German Train Crash

[Mike Storey]

??? Why speculate if it's best not to?

 

'cos that's what we do.........

[FelixM]

Mike, my point is that the cause of this accident certainly is not solely in a failure of technics. As you state correctly at one stage a human error must have come into play, and I appreciate your view on that a minimal probability stays.

 

I did not compare safety levels of different countries on here, that have been others commenting. I simply do not know enough about British systems. But you asked, and at one stage, I remembered this Guardian article about the creation of the Eurostar: http://www.theguardian.com/artanddesign/2016/jan/26/how-we-made-the-eurostar

Citation: "And when it came to safety, the British inspectorate said we had to show we weren’t going to kill more than one person every 15 years or something. They put it into numbers. My French colleagues couldn’t work out what the hell this UK logic was all about. Railways aren’t like roads; accidents are rare. You don’t have the same statistical base."

 

It may not be representative, make from it what you want. But I am only going to answer questions regarding German signalling from now.

 

Good night!

Felix

[david.hill64]

Mike, my point is that the cause of this accident certainly is not solely in a failure of technics. As you state correctly at one stage a human error must have come into play, and I appreciate your view on that a minimal probability stays.

 

I did not compare safety levels of different countries on here, that have been others commenting. I simply do not know enough about British systems. But you asked, and at one stage, I remembered this Guardian article about the creation of the Eurostar: http://www.theguardian.com/artanddesign/2016/jan/26/how-we-made-the-eurostar

Citation: "And when it came to safety, the British inspectorate said we had to show we weren’t going to kill more than one person every 15 years or something. They put it into numbers. My French colleagues couldn’t work out what the hell this UK logic was all about. Railways aren’t like roads; accidents are rare. You don’t have the same statistical base."

 

It may not be representative, make from it what you want. But I am only going to answer questions regarding German signalling from now.

 

Good night!

Felix

Germany, France and UK have different definitions of what level of risk is allowable. As noted above there is no such thing as absolute safety.

 

Germany has a concept of 'Minimum Endogenous Mortality'. Basically this means that the risk experienced by the rail user should be no greater than the risk of death by other causes. There is a number associated with this (I cannot remember it now) but basically it is an absolute limit. Any value higher and the system must be changed to reduce the risk. Anything below is acceptable without further review.

 

France uses GAMAB (Globalement Au Moins Aussi Bon): basically anything you do must not increase the current risk level.

 

UK uses ALARP (As Low As Reasonably Practical). Typically risks are assessed into one of four categories. Intolerable Risks must be eliminated or mitigated. This is similar to the German Limit (though the value need not be the same). Undesirable Risks also have to be eliminated or mitigated, but exceptionally may be accepted by an Operator (usually if they are time limited and the ability to eliminate or mitigate is not available). At the other end of the scale negligible risks need no further management action. Between negligible and Undesirable we have risks identified as Tolerable. These often form the majority of risk. Here there is a specific duty for management to review the risks and make sure that the risk is managed to be As Low As Reasonably Practicable (ALARP). The test for this is invariably financial: how much can we afford to spend to save a life (or a different value for multiple fatalities which are seen as less tolerable than individual risk). Here a Value For Fatality Prevented has been agreed by law and is a useful test. This was used by Railtrack to overturn BR's decision to implement nation wide ATP as an overlay to the signalling system: the costs were totally disproportional to the safety benefits.

 

This approach was seen as somewhat distasteful by other EU nations and the whole principle of using ALARP was challenged, not just for UK rail but for the whole UK approach to safety. Fortunately the Health and Safety Executive were able to mount a robust defence and the UK approach was upheld in by the European court.

 

So I am not surprised that your French colleagues could not understand it: at the time of Eurostar there were many such concerns.

 

Wrong side failures of safety critical interlocking are mercifully rare (I think the Washington metro crash a few years ago was the last one), but nothing is impossible and you cannot rule out at this stage a technical failure. I write this as somebody who is currently responsible for writing safety cases for signalling systems.

 

Far more likely is human error. Germany had a similar crash a few years ago when a driver disabled the Indusi system, resulting in a head on crash. From the information posted by Dutch Master, the culprit this time may be a signaller, but we must wait for the results of the official investigation.

 

Meanwhile my thoughts go out to the bereaved and injured.

[FelixM]

Several German media reporting now that the immediate cause was human error on the signaller. There are not yet any details on this however.

[Dunsignalling]

Several German media reporting now that the immediate cause was human error on the signaller. There are not yet any details on this however.

Unless German signalling is much more primitive than ours (not impossible), or there has been a major wrong-side failure (also not impossible), the system should not allow trains into both ends of a single-line section; whatever the signaller tries to do. The relatively old-fashioned Tokenless Block formerly used between Wilton Junction and Pinhoe certainly would not have done so.

 

Quite simply. clearing the signal at one end of such a section will lock/return that at the other to danger and, as soon as one train has entered, the signals controlling entry to both ends will be held at danger until it emerges. Signaller error would have to coincide with technical inadequacy or failure to let this collision happen. The greater probability is that one train entered the section against a red signal.

 

In this country, TPWS would quickly bring such a train to a halt (depending upon how fast it was travelling when the SPAD occurred) but again, in this respect, German signalling technology on secondary routes might not equal ours.

 

John

[meil]

Unless German signalling is much more primitive than ours (not impossible), or there has been a major wrong-side failure (also not impossible), the system should not allow trains into both ends of a single-line section; whatever the signaller tries to do. The relatively old-fashioned Tokenless Block formerly used between Wilton Junction and Pinhoe certainly would not have done so.

 

John

The Tyer No9 of 1912 would have done the same with its interlocking to the starting signal.

[APOLLO]

A tragic accident, condolences to all those involved.

 

Head on train crashes on single lines are a rare and mostly catastrophic occurrence. Usually (but not always) human error is the cause. The holes in the Swiss cheese aligning so they say.

 

The UK had a particularly bad one at Abermule (Wales) back in 1921 on the Cambrian Railway. Cause - wrong single line staff given to driver by signalman, and driver never looked at it to verify it was correct. 2 simple but deadly co-inciding acts.

 

The detailed interesting report is here http://www.railwaysarchive.co.uk/documents/BoT_Abermule1921.pdf

 

Short account http://www.railwaysarchive.co.uk/docsummary.php?docID=103

 

The world of railway signalling has moved on from Abermule in 1921 with the advent of electronics and computers. Human beings complacency however has probably not.(IF that was the major cause - yet to be determined).

 

We're all human and all make the odd mistake. Technology advances to make us less culpable (if this is the right word) to our mistakes. But, even if our machines & systems are 100% automatic, the human element (programming etc) will always be there. Nothing is 100% safe.

 

Let us hope the cause is quickly found.

 

Brit15

[Quarryscapes]

A tragic accident, condolences to all those involved.

 

Head on train crashes on single lines are a rare and mostly catastrophic occurrence. Usually (but not always) human error is the cause. The holes in the Swiss cheese aligning so they say.

 

The UK had a particularly bad one at Abermule (Wales) back in 1921 on the Cambrian Railway. Cause - wrong single line staff given to driver by signalman, and driver never looked at it to verify it was correct. 2 simple but deadly co-inciding acts.

 

The detailed interesting report is here http://www.railwaysarchive.co.uk/documents/BoT_Abermule1921.pdf

 

Short account http://www.railwaysarchive.co.uk/docsummary.php?docID=103

 

The world of railway signalling has moved on from Abermule in 1921 with the advent of electronics and computers. Human beings complacency however has probably not.(IF that was the major cause - yet to be determined).

 

We're all human and all make the odd mistake. Technology advances to make us less culpable (if this is the right word) to our mistakes. But, even if our machines & systems are 100% automatic, the human element (programming etc) will always be there. Nothing is 100% safe.

 

Let us hope the cause is quickly found.

 

Brit15

 

Although Abermule is the stand out Single Line tragedy, the one at Park Hall in 1918 shouldn't be forgotten either, Two signalmen, both clear a goods train to use a single stretch of line with the inevitable result. Investigations inconclusive, both crews have a token for the section, both have been signalled appropriately but somewhere between the two signalmen a fatal error occurred, whether in the equipment or human. 

 

http://www.railwaysarchive.co.uk/docsummary.php?docID=1482

[The Stationmaster]

A couple of things - while we do try to avoid speculation we can of course say, as some have, that 'so & so might have gone wrong' etc and I see nothing wrong in that in the way it has been put over in this thread but in reality hopefully the real answer will emerge from the investigation.  However one thing to bear in mind is that with trailable points in common use (still I presume) in Germany there is unlikely to have been any points run through.

 

That apart the immediate cause seems to come down to either human or equipment failure of some kind and we will no doubt learn in due time which it was or whether it was something else.  However one thing which does stand out in the very nasty scenes we have seen on tv is the remarkable resilience of the trains involved.  Obviously the damage to the leading vehicles is very severe but they seem to have deflected to the side of each (due to the curvature?) and the exterior state of the other vehicles looks to be pretty good considering a closing velocity which could have been as high as =125 mph.  

 

Clearly we don't know the state of the interiors in the seemingly lesser damaged vehicles or the extent to which they contributed to any injuries but the overall structural design of the trains seems to have stood up well to the impact of what is undeniably the most serious kind of collision trains could be involved in.  It is interesting to compare the consequences of this collision where - assuming the trains were both travelling at line speed - the impact velocity was =125mph with Ladbroke Grove where the estimated impact velocity was 130 mph.  There are of course some critical differences from Ladbroke Grove as it involved one train with a heavy power car colliding with a dmu whereas at Bad Aibling the trains were, or appear to be, of a similar type to each other; also at Ladbroke Grove the point of impact was on much nearer to straight track rather than the curvature at Bad Aibling so effect of the collsion would not be exactly the same.

[meil]

 

The UK had a particularly bad one at Abermule (Wales) back in 1921 on the Cambrian Railway. Cause - wrong single line staff given to driver by signalman, and driver never looked at it to verify it was correct. 2 simple but deadly co-inciding acts.

 

 

Brit15

 

Actually there were three - the third was the failure of the Tyer No6 instrument not to be interlocked with the starting signal. if any one of those three had not been present the accident would not have happened.

[Dunsignalling]

German media now report the following:

 

The signaller appears to have "disabled" the system to allow one train (which was late-running) to quickly reach a double track. (meaning, the next station) But (also apparently) he (or someone else, that wasn't clear) then cleared the train in the opposing direction to depart. A Dutch news site posted a video clip made by one passenger (who was unhurt) as he walked through the train.

 

WARNING: DISTURBING IMAGES!!!!

http://nos.nl/artikel/2085896-video-beelden-uit-verongelukte-trein-duitsland.html  (sadly I can't isolate the video clip from the page)

 

[edit: I did manage to isolate it: click (mp4) ]

In other words, he/they were able to "beat" the Interlocking. No signalling system would permit that to happen if it were designed to give safety its proper precedence over "efficiency". 

 

Overriding the Tokenless Block system referred to in my earlier post required the co-operation of two signallers acting in concert under third-party supervision.

 

Having agreed the course of action between themselves, their decision would be verified by the on-call signalling manager speaking to each of them separately. He/she would also obtain confirmation that the route had been set correctly and that all relevant signals were at danger with reminder appliances attached to their respective switches/levers.

 

The manager would then instruct ONE of the signallers to issue or dictate a written authority (Modified Block Working Ticket) to the driver of the train which was to proceed into the section by passing a nominated signal at danger but then to obey all others.

 

It was a time consuming procedure and sometimes caused delays of several minutes when it had to be done but it was about as safe as it is possible to be.

 

In the modern idiom, minutes equals money but, as my old Signalling Instructor used to say, "better twenty minutes late in this world than twenty years early in the next".

 

John

[martin_wynne]

German media now report the following:

 

The signaller appears to have "disabled" the system to allow one train (which was late-running) to quickly reach a double track.

 

This now being denied as obvious media nonsense. A "safety" system would be ludicrous if it could be turned off at will just because a train was a few minutes late. And how would that enable it to make up time anyway?

 

Martin.

[FelixM]

In the meantime while the newspapers and even the German Press Agency dpa use the term "growing evidece of" it is said everywhere that the interlocking was working properly and the signaller just guessed it was wrong. He override two safety steps to get the second train onto the line only to notice he was wrong when it already was too late.

[ejstubbs]

Human error by the signaller "rejected as speculation" by Bavarian police: http://www.bbc.co.uk/news/world-europe-35539089

 

Some of the reporting surrounding this dreadful accident is quite mind-boggling.  Take this from the BBC "What questions remain?" article linked from the one posted above:  "Trains are fast and unwieldy vehicles which cannot swerve to avoid obstacles."  The writer seems to assume that there are some people out there who may not be aware that trains run on rails...  I mean, really?

[Dunsignalling]

In the meantime while the newspapers and even the German Press Agency dpa use the term "growing evidece of" it is said everywhere that the interlocking was working properly and the signaller just guessed it was wrong. He override two safety steps to get the second train onto the line only to notice he was wrong when it already was too late.

It really should not be possible to override things to that degree without jumping through some very serious hoops. Any signalling system only remains safe if, in the event of needing to mess with it (equipment failures) proper procedures are in place for someone else to verify and authorise the actions to be taken before they are taken.

 

Apart from taking decision making out of the hands of one person, it creates vital thinking time. Not for nothing did some old stagers encourage youngsters to "stick everything to red and put the kettle on" when things went awry.

 

The problem with the modern trend to large signalling centres covering wide areas is that one person often controls both ends of a single-line section. That makes "doing it by the book" even more important than it was when you had a colleague in the next box to pick up the phone and tell you "you cannot be serious".

 

With increasing service frequencies, the really safe solution is the (re)doubling of single line sections but reality, unfortunately, often lags several decades behind need.

 

John

[Dunsignalling]

Human error by the signaller "rejected as speculation" by Bavarian police: http://www.bbc.co.uk/news/world-europe-35539089

 

Some of the reporting surrounding this dreadful accident is quite mind-boggling.  Take this from the BBC "What questions remain?" article linked from the one posted above:  "Trains are fast and unwieldy vehicles which cannot swerve to avoid obstacles."  The writer seems to assume that there are some people out there who may not be aware that trains run on rails...  I mean, really?

Yeah Really. Stand by an AHB crossing for half an hour sometime and just watch............

 

J

[34theletterbetweenB&D]

...  "Trains are fast and unwieldy vehicles which cannot swerve to avoid obstacles."  The writer seems to assume that there are some people out there who may not be aware that trains run on rails...  I mean, really?

 Never underestimate lack of awareness. I once collected evidence that convinces me to this day that most UK citizens men probably have a pre-Newtonian grasp of mechanics. They had all got to the destination where this occurred by driving, and would later pilot their tons of high velocity machinery homeward to all points of the UK and beyond. (Salesmen all  - or to further qualify successful salesmen - witty, charming, good grasp of the social niceties...)

[FelixM]

Dunsignalling, I fear you hit the nail on the head. The following diagram is from a study of the old Signalling School of University of Dresden, Saxony and evaluated how good students perform in their mind when they are forced to deal with railway incidents.
 

 

The students had been directly asked whether they are unchallenged (left, dark grey), appropriately challenged (middle, light grey) or overchallenged (right, white) when they take the role as a signaller and experience normal operation (top) or disruptions (middle). Bottom is average of both which is not of interest in this thread.

 

Don't get me wrong, this is not related to the Bad Aiblingen accident, but it does show that technical or operational problems/disruptions do exert a lot of stress on signallers. Disruptions and therefore stress as well as insufficient training of new signallers greatly enhances the probability of human errors.

 

The study where the graphic is from dates from 2005. I am currently using it for my master thesis on Berlin's Signalling School. We oversee the education of new signallers ca 3 month after their education has begun and they usually see proper interlockings with us for the first time. Luckily they only control model trains, but not just me often observe those newcomers worrying more about which button to press to get the train moving than whether it actually is safe in that situation.However I wish to make clear that with this post I do not intend to give a hint on what has happened in Bavaria yesterday.

[Reorte]

The problem with the modern trend to large signalling centres covering wide areas is that one person often controls both ends of a single-line section. That makes "doing it by the book" even more important than it was when you had a colleague in the next box to pick up the phone and tell you "you cannot be serious".

Isn't there a computer somewhere that'll tell you "you cannot be serious"?

[Colin_McLeod]

 

UK uses ALARP (As Low As Reasonably Practical). .................. Here there is a specific duty for management to review the risks and make sure that the risk is managed to be As Low As Reasonably Practicable (ALARP). The test for this is invariably financial: how much can we afford to spend to save a life (or a different value for multiple fatalities which are seen as less tolerable than individual risk). Here a Value For Fatality Prevented has been agreed by law and is a useful test. This was used by Railtrack to overturn BR's decision to implement nation wide ATP as an overlay to the signalling system: the costs were totally disproportional to the safety benefits.

.........

 

Ireland also uses "as low as reasonably practicable" where this is defined in Safety Health & Welfare at Work act 2005 Section 2(6) as follows:   I have printed the key phrase in red.

 

"(6) For the purposes of the relevant statutory provisions, “reasonably practicable”, in relation to the duties of an employer, means that an employer has exercised all due care by putting in place the necessary protective and preventive measures, having identified the hazards and assessed the risks to safety and health likely to result in accidents or injury to health at the place of work concerned and where the putting in place of any further measures is grossly disproportionate having regard to the unusual, unforeseeable and exceptional nature of any circumstance or occurrence that may result in an accident at work or injury to health at that place of work."

[TonyA]

Aircraft type "black boxes "were on board both commuter trains

 

I see in reports that there were three black boxes between the two trains, which I assume means two on one train and one on the other.

As the units appear to be a 3-car and a 6-car unit, does this mean that 6-car units always have two black boxes?

Does this imply then they are normally mounted in the second carriage rather than the leading one?

 

Tony

[Fat Controller]

I see in reports that there were three black boxes between the two trains, which I assume means two on one train and one on the other.

As the units appear to be a 3-car and a 6-car unit, does this mean that 6-car units always have two black boxes?

Does this imply then they are normally mounted in the second carriage rather than the leading one?

 

Tony

Is it possible that the two trains were formed of a single rake and a pair?

[Fat Controller]

Isn't there a computer somewhere that'll tell you "you cannot be serious"?

There probably is; the problem is they've already used 'Fatal Error Exception' for a relatively small transgression..

At work, we have audible alarms, as well as on-screen ones; if the signaller has to cancel an interlocking, perhaps on a work-site, then there is a VERY loud alarm, so there is no possibility of it going un-noticed by your colleagues.

[PhilJ W]

I see in reports that there were three black boxes between the two trains, which I assume means two on one train and one on the other.

As the units appear to be a 3-car and a 6-car unit, does this mean that 6-car units always have two black boxes?

Does this imply then they are normally mounted in the second carriage rather than the leading one?

 

Tony

 

 

Is it possible that the two trains were formed of a single rake and a pair?

The stock is made up into 3-car multiple units. Each 3 car unit has its own black box.

[roythebus]

Human error by the signaller "rejected as speculation" by Bavarian police: http://www.bbc.co.uk/news/world-europe-35539089

 

Some of the reporting surrounding this dreadful accident is quite mind-boggling.  Take this from the BBC "What questions remain?" article linked from the one posted above:  "Trains are fast and unwieldy vehicles which cannot swerve to avoid obstacles."  The writer seems to assume that there are some people out there who may not be aware that trains run on rails...  I mean, really?

Have a look at the Evening Standard video of the Croydon tram/car crash last weekend. It appears the tram swerved to avoid the car. the car actually knocked the tram off the track on a curve at Wellesley Road!