TPWS has cut SPAD fatalities to near zero — but its limits are showing

Sixty-six fatalities and 1,734 injuries in the 22 years before 2004; two injury accidents in the 22 years since. Those figures frame the verdict of a technical report published by the Rail Safety and Standards Board of Great Britain, titled 'Train Protection and Warning System – A History and Summary of Its Capability and Limitations', which the author of the underlying article helped prepare. The document sits on the RSSB landing page for RIS-0775-CCS AWS and TPWS Application Requirements.

The report was written partly because TPWS is now being proposed to solve problems it may not suit. Britain developed the system in the mid-1990s in response to rising harm from Signals Passed at Danger (SPAD). Automatic Train Protection had been trialled beforehand. Retrofitting rolling stock nationwide proved too expensive.

What is TPWS and how does it work?

TPWS extends the existing Automatic Warning System (AWS). By redesigning AWS with semiconductor components instead of relays, engineers added a train-stop function — applying the brake if a signal is passed at danger — and an overspeed function that brakes a train approaching a red signal too quickly.

The overspeed capability matters because mainline signals typically offer only 183 to 400 metres of safe overrun distance, formerly 200 to 440 yards. That is inadequate to stop a train passing a red at speed. Track-mounted loops transmit electromagnetic tones to the train to trigger either function.

After the Southall accident, the Health and Safety Executive mandated TPWS at all junction-protecting signals, plus buffer stops and selected speed restrictions.

An arming loop starts a timer — typically one second for passenger trains — and if the trigger loop is detected before it expires, the brakes apply. A train-stop loop pair sits immediately beyond the signal. An incapacitated driver may trigger the same brake application three or four times without increasing braking force.

What are the limitations?

The report identifies three structural constraints:

• The overspeed sensor (OSS) measures speed only at one spot. An accelerating train can pass below the set speed yet soon exceed a safe speed.
• The system needs a positive signal to act, requiring integrated monitoring of both transmitter and receiver.
• Permitted speed depends on loop spacing and the on-train timer, so different timer settings across train types cause problems.

Only two of ten available timer settings are used: one for passengers, one for most freight. The freight value is roughly 25% longer because of slower brake application, exposing freight trains to brake demands at speeds 20% below passenger thresholds. That raises the risk of unnecessary interventions, especially approaching speed restrictions.

Reducing the passenger timer to help high-performance trains carries its own cost: it raises the trip speed approaching buffer stops, potentially increasing collision speed if a driver is incapacitated. The brake itself works by venting the air brake pipe, isolating regenerative braking on some trains, and is held for at least one minute — leaving a train stranded after even a minor infringement.

How effective has it been?

Measured against SPADs, TPWS has exceeded expectations. Excluding Salisbury Tunnel Junction and Talerddig, where adhesion contributed, only two injury accidents have occurred since implementation in 2004: Elstow, with one fatality and 162 injured, and a buffer stop collision at Kirkby with one injury. Professional driving, developed alongside TPWS, cut SPAD event numbers by around 50%; TPWS reduced overrun distance and consequence severity.

Overspeed is different. No reliable data exists on overspeed volume because events go unrecorded unless reported or found in On Train Monitoring and Recording downloads. No confirmed case exists of TPWS intervening to reduce an overspeed consequence.

An RSSB study, T1014, examined publishing OSS set speeds to drivers to cut unnecessary interventions — which can encourage 'Reset and Go' behaviour, where drivers reset the system and proceed without signaller confirmation.

Can TPWS manage junction overspeed?

Recent events have seen drivers ignore the speed profile of a junction route, some close to becoming significant accidents. Using TPWS here faces serious difficulty. Britain uses route signalling, not speed signalling, and a driver who misreads a junction indicator will likely apply power — exactly the acceleration scenario the spot-speed OSS cannot handle. Multiple sensors would be needed per route, complicated further by freight trains' longer timers.

The report's conclusion is blunt: TPWS plus professional driving achieved its SPAD and buffer-stop objectives, but there is significant doubt about its merit for managing overspeed beyond red-signal approaches.