As a Londoner, my tube journeys are often delayed by "signal failures".

What constitutes a signal failure?

How do they occur?

Why do they happen so often?

Why can't they be effectively prevented?

  • 2
    Signal failures are not specific to The Tube, it's a term used in rail generally. I once took the train from Perth to Sydney in Australia, a trip of 72 hours. For the first third of the trip every single signal had a signal failure and the train had to proceed at walking pace! Dec 1, 2014 at 3:05
  • 1
    I wouldn't say they happen often. Consider the number of tube/rail lines and the flux of trains.
    – JoErNanO
    Dec 1, 2014 at 11:21
  • 1
    I was told (but can't back it up with any evidence or sources) that they tend to use "signal failure" as a catch-all reason to cover a number of problems they don't want to give the details of, such as passengers falling under trains and the like. I have however heard them say "...because of a fatality at...", but I've also heard them report a day of strike action as "a problem with the signalling system" (presumably that nobody was there to operate the signals).
    – Tyr
    Dec 1, 2014 at 17:05
  • <anecdotal evidence warning> In my experience, "signal failure" seems by far the most common, maybe 2/3rds of announced delays; the euphemistic yet extremely graphic "due to a person under a train" is the second most common, maybe 20% of announced delays, then there's "adverse weather conditions", "a passenger being taken ill", and some others that are each rare enough that I don't remember them. Dec 1, 2014 at 17:48
  • 1
    @Tyr: It's also used as a euphemism for "Someone has stolen the cables which control the signals", which sadly seems to be happening ever more frequently...
    – Nick C
    Dec 2, 2014 at 12:56

3 Answers 3


In lieu of a link-only answer...

A signal failure can refer to a number of situations... It can be down to a track circuit failure which means that the signaller cannot guarantee the safety of the train as he is unable to see on their screen or panel that the section of line is clear of any other trains. It could be down to one of the Train Protection Warning System failing which can cause trains to stop when it is not required, these are in place to stop drivers passing signals at danger and risking collisions, it is also in place on certain bends of tracks where there is a speed limit, this is to stop tragic accidents like the one recently in Spain. it can be the set of points which switch you from one line to another have failed in their position, there are thousands of different failures, it can be level crossings have failed.

Much as traffic lights regulate and control the flow of traffic on the roads, railway signals perform an analogous function on the railway network - not to permit crossing (which is illegal), but to prevent railway trains running into each other's rear. As a safety measure, the default position of any railway signal is "danger", that is, indicating a hazard ahead. Any signal giving an ambiguous display (one where the meaning is unclear) is, in the instructions of train drivers, to be regarded as "danger" until proved otherwise. In the days when signals were given using a cable-operated baton, once the baton froze solid in the "all-clear" position, causing a terrible train crash. Signals were afterwards directed always to stand in the "danger" position when at rest, a system known as a fail-safe

Source http://www.theguardian.com/notesandqueries/query/0,,-201845,00.html

For the other parts of your question, they appear to happen often because the system is large, complex, and old. Plus we don't want to pay for it. Overall, the record is impressive. Last year for example the DLR achieved the highest punctuality rate in 26 years.

For your last question, they will never be ENTIRELY eliminated, but year-on-year reductions are averaging around 10%.

Source https://www.tfl.gov.uk/cdn/static/cms/documents/annual-report-2013-14.pdf

Finally, if you want to make representations, you can contact their Stakeholder Engagement team and get the name of the passenger representative that handles stakeholder issues for your routes [email protected].

  • 2
    "Plus we don't want to pay for it"... I think train tickets in London are really expensive compared to other countries.
    – algiogia
    Dec 1, 2014 at 13:27
  • 5
    They're incredibly cheap compared to the rest of the UK, however, for a much better service. Part of the problem in London is that the sheer volume of traffic quickly causes knock-on problems: an issue which may delay 1 train in a remote area of Yorkshire could delay 20 or more in London, simply because there are so many more trains in a small area in a short space of time
    – Jon Story
    Dec 1, 2014 at 16:18
  • 1
    @Jon is right - and even though transport capacity is being expanded, it still can't keep up with London population growth: theguardian.com/uk-news/2014/sep/21/… And the volume of use is enormous, Waterloo tube station (the busiest) had nearly 90 million passenger-journeys in 2013, about 300,000 every weekday (that's just the tube part of Waterloo). tfl.gov.uk/cdn/static/cms/documents/…
    – A E
    Dec 1, 2014 at 16:27
  • 1
    @JonStory, what you wrote about the delay in Yorks is spot on!
    – Gayot Fow
    Dec 1, 2014 at 16:42

A signal failure is the failure of a signal and associated equipment.

The most important thing to realize is that a train at full speed cannot stop on sight so operators need to know that the way ahead is free. There are several different systems to deal with this difficulty and regulate traffic on railroads but most of them depend on signals in one way or another. If a signal is not in the “all-clear” position, the train has to stop or at least slow down.

Presumably the signal itself can fail for all the usual reasons (power outage, broken lamp, vandalism…) and it is also typically designed to stop traffic if for some reason (malfunction in a sensor used to detect the passage of trains, communication problem…) it cannot be determined whether the next block or section of railroad is free or not.

But unlike traffic lights on the road (which are less numerous – relatively speaking – and can easily be ignored by drivers or overruled by the police), you can't just ignore a railway signal until the problem is solved. There are big safety implications and railway companies tend to be very hierarchical so procedures have to be followed and an individual operator certainly cannot let a train run on his or her own initiative. Crossing a “closed” signal can mean dismissal.

When traffic is very dense, everything has to work perfectly and any issue will have long-lasting effects down the line thus increasing the impression of unreliability. Beyond that, it is far from trivial to run a train network but ultimately regular problems often boil down to a lack of investment in the maintenance and modernization of the network.

I don't know much about the London underground specifically but Wikipedia suggests part of it relies on a signalling system installed in the 1940s and it is being upgraded, which suggests that signalling has been recognized as a problem.

  • 3
    Usually it is not a signal itself that fails -- in a typical failure situation the signals themselves are functioning excellently, but some other part of the rather complex machinery that creates the information that the signals display has a fault. The rest of the system, when functioning as designed, reacts to the failure by keeping the signal at red. Nov 30, 2014 at 23:21
  • 1
    @HenningMakholm Yes, that's why I mentioned sensors as well… I will try to clarify that a little more.
    – Relaxed
    Nov 30, 2014 at 23:25
  • The sensors to determine the location of the the trains are just a small voltage applied to one of the running rails. The train's wheels and axles complete the circuit to the other running rail and the system knows that that segment of track has a train on it.# Dec 1, 2014 at 1:35
  • 4
    @DavidRicherby: In new installations it is becoming common to use axle counters instead of track circuits. They magnetically detect the passage of railway wheels past fixed locations on the track, and a central computer then keeps track of how many wheels are present on a given stretch of track. A major advantage of axle counters is that they are less prone to failure than track circuits, but the rest of the signaling equipment has to be designed to work with them, so it's not a plug-in replacement you can just roll out without resignaling everything. Dec 1, 2014 at 13:16
  • 1
    @DavidRicherby -- even something as seemingly simple as a track circuit can be triggered by things other than trains. Wet tunnel conditions, broken rails, or wiring faults can all cause the system to think there's a train there. The paranoia is good though, because the last thing you want is a train getting a green light straight into the rump of the train in front of it. Dec 3, 2014 at 2:55

I'll try to explain in layman terms. Also see Wikipedia article for more details.

The idea is that the whole track is divided into sections (several hundred feet long in underground). When a train enters any section its wheels electrically connect the rails and this can be electrically detected and semaphores can be adjusted automatically and the data on track being occupied can be seen by nearby trains.

If there's a train on some section then the section is considered occupied and the semaphore in front of that section will display red (prohibitive) light - autopilot-like system in the train will detect that electrically and stop the train before that light. One or several sections before the occupied one will have semaphores displaying yellow (warning) light - autopilot-like system in the train will detect that electrically and slow the train down in those sections. Sections "far enough" from the occupied ones will have semaphores with permissive light and there the train can roll at high speed.

Now note that there's a lot of trains on the same track and so "far enough" can be just two or three sections between trains. Should any of those trains suddenly stop all the rest will have to queue. Should any piece of wiring used for detecting track being occupied fail all the other components immediately act paranoid and assume the track is occupied and stall the trains. One that happens the pilot has to contact the dispatcher and manually run the train being verrrry careful and paranoid because the tunnels have a lot of curves and visibility range is very poor.

So basically a minor fault in that complex system immediately makes everything act paranoid and that slows thing down greatly.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .