield [RAIB Report 11/2023].
These incidents follow a ‘serious irregularity’ at Cardiff station (2017) on completion of re-signalling and track re-modelling works [RAIB Report 15/2017] and a slow-speed train collision at London Waterloo station (2018), the result of uncontrolled wiring added to the signalling system which was not detected during testing [RAIB Report 19/2018].
These incidents have arisen on ‘tried and tested’ systems which we, as an industry, are very familiar with, having developed standards, procedures, and competencies to manage safety risk which may emerge through the design, test, operations, and maintenance of these systems. However, we now have a fresh challenge in that the digital era of the railway brings new safety issues as we deploy complex, high-integrity, software-based systems to improve the capacity and performance of our railway networks. This not only presents a period of learning for engineers and designers, but also for railway organisations and infrastructure managers who must safely manage the operations and maintenance of those systems without having the benefit of empirical knowledge that has helped improve the safety of such systems over decades of operational ‘in service’ experience.
As an example, the European Rail Traffic Management System (ERTMS) on the Cambrian Line, which served as a pilot installation of ERTMS for the UK, replacing conventional lineside signalling, had a safety incident in 2017 [RAIB Report 17/2019] where the Temporary Speed Restrictions (TSR) data failed to be uploaded following an automated computer restart the night before. The display screen at the control centre incorrectly showed the TSRs as being loaded for transmission to trains. The error was only detected when a driver realised they had passed over a level crossing at 80 km/h (50 mph), significantly exceeding the temporary speed restriction of 30 km/h (19 mph) needed to give adequate warning time for level crossing users.
This problem is not unique to the UK, Hong Kong Mass Transit Railway (MTR) had an incident in 2019 resulting in a collision of non-passenger trains which was caused by software implementation errors made on a new signalling system during the process of performing a software change.
The Railway, similar to many other safety critical industries, has the (virtual) ‘handbook of safety’ built over many years, and reviewed and updated after every accident and near miss. Arguably the most significant set of changes to UK railway practice occurred after the Clapham Junction accident in 1988, but what is of concern is whether those hard lessons learnt after Clapham are now being forgotten through fading corporate memory, and whether new lessons are being learnt at the rate needed to keep pace with the introduction of new and novel technologies and operational practices.
The UK railway industry operates to the two principles of:
Any new system must be safer than that currently in use; and,
The risk is to be reduced so far as is reasonably practicable (better known as SFAIRP).
Whilst appearing very simple and straightforward, applying these two principles reveals a greater number of problems, for example (this is not an exhaustive list):
A railway is a system of systems, changing one part may have emergent properties on another part. That part may be mechanical, electrical, software or human, and the humans include not only the railway staff, but also the passengers and other members of the public.
Accurately assessing how safe one part of a system is, particularly in the railway industry, can prove difficult when you consider the very low numbers of accidents, lack of reporting, and use of legacy systems that had no safety targets to meet when they were installed.
Post-privatisation there was a lot of staff churn and loss. A significant amount of ‘railway knowledge’ was (and still is) maintained in people’s heads. When they leave the industry, that knowledge can be lost forever.
These issues are compounded by the increasing complexity of the computer systems managing the safety critical functions on the railway, the use of software programmers who ‘know the machine’ but do not necessarily know (or understand) the nuances of a railway system. Additionally, the geographical centralising of more and more railway operations and functions means that less people control a greater area with the associated reliance on computer-based systems to reduce the workload to manageable levels.
So, are we Maintaining Railway Safety in an Ever-Changing World? This Special Issue of the SaRS Journal invites safety and reliability experts from industry and academia to present their views on such an important subject.
In our first paper, (King, Nicholson and Schmid) provide a systematic literature review to demonstrate the importance of a robust approach to Reliability, Availability and Maintainability (RAM) in the development and delivery of safety-critical systems being deployed through complex rail mega projects.
For our second paper, the differences and similarities on how safety is managed on Road and Rail is explored by (Kantartzis, Bishop, Kappeler, Nkomo) as they discuss parallels and potential synergies which could be drawn from the two industries, both of which are at a time of change and innovation with respect to the systems and practices which manage traffic on their respective infrastructure.
The third paper (Elsmore, J. Harding, Lee, Sharma, K. Harding, Hobden) addresses one of the major risks for railway safety at stations, the Platform-Train Interface (PTI). The authors discuss their approach and methodology for the analyses of incident data and application of mathematical modelling to identify appropriate safety measures to mitigate this safety risk.
In our final paper, (Boussif, Tonk, Beugin, Collart-dutilleul) analyse the changing role of the driver through remote driving of freight trains and propose a risk assessment method for managing operational safety risk which may be affected by an increased use of autonomous systems on our railways.
As a reader of this journal, and perhaps a member of the Safety and Reliability Society, what are your views? Are we Maintaining Railway Safety in an Ever-Changing World or are we potentially sleepwalking into another major railway accident?