Providing a train protection system to ensure drivers correctly observe lineside signals has always been difficult to justify in financial terms. The Great Western Railway (GWR) did have a system called ATC (Automatic Train Control) with contact ramps in advance of signals to give the driver an indication whether that signal was at danger or clear. This was rolled out on the Great Western main lines. The London, Midland and Scottish Railway (LMS) had a contactless system installed on the London Tilbury and Southend (LTS) route before the Second World War, mainly because of the risk of dense fog in the Thames estuary and the difficulty in seeing signals.
It took the Harrow accident in 1952, with 112 deaths, to force British Rail (BR) to adopt the Automatic Warning System (AWS) across the network even though this took decades to fully implement. It was very similar to the LMS system and was certainly better than nothing. The main weaknesses were that the system only distinguished between green and any other restrictive aspect, and a warning could be cancelled by the driver before any brake application occurred. Thus, a train closely following other trains in a congested area, and seeing yellow or double yellow signals all the time, would receive a warning which the driver habitually cancelled thus negating a real warning to brake and stop.
These weaknesses and a growing number of Signals Passed at Dangers (SPADs) led BR to investigate an improved Automatic Train Protection (ATP) system in the 1990s. The supply industry offered up several bespoke products and two were purchased for trial, one on the GW main line from London out to Bristol, the other on the Chiltern Line from Marylebone to Banbury. In parallel, BR in conjunction with Redifon, an electronics company primarily in the business of simulation systems, was developing a less expensive system that could enhance the capability of AWS. Neither this nor the proprietary systems would be cheap to install nationwide and getting the funding was always going to be a protracted exercise with the government paymasters.
Two subsequent accidents changed everything: Southall in 1997 with seven deaths and Ladbroke Grove in 1999 with 31 deaths, both due to signals being passed at danger. These accidents resulted in public enquiries, requiring an investigation into train protection systems to be undertaken by an independent assessor. Sir David Davies from the Royal Academy of Engineering was appointed to the task. His report took account of the various systems on offer and also the forthcoming ERTMS and its European Train Control System (ETCS) sub system, then under development as a European standard.
The recommendation was to proceed with the short-term deployment of the BR/Redifon system known as TPWS (Train Protection and Warning System), pending ETCS becoming available as a standard product. This was a wise decision as ETCS was never going to be a quick project. Sir David’s premise being that it might take ten or more years to implement. With hindsight, even this was wishful thinking as only now are main line ETCS projects in the UK beginning to happen. Some signalling suppliers viewed the recommendation with dismay as they had reservations about the integrity of TPWS and, more obviously, a missed opportunity for their order book.
The system was designed as a low cost add on to AWS that would assist in reducing the number of SPADs. It consists of two sets of inductive grids positioned between the running rails and placed approximately 300 metres apart, the first to detect whether a train is running at too high a speed (the over speed sensor), the second to instigate a full brake application if the train has not slowed sufficiently to stop at a red signal.
TPWS was only fitted at high-risk signals, mainly those where SPADs regularly occurred, or control signals protecting a junction or station. All remaining signals relied on AWS for their protection. The onboard equipment required a TPWS display in the drivers cab but the interrogator equipment under the train was designed to fit in the same space envelope that AWS occupied, thus saving significant cost. As such, the system comes in the intermittent category similar to ETCS Level 1. Initially, the system was only truly effective for train speeds up to 60mph and some more traditional signal engineers regarded it as not truly fail safe.
Subsequently, and to overcome the train speed limitation, an enhancement known as TPWS + has been developed with a further set of grids placed around 800 metres before the over speed sensor. This enables train speeds of up to 125mph to be sensed and monitored, with brake applications made if necessary. Routes with high line speeds have had signals fitted with this extra precaution.
The result has been very successful and the number of SPADs occurring has been significantly reduced such that these are no longer at the top of the list for accident risks. This has also had the negative effect of making the business case for ETCS deployment that bit more difficult.
Further TPWS development
So, what more needs to be done? There is a general recognition that ETCS roll out on main lines will take decades to implement and it is highly likely that it will never happen on secondary and rural routes. Can the benefits of ETCS Level 2 be achieved by further development of TPWS by making it a continuous monitoring system? Thinking within Network Rail certainly suggests this to be possible and a number of initiatives have emerged. Sixty-five options have been considered with analysis reducing these to around five. One such option proposed by Thales was an upgrade to the on-train TPWS to enhance the protection, and early designs have now reached the state of physical trials to prove the practicality and begin to understand the likely cost.
The ‘CS’ in TPWS-CS stands for Continuous Supervision, meaning that the system will continually monitor the train speed and other elements as well. But how will it do that?
For the onboard equipment, there is already a radio link courtesy of GSM-R. In addition, the train will require a GPS connection, a radar, and an inertial measurement sensor. The existing TPWS display − Driver Machine Interface (DMI) − will be adapted to accommodate the new information to be shown.
Infrastructure elements are also needed, these being, first, a ‘state of the railway compiler’ (SoRC), primarily to interrogate the signalling interlockings in the area and which will be located in a convenient trackside equipment room. The SoRC is also an output of the Network Rail options analysis through work with Park Signalling.
Second, there will be a series of trackside processors also located in trackside equipment rooms, which will connect to the SoRC and communicate with the on-board systems via GSM-R. Through a combination of these three ideas − upgraded TPWS, the SoRC and the trackside processor − a combined system can be created with comparable benefits to ETCS Level 2 without the full complexity.
As the train progresses on its journey, the information gleaned from the interlockings will enable a Movement Authority to be generated which will not be shown to the driver who remains observing the lineside signals. This authority will take account of signal aspects, train speed, speed restrictions, station stops, and junction divergence. Should the driver not be responding to the required speed, then an alert will be displayed via the TPWS DMI and, if not responded to, braking will occur. The behaviour of this system would be aligned with how one would expect an ETCS Level 2 system to behave. The main objective of this system is therefore to enable a continuous background supervision against a movement authority which is protected via the existing TPWS on the trains. This may lead to defensive driving techniques being able to be avoided and provide performance improvements. This potential performance improvement is currently being modelled.
In addition, the availability of data from the sensors included in the onboard system could enable new applications, such as detection of track or lineside irregularities and the potential for supervised movement authorities within worksites to improve track worker safety.
Testing the system
One requirement that triggered the initial initiative to introduce these sensors was the existence of user worked level crossings (UWCs) on lines with long block sections. If a member of the public rang to ask if it was safe to cross, the exact whereabouts of a train gave the signaller a dilemma as to whether to grant permission. Calculating a continuous train position using only on-board sensors provides a potential solution to this problem and when investigated it was quickly realised that such a facility would have many other benefits.
To date, a Class 143 Pacer train has been equipped but since these are no longer in normal passenger service, a Class 150 DMU from Great Western has also been fitted, which will be the concept proving train. In 2022, a set of demonstrations of some of the additional applications is planned on the heritage West Somerset Railway as this can be carried out in non-traffic periods without the need for a main line possession. Additionally, while the fitted Class 150 is operating in normal passenger service the installed system will be monitored to understand how the upgraded TPWS-CS would behave.
The business case is evolving. The advantages of having a continuous monitoring system can be seen but the financial benefit is more difficult to quantify. The basic trackside TPWS system will remain as is, but once fleets are fitted both safety and operational benefits begin to be unlocked. The cost of fitting the train fleet will be an important factor but using the existing TPWS hardware and on-board positioning sensors will help keep costs down.
The normal process of equipping a ‘first in class’ to learn what has to be done will pave the way, whereafter fleet fitment can commence. The initial emphasis is expected to be on secondary and rural routes as these are never likely to have ETCS. A broad order estimate for the appropriate fleets is significantly less costly than a full ETCS deployment. A big advantage over ETCS where signals have been removed, is that trains not equipped with CS can still operate on any route using the existing TPWS protection.
Fitting the infrastructure will also need to be determined and negotiating with Network Rail to keep these costs at an acceptable level, particularly the interfacing to existing signalling assets, could be a deciding factor.
This will be an interesting initiative to see how and if it develops. Network Rail is reported as keen to push on with it or something similar, so Rail Engineer will keep a watchful eye.
Thanks are expressed to Trevor Rolfe and Trish Shanahan from Thales for explaining the system proposal.