Lessons from UK Ladbroke Grove crash 25 years on, plus Odisha crash

Amtrak Unlimited Discussion Forum

Help Support Amtrak Unlimited Discussion Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
I came across this BBC report on a UK train crash, which has a separate video embedded about the Odisha Indian train crash, and rail safety worldwide. (The A.I. content in the video is a bit tedious, but overall the video is informative.)

https://www.bbc.co.uk/news/articles/cje3kd8y1j0o
I wonder, how extensive is the adoption of TPWS in the UK? I was under the impression that there are still lines that only have AWS, although they may be ones with little traffic or freight only. (TPWS is an ATC type system that will stop a train if it passes a red signal, whereas the older AWS only provides a warning that the driver must acknowledge else the brakes are applied.)
 
When I worked at B.R. over 40 years ago, AWS was the common safety standard. I have no information about more modern systems, although the video mentions a new "digital" system due in 2025. I wonder if that is going to be similar to that new Indian system called Armour?
I understand that the London Underground has a mechanical system where a lever will rise when the signal is red, and "catch" a stopcock on the train passing a red signal, and thus apply brakes automatically.
It seems that both the major accidents mentioned were caused by incorrect track and signaling repairs and maintenance as primary cause?
 
I came across this BBC report on a UK train crash, which has a separate video embedded about the Odisha Indian train crash, and rail safety worldwide. (The A.I. content in the video is a bit tedious, but overall the video is informative.)

https://www.bbc.co.uk/news/articles/cje3kd8y1j0o
I was denied access to the video due to my location allegedly 🤷🏻

As for the lessons learned from the Bahanaga Bazar accident here is a good article (over and year old article now):

https://frontline.thehindu.com/news...rain-accident-at-balasore/article67039892.ece

There was a similar accident at lower speed involving an Amtrak train in Michigan for surprisingly similar causes sometime back. In that case an Amtrak train inadvertently went into a siding at speed and derailed even though it had a clear signal on the main line. It happened because of improper wiring in a connection hut that S&C staff were working on.
 
Last edited:
The Frontline item draws the same broad conclusions as the video does. 👍
The interesting thing is that unless these human actor protocols and operating procedures are fixed no amount of PTC or Kavach will prevent these sorts of accidents since these automation systems would be perfectly happy with what are essentially inputs at variance with reality (absent independent means of verification of consistency), i.e. not reflecting reality. In automation systems, for them to work right, the real state of the system must be reflected accurately to the system so that actions that it takes based on that information, that is reified into reality are consistent with reality. If not then things like train getting a green signal on the main runs off into a loop and crashes into something will happen. Unfortunately human beings can be very resourceful in screwing up.
 
That kind of problem can be prevented with what NASA calls "dissimilar redundancy", such as using GPS to verify all trains are on the expected tracks and are traveling at the expected speeds, but this is very expensive and won't fix software problems in the computer system that is controlling everything. For that, you need separate, dissimilar systems that vote, which is even more expensive.

And if everything is working normally, none of these safety systems will ever kick in, and at some point some bean counter will ask "why are we doing this? Doesn't the system KNOW where all the trains are?", and the original designers having long since moved on, the current maintenance staff will not know why the system was built that way and will do something they think is harmless, like replace the GPS feed with a clone of the track sensor feed, which will work fine until a track sensor fails at a bad time. Meanwhile, the railroad will save lots of money by not maintaining the redundant systems :eek:

The interesting thing is that unless these human actor protocols and operating procedures are fixed no amount of PTC or Kavach will prevent these sorts of accidents since these automation systems would be perfectly happy with what are essentially inputs at variance with reality (absent independent means of verification of consistency), i.e. not reflecting reality. In automation systems, for them to work right, the real state of the system must be reflected accurately to the system so that actions that it takes based on that information, that is reified into reality are consistent with reality. If not then things like train getting a green signal on the main runs off into a loop and crashes into something will happen. Unfortunately human beings can be very resourceful in screwing up.
It is very hard to make something *****-proof because idiots are very ingenious. It is even harder to make something resistant to mistakes made by intelligent people who generally know what they are doing.

I was denied access to the video due to my location allegedly 🤷🏻

As for the lessons learned from the Bahanaga Bazar accident here is a good article (over and year old article now):

https://frontline.thehindu.com/news...rain-accident-at-balasore/article67039892.ece

There was a similar accident at lower speed involving an Amtrak train in Michigan for surprisingly similar causes sometime back. In that case an Amtrak train inadvertently went into a siding at speed and derailed even though it had a clear signal on the main line. It happened because of improper wiring in a connection hut that S&C staff were working on.
One very sad thing in the article is the last sentence: A month after the accident, more than 80 of the people killed had not yet been identified.
 
When I worked at B.R. over 40 years ago, AWS was the common safety standard. I have no information about more modern systems, although the video mentions a new "digital" system due in 2025. I wonder if that is going to be similar to that new Indian system called Armour?
I understand that the London Underground has a mechanical system where a lever will rise when the signal is red, and "catch" a stopcock on the train passing a red signal, and thus apply brakes automatically.
It seems that both the major accidents mentioned were caused by incorrect track and signaling repairs and maintenance as primary cause?
I understand the mechanical system, on the London Underground has since been replaced by a more modern electronic system. I'm not sure if all lines have switched over yet, or whether it is still work in progress. Anyway, this is stated as being one of the reasons that heritage trains cannot be operated as freely as they could in the past, including the highly popular annual "Steam on the Met" event.

And if everything is working normally, none of these safety systems will ever kick in, and at some point some bean counter will ask "why are we doing this? Doesn't the system KNOW where all the trains are?", and the original designers having long since moved on, the current maintenance staff will not know why the system was built that way and will do something they think is harmless, like replace the GPS feed with a clone of the track sensor feed, which will work fine until a track sensor fails at a bad time. Meanwhile, the railroad will save lots of money by not maintaining the redundant systems :eek:
The issue of de-skilling and absence of institutionalized memory is indeed a big problem. I have seen plenty of examples in my own career with things being thrown away or ripped out because nobody could work out what they were there for, only to result in something going haywire many years down the line because of a special situation this seemingly redundant code or sensor was installed to prevent. And this even though it was fully documented (but who bothers to check paper documentation).
 
I understand the mechanical system, on the London Underground has since been replaced by a more modern electronic system. I'm not sure if all lines have switched over yet, or whether it is still work in progress. Anyway, this is stated as being one of the reasons that heritage trains cannot be operated as freely as they could in the past, including the highly popular annual "Steam on the Met" event.
It varies depending on the line. The subsurface lines (Metropolitan, Circle, District, H&C) are getting CBTC as part of the "4 lines modernisation programme" which will eventually enable ATO on these lines. The Northern, Jubilee, Central and Victoria Lines already have ATO. The Bakerloo and Piccadilly still use the older fixed block signaling with mechanical tripcock operation as described. The Piccadilly is getting new trains which are future proofed for ATO operation.
 
One very sad thing in the article is the last sentence: A month after the accident, more than 80 of the people killed had not yet been identified.
This has always been an inherent problem of every regular Mail/Express train having four to six General Second Class unreserved cars. These cars are typically incredibly overcrowded and the identity of people in them, their origination point or destination, are usually not known, unlike in reserved coaches. For the identifiable bodies, unless someone comes forward to identify there is little guidance on what to do. Of course for bodies not in an identifiable state that is a lost cause without a priori knowledge of who was in those cars. Things have improved considerably in identification with DNA, before that it was Dental Records which was hard to come by for the typical demographic of the travelers in General Second Class (GS). Even then some genetic relative has to be found first.
 
This has always been an inherent problem of every regular Mail/Express train having four to six General Second Class unreserved cars. These cars are typically incredibly overcrowded and the identity of people in them, their origination point or destination, are usually not known, unlike in reserved coaches. For the identifiable bodies, unless someone comes forward to identify there is little guidance on what to do. Of course for bodies not in an identifiable state that is a lost cause without a priori knowledge of who was in those cars. Things have improved considerably in identification with DNA, before that it was Dental Records which was hard to come by for the typical demographic of the travelers in General Second Class (GS). Even then some genetic relative has to be found first.
Surely if a relative goes missing unexpectedly and unexplainedly at about the time of a major incident like this, families, parents, children, whatever, might put one and one together and come forward, thus at least populating the list of possible victims. It would then be for the forensics to try and match those to the actual fatalities.
 
Surely if a relative goes missing unexpectedly and unexplainedly at about the time of a major incident like this, families, parents, children, whatever, might put one and one together and come forward, thus at least populating the list of possible victims. It would then be for the forensics to try and match those to the actual fatalities.
Apparently it is not unusual to have no one come to claim anything for significant number of travelers for whatever reasons that are unknown. That is the reality notwithstanding what one would like it to be.
 
Back
Top