Initial RAIB Report on Tram Accident Released

1. My sympathies to all those connected with this incident.

   In addition to the “why did this happen?” part of the investigation, will the RAIB also look at why several people died or were seriously hurt as a result?

   The outcome of this accident compares very poorly with the much higher speed derailment at Greyrigg.

   Initial reports indicate that the fact the windows shattered / fell out might have contributed to several of the deaths and injuries.

2. Highlighting the curve with further signage and / or the placement of an intermediate speed restriction between the 80 km/h and 20 km/h sections is only the very least that should be implemented.

   To ensure that a similar cannot be repeated, a simple overspeed alert (aural and visual) via a software upgrade in the driver’s cab, and the introduction of automatic train protection, similar to heavy rail, is a necessity.

   Let’s see if the RAIB stipulates this or something similar in their final report.

3. Is it normal in rail applications to have such a dramatic reduction in speed limit so close to the obstacle that the limit exists to protect? I’m aware that light and heavy rail drivers have to have route knowledge before they’re allowed out, so should know where they need to brake early which makes comparison to roads not entirely valid. However, I’ve never come across a speed limit reduction of over 30mph or 75% in one go on the road network, and where there are sudden changes — typically 60mph -> 30mph as a main road passing through open country arrives at a town or village — there are increasingly 40mph ‘buffer zones’ or ///, //, / warnings of the lower speed limit in order to give drivers ample time to decelerate safely.

   We won’t know the outcome of the report or its recommendations for some time, but the cost/effort required to add 60km/h and 40km/h signs at appropriate places would be trivial, and if put up at points where the tram should be braking to those speeds anyway in normal operation, they would not have any impact upon journey times.

4. There are numerous road locations where the speed limit drops from 70mph to 30mph. Typically that is a dual carriageway that goes into a built-up area.

5. Whilst the causes of the derailment seem reasonably clear, what is important is to establish why there were so many fatalities and serious injuries. The conventional wisdom is that most harm is caused by passengers being thrown into internal fittings such as luggage racks and tables; trams lack these features in particular, so the fittings in this instance that did cause such harm need to be specifically identified and probably redesigned.

6. @ Moosealot – it may well be trivial to put up some signs. However the impact on the timetable and cost of operation may not be. I am not suggesting that overrides safety just that we know from many articles on here that even tiny adjustments to line speeds can have disproportionate impacts overall. I see the BBC are reporting that the Mayor has said three more speed restrictions are to be imposed elsewhere on the network and the use of limiters will be considered. ASLEF are also making some pointed remarks too about wider issues.

   I think we need to be a tiny bit careful about suggesting railway solutions on a tramway. Not everything will match across seamlessly given the very limited signalling that exists on Tramlink. We also need to keep in mind that this tragic incident is thankfully almost unprecedented on any modern tramway system. There may be things that need fixing on Tramlink but let’s not start to turn a safe tramway into a railway for no good reason.

7. There may be things that need fixing on Tramlink but let’s not start to turn a safe tramway into a railway for no good reason.

   Seconding this.

   Ultimately trams != trains, and it’s important not to automatically start trying to turn the former into the latter.

   What will need to come out in the wash though is whether in some way (in staff training, signage or technology) Tramlink has drifted too close to taking on some of the risks of railway operation (in return for the benefits) without the compensating safety requirements. Time, and the ever-thorough RAIB, will tell on that one.

8. As above its the severity of injuries that worries me. 7 deaths and 51 serious injuries suggests that only 2 people on the train escaped with no/slight injuries. For a derailment this seems very high. When initial reports came through I wasn’t expecting any fatalities so for there to be 7 should warrants through investigation.

9. Graham: Only the immediate cause – excessive speed – is reasonably clear. There are many possible reasons why the tram was not slowed, though publishing any speculation before they are investigated is probably unhelpful and possibly illegal.

   Whereas discussion about the severity of the consequences could indeed, as you suggest, be of great interest, and is also unlikely to hinder the official investigation, which will also be looking hard at this.

   A further, and quite different point is the underlying design reason for the sudden change from a very straight section to a sharp curve. As can be seen from the photographs (and is probably well known locally), the unusually straight section, with tunnels, is part of a re-used railway line, whereas the curve exists because trams in both directions from this line were redirected into Croydon. And the reason for that centres on the ingenious way that a line – very lightly used mainly because it did not serve Croydon, was repurposed into two tram lines which do.

10. I am reminded of the circumstances of the Santiago de Compostlla accident in 2013, where a train derailed after failing to reduce speed from a high speed run through a series of tunnels in order to negotiate a tight curve. The absolute speeds in question were much higher (120mph to 50mph) but the relative difference is actually smaller (a 62% reduction rather than 75%). Different in detail, as the RENFE accident was on a line with speed controls approaching the accident location, but not on the stretch where the derailment actually occurred. From reports of the investigation, it would seem the driver was expected to control the speed but suffered a lapse of concentration – possibly losing count of the tunnels.

    The curve at Sandilands would not be unusually tight on a street running section, but is unusual for a reserved track section.

    Many modern cars are fitted with speed limit recognition systems, either based on satnav data or on pattern recognition of road signs, and also with automatic speed limiters. It would surely not be impossible to fit such equipment to trams.

11. @Malcolm
    “the ingenious way that a line – very lightly used ”

    Very lightly used indeed – it had no service at all after 1983, nearly a decade before Tramlink was given the go-ahead

12. Sympathy to all who have been affected, directly or indirectly, as a consequence of this accident. Obviously the fact that the tram was proved to be in excess of the speed limit is a major factor but let’s not prejudge and assume reasons for the excessive speed. Let RAIB complete their usual thorough investigation to determine the facts. However some immediate consideration should be given as to what caused the fatalities and critical injuries in what was a relatively low speed (44mph) accident for a road or rail vehicle.

13. Regarding the high death count. There have been harrowing reports in the media regarding the injuries that sere inflicted on some of the casualties / dead and the injuries included crushing as the tram slid across on its side. I assume that, as the windows were broken / missing, some people were in direct contact with the ballast and whatever rails the tram passed over. Also people potentially being trapped between the side of the tram and the track.

    The tram was quoted as moving for about 25m after derailing before it came to a halt. That is a lot of sliding in a very short time.

    Judging from the position of the tram when it came to a rest, as clearly shown in one of the overhead photos in the media, it would appear that the rear travelled the most. It looks like the front derailed and travelled a relatively short distance before coming to a halt whilst the rear continued travelling about 50 degrees anticlockwise. Possibly the front was stopped by the adjacent track (shown buckled) or pylon / equipment and the momentum carried the rear along before ending in its final position. I would imagine that, assuming passenger loading was evenly spread, more injuries would have occurred to passengers in the rear car than in the front.

14. @John Bull,

    Correct, there should not be a rush to impose all heavy rail regulations onto light rail.

    However, given that no automatic speed inhibitor nor overspeed brake application system exists on Tramlink, this is clearly an operational flaw that has not been designed out.

    Given the lack of safeguard on the approach to a substantial change of direction after a relatively high line speed section, and the reliance on a human operator maintaining appropriate level of alertness, this (like the RENFE incident highlighted by timbeau) was an accident waiting to happen.

    The technology is available and should be employed at the most vulnerable locations on the network.

15. I am sure we are all trying to make measured comments and I acknowledge Malcolm’s stricture to avoid speculation. Can I therefore make a general observation. It seems to me that in a long straight length of track comprising open ground and then tunnels, the usual visual clues which we all rely on to indicate speed are lacking. (Yes, I know the driver has a speedo but he also has other things to look at). Whether this had any effect on the tragic accident is for the RAIB to determine but I note that initial media reports referred to suggestions that apparent excess speed in this area was by no means unknown.

16. People have suggested that the Croydon Trams and presumably other tram systems in the UK should be fitted with some form of automatic speed supervision.

    Can anyone advise is the UK in a position to do this or are trams subject to European Approval and thus would require a change to the European Directives / Regulations to impliment this?

    I had thought that Trams were approved under the European Bus & Coach legislation but a search of the VCA (who do UK Road Vehicle Approvals) at http://www.dft.gov.uk/vca/index.asp fails to find any mention of tram. I can’t find a copy of the European Bus and Coach standards so have been unable to confirm or deny if they cover trams as well.

    It appears that the new Croydon Trams that were originally intended for Oslo appear to have been able to run in Croydon presumably with limited modification. This makes me think that they are built to some European Standards.

    A search of some of the EU legislation for ‘tram’ shows some possibly national requirements ( mostly German) for tram systems.

    They are not trains as far as I am aware – at least from EU TSI stance.

    Anyone got any more / better knowledge of the legislation / approval framework ?

17. Re Littlejohn,

    The RAIB are certainly looking at the effect of tunnel lighting and other visual effects but that will be very slow burning part of the investigation.

    Re Rogmi (& Anon 1800 and Flare),

    Agree with Rogmi’s analysis hence why I’m not surprised at the high casualty levels (I would have written something similar). With heavy rail train derailments the track curvature is far less so the net direction of motion is closer to along the length of the train (and it usually remains upright for while too and loses speed – see Hatfield, Greyrigg, Watford) than along the height of the tram as in this incident, particularly in the rear car, so lots of the measures adopted for heavy rail injury reduction won’t have worked. The tram design also pre-dates the era of attempts to better retain passengers in the cars in the event of an incident. With the front of the tram stopping presumably due to combined effect of hitting the mast and deforming the up track from Beckenham the retained momentum and resultant impulse will presumably have caused the rear to travel further across the ground increasing the injuries sustained there.

    The solutions to making tram derailments more like heavy rail derailments and hence safer is either in reducing the max speeds round tight corners or relaxing the curvature round tight corners so the net direction of motion in a derailment is along the train. This has some fundamental effects on future light rail schemes or extensions in the UK especially around greater land take for smoother corners or not and schemes not being viable (several features of the proposed Sutton schemes are now effectively in the spotlight). The stability of low floor tram designs at higher running speeds may be questioned.

    With the Croydon trams running slower at four places on the network after this incident this will increase the overall tram requirement at todays service levels and hence more new trams or reductions in the current service level given TfL current budgetary “issues” especially as it is wise to assume the tram will be written off.

    Speed limiting using GPS and balises for the larger covered sections (e.g Sandilands tunnels and under the shops /car park Wimbledon ) should be relatively easy to implement as the technology is already exists for other applications in the heavy rail industry for precise location data without requiring signalling (e.g. SDO/CSDE on Bombardier rolling stock).

18. Looking at the pictures, the tram went from travelling at 43.5 mph to stationary in a very short distance (almost the length of the tram itself), plus it turned over, so the high number of injuries shouldn’t be that big a surprise, as the unsecured passengers would have hit the internal fittings and each other at quite high velocities.

    I’m not sure what chance anyone standing next to the doors would have had, any other passengers standing nearby would have fallen on you and pushed you through the full length glass.

19. @Littlejohn
    “in a long straight length of track comprising open ground and then tunnels, the usual visual clues which we all rely on to indicate speed are lacking.”
    As it was some time before dawn, there would have been ever fewer visual clues than usual. Could it have been that the end of the tunnel was not as obvious as it would be in daylight?

20. As for the high numbers of injuries,
    1. When was the last time a passenger train derailed and fell on it’s side with a reasonable number of standing passengers inside?

    2. How many accidents have involved open coaching stock with a reasonable number of passengers.

    Also it should be noted the large side lights that are not normal in UK passenger train stock. Similar designs of tram / train are now used in Europe (Switzerland and German) on the mainline. I seem to remember that high numbers of injuries have been a feature of the last two crashes.

    While the use of the disused railway line seems a good idea, I have always wondered if a better arrangement could have been used to link with East Croydon, than the tight curves and humpback arrange meant at Sandilands.

21. @John M

    I’m not sure what was in force at the time that Tramlink opened (possibly covered by the Croydon Tramlink Act 1994?) but vehicle approval would seem to be one of the functions of the Independent Competent Person required by ROGS – formally The Railways and Other Guided Transport Systems (Safety) Regulations 2006 (as amended).
    http://orr.gov.uk/__data/assets/pdf_file/0020/2567/rogs-guidance.pdf

22. Re timbeau,

    The tunnel is lit so fairly obvious when you reach the end in the dark…

23. A shocking turn of events for us locals. I have always enjoyed the convenience and speed of the trams, and it never crossed my mind that something like this could occur, nor have I been thrown around as a passenger.

    Rapid changes of speed such as 80 to 20 kmh are quite common on Tramlink, not limited to this corner, but at other tight curves, junctions and road crossings.

    The RAIB report states they will investigate ‘any previous over-speeding incidents at Sandilands Junction’ among other factors. The BBC are reporting that passengers had complained the week before the crash about a tram going uncomfortably quickly around the same corner http://www.bbc.co.uk/news/uk-england-london-37979121.

24. Most car sat nav systems (and indeed, the sat nav apps in many of our phones) will flash a warning and/or sound an alert if the vehicle is travelling significantly higher than the limit. Even relatively ordinary new cars (such as a Ford Mondeo) can “read” road signs – including temporary speed restriction signs – and display the limit in the sightline of the driver. This is reliable, effective, technology that could be fitted for very low cost. (And it’s not a “heavy rail” solution).

25. JohnM: It would be interesting to know more about the relevant regulations. But as far as I understand it, such regulations, whatever they are, would not prevent the fitting of any additional safety devices. Regulations would /mandate/ many things for safety, but I would have thought that an operator would be at liberty to add further gadgets without the regulations having to be changed.

26. Sure, trams are not trains, and trams can do some things that trains simply can’t do (or at least can’t do without quite cumbersome exceptional arrangements).

    But when a rail vehicle is going at 80 km/h down a stretch of nice straight grade-separated track, with neither street lighting nor any road markings in sight, then in many ways it’s behaving more like a train than like a tram — even if it’s still running “on sight” and the driver is prepared to avoid another tram that might be stopped in front of it (which would be assumed at least to have functioning tail lights).

    Perhaps too little thought has been given to the interface between act-like-a-train territory and act-like-a-tram territory, for making sure that drivers have sufficiently conspicuous cues when they need to switch mode. At least, a single reflective trackside sign does sound like a rather meagre indication. And perhaps the mindset that “trams are trams and this is a tram system all over” has made it more difficult for planners to realise that there is a mode switch to warn of at all.

    (This is not exclusive to the Croydon system, of course — probably most tram systems from the last several decades feature railway-like stretches where the topography allow it).

27. In some countries speed controlled traffic lights are used. They’re very common in Spain at the entrances to rural towns. The lights stay red till you drop to say 50kph then typically flash amber. You risk getting done for both speeding and jumping a red light if you fail to comply.

    Surely simple and quick to set up on the tramway (compared to modifying all the vehicles which can come in time) with loss of employment being the sanction for non-compliance.

28. Regarding automatic speed restriction- it is perhaps worth noting that TfL are fitting buses on routes in central London with speed restrictors, accurate enough to work with the mesh of different limits based on the London Digital Speed Limit Map http://content.tfl.gov.uk/digital-speed-limit-map.pdf . That gives a precedent that isn’t heavy rail.

    Also on not-heavy-rail, on the not-restricted Cambridgeshire Guided Busway there have been two incidents in the past year that have been attributed to drivers speeding through a low limit section, with both drivers sacked. Unfortunately the busway isn’t covered by the RAIB, or really indeed any specialist body, so there’s no public independent investigation and we won’t find out about any habitual speeding at the locations- whereas the RAIB have clearly said they are looking at whether this is the case in the Croydon incident.

    Perhaps more generally public transport that isn’t Heavy Rail could do with some stricter regulation, both in paperwork and mechanical means?

29. @ Al S – I’m going to be controversial here and say I’m not remotely convinced that we need loads more regulation and whizzy systems. What we need and I hope we will get is a rational assessment of all root causes and effective solutions identified to remove or substantially reduce risk together with improved ongoing management, assessment and review processes (where required). I don’t consider the proposals for the bus network to have any great relevance as I’m not convinced the bus network is as “dangerous” as some campaigners and politicians would have us believe.

    The use of speed limiters is likely to be pointless given the ludicrous levels of congestion and highly complex street pattern bus drivers have to cope with, never mind the spread of 20 mph limits and road humps. I’d be astonished if many buses ever exceed the applicable speed limit in London. I am also not convinced that those accidents that do take place are solely the fault of bus drivers. There are far too many instances of people wandering around unaware of their surroundings because they’re plugged in a musical device or have their head gaping at a phone. All the vehicle technology in the world won’t solve this. Where drivers *are* at fault then the law and due process with the operators should take its course. In all other instances where the driver and operator are not at fault then that’s where the efforts should be directed and not on pointless on vehicle technology.

    I know we all want to rationalise and understand this incident but I am a bit concerned that we are seeing lots of “statements of the obvious” from all quarters when it’s not at all obvious what happened or why. We have one key statement from RAIB and a lot of conjecture about what may prove to be quite a complex and involved sets of factors. I say this solely because I have read a fair number of incident investigation reports and had specific safety responsibilities in my last job. Beyond that I’ll wait to see what RAIB find and conclude.

30. @Henning Makholm, 16 November 2016 at 22:15

    I advocate a very clearly visible advanced warning board, positioned at service braking distance ( in this case about 180m) from the start of the restriction to show drivers where they MUST have started braking for these major reductions. That could be supplemented by a ‘Tram AWS’, activated by a simple modern passive transponder located between the rails. At it’s simplest such a transponder could trigger a cautionary alert and reminder in the cab with an AWS-like requirement to acknowledge within a time to avoid an emergency brake application. As older trams get replaced or upgraded, more sophisticated behaviour could be specified, triggered by the same transponder, i.e reduce speed to a given target within a given elapsed distance or time. These AWB installations would be applied sparingly on a risk assessed basis only at sites where there is a particular hazard, such as approaching Sandilands, Note the section in question specifically excludes pedestrian by full fencing and provision of signs and level crossing style ‘cattle grids at its extremities. Note also that the speed limit commencement signs used throughout the network are small unobtrusive diamond shaped plates with grey borders, and are not particularly visible.

    This video includes a journey through the tunnels between Lloyd Park and Sandilands.: https://youtu.be/lhuogCAh6Pg?t=11m50s

31. For me the key phrase in the article is “In order to reduce its speed to the 20km/h by the time it reached the reflective speed limit sign, it would have to brake at its full service rate 180m in advance of it”. While it may be tempting to come up with all kinds of technological methods to enforce speed limits, by the time the tram was actually exceeding the posted limit (when it passed the sign), an accident was already inevitable.

    Having to slow from 70mph to as low as zero mph is common on the road network – it happens often when a dual carriageway meets a roundabout, as well as on motorway sliplanes (and there have been fatal accidents where coach drivers have failed to slow down enough in this kind of location and coaches have overturned as a result).

    Sometimes there is a formal change in speed limit or an advisory speed posted, but mostly a range of more subjective measures (road geometry, countdown signs, surface treatments) are employed to slow drivers down. Presumably there is a body of research into which methods work best. The key thing is that plenty of advance warning is always given – the highway equivalent of Mark T’s “start braking now” sign.

32. The cheapest do something solution is probably one of the solutions used in Dublin:
    multiple speed-limit signs where needed but also with the distance till the speed limit starts on the signs e.g. 30kmh – 200m, 30 kmh – 100m etc. I think they also have intermediate speed sections before going from max speed to tight bends.
    The Croydon speed limit sign is actually after the start of the bend as can bee seen in Mark T’s linked video.

    Also interesting to note in the video is the pile of “scrap” rail when the video was taken in August ’16 directly under where the tram landed. It would be interesting to know if this was still there when the derailment happened and if so whether this contributed to the injuries sustained.

33. @ ngh 19:58
    “several features of the proposed Sutton schemes are now effectively in the spotlight”

    Please elaborate.

34. Tom Westmacott wrote :
    “nor have I been thrown around as a passenger”

    Well I certainly have. Very early on, if not on the opening day, I made the mistake of getting up from my seat just as an outbound Beckenham entered the curve after leaving Arena and I landed on the seats opposite. Soon learned to stay seated.

35. @walthamstowwriter. The borough I live in has a 20mph limit (save on 5 tfl roads). Buses routinely exceed this 20 limit – as indeed does all other traffic. I was driving locally yesterday, and as I attempted to hold my speed to 20 (almost impossible on a wide low traffic road), the bus ahead accelerated away from me. I really do think that the adoption of GPS limiters is going to play havoc with timekeeping on some routes.

36. I think the need to make drivers aware that speed limits on rail networks are there for a reason and need to be adhered to. I witness on a daily basis the total disregard of restrictions by drivers and a total lack of management interventions or any type of punishment for repeat offenders. Maybe just maybe people/drivers eyes may be opened to the devastation excessive speed can cause.

37. Walthamstow Writer, Island Dweller,

    I’d be astonished if many buses ever exceed the applicable speed limit in London

    I concur with Island Dweller. In Croydon, George St East is 20mph and, as the road bit outside Croydon station is bus and taxis only, nearly all the vehicles that use it are buses. They don’t always keep to the limit and certainly didn’t pay much notice at all when it was introduced.

    One of the ideas of 20mph speed limits is they effectively become self-enforcing if enough drivers voluntarily (or are forced) to adhere to it. On busy, but not slow, routes limiting the buses to 20mph would do a lot to make them more effective.

38. I wonder if the tram driver might have suffered from tunnel vision?

    I’m with those who don’t advocate an over-reaction and the introduction of (expensive) electronic measures…..

    If some kind of “rumble strip” can be fitted to the track to indicate when the driver needs to start to slow, then that is what should be used, If extra signs are required, then use those. If something can be done with lights, then use that. But fitting all sorts of additional systems for a single incident, to me looks like a massive over-reaction…

    I have travelled along that stretch of Tramlink (in daylight) and did wonder about the signage, as I only spotted the sign at the curve and nothing in the tunnel before hand, but then I did not have a front row seat…. I also wondered about light from the rising sun, but it was too early for that…

    Anyway, it will be good to see the final report, knowing RAIB reports in the past it will be factual and honest.

39. @Walthamstow Writer – on your point about bus fatalities, police will only get involved if there is clear evidence of criminality and there is no thorough and dispassionate analysis of the root cause as is done by the RAIB for rail incidents. You are speculating that a large number of bus fatalities are the victim’s fault but can you point to any analysis or investigation that has been done on this?

    Over the last 10 years, on average there has been 2.3 “Croydons” every year in terms of the number of people killed by London buses but there is a stark difference in terms of the level of scrutiny done between the two modes of transport.

40. Among the automated announcements on the trams, there is one (not Nicholas Owen, a female voice) which warns “trams can accelerate and brake suddenly” or something similar. In other words, it is a policy of the operating company that they expect drivers to accelerate and brake suddenly.

    Regular passengers get used to the way the trams are driven, so they would not be on their guard when sudden braking begins to occur, giving them less time to hang on to something secure.

41. I have never been involved in a bus incident due to the bus travelling at excessive speed, but have been involved in a couple of incidents where passengers were thrown off their feet because of private car drives suddenly turning in front of buses.

    A number of press reports have quoted passengers suggesting that the tram driver appeared to be taken ill in the moments before the crash, so this may have been a case of incapacitation rather than lapse of attention.

42. @WW
    I was specifically suggesting a cheap and easy improvement that would not impact journey times. If it is generally accepted that the tram needs to start braking from 80km/h 180m before the curve, then at -100m the tram should already be down to 60km/h* so a 60km/h limit at that point would not affect the timings of a tram running in normal service.

    If the driver saw the sign 20m away, traveling at 80km/h, then the tram will have 120m to slow. Application of the emergency brake at that point would be likely to get it below 30km/h for the curve, which should be survivable without derailing.

    * Working: 22.2m/s -> 5.6m/s assuming steady braking gives an average of 13.9m/s, which equates to 13s to cover 180m giving a constant deceleration of 1.3m/ss. This sounds reasonable for an in-service tram. It will thus take 4.25s to decelerate from 22.2m/s (80km/h) to 16.7m/s (60km/h) in which time the tram will have travelled 82m.

43. It is good to see the measured comments in this forum. I found the comments from ASLEF rather distasteful –

    “While individuals will be held accountable for their actions, it is clear that the lack of adequate safety systems were at the root of this dreadful accident,”

    and

    “The operation of Tramlink should now be taken under the direct control of TfL …”

    I dislike the attempt to move the blame away from their member, even before that has been fully investigated, and then try to use the accident for political purposes. RAIB will determine the cause of the accident, whether it was the driver or lack of safety systems, and make suitable recommendations.

    Ironically, with driverless technology progressing so quickly, the “safety systems” of the future are likely to remove ASLEF members entirely (please, no comments on the merits or otherwise of driverless systems).

44. @Philip
    I witnessed a very close shave at a pelican crossing on Pentonville Road last year. The bus was either overloaded (it was a tube strike day) or speeding or more probably both, and it was definitely going downhill; the bus’ stop lights came on pretty much the instant that the crossing lights went amber, and I could see it dive under heavy braking, so the driver clearly had the brake pedal on the floor… but it still ran through the red light and came to a halt with 2/3 of its own length too late. The pedestrian using the crossing was wearing earphones and had started crossing, but saw the bus in the nick of time and jumped backwards out the way with less than 0.5s to spare (if she’d been taller she would have been hit by the wing mirror). The thing that did surprise me is that while the driver must have known some distance away from the stop line that he wasn’t going to make it, he didn’t use his horn.

45. It might be worth adding that just as the new Stadler trams are based on a design for Oslo among other places, the original fleet was based on a design for Köln, the main difference being that those have a somewhat longer central section.
    On the issue of signage, of course tram speed limit signs are fairly small diamond designs, I assume to distinguish from any road speed limit signs which could be adjacent. From my recollection of the layout – I should say that I live locally and use the system frequently – I’m not convinced that the speed reduction sign would be very visible from 180m, the point at which service braking should start according to the RAIB report. By comparison on a trunk road one might well encounter an advance warning sign for a major speed reduction. Whatever the final report says I can understand why additional signage or a stepped speed reduction is being suggested.

46. RayL: re a warning that, “trams can accelerate and brake suddenly” or something similar. In other words, it is a policy of the operating company that they expect drivers to accelerate and brake suddenly.”
    Most road vehicles can accelerate and brake suddenly, in fact brakes must be able to work suddenly, and the driver of any vehicle needs to be able to apply them suddenly. That does not make it a ‘policy’.

    @ Jim Cobb: “RAIB will determine the cause of the accident, whether it was the driver or lack of safety systems, … ” or, one might add, possibly a bit of both, or conceivably a hitherto unforeseen or undetectable issue that escaped the attention of systems and driver. One could interpret ASLEF’s comments to be intended to reassure their members. The choice of words is open to debate, it usually is when there are quick reactions to things which will take a lot of unravelling, but such debate, in my view could be ‘political’ on the part of their critics.

47. @Moosealot
    Both the law and the Highway Code say that you should not brake sharply if a light goes amber in front of you if you are so close that to do so would cause an accident. If the bus braked immediately the lights went amber but still overshot the crossing by 2/3 of its length, it was too close to stop safely, especially if the bus was full-and-standing.
    So unless the bus was going much too fast, the amber phase was too short.

    @Southern Heights “I also wondered about light from the rising sun,”
    Unlikely – the tunnel is aligned slightly east of north, and would never have the sun shining through it. On the date in question, sunrise would be roughly at right angles to the alignment of the tunnel.

    @Retrogordy
    “I think the need to make drivers aware that speed limits on rail networks are there for a reason and need to be adhered to. I witness on a daily basis the total disregard of restrictions by drivers and a total lack of management interventions or any type of punishment for repeat offenders.”

    Where are you seeing this? LUL or National Rail?

48. The video explains quite a bit – with the only 20 sign being directly before the curve. Much as route familiarity plays a vital role it’s all too easy to see why they are adding further speed limit signage in the first instance. This will at least reduce the risks down from ‘genuine’ mistakes to flouting the rules or incapacitation. Regardless of which of the three was the case here, other incidents (e.g. Moorgate) must point to some form of automatic protection against the latter two being implemented – particularly with the terribly high casualty rate here in mind.

    The only time I have experienced alarming speeding was many years ago on the Camden Town junction, when I felt lucky that the train stayed on the tracks after a massive sideways jolt on the points. I would say that the auto-speeds on the junction these days are generally more cautious than the manual days.

49. Sydney’s light rail system, very similar in concept to Croydon’s, was originally equipped with the Ebicab ATP system. It used balises and wheel revolutions for location detection, and would brake any tram that was over speeding or passed a signal (although not an on-street traffic light) at danger.

    Obviously we cannot speculate on the particular cause of the speeding in Croydon’s derailment, but in general I believe that trams pose a particular challenge. As mentioned already, changes in speed limits can be dramatic over short sections. The high performance of trams probably leads to somewhat less cautious behavior from drivers versus those driving a train.

    Those who haven’t driven a tram may not understand how the tiredness of extreme shift work is compounded by the exceptionally comfortable driver’s seats and gentle rocking motion of the tram. Several (Sydney) drivers I spoke to admitted in private that they had been woken from a brief slumber by the ATP blaring away.

    I find the suggestion of installing extra signage to be laughable. Signs tend to become ‘invisible’ after the first few viewings. TPWS traps would be more useful, but the benefit could be limited, since the high performance of trams may lead to drivers reapplying power after a speed trap but before the hazard if it is overly restrictive.

    A basic ATP system may be expensive, but so are accidents. It should be fully considered, and not immediately dismissed by the ‘trams are not trains’ brigade.

    As for Sydney, they removed ATP from their system after the most recent extension. I’ve been waiting to hear of a Croydon style incident there, mercifully it hasn’t happened so far.

50. A potential problem with many modern tram systems is the mixture of on-street running and reserved trackway. Other than speed limits, is there any difference in how trams are operated in the two modes? Manual driving, on sight, seems to be the norm on reserved trackway as well as on the street. Is there any safety system to prevent a tram being driven at 50mph down Addiscombe Road and George Street, or for that matter Edinburgh’s Princes Street, or Manchester’s Market Street, should a driver take it into his head to try?
    Any system which automates the reserved trackway sections would have to be able to transition in a foolproof and failsafe way onto street running.

51. @Anonymous
    “Those who haven’t driven a tram may not understand how the tiredness of extreme shift work is compounded by the exceptionally comfortable driver’s seats.”

    Quite a difference from first-generation trams – tram drivers were not provided with seats.

52. @timbeau: Removing the driving seat would allow for more room for passengers as well…. 😉

53. Regarding further speed restrictions, I cannot see how putting up ’60’ and ’40’ signs at appropriate distances in advance of the ’20’ reduces capacity. Drivers have to start braking some time before the ’20’ signs anyway to drop from 80 to 20, having two additional signs at the appropriate points (i.e. not 300m but at the right points of the braking curve within the last 180m before the ’20’ sign) may actually help drivers with braking at the right time.

    Having said that, if the tram derailed with 70 kph then clearly additional signs would have made no difference as it is not an incorrect braking curve that is to blame but the absence of meaningful braking, something no speed restriction signs can solve.

54. Anonymous at 16:46: I agree with you about the way signs tend to become invisible after first viewing, particularly to someone who passes the area many times a day. I wouldn’t quite go so far as to describe the suggestion of extra signs as “laughable” – for one thing it might help to appease the “something must be done” brigade. But whatever it was (illness, distraction, or something else) that prevented the driver from using his route knowledge to decelerate would very likely have also prevented him from seeing any number of advance warning signs.

    It would be very interesting to know exactly why Sydney removed ATP from their system.

55. The notion which timbeau raises, of a crazy tram driver going for 50mph in Addiscombe Road is hopefully highly improbable, but perhaps we should note that there is likewise no mechanical device preventing a crazy bus driver from doing the same thing.

56. Can I clarify the point made earlier about the Independent Competent Person (required for non Interoperable railway jobs where there are “new or novel” risks – heavily summarised)? The ICP does not “approve” projects. The ICP does an independent safety review and will expect his/her recommendations to be carried out. However the responsibility for approval rests with the transport undertaking. ORR usually gets involved as a new vehicle almost always requires a significant alteration to the transport undertaking’s safety certificate.

57. My line of work became heavily regulated about twenty years ago due to concerns about photosensitive epilepsy, a type of epilepsy, in which all, or almost all, seizures are triggered by flashing or flickering light. This was due to an incident with a group of children in Japan after a particular programme was shown. This is why news programmes now often warn viewers about reports that include images of flash photography.

    The relevancy to the topic is the way that the tunnels are lit. If there are bright lights and darker areas at close spacing then there is apparently a risk of photosensitive epilepsy for the driver and passengers which may be more likely in hours of darkness outside the tunnels.

    There is also a view that some RTAs in France on the traditional long straight avenues of trees are caused by this phenomenon.

58. @anonymous: 16:46: I find the suggestion of installing extra signage to be laughable

    Extra signage is exactly what the RAIB have suggested in their interim report – do you consider the experienced safety professionals there to be laughable? If so, why?

    @Malcolm: whatever it was (illness, distraction, or something else) that prevented the driver from using his route knowledge to decelerate would very likely have also prevented him from seeing any number of advance warning signs

    As a pedantic but perhaps important point, the report makes it clear that driver did in fact apply the brakes, albeit nothing like enough. This might seem to cast doubt on scenarios where the driver was entirely incapacitated.

59. Ian J: you are right about total incapacity being probably ruled out. However, I do not think that this invalidates my point about how extra signage would probably have been useless in this case, as there must have been some reason for the over-late braking.

    I do agree with you, though, that “laughable” is too strong a word for the suggestion of extra signage – apart from giving public reassurance, it might, just might, prevent another slightly different accident, even though it probably would not have prevented this one.

60. @Malcolm: extra signage would probably have been useless in this case

    Unless the cause of the failure to brake in time was locational disorientation – the driver being fully conscious and able, but mistaken about where on the route he was.

61. Ian J
    But .. IF (perhaps) the driver temporarily seized, because of the light-flicker effect in the tunnel [ Which can be very disorienting ] then on beginning to recover, he would, of course, apply the brakes – but it would be too late by then.
    
    Kit Green is entirely correct concerning the effects such flicker phenomena can cause – I have seen it work on other people & a friend crashed-&-totalled a Saab, driving in a forest, because of said effect.
    Nasty.

62. Ian J: True, and a good point.

63. “Laughable” depends on your sense of humour…
    But do we have any evidence that increasing the number of signs increases the safety?
    What proportion of people fail to read or notice signs?
    At what point do the number of signs actually decrease safety?

    The Aviation industry has been looking at the safety checklists as they have found the longer lists take too long to work through, and delay, potentially life saving action. see US Airways Flight 1549.

    To often safety professionals seem to work to the idea, that when the weight of safety paperwork weighs the same as the train the train will be safe. But is this true… In the chemical industry as the weight of the safety paperwork increases so does the examples of work being exported off site. At University I was paid to run experiments that were banned under the safety culture at the time in industry. Now we happily export manufacture and research to China, and the safety professionals claim we are safer, well yes ‘we’ are safer, but the risk has been exported and could be much greater.

64. Anonymous(16:46) here.

    @Ian J: in hindsight, “laughable” was too strong a word, and I apologise for it. @Malcolm referred to appeasing the “something must be done” brigade, and I was going to use that exact phrase in my original post. By doing so, the impetus to do something that is substantially more helpful, such as installing active safety measures, is reduced.

    Should a similar incident occur in the future, the authorities could point to the signage as evidence that they have done all they could to manage the risk, leaving the driver to shoulder all the blame. This is a particular bugbear of mine, humans are frail, and even those that don’t intentionally engage in negligent behaviour can suffer lapses. Train drivers have TPWS or TASS or train stops, signallers have interlocking and lever collars. Tram drivers have very little equivalent technology to assist them, and any mistake they make can have disproportionate consequences.

    And yes, bus drivers don’t either. However they at least have control of their steering which would help to improve alertness, and may be able to reduce the severity of an accident through steering unless they have lost consciousness entirely.

    @Malcolm, I don’t have a first hand account of why the ATP was removed, I believe the lineside equipment was installed on the latest extension but the new trams were never equipped and so the system was decommissioned. There was a perception that the ATP slowed down operations, and that was true in so far that it was poorly configured, for example speed limits were overly cautious and earlier than necessary. It was a highly configurable system, there just didn’t seem to be the will to get it working properly. As far as I’m aware they still use lineside signals, I have no idea how (or indeed if) they now mitigate SPAD risks. There are plenty of blind spots on the line, some involving crossovers, where a driver passing a signal at danger assuming the line was clear could cause some very serious trouble. At least Croydon has the ‘strobe light’ system protecting the single track sections.

65. @Ian J, 18 November 2016 at 03:59

    It did strike me in the video that there was no signage or other positional information inside the tunnel, through which it might thus be difficult for a driver to accurately gauge where to start braking. Additional ‘stepped’ speed restrictions on approach to the 20 aren’t really helpful except to provide proxy positional clues. This is why I prefer a more explicit advance warning that helps the driver in showing where to start braking. A new type of sign could be designed for better visibility than the standard speed sign, maybe with a light built in for tunnel/night operation. On the heavy rail main line, AWBs for speed reductions, like distant signals, are among the most important trackside objects, the only chance of getting sufficient advance notification of a required speed reduction or impending stop. In the limited areas where operators are tacitly expecting trams to be ‘driven like a train’ then at a very minimum some form of advance warning of major speed changes seems wise. This is not a slippery slope ‘back’ to expensive fixed block signalling on tram reservations like the original Manchester Metrolink. Trams following each other on line of sight are still comparatively safe. It’s the sudden unexpected change in infrastructure characteristics that is the danger here. I think AWBs are the most promising answer with the possibility of adding a ‘tram AWS’ system for positional vigilence. The AWS, based on digital balise technology, could be developed incrementally, first just providing a cab warning, then adding various levels of supervision such as:
    – Acknowledge within time or trigger emergency brake.
    – Supervise speed reduction within parameters or trigger emergency brake.
    – Take control of traction and braking to enforce reduction.
    The additional functionality levels could be introduced after an initial simple warning only system was introduced That would permit some additional protection to be provided ASAP without waiting for extensive system development and rollout on existing rolling stock. The more sophisticated behaviour beyond simple emergency brake interventions might have to wait for the next generation of vehicles.

66. When this happened I took a look at video of the track and compared it to video of other tram systems, especially in Brussels. I was left with one question: Why are there no check rails?

    It would probably have been insufficient to stop it toppling, but it might have helped.

67. Speed indicating signs are commonplace on our roads. They accurately display the speed of approaching vehicles. I wonder what is wrong with installing such a sign at a suitable trackside location on the approach to the curve, equipped to sound a loud klaxon, illuminate a red STOP light and/or even place a detonator if the passing tram is exceeding the proper speed. Just a thought. Better than sleepwalking into another disaster.

68. Re Alison W,

    Check rails are really ineffective for small diameter wheels (e.g tram ones) at any speed so no point in fitting them! The only sensible way to stop it toppling would be to have a wall between the 2 tram tracks equivalent to concrete bridge protection structures on new /upgraded bridges to deflect the tram round the corner rather than topple (the croydon trams are slightly lighter than the heaviest HGVs).

    Some of the Brussels tram network was designed to be converted to metro later so has unusual track specifications for a tram system (Ditto Dublin segregated sections especially the Green line)

69. In response to Malcolm’s comments about what changes an ‘operator’ can make as I understand it the situation is quite complicated. When a ‘product’ is CE approved manufacturers can’t put the item on ‘the market’ if the product is non-compliant. As an aside this is why new escalators on London Underground have cleated step risers as required by EN115 rather than the plain risers required by the LU specification as they were believed to be safer.

    If a product is modified it may need to be re-certified and would not necessarily pass if the modification is not in the original standards. There are also EU ‘common market’ rules that require that products are the same in all countries. It is all very complicated. It may be easier when we leave Europe unless we wish to stay in the common trading area.

    In terms of the comments on road speed limits above road speed limits are not set to prevent cars leaving the road at sharp bends. Think of all those minor roads with a 60mph speed limit and right angle bends let alone junctions where it would be impossible to turn left or right at 60mph. Drivers are expected to ‘read the road’ and control their speed even if they have never used the road before.

    In response to the comments about Braking Rates I understand that trams have to meet the same braking rates as road vehicles which is why they are fitted with magnetic track brakes ( also like road vehicles the brakes do not have to be ‘fail safe’ which is why they are electrically energized to stop). I am not clear what the braking rates on the off road sections have to be mostly because I don’t understand the legislative framework as in my first paragraph above.

    Hopefully the RAIB / ORR reports will clarify this.

    In response to the apparent failure of the glass during the accident this has been covered by some of the RAIB reports with a tension between laminated glass to keep passengers within the vehicle during a crash and smashable tempered glass in case of fire. Buses I believe have tempered glass in the side windows – based on the hammers or green smashing buttons on modern buses.

70. With regard to the comments including mine about speed control TfL have been trialling a system on buses see https://tfl.gov.uk/info-for/media/press-releases/2015/june/london-buses-to-trial-speed-safety-technology.

    This appears to have been going on for some time as a search for ‘Intelligent Speed Adaptation technology’ will show initial reports going back to 2006 see the survey of technology on the TfL website. There is also a large amount of EU research though this is mostly dedicated to speed limit compliance rather than than Maximum Safe Speed compliance that would be vehicle and road condition dependent.

71. Regarding the “flicker” effect of tunnel lighting and trees – this is very well known in the industry and it is the reason that Eurostars and the tri-bo Eurotunnel locos do not have proper side windows.
    I’ve been through the Tunnel in the cab and the strobe effect of the tunnel lining, let alone lighting, is quite noticeable.
    How they deal with this on the Tube I know not.

72. As I understand things it is rare for a tram that derails to overturn if they are single deck. Usually they derail, remain upright and grind to a halt or hit an obstacle. Looking at this case it would seem that the rails were quite proud of the surrounding surface, It is more usual to build up the surface to close to rail height. I suspect that if this had been the case things might have been different. It has been suggested elsewhere (LRTA discussion group) that the deep windows and high mounted seats may have been a factor in the number and severity of the injuries. I recall a survivor of, I think, the HST crash at Southall describing being dragged along the ballast when the window broke.

73. Normally there would be superelevation of such a curve, but the junction makes it impossible in this case – the two curves would need to lean in opposite directions.

    A redesign of the junction to put the points further from the curves would solve this, but it would mean you would need four tracks over a some distance.

74. timbeau: …. which would of course make the curve radius even smaller! (It’s the diamond that is the constraint, not the (symmetrical) turnouts).

75. @JimJordan, These single deck trams have a lot of the heavy equipment on the roof rather than under the floor to allow low floors for easy boarding. You can see from the pictures of the derailed train that the only heavy items are the bogies and traction motors. Pictures on the Internet show the equipment on the roof. As a result the center of gravity will be higher than something like a tube train where the equipment is below the floor. The higher center of gravity makes the vehicle more likely to overturn (and results in lower speed restrictions on curves).

76. @TJ: At what point do the number of signs actually decrease safety?

    Probably at a considerably higher number of signs than the number in place on Tramlink before the accident. Placing warning signs in advance of dangerous points, rather than at the point of danger itself, has been standard practice on the road network for many years – you would never have a roundabout on a 70mph dual carriageway without some kind of sign (and often “countdown” markers) well ahead of it, and the same applies to sharp bends etc. It is hard to argue that such signs, designed very consistently and backed up by much research into psychology and perception, reduce road safety.

77. Roads are very different to tram tracks though, and tram drivers are supposed to have detailed route knowledge, unlike car/lorry drivers using a dual carriageway with a roundabout.

78. You will probably know by now that tram services were fully resumed yesterday.

    There are additional speed restrictions on the approaches to Sandilands junction and repositioning of the 20kph warning sign further back from the curve where the accident occurred. The former 80kph limit up to the 20kph restriction has been reduced to 60kph.

    I have also been informed that, apparently, there are three other locations where speed restrictions have been applied.

79. @ IanJ
    “Probably at a considerably higher number of signs than the number in place on Tramlink before the accident.”

    I would agree that a few more signs might improve safety, but we know from the airline industry, in many accident reports it has been stated that too many warnings cause confusion. Which warning is the most important one? At what point do more warnings become meaningless? At what point do they become a distraction?

    The same is also found in my profession, chemistry, certain warnings on quite harmless things have caused a popular idea that they are dangerous. This has helped mask far greater dangers that no one discusses or warns about. A woman died in the US a couple of years ago because of this.

    Too often these days we see these knee jerk reactions, that something must be done, new signs, new restrictions… But does this work? do we really reduce risk by moving a sign, or do we transfer it elsewhere?

    What is the risk of travelling by tramlink, or trams in general, I would suspect very low. Will these new restrictions inadvertently reduce capacity, and causing overcrowded trams or worse people to use less safe methods of transport?

80. @TJ, that’s a pretty fatalist issue: If anything is done it might conceivably possibly have an undesired effect; therefore every proposal to change anything should be shut down immediately.

    You can dismiss any proposal to improve anything anywhere as a “knee-jerk reaction” — but that is in itself a knee-jerk reaction.

    If you want to argue that making the particular curve this accident happened in easier to see for drivers in darkness and bad weather would be a bad thing, then by all means do that. But please don’t just trot out the “doing anything ie bad in principle” argument.

81. Looking at the video that Mark Townend posted on 16 Nov. at 23:58 it seems to me that the gaps between the tunnels is quite short. I wonder how visible these gaps are when it is dark. Someone suggested earlier that the driver may have become disorientated.

82. I live inside the area which was cordoned after the accident so understandably I take a close interest in this.

    It seems to me that on the open sections of track neither the usual safety regime one would see on trains, nor that usual on highways, applies. That had to be wrong. If trams on those sections are regarded as trains, then they should be required to fit the safety systems used on heavy rail. If instead they are treated as road vehicles then there should be much more signage and should be trackway markings underneath the tram. At present trams on such sections seem to be less stringently controlled than either road or rail.

    That begs the question. If nobody seems certain which safety regime applies, how can the inherent safety of the design of the diamond have been properly assessed? And how can speed limits be set appropriately? If the argument is that trams are a special case, neither road nor rail, then the vast experience gained from those networks is not applicable and one would then expect much more conservative junction design and speed limits to reflect that lack of knowledge.

83. The location is a busy junction. I assume it will be in the report but when I first saw the scene there was a tram in the opposite platform at Sandilands. (It might have pulled in later to evacuate of course.) The incident was awful but it is lucky it wasn’t much worse with another tram on the junction, either cleared from Sandilands to Addiscombe or, had the signals been against the tram which crashed, a tram running between Addiscombe and Sandilands. So far the comments seem to relate mostly to the open stretch running into a curve, but we need to remember that the tram might not have needed just to slow to 20km/h but thereafter stopped entirely.

84. Is anyone familiar enough with the vehicles to know if they have a dead man’s handle?

    Given (one) line of enquiry being temporary incapacitation, I’m surprised to have come across no mention either way.

    Though of course we know dead man’s handles are not infallible (Moorgate, and non-injuring incidents), one would expect it to reduce risk.

85. @ Londoner – 19 November 2016 at 16:30
    Is anyone familiar enough with the vehicles to know if they have a dead man’s handle?

    From what I recall of the assorted news reports and footage, their dead man’s handles are each fitted with a button on the lever which requires positive thumb pressure.

86. The question then is whether the system is set up so that if the driver misses activation of the dead man’s handle and the tram was at regulation speed at the last check (not the check which is missed), will the tram then stop in time.

87. Kate says “If nobody seems certain which safety regime applies”

    There may be some uncertainty here among people commenting. But the legal position is quite clear. Trams are neither trains or road vehicles, they are trams, and they have their own specific safety arrangements.

    However, it is of course perfectly valid to make comparisons between the different modes. But when doing so, the comments should take into account the inherent differences. The one which, it seems to me, is sometimes neglected when comparing trams with road vehicles is that vehicles on reserved tram tracks are all driven by professional drivers with very detailed knowledge of the route. So the rather limited amount of signage (compared to roads) is understandable.

    I am not saying that the signage should not be improved. Improvements might be appropriate – not just here, of course, but all over all tram networks world-wide. But the case – if there is one – for improving the signage has to be made specifically and rigorously, not by just adopting road standards willy-nilly.

88. Given recent comments (not here) about a driver “Nodding off” …
    What are the roster / work schedules for Tramlink drivers?
    I’m reminded of mid-Victorian accidents, where it transpired that the parties involved had been working for far too long without a break ….

89. Responding, tangentially, to Greg’s query, I would be very surprised if the sort of shifts reported to have applied in mid-Victorian times are even remotely approached today. All transport workers are tightly regulated.

    But there are of course other reasons for “nodding off” apart from 18 hour shifts. Many transport workers report that their shift patterns, while compliant with all the rules, still disrupt their body clocks to an extent that they find unreasonable. Early starts or late finishes, very common throughout the industry, are onerous. Mixed patterns, with frequent changes of start time, are notorious. Sleep apnoea has also turned out to be a frequent factor in accidents involving professional drivers. And there are many other issues.

90. @ Londoner – there have been reports in the media that the Croydon trams have a safety system whereby if the driver releases the “handle” the brakes will be applied. I’ve driven a tram on reserved track in Oslo and can confirm the same thing applied there. I have also driven almost all of the Tyne and Wear Metro system (during the test phases) and, again, you have to keep the handle pressed down or else brakes activate. T&W Metro trains are a derivative of an old German light rail vehicle.

    Much as I would like to comment further on remarks above I don’t think it would take the debate forward. We need the official report to be published which will identify what happened. To be frank I don’t think endless speculation is getting us very far.

91. I think we should steer well clear of speculating why the driver didn’t brake at the point where he was supposed to have done so, lest we find ourselves on legal thin ice……for now, all we can do is state the one clear fact (the tram was travelling too fast at the point where it negotiated the curve), and restrict ourselves to discussing everything aside from what the driver was (or wasn’t) doing at the time.

92. This is definitely a completely trivial point, but you describe the incident as “one of the most serious rail accidents in London since the Moorgate accident” (as have others), and it just seems a bit odd and arbitrary to me to specifically single out Moorgate as one of many deadly accidents in London’s history, rather than the more definite “most serious rail accident in London since Ladbroke Grove”.

    I’m sure the author isn’t trying to draw a specific parallel with Moorgate, it just struck me as strange that the example used wasn’t the most recent (and therefore the one the highest proportion of readers will have been around to see happening in the news).

93. @Anon
    Moorgate may be being cited because it was on London Transport – the more recent examples were both on National Rail.

    http://www.bbc.co.uk/news/uk-england-london-38040186
    Not the same driver, be it noted.

94. I think everyone should refrain from the phrase “nodding off” to describe a partial loss of consciousness while driving. This could equally be described as passing out or fainting – without the judgemental overtones. It’s hardly as if a driver fancies a nap at the controls. There are plenty of potential reasons for such behaviour apart from feeling a bit tired.

    I am sure that this will be properly investigated. It just requires patience.

95. Man of Kent: with regard to your commet above about ROGs I beleive Tramways are excluded from ROGs – the ROGs guidance on the ORR website has a link to a document that lists ‘railways’ in terms of the ROGs definition into several classes, Mainline Railways, Heritage Railways and Tramways amongst them. The Croydon Tramlink System is listed in the ‘Tramway’ part though there is a rather confusing general statement about exemptions.

    As I result I don’t think the system would have had or required an ICP. In any case a major part of the duties of the ICP are to confirm that the system confirms to the relevant regulations / specifications .

96. With regard to my comment above I now think I am wrong – some parts of ROGs do appear to apply to tramways but not all parts – there is a table in the ORR guidance document that shows what is applicable.

    From the Tramlink act it appears that the system complies with most of the 1870 tramways act though some amendments are included in the Tramlink act. There are also some exclusions for a couple of railway acts. The 1870 tramways act does not appear to have any specific safety requirements. The Tramlink act allows the secretary of state to make provisions about the safety of the Tramlink System.

    It appears an ICP would be needed for the introduction of new vehicles but the situation for the original construction would depend on the dates of the construction of Croydon Tramlink, ROGs and the predecessor acts.

97. We have spoken with some Tramlink regulars and they agree with our own impression that, since the incident, across the network trams are now being driven less aggressively and are braking sooner. Draw what conclusions you will…

98. I think people need to stop overreacting.

    The last time I remember a tram overturning on a curve anywhere in Europe was in 1993 in Poznan, Poland – where a driver thought the points were set to ‘straight ahead’ instead of ‘right turn’ and ploughed into a sharp right curve at 40kph, toppling the tram onto its side and smashing it into a house, roof-first. Nine fatalities if I remember correctly:
    http://poznan.wyborcza.pl/poznan/1,36001,14566946,_Mama_kazala_jechac_tramwajem__ale_poszedlem_na_deser_.html

    That’s 25 years between such serious accidents involving overspeeding on corners across a continent with over 100 active tram networks in varying states of repair. Is it really necessary to implement a system preventing such a rare type of accident? You may argue that the lives of those tram passengers would be worth it. I’d counter-argue, that if such a system was mandated, those people wouldn’t have been tram passengers, but would have spent their lives commuting on far more dangerous buses instead. With the additional cost of ATP the Tramlink scheme (along with all other light rail projects in the UK) would have been dead in the water.

    The high death/injury toll stems from the fact, that trams are just not built with falling over in mind. That’s because they pretty much never ever do. They are typically one of the heaviest vehicles on the road, and despite some equipment located on the roof of the vehicles, their centre of gravity is still comfortably low.

    I genuinely hope RAIB maintains a common-sense approach to safety and does not overreact, as the press – and indeed many writing here – have. A reminder to slow down with bright flashing lights is all that’s needed.

99. @Straphan

    Looking at the damage to the Polish tram, we can at least be grateful that the tram in Croydon appears to have had rather greater structural integrity.

100. @timbeau: indeed, although remember the one in Poland first toppled onto its side and then smashed into the house on the corner. In Croydon the tram (thankfully) had a clear run once it overturned.

101. Also, remember the newspaper pics in the article I linked to show the tram only after the fire brigade had finished cutting everyone free from the wreckage. This being Poland in the early 1990s, I think the fire brigade deployed their hydraulic-powered shears first, and asked questions later…

102. Straphan: I’d encourage fire crews to do that at all times and in all places, though “Is the electricity off?” is a life saving prior question. Especially in conductor rail territory.

103. @Old Buccaneer: I think what I was trying to say is that the Fire Brigade was very prolific with the use of cutting equipment (particularly as the tram was quite full at the time of the crash); not that it was reckless.

     Also – apologies for my lapse in memory – the accident in question led to five deaths, not nine.

104. There is a list of tram accidents on wikipedia at https://en.wikipedia.org/wiki/List_of_tram_accidents as usual there is no guarantee it is complete or correct.

     The numbers in the UK fell drastically but then so did the number of trams in operation

     A similar accident occurred just the other side of Croydon on 1st April 1907, in this case the driver and conductor lost control of the speed of the tram on the downhill section before the left hand turn into Ruskin Road Carshalton / Wallington. (The crash was right on the borough boundary so some reports show Wallington and some Carshalton).

105. @JohnM: given that list does not contain the Polish accident I have linked, but does have accidents of a similar severity (in terms of no. of casualties) I’d say it’s probably a tad incomplete…

     Interesting reading, though. Does appear to confirm what I wrote above: no similar accident since the 1990s in Europe.

106. Strachan, there is a substantial difference in weight (mass) between a tram and even a double decker bus – and therefore in momentum. There are very good and logical grounds for saying that trams should be subject to a tighter safety regime than buses.

107. straphan 15:25

     In support of your argument is the fact that on rail it is nearly always the case that speed restrictions on bends are there for passenger comfort or, just possibly, wear and tear on the track. It is probably no exaggeration to say that in general a train or tram could be travelling at least at double speed and no cause any real danger – just discomfort. There are exceptions of course. Camden Town springs to mind.

     As implied by you, it is probably very rare indeed for a train or tram to overspeeding to the extent that it was on this occasion. Rational thought may suggest that having trams as a cheap and cheerful realistic alternative to heavy rail will save lives on the road. The odd life lost in lost of individual events does not have the same publicity impact as a lot of lives lost in one crash – even if it is the only one since Tramlink opened.

108. @PoP: Thank you. Precisely the point I was making, only – as always – far more elegantly put.

     @Kate: Mass and momentum have nowt to do with it. By your logic the mode of transport with the most safety systems should probably be shipping…

     What does matter is that trams have over the past decades proven themselves already to be a far safer mode of transport than buses or cars. When deciding on the implementation of new safety systems, there is a cross-industry consensus that the ALARP principle should be applied. Namely, that risk must be kept As Low As Reasonably Practicable.

     The cost of a blanket installation of ATP systems to all light rail networks in the UK, as well as requiring their fitment in any new networks will make them financially unviable. It will condemn people to continue to travel on a mode far less safe – buses. Or it will condemn them to much higher fares – making them more likely to abandon public transport altogether in favour of cars. I consider that to be counter-productive.

     Given the past safety performance of trams not just in the UK but across the continent suggests, that practices already applied only produce one fatal accident of this type every two decades or so – across over 100 tramway and light rail networks in around 30 countries. To me that risk is more than acceptable. I can only hope others see the light on this as well.

109. Straphan – although I mostly agree with your comment, the phrase “To me that risk is more than acceptable” could be better worded, particularly from the point of view of those who have been, directly or indirectly, affected by the recent accident. Perhaps it could have been something like “that risk is the unavoidable price which has to be paid to achieve the other benefits which widespread tram use can and does bring”.

110. @straphan & timbeau – Looking at the photos in that Polish article, one is reminded that the tram in the accident was likely built during the communist era when all was not what it seemed, with flimsy bodywork forming part of that (until unintentionally revealed, that is). However, I am also reminded of those photos often appearing in the press over here of our double deck buses with rooves peeled off like the top of a tin of sardines because of an encounter with low-hanging bridges and tree branches. A recent London example here (Kingsway):

     http://ichef-1.bbci.co.uk/news/640/media/images/80735000/jpg/_80735521_80735520.jpg

     All there is down Kingsway are signs warning of “low-hanging trees” but the buses ought to be able to get through Kingsway, n’est-ce-pas?

     In any case, I suggest that a Croydon tram has greater structural integrity than that to be found on the top decks of our buses.

111. @Graham F
     “the buses ought to be able to get through Kingsway, n’est-ce-pas?”

     They can, but drivers are warned not to get too close to the kerb when there are trees about.

     Modern tram roofs contain a lot of electrical equipment – necessary because of the low floor configuration. The roof needs to be structurally strong enough to support this gubbins. Many modern single deckers, especially hybrids and electrics, also keep equipment on top. This is not an option with a double decker, which keeps all the oily bits tucked away at the back of the lower deck, and very lightweight construction for the roof, in order to keep the centre of gravity low.
     Even allowing for the fact that they are largely full of fresh air, the tare weight of a double decker bus is not much more than a single decker – although the 7 tons of a Routemaster is rarely achieved these days (a full Routemaster weighs the same as an empty Borismaster)

112. @Malcolm: you’re right, and apologies if I caused any offence. Trying to counter irrational fear with statistics does tend to make one sound insensitive.

     The point does remain, however, that currently the risk of a passenger dying in a tram accident is far, far lower than the risk of dying when using other modes of transport which many of us are happy to use on a regular basis. If you are looking to prevent significant numbers of deaths and injuries, I once again recommend you look at other modes first and try not to make the construction of routes for what is already the safest mode of on-street transport unaffordable.

     @Graham Feakins: bear in mind the photos were taken once the rescue operation had been concluded and the fire brigade had done a thorough makeover of the vehicle using hydraulic shears and circular saws to cut people free. Also, the tram crashed into a building with its roof when it was doing almost 40kph/25mph. Having said that, the Konstal 105N family of vehicles is generally not the sturdiest or safest of designs as far as trams go.

     Modern trams are generally built to be more sturdy due to more of their equipment migrating onto the roof. They will also have significant crash protection built into their sides and into each driver’s cabin. This makes sense, as the vast majority of tram accidents are frontal collisions (either with road vehicles or other trams); or ‘side-swipes’ (where road vehicles – especially large ones – crash into or swipe against the side of the tram). I don’t think much thought is given in to the vehicles rolling over – purely because it is such a rare occurrence.

     Thing is, though, that unless you increase the strength of the windows to bulletproof (so that people do not fall out under the tram as it slides on its side); and unless you introduce safety belts for all passengers seated and standing (so that they do not hit their heads against the ceiling) I’m not sure there is much else you can do to prevent deaths in these situations.

113. Kate and Straphan,

     Regarding mass and momentum …

     You could look at it in the way my petrolhead friend looks at crashes. What matters is not the speed of the crash. It is the rapidity at which your vital organs come to a halt that kills you.

114. “What does matter is that trams have over the past decades proven themselves already to be a far safer mode of transport than buses or cars. ”

     What is your evidence for that? Certainly I would counter that in Croydon buses have proved safer than trams over the past decade. A quick Google for Nottingham gives the impression that there too buses are safer than trams, certainly so over the last 12 minths. So what is the basis for your claim? To me it looks totally wrong.

115. PoP – you are missing several points though.

     1. It is not just passengers on the vehicle who have to be considered. We also have to consider impacts with pedestrians and other vehicles (or trams) and there momentum is the major factor.

     2. Momentum – particularly conservation of momentum – matters when it comes to making a tight bend – the heavier a vehicle is the harder it is to take a bend at speed and, in the case of a tram, the more mass a tram has the more likely derailment is for a given speed and radius of curvature. I would guess momentum matters when it comes to the risk of derailment at points too.

     3. Momentum affects how much structural damage a vehicle or tram suffers on impact, either a head on impact or how much the sides or roof will be torn up if they come into contact with the ground.

     So I believe that momentum is a massive (sorry, could not resist the pun) risk factor. I see no reason why a tram should be exempted, for example, from the safety requirements which apply to a Pacer train for example in terms of signalling, driver knowledge etc. The two run at a similar speed on rails and for open stretches of tram track it seems inarguable to me that they should operate under identical safety rules. Obviously trams when running on street, in town centres, are different.

116. @Kate
     While your request for evidence regarding the safety of trams vs buses is a fair point, there is a question around what does ‘safe’ mean? Do we measure fatalities? Major injuries (if so, what counts as major)? Do we count only people on the tram/bus in the total, or third parties as well? What about accidents that the driver of the bus/tram could not avoid due to third parties’ erratic behaviour? Number of collisions regardless of damage?

     I would also suspect that bus companies can get away with not reporting incidents far more easily than tram operators can: the only time I’ve had a serious injury after being hit by a bus (8x broken bones) the bus company in question stated that their drivers were highly trained, therefore couldn’t possibly have been driving on the pavement. The other times I’ve been hit by buses (twice on the pavement, once on a zebra crossing), the bus hasn’t even stopped. I’ve never been hit by a tram, but would imagine that the RAIB would become involved in any collision where there’s a tram involved, so it would be much more difficult to just brush it under the carpet.

117. @kate
     “I see no reason why a tram should be exempted, for example, from the safety requirements which apply to a Pacer train ”
     Unlike a tram, which is unlikely to have a collision with a closing speed much above 80mph, a Pacer has the potential to be involved in a collision with an IC125, at a closing speed of 200mph, involving more than six times as much kinetic energy.

     ” in Croydon buses have proved safer than trams over the past decade. A quick Google for Nottingham gives the impression that there too buses are safer than trams, certainly so over the last 12 months. ”
     It does of course depend on the timescale and geographical data you select. What if you take accident statistics for Croydon for the years 1999-2015? Or include not just Nottingham and Croydon, but all tram and bus mileage in the UK.
     Indeed, taking the incident in Nottingham last July alone, one would deduce that it is safer to be in a tram than in an ambulance
     http://www.nottinghampost.com/three-people-taken-to-hospital-after-tram-and-ambulance-collide-in-nottingham/story-29537674-detail/story.html

     Relying on news reports or official statistics can be unreliable anyway, as bus accidents are less likely to be reported – as Moosealot’s experiences would seem to illustrate.

118. I suspect the bus/train health stats are also distorted by air quality issues. Buses pump out fumes that shorten lives on the street, tram electricity generates pollution elsewhere. I wonder if air quality life reduction dominates over impact effects.

119. @Kate 13:34

     1. Impulse rather than momentum: the violence with which the other party’s velocity is changed. Let’s say a 1500kg car hits a 75kg pedestrian at 30mph, and the pedestrian is stationary; the collision is going to be pretty inelastic, so afterwards the combined pedestrian+car will be travelling at 28.5mph. Repeat with a 15000kg bus and the combined pedestrian+bus will be travelling at 29.9mph. Repeat with a 45-tonne tram and the combination will be travelling at 30.0mph after collision. The difference in impulse felt by the pedestrian is barely noticeable between a car and a bus and not noticeable at all between the bus and the tram. In fact, the material that the front of the vehicle is made of — and its ability to deform under impact, especially at the height of essential organs — is of far greater consequence to a pedestrian collision than the mass of the vehicle.

     2. Velocity rather than momentum is important here. The lateral force applied to the tram’s wheels by the rails is proportional to mvv/r, while the downward force of gravity resisting the toppling is proportional to mg. Keeping centre of mass and wheelbase constant, that tram would have toppled at that speed regardless of what it weighed.

     3. Actually those would be kinetic energy (mvv) not momentum (mv).

120. @timbeau
     If you count passenger deaths per km travelled, I would suspect that ambulances are the least safe method of transport going, although there are probably some externalities in there that might want to be considered…

     @John B
     Which all goes back to how do we measure safety? In terms of total expected years of quality life for the average member of the population nationwide, the most cost-effective way to make trams safer might be to replace coal-fired power stations with combined-cycle gas turbines…

     Although I don’t think this will appear as a recommendation in the final RAIB report.

121. Re Moosealot,

     Completely agree, I pointed out that it was impulse that mattered last Wednesday when the discussion was around injuries due to hitting internal fittings.
     The only way to keep the tram upright is to build a suitable concrete wall parallel to the rails so close to the tram that the tram can’t topple and has to remain upright as it scraps round the corner on the concrete wall (Hopefully first contact with the wall as the tram attempts to topple at the heavily engineer ceiling level where all the equipment is). For a 36tonne Croydon tram the standard spec. deflection wall protection for new build or overhauled over bridges on roads could be used.

122. Kate, Moosealot

     With anything, but smashing lumps of concrete is a good example, the ability of something to stay intact on impact is related to its ability to absorb or dissipate energy (½mv²) – not its ability to absorb or dissipate momentum (mv).

     If you are pinned against a solid concrete wall it doesn’t matter too much whether it is done by a bus, a tram or a train. The outcome is probably the same.

     As Moosealot points out, without an obstacle in the way, you are probably going to be propelled at roughly the speed at which the vehicle is travelling. It doesn’t matter too much what the vehicle is. Your best hope for survival is that you are thrown to one side.

     A dramatic car v pedestrian collision where the pedestrian hits the bonnet and then possibly the top of the roof is probably a relatively good scenario as a lot of the kinetic energy is absorbed fairly harmlessly in denting the bonnet and the roof although probably not good from a whiplash perspective.

123. Re PoP,

     “With anything, but smashing lumps of concrete is a good example, the ability of something to stay intact on impact is related to its ability to absorb or dissipate energy (½mv²) – not its ability to absorb or dissipate momentum (mv).”

     Exactly – Which is why the comparative breaking ability of impact tools to do that is measured in Joules (Energy)…

     Impulse is key to how the objects move post impact.

124. @Kate:

     Chart 2 on p.6 here:
     https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/533293/rrcgb-main-results-2015.pdf

     Bus statistics for 2015: 188 casualties per billion passenger km; 0.1 fatalities per billion passenger km

     http://orr.gov.uk/__data/assets/pdf_file/0007/22876/rail-safety-statistics-2015-16.pdf
     Tram statistics for 2015-16: 44 injuries;

     https://www.gov.uk/government/statistical-data-sets/lrt01-ocupancy-journeys-and-passenger-miles – table LRT0103
     Tram statistics for 2015-16: 1099.8 million passenger-km (all light rail networks excluding DLR, which isn’t really a tram)

     Hence, trams in the UK showed about 40 injuries per billion passenger-km.

     That means that trams in 2015 were approximately four times safer than buses in terms of passenger injuries.

     I’m very confident you will get similar ratios in other years and in other countries across Europe.

125. Plus – as Moosealot pointed out – there is a tendency for bus companies to file most bus accidents in the ‘these things happen’ (or ‘wasn’t us, guv’) category, whereas the moment a pedestrian walks into a tram the RAIB does six months’ worth of investigating, stopping just shy of performing a colonoscopy on the driver:

     https://assets.publishing.service.gov.uk/media/547c907040f0b602410001bb/R012006_060301_Staniforth_Road.pdf

126. @PoP
     The total KE [Kinetic energy] of the vehicle is not an issue. I’d far rather be hit by a 1,000 tonne freight train at 10mph than a 1-tonne hatchback at 40mph, because while the former has over 60 times the KE of the latter, the KE that will be transferred to me is what is required to accelerate me to near-enough the vehicle’s pre-collision speed, so will be 16 times lower.

     Even that’s not the whole story, because the energy that is transferred to me by the collision is unimportant: it’s how fast that energy is transferred that is important. If there is something relatively soft and deformable on the front of the vehicle, that will take some time to deform, thereby spreading out the duration over which the energy is transferred to me, so the rate of transfer of energy is reduced and I’m more likely to survive.

127. Moosealot,

     Totally agree. I was mainly trying to make the point that kinetic energy is the issue not momentum. It is how you dissipate that energy that matters. Hence my earlier comment about the rapidity at which your vital organs come to a halt (assuming you are in the crashed vehicle which was what was being discussed).

     If you are outside and stationary, or nearly so, then obviously the critical thing is the rate at which your organs accelerate – or how much and how quickly they deviate from being stationary that matters. So yes, far better to be hit by a freight train at 10 mph than a bus at 50 mph.

128. Kate
     ” I see no reason why a tram should be exempted, for example, from the safety requirements which apply to a Pacer train for example in terms of signalling, driver knowledge etc. The two run at a similar speed on rails and for open stretches of tram track it seems inarguable to me that they should operate under identical safety rules. ”

     I see no reason why these railway safety rules should not be applied to for example to car drivers. The roads would be so much safer for everyone if,
     1. Cars had automatic brakes to prevent drivers jumping red lights (and railway level crossings)
     2. Cars had automatic systems to prevent them travelling too close to the vehicle in front.
     3. Cars had automatic door locking to prevent exit/entry unless at a designated safe place.
     4. Drivers had regular alcohol /drug tests before driving.
     5. Car safety critical systems regularly inspected.
     6. On car systems and fixed checks to prevent speeding.
     etc.

     With these systems in place we could be saving one life per week, add in the decrease in pollution and it may be two or three. Also with these level of checks for car drivers we will also solve the problem of cars coming into contact with trams!

129. TJ, you could look at designing the safety critical systems to fail safe principles whilst you’re at it…

130. @Kate – per contra, it may be “inarguable” that a Pacer should run under tramway rules on branch lines… So much cheaper.

131. In terms of the bus – tram safety comparisons perhaps a better choice would be tram coach comparisons as most coaches are single deck though with a high centre of gravity. Also the governed speed of 60 mph on a coach is only slightly above the maximum speed on Croydon Trams. There have been quite a number of coach crashes which resulted in deaths either as the result of trying to take bends at speeds capable than the coach was capable of, driver fatigue / illness or poor vehicle maintenance. I don’t have a complete list but Dibbles Bridge in Yorkshire is one black spot with 7 killed in 1925 and 32 killed in 1975.

     There is a ratehr old sumamry paper of relative accident rates at http://www.hse.gov.uk/research/rrpdf/rr073.pdf but only covers upto 2001. Not a single tram but how many wer operating before 2001 ?

132. TJ,

     I think you will find quite a number of these on European Proposals to improve road safety but as we are leaving Europe these may not get implimented – it depends on the terms of any trade deal with Europe.

     Part of the reason for not implimenting the systems yet is that they don’t want to use location data from a US military controlled GPS system which can be withdrawn or changed at any time leaving all the vehicles in europe disabled. The plan I beleive is to use the European Galileo navigation system that is under EU control.

     The other problem is GPS is quite easy to jam so someone with a small jammer can stop all the road traffic in an area. It would be sensible to have it default to 20 or 30 mph when not located which at least would allow some movement.

133. Just to add to Straphan’s earlier provision of statistical info re accident rates it is worth saying that TfL publish bus passenger accident stats plus quarterly HSE reports and annual reports that have a range of measures / metrics in them relating to the safety performance of the modes TfL control / procure as well as employee / supplier / project / road safety performance stats. There are also a series of action plans for Bus Safety, Road Safety and Health (in the widest sense and applicable London Wide). This can be found in the “Publications and Reports” section of the TfL website. As for some of what else has been said recently in this set of comments then I’ll just roll my eyes in disbelief.

134. Graham, yes maybe it is the rules for pacers which should change.

     @straphan

     That comparison is obviously wrong. You are including bus journeys which have no comparison on trams such as certain rural routes which are on steep hills or on narrow roads. Trams go through fully pedestrianised areas: buses generally don’t. Buses suffer because Councils don’t keep roads clear of obstructions like overhanging tress.

     Your figures presumably also exclude the Croydon crash.

     I suspect a fair comparison might be somewhat close but I think buses would be shown to be safer than trams over the period. However, buses have been modernised quite a bit and I suspect the safety figures for buses have improved.

     Still, if the safety of trams is comparable to that of buses then they must be much less safe than trains so there is clearly a case for doing something for trams on rail-like sections.

135. JohnM – there is certainly a case for including coaches in the comparison.

136. @ Kate – I think you need to be very careful about your comments without the benefit of statistics and data to support them. You seem to be displaying a distinct anti tram bias in the light of this one serious incident. Tramlink has an exemplary safety record overall as do other tram systems. You can’t get away from that no matter how uncomfortable that may make you feel. I see little or no case for making trams adopt the plethora of main line railway systems, many of which will be completely incompatible with street running. Therefore if you push your argument to its maximum extent it would probably result in every tram system closing given they all have elements of street running. The alternative will be to ban sharing of all street running with other motor vehicles – can’t see that being popular or feasible. As said earlier if you end up closing tram systems you will push people on to modes which are vastly less safe. I know it is hard to be dispassionate about safety issues but we have to be like that to ensure a proper, rational analysis is applied that properly identifies and weighs the risks against the legal obligations. As others have said the real target for safety improvements and better training / monitoring should be drivers of cars, vans and lorries. However as a society we’re quite prepared to “accept” the carnage on our roads from motorised transport because we place a higher value on mobility than safety. That’s not the message you want to hear if you’re a victim or a grieving relative but that is the harsh reality of our (and almost every country’s) road transport policy.

     I would also add that if you were to believe some of what is said in some quarters about London’s bus network you’d think every bus driver was a trained killer and buses were mowing down hundreds of people a week. I’ve never experienced an accident in a bus which resulted in injuries to anyone. I consider travel by public transport to be vastly safer than using cars, motorcycling or cycling (I’ve used the latter two to commute so know what rush hour conditions are like). The only two times I’ve had to be in hospital overnight were the result of accidents while using a two wheeled mode.

137. Surely a relevant question would be

     “Where else on the segregated Tramlink network (terminals excluding) could such an overspeed have caused such a derailment?”

     Yes there are sharp corners elsewhere, but none demanding a 45/50 degree turn on a very sharp gradient, with a real risk of a collision with another service in the event of a dereailment (it being pure chance that there wasn’t a corresponding ex-Woodside service coming the other way at around the same time – or indeed pausing at the signal for the points into Sandilands station, around where the unfortunate tram ended up…)

     I have always thought that Sandilands junction was inherently unsafe for this reason. The underlying quality of the original ’90s infrastructure likewise has caused concern. I’ve never taken it from the New Addington direction but have frequently from Elmers End / Beckenham. The last time I took a tram over this section back in August, the screeching of the unit climbing the gradient was very audible!

     IMHO the ‘bridging’ infrastructure as built at the time the system opened – as with Sandilands Junction – was done on a budget, and it shows. This was evident with the section up to BJ through South Norwood Country Park where the track as initially laid – notoriously – looked more like a 50p piece…along with the clearances past the bottom of a residential garden into the tram station at BJ – tight to say the least…

     [just realised I have answered my own question – the (only) other such location being the sharp curve from Birkbeck down through the park…?]

     The only ‘bridging’ section I would hesitate to suggest was done anywhere near appropriate engineering standards was the flyover at West Croydon.

     As per TJ 16/11, I wonder if a more suitable location for merging the two routes would have been at Bingham Road – especially since the embankment which originally carried the railway was flattened to street level at vast expense!

138. @WW – well said. For Kate’s benefit, she might like to consider that we tolerate on our roads manually-steered vehicles weighing much more than trams, capable of higher speeds, and with far fewer built-in safety systems. There is no call for these to be subjected to ever more constraining safety systems or for the drivers to be subject to ever more rigorous vetting. Basically, you can’t eliminate risk, even with fully automatic systems, in life generally; it’s a question of what is acceptable. Here we enter a discussion about the value of life, and so on. Kate fails to explain – because she doesn’t provide any data – why similar risks on different modes should be valued differently.

139. Tim
     “I have always thought that Sandilands junction was inherently unsafe for this reason. ”

     While I agree that the arrangements at Sandilands are not ideal, do we have any evidence that the junction is ‘inherently’ unsafe?
     1. How many trams have been through the junction / curve without incident?
     2. How many passengers?
     3. What is the fatality / injury rate?
     Even with the latest accident the risk may still be very low, and we really should add in the other tight curves with speed restrictions into our calculations.

     It may be fair to say that the junction is ‘potentially’ unsafe but then that would be true of most of the network / bus network etc.

     The reasons for the arrangement at Sandilands were made at the time of planning and construction, this was done to avoid as much disruption as possible to the road network. As far as I remember it was considered untenable to have more than one road into Croydon from the east given over to street running and to make the street running section as short as possible.

     A junction at Bingham road would also have involved a tight curve from the New Addington direction and a greater distance of street running, and I note we have had more than a few incidents between cars / lorries and trams on the short section between Sandilands and East Croydon.

     The original infrastructure has served us well, with more money it could have been better, but money is not in unlimited supply, even if some people believe that others can pay the higher taxes to fund their favourite project, but without offering to pay anymore themselves.

140. Re the safety rules for trams vs. pacers – as has already been alluded to, compared to most other rail vehicles, trams have a vastly superior braking performance. I don’t know the exact requirements in the UK either, but I assume they don’t differ too significantly from continental practice and Germany e.g. requires 2.73 m/s² at 70 km/h (= 69 m stopping distance). I don’t know any precise figures for main line railways, but I think as a rule of thumb they are unlikely to exceed 1 m/s² by much, i.e. their emergency braking rate is comparable to a tram’s service braking rate. I’d think that goes at least some way to justifying the relaxed requirements for trams.

     In the Croydon case that would mean that after passing the service braking point at full speed, there would have been another 4 seconds in which an emergency brake application still could have brought the speed down to 20 km/h before reaching the curve.

141. TJ
     With these systems in place we could be saving one life per week
     Lets correct that, shall we – apparently, the average road death rate is 5 a day.
     So, your statement should read:
     On average, with these systems in place we should be saving 28.9 lives per week

     ( i.e one death every 10 weeks or so. )
     Which puts the whole thing into a proper context, or so I think, anyway.

142. @JanH – 1.5m /s/s was thought to be maximum acceptable figure for service applications in BR days. Anything higher tended to deck the punters. I am genuinely surprised at how high the German figure is.

143. Tim Semken,

     it being pure chance that there wasn’t a corresponding ex-Woodside service coming the other way at around the same time – or indeed pausing at the signal for the points into Sandilands station

     Not pure chance as I am pretty sure they try to timetable this not to happen. But a possibility nevertheless. Probably the only saving graces due to the time of the derailment are that the tram wasn’t busier than it was and it was the only vehicle involved.

     One could speculate as to whether a New Addington bound tram being there at the same time would have made it worse or, in fact, saved lives due to a more controlled deceleration and an increased chance of the affected tram remaining upright.

144. @Tim Senken
     “it being pure chance that there wasn’t a corresponding ex-Woodside service coming the other way at around the same time ”

     There must surely be signalling arrangements to tell the drivers which way the points are set/which westbound tram has priority.

     Christian Wolmar in today’s RAIL seems surprisingly unaware of why there is such a tight bend on a converted railway line. Maybe he has been distracted by his by-election campaign.

145. @Kate: That’s the best comparison I could come up with, which – if anything – is biased in favour of the bus. Rural areas tend to be less accident prone due to fewer vehicles and pedestrians interacting with each other, as well as fewer external factors to distract drivers.

     If you think my comparison is inadequate, you are welcome to prepare your own – I draw your attention to Walthamstow Writer’s suggestion of using TfL bus safety data. Just please do use some data rather than making blanket statements that trams are unsafe without basing it on anything else than what appears to be your gut feeling…

     @Tim Semken: Sandilands junction is perhaps located in a difficult spot, but is it really dangerous? If it was, this would not have been the first derailment/accident at this location. I also fail to see how the downhill gradient is an issue – trams will usually need to stop at Sandilands stop before the junction, so they will not really be able to gain much speed approaching the curves.

     As for the other two approaches, additional warning signs e.g. with flashing lights would not go amiss, but anything else is just overkill.

     @JanH/Graham H: The 2.73 m/s^2 figure surely refers to emergency braking. At that sort of rate even seated passengers may fall out of their seats – never mind those standing…

146. @timbeau:
     http://www.croydonadvertiser.co.uk/test-trams-make-trips-through-sandilands-as-tfl-prepare-to-reopen-tram-lines-after-fatal-derailment/story-29893886-detail/story.html

     2nd photo from top / 1st photo below video: signal protecting the junction is clearly visible.

147. Warning signs/lights at braking distance are a good idea, not only as a safety measure but to give drivers confidence where to start service braking for an upcoming speed restriction or other hazard. You don’t need them at every speed change, only on segregated reservation at significant sites where the reduction is (say) over 33%, approaching junctions, single line sections, busy traffic crossings, and where the risk is not already mitigated by a station stop immediately before the hazard.

     I don’t think it’s tenable for trams to completely reject the trend in both railways and highways for more supervision and automation of the driving process, however I fully agree the full supervision of a heavy railway style ATP system would be complete overkill. I can’t see how it would work anyway on a tramway as there’s no widespread signalling infrastructure to plug into. A simple cab warning at these new critical advance warnings would be relatively simple to engineer however and would require no significant technical ecosystem to support it . Trackside, there would be a fixed balise at each site, a compact and rugged radio transponder fitted between the rails). A fixed balise is passive, fully self contained and requiring no external power supply or processing equipment to do it’s job, which, like an RFID (radio frequency identity) tag, is to return a fixed data message when a reader on a vehicle passes near by and interrogates it. On board, a control unit would interpret the message, generating suitable warnings in the cab and making interventions if any are incorporated in the system. As a minimum the response would be a visual warning and an intrusive audible alarm the driver would need to acknowledge to silence. Further enhancement could include forced service braking to a stand (not the full tram emergency stop mode!) if the alarm was not acknowledged or if sufficient slowdown was not detected within a set time or distance. At its most sophisticated an ‘ATO like’ response could take over control of the vehicle to enforce the required speed reduction.

     Of course once there’s a balise reader and a connection to traction and braking, then trainstops using the technology at signals could be a viable next step, protecting junctions and single lines more effectively. Non of this would be ‘Full ATP’ in the main line sense because I’m not trying to use traditional signalling concepts to safely separate following trains so need no widespread infrastructure of headway signals, train detection, interlocking, failsafe 100% radio coverage, etc[*]. There’s a sophisticated moving block system for that already, called ‘line of sight’. It even works with road traffic, and might be enhanced one day with measures such as forward radar and vehicle to vehicle communications, as these technologies mature rapidly and transfer from highway applications.

     Automatic alarms at distant signals were some of the earliest cab warning signals employed on mainline railways in Europe and the USA, soon after early subways devised mechanical trainstops. The GWR developed it’s ATC system over a century ago and had all it’s main lines covered by the end of 1930s. The BR AWS system, widely rolled out from the mid 1950s after the appalling Harrow disaster, was functionally similar, triggering a cab alarm and requiring an acknowledgement without which a brake application would be made, although relied on different technical principles. On many railway administrations there was a resistance to using such cab warnings for civil speed restrictions in addition to adverse signal aspects. After a number of severe overspeed derailments in curves in UK, BR relented and started to use fixed AWS magnets on approach to the most severe speed restrictions, and portable magnets for temporary slacks. Often, the severe speed restrictions with advance warning boards and AWS also get TPWS speed traps today. I think speed traps have no place on tram systems but an advance warning system with the potential to enforce fixed speed reductions in the future could be realistic combined with ‘drive on sight’.

     [*] Plainly there is some fixed signalling and interlocking on Tramlink, but it is limited to junctions and single lines. There are also some tram signals controlled by road traffic light controllers at road intersections on shared infrastructure.

148. The reason why cars aren’t more regulated is that the capacity of each vehicle is small. In contrast a double decker bus can carry 65 passengers so public service vehicle requirements apply. A busy tram can carry well over 100 passengers so further safety standards should apply.

     I know there are those who will fight that but the public simply won’t understand why a tram running on an old heavy rail route, which looks and behaves like a train at that point, should not be required to meet the safety standards of a train. If anyone thinks they are going to win that argument in the court of public opinion, I suggest they are being far, far too optimistic.

149. Thank you Mark. I think that makes perfect sense.

150. @Kate: thankfully it is not the court of public opinion that decides on such matters. Unless, of course, we merrily continue with the wholesale dismissal of expert opinion, data and hard facts that has already done so much damage on other occasions?

     The simple answer is: advanced safety systems are not required for trams, because their light weight and braking capabilities are such, that the braking distance is shorter than the line of sight of drivers. Unlike trains, they are therefore able to slow down or stop before an obstruction without the assistance of complex signalling systems.

     @Mark Townend: what you describe is precisely the kind of safety system scope creep that needs to be nipped in the bud if the construction and operation of tram routes is to remain viable. The cost of fixed balises and vehicle-to-track transmission systems is far from trivial.

151. @straphan,

     Not costless clearly, but there would be no requirement for widespread radio coverage or continuous balises, only a single unpowered unit at the fairly small number of risk assessed locations. At its simplest the system could be limited to a simple alarm and acknowledgement button in the cab driven coupled to a control unit and balise reader, with no connection to the traction and braking systems on board at all. Built into a new tram, the cost would be fairly trivial and might incorporate some greater protection features as described. I certainly do not see it as ‘scope creep to be nipped in the bud’, rather a safety improvement idea suggested in good faith after giving some thought to practical systems engineering.

     There was much opposition to the original and expanded uses of AWS, to TPWS, to no doubt to every safety improvement ever proposed on any network. If light rail is to expend to any large extent in the future we mustn’t be blind to future trends, and I can’t see that one can make an exclusive case for trams alone to have no future improvements to driver supervision and safety when both road and rail vehicles are gaining better systems. Are tram driver somehow super human and cannot make mistakes?

152. let’s look at road vehicle solutions.

     There are several radar systems around to detect overspeed or other reasons to stop.
     http://l7.alamy.com/zooms/9993e1a7c5ef432095d07145ba57288b/speed-smiley-sid-radar-speed-indicator-device-sid-slow-down-30mph-bcy0c0.jpg
     Often with a “cross face” symbol if you are going too fast, or a smiley one if you are not (- I imagine!)
     Moire immediate reasons to stop are sometimes covered too.
     https://coe-sufs.org/wordpress/wp-content/uploads/2014/02/Initiative-37-2.png.jpg

     There are also onboard systems available on quite mainstream cars that can provide in-cab reminders of the current speed limit, either through satnav data or more commonly pattern recognition of traffic signs
     https://www.carwow.co.uk/guides/glossary/how-does-road-sign-recognition-work

     If it can be fitted to a Ford Focus, surely it can be fitted to a tram?

153. straphan: I’m not so sure about that. Public opinion is important, and some of the “damage” to which you refer can perhaps be blamed on the experts not paying enough attention to public opinion in the past. Although I think you may be right on balance, your arguments are not conclusive. For instance:

     “Inertia reels are precisely the kind of safety system scope creep that needs to be nipped in the bud if the construction and use of seat belts is to remain viable. The cost of these complex and unreliable devices is far from trivial.”

154. @Kate -the difference between cars and trams is that the contents of the cars are more likely to die, passenger mile for passenger mile, than the contents of a tram (or a bus for that matter). This is because trams are inherently safer than a car for a whole variety of reasons. Safer means that pro rata fewer peole die /are seriously injured. And how closely does the tramway have to follow the railway route before it becomes that railway?

     @Malcolm/Kate – yes, let’s all vote for safety measures – we are all experts now, facts are elitist etc etc. And I’m also voting for my surgeons and doctors – I don’t need experts when I can see that the chap next door has a smiley face and can do the job err, clearly.

155. There are some articles in RAIL magazine about the accident. One statement that they have made is that if the magnetic track brake had been applied this would have provided sufficient force to prevent the vehicle falling onto it’s side.

     Any thoughts on this ?

     Has anyone done the calculations ?

156. @ Kate – I find your line of argument and apparent desire to see the “court of public opinion” take precedence over rational analysis mind boggling. We’ll be holding seances and invoking witchcraft for transport policy making next. No matter what we think of “public opinion” it is very rarely right when it comes to things like public transport accidents. All you get is hysteria and disproportionate demands while the carnage of death and injuries from private personal transport carries on without restraint.

     Why don’t we just wait to see what the investigators find? When we have a definitive set of causal factors, identified risks and proposed solutions then by all means attack whatever failings are found and those who are responsible. I think those demanding the use of more interventionist technology may well get a rather nasty shock when RAIB report. I suspect a whole load of other things will be identified as causal factors and not the inherent design of tramways.

157. I find the excuses made to ignore implementing some additional on-board safety measures on the grounds of cost and/or complexity to the Tramlink operation at key weak points on the network to be disingenuous.

     Let us not pretend that this was an entirely unique occurrence – there were complaints by passengers about another service travelling uncomfortably quickly around the same corner the week before the crash.
     While there will be sharp focus drawn to this incident at the Sandliands location for the next while, invariably *some* complacency will creep in over the next few years of operation, as it tends to do after repetitive processes pass without incident.

     Demanding a safety system to ensure that passengers are not reliant only on a human operator and risk a similar incident at the Sandilands junction and other corners with similar sharp turns on the network, should be the bare minimum as an outcome from this tragedy.

158. @Anonymous – no, it’s not disingenuous at all to consider safety measures in relation to cost. Clearly, you could always think up further safety measures in any situation at whatever cost. I’m afraid the question then is whether they are worth it. And at that point, you consider the cost in relation to the value of the lives saved and the injuries avoided. This may sound callous but some means of ranking or prioritising spend on safety measures is unavoidable, otherwise – to take an extreme example – you can spend £100ms to save a single life.

     Whether any of the measures mentioned earlier in this thread are worth doing will depend on the assessment of risk to life that emerges from the final investigation and the cost of dealing with that.

159. @Graham H
     “This may sound callous but some means of ranking or prioritising spend on safety measures is unavoidable”

     To put it a slightly less callous-sounding way – what is the best way of spending the money available that will save (on a statistical analysis) the most lives.

160. @timbeau – that’s not quite right in the sense that the valuation of life provides a cap even if funds available are higher.

161. Graham
     So you are essentially arguing that if the sums show a safety modification costs more than expected compensation after an accident, factoring in the likelihood of an accident, then the safety modification should not be made. And, of course, arguing that trams are safe makes that comparison to fall even more against improving safety. That is the argument several African ferry operators follow which is why there have been disasters but, hey, little if any compensation will be paid so the accountants are happy.

     You are also left with the absurd situation that a stretch of track, say the Bromley North spur, could be run by trams, replacing trains, and the safety regime would be reduced.

162. @Kate
     “So you are essentially arguing that if the sums show a safety modification costs more than expected compensation after an accident, factoring in the likelihood of an accident, then the safety modification should not be made.”

     not at all – the value of a life used for these calculations is not just the compensation that would be paid to the grieving relatives.

     In practice, a cheap tramway that gets people off the roads is safer than a rail line that is unaffordable and doesn’t get built at all, however safe the latter would have been had it been built. Likewise, replacing an extremely safe but little-used rail branch by a much better-used tram line, albeit not quite as safe (whether Bromley North or Wimbledon- West Croydon) may increase overall safety if those extra people migrate from a less safe mode.

     Contariwise, there have been cases where level crossings have been closed “to make the railway safer”, but the pedestrians now have to make lengthy walks along footway-less roads and are much less safe in consequence.

163. Please note cost was the reason I suggested an intermittent warning system, applied only at the highest risk sites selected by analysis, and using passive transponders with no trackside power or other ancillary equipment requirements. It is also why I suggested that on existing rolling stock at least initially the system might be ‘indicating only’ in that no intervention to traction or braking might be provided only an increasingly strident alarm persisting in the cab until the driver acknowledges it. Even at this level of functionality it could provide a useful level of ‘positional vigilance’. With or without such an on board system, I think the most important safety measure would be advanced warning indications at braking distance, something I believe may have been implemented already according to this post on railforums:
     http://www.railforums.co.uk/showpost.php?p=2781968&postcount=738
     (Note I have seen no other source to confirm this).

164. Kate -yes, I am saying that safety expenditure is capped at what is determined by the value of life (not, by the way, the same thing as any compensation paid out,which may determine a different limit, driven by insurers’ legal costs).If there were no way of fixing an upper limit, then youwould face irresistable demands for safety spend without limit. It’s exactly the same problem/solution that is faced by, for example, NICE, or. closer to home, those seeking to justify spending on transport investment. It doesn’t mean that you don’t spend anything on improving safety,merely that it’s not as long as the proverbial string.

     It is certainly true that the value of life is very low in some countries. As part of a study for the Japanese Investment Bank, I had to look into the value of life in India. Life was valued there at less than £250 in some circumstances. I wouldn’t be surprised therefore if it was given a low value in Africa.

     The replacement of one mode by another has nothing at all to do with different values of life – these are the same in the UK for all modes. The choice of safety regime for each mode is driven by assessment of risk. So it’s perfectly logical for a tramway safety regime to be less rigorous than a railway one because the risks are different. [It’s arguable that the present level of safety in the railway industry may be above the level indicated by the value of life because it has been accumulated over several centuries for much of which risk assessment was very simplistic, if it existed at all, but that is a controversial view…].

165. Re John M,

     I did a few quick calculations – Assuming they have the standard Knorr Bremse tram units, the max speed the magnetic track brakes (based on some assumptions including # passengers, number seated vs standing etc.) could have retained it on the tracks I got was 63.4kmh and the unit was doing 70kmh so still probably going over…

     Re WW,
     “I think those demanding the use of more interventionist technology may well get a rather nasty shock when RAIB report. I suspect a whole load of other things will be identified as causal factors and not the inherent design of tramways.”

     Agreed, I suspect changes to the driver shift patterns will be near the top of the list.

166. I cannot better Mark Townend’s excellent proposal. The increasingly strident alarm will overcome the ‘familiarity breeds contempt’ argument. And his alarm will not be so costly & disruptive that it prevents more tramways being built & people having to continue to use much more accident prone roads. I hope ORR people see these posts & specially Mark’s!

167. Graham, presumably you would support the publication of the value placed on life so that the travelling public can also make an evidence based decision whether or not to travel? And so that they can vote for politicians based on what they propose for the value going forwards.

     Also, you are being disingenuous. You are averaging over the entire network – indeed I think over all trams in the country – which gives the lowest possible risk value. That though is not how things work on the roads where there are well-recognised blackspots which receive additional investment in safety because of the higher risk identified. The statistics now show that the mean time between failure at the Sandilands junction is 16 years which suggests a very high level of risk – a blackspot. And if you believe in a statistics driven approach there is no need to wait for the RAIB report before making that assessment because the report cannot change the statistics. Applying your logic, that means there should be very considerable investment at Sandilands. You cannot have it both ways. If you believe in a dispassionate approach based on statistics then, if there is an incident anywhere on the network, those statistics are likely then to justify quite considerable investment on additional safety measures at that location. And Mark Townsend has made a proposal which shows what could be done just at Sandilands. That might not be the best proposal – RAIB might come up with something better – but for the purposes of our discussion it proves that something could be done.

168. Kate – a blackspot is known as such because they are higher risk. They receive greater spending because the number of lives saved is higher so the value of those saved lives gives a greater spend. It really is that dispassionate and it has to be. That’s why professionals do the calculations not elected politicians. One incident on its own does not necessarily sway anything. If it did bin lorries would have countless extra safety features to stop drivers blacking out as they caused the deaths of a number of people in Glasgow last year.

     Graham H – I’m sure I read a few years ago that the value a signed to a saved life on the railway was something like 5 times that on the road. Otherwise even TPWS would have struggled to get through.

169. Kate: I detect a certain shift in my understanding of what you are advocating. I thought at first that you were indicating a preference for all tramways to be subjected to heavy rail safety features. That preference is very susceptible to the argument, which was advanced, that such rules would prevent new tramways being affordably built, and would thus force passengers onto less safe modes. And in discussing that proposal, “averaging over all trams in the country” would be quite appropriate.

     But you have now moved to specific targeted actions to deal with specific blackspots. All well and good, particularly if some historic “near misses” at Sandilands can be found in the records. (One accident, even a nasty one like this one, doth not a blackspot make). But not quite the same proposal as we started with.

     (Incidentally, I don’t think that a measure of “mean time between failure” has much meaning when applied to rare things like accidents. One accident at Sandilands would have given us exactly the same indication of the “dangerousness or otherwise” of Sandilands whether it occurred 16 days, 16 years, or 16 decades after the line was built). You can only determine a mean value of something when you have a plural number of items to average).

170. @Kate: Graham, presumably you would support the publication of the value placed on life so that the travelling public can also make an evidence based decision whether or not to travel? And so that they can vote for politicians based on what they propose for the value going forwards.

     Back in the 1990s the values were indeed published – from memory it was about £2 million per life saved in rail accidents, and about £200,000 per life saved in road accidents. Even with this discrepancy, it proved impossible to justify the amount of money needed to install TPWS, so a fudge was added that it was justifiable to spend a higher amount per life to avoid multiple-fatality accidents.

     As for where these values came from – my understanding is that they are based on public opinion surveys where people were asked how much they were prepared to have spent to avoid road and rail fatalities, and the figures (complete with road-rail discrepancy) were simply an average of their responses.

     The statistics now show that the mean time between failure at the Sandilands junction is 16 years

     A clear example of why sample size matters in statistics: you are drawing conclusions from a sample size of 1.

171. Kate, in particular – & everbody …
     Let’s put this into context
     Last year 11 people died, 287 were left with serious injuries and 4,647 with slight injuries [In hit and run incidents in London]. And that is not counting “other” road accidents.
     At which point one wonders if spending any money at all on “extra safety” for the trams is worth it [ Rhetorical question, actually ]

     I suspect that a “flashing light” or siren overspeed warning, plus ngh’s suggestion of shift-pattern changes, plus, perhaps turning off the lights in the tunnels (flicker effect) should be sufficient.
     Wait & see.

172. Re Ian J,

     “The statistics now show that the mean time between failure at the Sandilands junction is 16 years”

     A clear example of why sample size matters in statistics: you are drawing conclusions from a sample size of 1.

     Exactly – very poor attempted use of statistics. The actual rate could be 1 in 1000 years but the accident has occurred in year 16. There are lots of very advanced stats textbooks that cover how you actually deal with it based on the work of Simeon Poisson and later Ladislaus Bortkiewicz* (the later investigating the number of soldiers in the Prussian army killed accidentally by horse kicks)

     * Author of The Law of Small Numbers

     —-

     Re my calculations above, the Bombardier tram design involved in the accident had an older design of magnetic brake with lower holding force than the current specification (which I used to do the calculations) so recalculation suggests maximum non-toppling speed on that curve with them applied to be just 37.6kmh (Assuming no cant or cant deficiency due to the near by points and crossing)

173. @Kate -as IanJ remarks, such figures are easily available in the public domain.and you would certainly use such averages when planning a system from scratch and choosing what sort of safety regime to use, but – as IanJ implies – railways (and tramways) have been accumulating statutory safety regimes since at least 1844, well before the concept of vfm was invented, whereas roads have been thought to need only the sketchiest of safety regimes until very much more recently. (Even the Rule of the Road seems to have been formalised only in the C19, although the practice of keeping left, or as it may be, right, seems to have been very much older; to judge from C18 vedute paintings such as those by Canaletto, keeping left seems to have been the preference but hardly consistent* – there are plenty of records of incidents between rival noblemen’s bravos about whose coach should have preference. Not much of a system, then, and I dare say your average bravo was considered to have a low value of life… )

     One of the consequences of running a safety system over a long time, as with any system really,is that you accumulate patches to fix specific problems without redesigning the system from scratch. In the case of railways and tramways, a lot of those patches have been location specific and designed on a pragmatic and, until the last decade or three, unappraised (at least in vfm terms) basis.

     I suspect that the origin of the actual value of life numbers also has something to do with the cumulative outcome of litigation about accidental deaths, loss of earnings, and so on but I would defer to a specialist (or even a non-elected expert) in the matter.

     _______________________________________________________________________-
     * At least one mediaeval Pope introduced a one way system for pilgrims crossing the Tiber in Jubilee years and it is thought that the Romans had a one way system for some of the Alpine crossings but – these are highly location specific and certainly unappraised in vfm terms!

174. @ngh – 🙂 (You can’t say this hasn’t been a wideranging thread, drawing in Prussian horsekick deaths and Canaletto and the case for TPWS…). Probably time for the “two bombs” joke…

175. @IanJ
     CBA for road schemes also includes an allowance for ‘pain, grief and suffering’ in the case of accidents and fatalities. By memory (and I could be wrong on the precise number but not, I think, on the scale) it was £5m for a fatality. £200,000 for a road fatality would seem very low indeed. Allowances of this sort are needed because the direct economic impact of injuries and fatalities would vary by the person. Indeed, the strict economic impact of killing an elderly pensioner is probably negative. Such an input into assessment would clearly be unacceptable so ‘pain, grief and suffering’ ensures that there is always a real economic benefit from reducing deaths and injuries.

     John Adams explains this all very elegantly in his essay ‘And how much for your Grandmother?’.

176. Alongside the value of lives and injuries, is not the cost of emergency services’ response and costs of downtime following an accident not also taken into account when assessing the worth of any safety improvements?

177. I am very glad to see common sense still prevails on this forum, especially in light of it quickly disappearing elsewhere.

     Just to play devil’s advocate: Prague, which has one of the largest and busiest (number of vehicles in service per day) tram networks in the world, has just announced it is to trial a GPS-based speed control system across its network. I suppose if there are no balises to be installed, it might be reasonably cheap to implement:

     http://www.metro-report.com/news/news-by-region/europe/single-view/view/praha-tests-automatic-tram-speed-control-software.html

178. John U.K.

     I believe it is, as is subsequent disruption (directly and maybe to the economy as a whole).

     quinlet,

     Many years ago I answered a question in my Chartered Institute of Transport exams about the true cost of accidents. I remember arguing that, if one was to do the job properly and fairly, if a pensioner with a reasonable life expectancy who was a liability to the state was knocked down and killed then we really ought to give that accident a positive value. I also argued that we ought to take into account the cost of the person to society so injuring an RAF fighter pilot with life changing injuries would cost £millions if properly accounted for since training the person is a mult-million pound cost and they would need to be replaced at the state’s expense.

     I passed so I presume they thought my answer reasonable.

     In response to the “you can’t put a price on human life” argument the fact is the NHS does this and even if all government income were spent on the NHS they would still have to make harsh choices and a financial criterion is the main one we have if we are to act rationally.

     Once you accept it is inevitable in the NHS then it is hard to argue why it shouldn’t apply to the DfT.

     We seem to have the classic case of:

     If we introduce B to replace A lives will be saved. But we know B is a flawed solution so we really ought to introduce C. However, C is unaffordable so if we insist on C then we get stuck with A. B is an imperfect solution, we all know that, but we have to be pragmatic.

     Interesting when Tramlink opened it was argued that that the injuries caused were fewer than the injuries saved by the modal shift to trams – which on opening was substantial. I suspect this is still valid when applied to deaths but 7 deaths and people injured with life changing injuries in one event his the headlines in the way that a combination of road traffic collisions doesn’t.

179. @Mark Townend 2236
     Retrofitting extra equipment to vehicles means ensuring the whole electrical system is compatible, and that there are no unintended consequences with interfacing with the existing systems (now 17 years old on the original trams). Lineside equipment which flashes up a very prominent “slow down” warning to any vehicle approaching too fast would require no mods to the vehicles themselves.

     One of these should do the trick
     http://www.autoblog.com/2013/05/15/clever-water-stop-sign-prevents-tunnel-crashes-in-australia/

180. @quinlet: you are right, I remembered too low a figure: in 1994 it was £784,000 for a road fatality. Police and property costs etc made that up to nearly a million. I’m quite sure that the rail values at the time were higher.

     Willingness-to-pay values vary a lot by country with the general and slightly paradoxical rule of thumb that the safer a country’s roads, the more people are willing to pay to eliminate fatalities.

181. @PoP: if a pensioner with a reasonable life expectancy who was a liability to the state was knocked down and killed then we really ought to give that accident a positive value

     An argument notoriously used by Philip Morris to argue that high smoking rates were a net benefit to the Czech economy. It was to get away from the awkward implications of this kind of thinking (for example, should more money be spent preventing accidents to First Class passengers because they earn more?) that the willingness-to-pay methodology was developed.

182. One point made by Anonymous yesterday @ 19:44:

     there were complaints by passengers about another service travelling uncomfortably quickly around the same corner

     This is subjective…. One mans “uncomfortably” is another man’s fun ride!

     But back to truth, if the accident hadn’t happened nothing more would have been made of those. It’s just that with the internet and our social connectedness, we’re more likely to find out about these complaints these days…

     See also this here. I’ll see if the recording of the talk is available. It’s very enlightening!

183. Anonymous and SH(LR). I would be very surprised if RAIB hasn’t downloaded all the data from the entire tram fleet’s “black boxes” and all available on- tram CCTV. This will allow them to replay virtually all journeys for the time period these things retain information (eg 14 days, but I don’t know the true value). From this, they should be able to track the level of compliance with all speed restrictions. It’s a little time consuming but can be done – my colleagues do it all the time.

184. @100andthirty: Yes, that is standard RAIB practice for every incident studied – OTMR readings are compared for a number of runs across the incident location to determine whether and how the particular run when the incident occurred differs from runs which were made without incident.

185. @JanH – 1.5m /s/s was thought to be maximum acceptable figure for service applications in BR days. Anything higher tended to deck the punters. I am genuinely surprised at how high the German figure is.

     Yes, I was referring to the requirements for the emergency brake. The service braking probably varies somewhat across different operators and vehicle models. I think in my hometown it tends towards 1.6 m/s² – and you’re right, if a driver is utilising that to the maximum, even that can feel already high enough as it is if you’re actually standing.
     So of course, normal operations will be planned around the service braking rate (and maybe not even necessarily the full 1.6 m/s²), but what I was trying to say was that the increased emergency braking capability affords an additional safety margin when compared to normal railway vehicles.

186. I was expecting the sample size argument to rear its ugly head. So essentially the argument being advanced is.

     There should only be safety improvements when statistics demonstrate a cost case for those improvements. We therefore will need to wait until there had been a statistically significant number of incidents. Of course, we will average over the national network because that gives the least incident rate.

     To me that proves that the argument that safety for trams should be based on statistics is flawed and inapplicable, which is what I have been trying to show. Instead we should rely on the figures for rail based transport as a whole and apply the safety regime for that model of transport. That is statistically reliable.

     This is a game of pick your population to prove your case. I not not see how picking a population and then arguing that the sample size for that population is too small has any credibility.

187. The current mainline railway Value of Preventing a Fatality (VPF) is £1,826,000.

     This figure is only intended for supporting ALARP decisions. If the cost of an action is less than the reduction in safety risk, using the VPF figure, then this suggests that the action is “reasonably practicable”. However, public concern about particular hazards should be considered leading to precautions that may not be justified otherwise.

188. Better than a speed trap sign might be a more strident sign for all trams at the position where they should all start braking.

     The simpler on-board version has no technical interfaces to traction/braking at all and is all independently wired. There is an increasingly strident alarm, an acknowledgement button and a visual caution reminder that extinguishes after a time. Only a little power is required, and space to accommodate the equipment, including ergonomic positioning of the lamp and switch on the desk and optimal mounting of the antenna unit beneath the vehicle. Within those limits a retrofit to the older trams could be a fairly low risk endeavour.

     Not costless though, and it might not be a good idea at any price. However, with many fewer interfaces, cab fitment should be cheaper than for mainline TPWS (itself 6x cheaper than ‘full ATP’ it was claimed at the time), and the trackside should be cheaper than TPWS too, with no requirement for trackside power or communications infrastructure. This is not full ATP by any stretch of the imagination, it is a very basic application of LS (limited supervision) techniques.

189. Kate,

     Er, no. Not as simple as that. You conduct risk assessments and use those to statistically determine the likely rate of future incidents. Past performance, I believe, is used to look at trends but not to consider future dangers.

     There was an excellent example from Crossrail when trying to mitigate delayed ATO on the lines into Paddington. RMT did its usual claptrap about safety being deliberately compromised and opposed it (which would have made the situation even worse). Chiltern thought it was great because a better form of TPWS (Train Protection and Warning System) would be introduced which would be safer for them.

     What it came down to was: how safe can you make TPWS? This is given that you can extend it to TPWS+ and even further if desired. The study concluded that they had got to a point where further increases in passenger rail safety were more than offset by the increase risk to track workers installing the signalling – even under full possession there are risks. So good news for rail workers (even if the RMT didn’t think so) but bad news for passengers.

     The point is relevant to trams for any proposal to mitigate risk by introducing further protection. If that further protection has to be maintained, at a small but probably fairly quantifiable risk, then it is essential that the benefits of improved safety to passengers outweigh the additional risk to those who are working on track. Don’t forget there are 750V around as well as trams. You can turn the power off and on again but even that involves risks including risks of misunderstanding.

190. Re Kate,

     “I can not see how picking a population and then arguing that the sample size for that population is too small has any credibility.”

     It is very credible in statistics and why lots of statistical maths has been developed for such analysis, confidence levels for 1 occurrence are usually never good.

     1 incident in 660,000+ tram journeys round the curve so far, definitely in the small numbers statistics e.g. Boltzmann rather than Gaussian statistics

191. @PoP
     “You conduct risk assessments and use those to statistically determine the likely rate of future incidents. Past performance, I believe, is used to look at trends but not to consider future dangers.”

     Past performance, and in particular single incidents, may also be used to identify previously unrecognised, or poorly understood, risks. Clapham Junction 1989, Tay Bridge 1879, Armagh 1889, Salisbury 1908, gauge corner cracking, Abermule come to mind.
     At the other extreme is the Mexico Inn saga, where a single incident caused the closure of a facility that had existed for 160 years.
     http://www.stopnetworkrail.org.uk/case-studies.html

192. @Kate: Based on your counter-proposal it is logical to invest infinite amounts of money to prevent accidents from happening in modes of transport which have had no significant accidents precisely because there is no reliable data available.

     How very logical…

193. timbeau,

     If you are going to mention Mexico Inn yet again then I will once more draw attention to the German Tank Problem yet again. This is all about how much reliability you can give to insufficient data.

194. Kate,

     Based on an earlier comment, I think you need to read The Improbability Principle or at least “Look Inside”.

195. There are some safety measures that are taken on principle rather than through risk assessment. For example, in TPWS there are rules that determine when a speed trap needs to be installed. These are not subject to risk assessment. For speed limit enforcement in TPWS, the rule is:

     TPWS shall be provided on the approach to speed restrictions where the permitted speed on the approach is 60 mph or more and the speed restriction reduces the speed by at least one-third.

     If this rule were transposed to tramways, but for speeds over, say 40mph when approaching a very significant speed restriction, I suspect there would only be one or two places affected on the Croydon tramway. Please don’t think I’m proposing TPWS, but a number of commentators have proposed comparatively simple systems that would be reasonably cheap to apply, and I can think of ready developed systems that could be adapted requiring just RFID tags on the track where the speed limit needs to be enforced and an electronic box on the tram that simply evaluates ……IF I see the RFID tag, AND the speed is below XXX AND the tram is braking, I will do nothing, OTHERWISE I will apply the brakes .

     Putting all this in the context of the likely overall “cost” of this incident which I fear will top £50m, it will be worth spending a significant sum to avoid any repetition.

196. OTMR On Train Monitoring and Recording, the railway equivalent of aviation’s ‘black box’.

197. @poP
     German tank problem – thank you for the reminder about that fascinating application of statistics.

198. Had a rare ride of Tramlink yesterday afternoon / evening (I usually cycle).
     Curve from Birkbeck to Harrington road has a new sign that implies a 15 (mph?) restriction, Tram travelled round at an estimated speed of 5 mph my walking pace!

     I note the the cant in the track appears to have been set for a higher speed.

     Speed restriction sign at Sandilands did not appear to be a new one 20 (mph?) Tram travelled around at an estimated speed of 15mph…

     Apart from a few pale faces and a group of school girls talking about the accident, things seemed to be running normally. Lots of signs up about tram safety but no one was reading them!

     I suspect that interest in the accident report by the locals will be low!

     Anyone worried about the speeds formerly run by those heavy trams around curves on Tramlink, should try the Bakerloo south of Baker Street, now that is exciting!

199. 100andthirty: The logic on your electronic box is slightly wonky, but we get the general idea.

     (2 problems – if the OTHERWISE refers back to the IF then the brakes will be applied all over the network except at the restriction, and even when that is fixed, you should not apply the brakes if the speed is already below YYY, or a tram which happened to stop and restart just before the tag will never get past it!)

200. @TJ
     The speed markers on the Tramlink are kmh, not mph. hence signage of `80` etc even between Phipps Bridge and Morden Road, denoting 50 mph in old money. So Sandilands was always 20 kilometres per hour. I explained that to a friend who wondered why the BBC were quoting 12.5 mph, as he was wondering why the half mile per hour! I did read that the limit through the Sandilands tunnels has been reduced from 80 to 60 kmh.

201. Malcolm……. yes, thank you. Will go back to the “logic classes for ancient relics”

202. @Kate: Instead we should rely on the figures for rail based transport as a whole

     Isn’t that just picking a population to make a case? If rail transport as a whole is not representative of conditions on a tram network, then you can’t meaningfully extrapolate as the sample will not be representative.

     To get a more representative larger sample, two approaches come to mind:
     (1) Expand the sample beyond the UK to light rail networks worldwide. The technologies and vehicles used in light rail networks in other countries are very similar (much more so than on the mainline railway)
     (2) Treat near misses as proxies for accidents – this has been done quite effectively in reducing the risk of signals passed at danger where all SPADs get reported and investigated whether they lead to an accident or not. Hence looking at past incidents on Tramlink could be useful.

203. DfT provides guidance for the cost of road crashes. There are links within this 2012 document http://www.gov.uk/government/uploads/system/uploads/attachment_data/file/254720/rrcgb-valuation-methodology.pdf that provide current values. Intriguingly it seems road fatalities are now valued higher than the figure for rail provided by James Forbes at 1411, at £2,005,664 (lower for local roads, higher for motorways).

     The table also gives values for serious, slight and non-injury accidents. The highway authority for which I once worked prioritised action based on both seriousness and frequency – so one serious accident may not have as high a priority as multiple slight accidents at the same location (i.e. broadly in line with PoP’s comment at 1445).

     I suspect most highway authorities have followed similar practices – the numbers killed in road crashes have declined dramatically over the last 10 years (3201 to 1775 deaths per annum, according to DfT figures), while other categories have declined, though less dramatically (albeit generally against a higher volume of traffic, and arguably improvements in vehicle design).

     The standard response to people saying “something must be done” was to use the crash stats to determine whether an intervention could be justified (usually no, leading to accusations of “someone is going to be killed”). Even when it was, if no pattern of crashes was discernible, individual incidents are difficult to design measures to deter repetition. (By the way, everyone wants slower speed limits where they live, but no restriction anywhere else they drive).

     No doubt RAIB will follow some kind of similar procedure, albeit that the incidence of rail crashes is far, far fewer than for road vehicles, and some of the statistical-based analysis will not be on a very wide base.

204. Re: “The Improbability Principle”: the gist of it is here:
     http://improbability-principle.com

     Thanks PoP.

205. Re Man of Kent,

     Another thought from the brains statistical archives*, as soon as you get a suitable number of events (25 was a good start and 40 + was ideal) from memory for massive population say 1m+) you can start to look at clustering statistics (at multiple not just 1 location say all junction) and provided you have some background on the cluster mechanisms (say accidents at T Junctions) could can easily come to far higher confidence levels as to where there is actually an inherent vs random correlation for incidents and also prioritise mitigation measures based on the quality (especially including multi-cause correlation) of clustering at each location.

     Speed Camera placement was more simplistic and didn’t use clustering stats and camera placement needing an increase in the number of accidents which probably would have decreased anyway and low and behold speed camera had a very high probability of reducing accidents most of which wasn’t due to their placement but the randomness! Hence the more recent TfL and some LBs focus on junction design and awareness (the 77% TfL adverts seen on the back of buses etc.).

     *must re-read chapter 8 of PhD thesis

206. @ngh
     the phenomenon of “regression to the mean”. Whether the outlier is at the good or poor extreme, it is likely to be closer to the average next year. And whether you put a speed camera there or not is unlikely to make much difference, as most accidents involve other factors than simply speed.

207. Re timbeau,

     Exactly, but you have to placate road safety campaigners and fund policing some how!
     The problems with focusing on speed as a cause /joint cause for a good 10-15 years was that they somewhat ignored other causal factors that cost money to sort rather than generate revenue. Things have been changing, for example Westminster increasing the length of double yellows at certain T junctions has reduced accidents due to improved visibility /space but also reduced parking revenue in addition to the cost of the changes.

208. @ngh, timbeau
     There have actually been some very detailed academic studies which show that regression to the mean is not the main reason for accident reduction at camera sites and that speed control really does reduce accident numbers and severity. While speed isn’t the only cause, by a long chalk, it is still important.

209. Driverless cars are coming to a road near you over the next several years. It will be the biggest change in road use for a century and will impact the whole safety regime. They are liable, for instance, to reduce speeding as a major factor in most places but things like roadworks could introduce whole new types of risk.

     It will be interesting to see how it affects accident rates but if, as hoped, they fall substantially then rail-based transport might then not be able claim a safety advantage over the road. That in turn could affect passengers’ expectations of safety systems on trains and trams.

210. “That in turn could affect passengers’ expectations of safety systems on trains and trams.”

     I don’t think about the relative safety when choosing my mode of transport, and I have worked on the railways as a safety critical member of staff. I would be surprised if more than a tiny minority of passengers consider safety as a major factor when choosing the method of transport.

     I suspect that cost and speed are the most important factor, followed by reliability. As for trains and trams, passengers are grateful to be able to board during the peak, with the demand outstripping the supply for the foreseeable future safety is not going to get much of a look in.

211. The biggest problem is that they are presently incapable of communicating with the living creatures that comprise the other road users, whether on foot, cycling or driving non-autonomous vehicles. This is compounded by the fact that they cannot obey the hand signals or other instructions of the police and other authorised persons.
     In real terms, I believe that they will either cause injury by asserting their right to proceed or more likely be brought to a standstill by cyclists and pedestrians who discover that they will always stop and give way to all other traffic.

212. @Nameless – hence the earlier suggestion on this site that “playful” children (and less playful criminals) will soon discover that traffic can be halted with a football or similar.

213. Graham H
     A tin can, whirled round on a piece of string …
     [ As used by “playful” persons to over-stimulate speed cameras, I am given to believe! ]
     Otherwise known as “The poor man’s corner-cube reflector”

214. As a number of commentators here have mentioned, “bus accidents are less likely to be reported” as Timbeau wrote.

     I took a look at some other posts from a blog mentioned under the Croydon Tram accident LR article, and found this excellent piece entitled “TfL and the Mayor’s Reporting on the Croydon Tram Crash is a Model for Transport Incident Reporting: Why Haven’t They Been Doing the Same for Bus Crashes?”

     http://saferoxfordstreet.blogspot.ca/2016/11/londonbuswatch-tfl-and-mayors-reporting.html

     A real eye opener of the double standard for bus and rail incidents, with a call for transparent reporting treatment and reporting of bus accidents like those for rail accidents.

     Part of the issue methinks is the societal and general culture of auto accidents being acceptable as part of the mobility boost provided by motor vehicles, and that auto accidents generally cause lower KSI (killed or seriously injured) numbers than rail or aircraft accidents. But in the aggregate, autos generate far more KSIs. Only with TfL, other cities’ transport operators, DfT and Police departments around the country will the culture of acceptable auto accidents be changed. We are already seeing a change with cycle lanes and Cycle Super Highways, as cycling advocates have rightly pushed for much safer road conditions for their disproportionately affected mode.

215. @ LBM – I’m going to be a tad controversial and say be careful in promoting a one man campaign to rubbish London’s bus network and its operation. If the man behind that blog had his way every bus would have a man with a red flag walking in front of it. I know he suffered serious injuries as a result of a bus wing mirror hitting him but I am deeply sceptical about his motives and his ability to see the wider picture. If he was really concerned about road related death and injuries he’d be directing his efforts at car drivers rather than an all out campaign against London’s buses.

216. auto = motor car

217. “The biggest problem is that they are presently incapable of communicating with the living creatures that comprise the other road users, whether on foot, cycling or driving non-autonomous vehicles. This is compounded by the fact that they cannot obey the hand signals or other instructions of the police and other authorised persons.”

     Ah, I see you are familiar with the typical Croydon driver.

218. I think the nearest equivalent to the RIAB for road traffic was the local Road Safety Committee. I wonder how many, if any, of these have survived LA cuts. They did not themselves investigate accidents/incidents. They did however look evaluate reports and make suggestions about improvements to reduce the number of accidents. There is no longer anybody to independently assess individual accidents. I think that even the police have their own agenda these days.

219. A second interim report on the Sandilands tram crash was published by RAIB:

     https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/592265/IR012017_170220_Sandilands_Jn_IR2.pdf

220. The BBC has now also run a piece on issues surrounding deadman’s handles on trams:

     http://www.bbc.co.uk/news/uk-england-london-39457148

221. The ORR are prosecuting TfL, Tram Operations and the driver for health and safety. https://www.theguardian.com/uk-news/2022/mar/24/driver-and-tfl-face-prosecution-over-croydon-tram-crash

222. This: https://www.orr.gov.uk/search-news/orr-launches-prosecution-over-sandilands-tram-crash
     is the ORR press release.

Comments are closed.