Bletchley Derailment

[tirphil]

This 'accident' appears to point to a need for more automation of train operations. Automatic systems are very good at observing speed limits (or more correctly, operating the locomotive at the correct speed). Human beings are actually not very well suited to such low level but vital tasks.

Example: sit someone in front of a panel on which there's red light that ocassionally illuminates. Provide them with a button that must be pressed each time the light comes on, but not be pressed if the light does not come on. A very simple task, but it won't be long before the human either misses a time the button should be pressed, or presses it when the light is not lit.

However, for almost no money, I can provide a bit of automatic kit (some wires and a relay would do it) which would 'press the button' with absolutly no error.

Humans are very good where descisions and judgements have to be made when the evidence isn't always 100% clear. machines are not so good at that.

Humans are poor at simple black and white tasks which a machine will never fail to get right.

ERTMS which as a system 'could' have prevented this incident (we cannot tell until the final report) has cost over £140 million to install on the Cambrian. This is hardly "for almost no money"

 

[CosherB]

Automation is only as good as the technology and programming behind it. The class 66 for example has a slow speed cruise control setting, but will usually overspeed before settling down.

So if you select it for say 5mph, you will find it overspeeds up to 7 or 8mph for some distance.

Which merely shows that's a poor bit of kit. It wouldn't be rocket science to develop one that does the job properly.

How does your "automation kit" cope with the different braking characteristics of different trains, locations and rail head conditions? Braking is not a black and white task.

With consumate ease. Control systems these days are very well understood and successfully control processes far more complex than that.
--- old post above --- --- new post below ---

ERTMS which as a system 'could' have prevented this incident (we cannot tell until the final report) has cost over £140 million to install on the Cambrian. This is hardly "for almost no money"

Oh do pay attention! I said i could do you a 'button presser when the light comes on' for almost no money, not an automatic train control system for almost no money!

 

[Geezertronic]

Which merely shows that's a poor bit of kit. It wouldn't be rocket science to develop one that does the job properly.

No offence, but the small bit of kit you're proposing wouldn't cut it in the real world. You'd need something much more robust than that

 

[tirphil]

With consumate ease. Control systems these days are very well understood and successfully control processes far more complex than that.

Unfortunately ERTMS does not cope with the different braking characteristics of different trains, locations and rail head conditions very well at all. When wheelslip occurs the system doesn't respond very well to it and rolls over resulting in total loss of signalling over the entire area covered by the controlling SCC. It is a system that needs more attention.

 

[A-driver]

Let's face it technology is responsible for as many incidents as humans are. Look at the number of signal failures, cat B spads etc. Not all safety related but cause significant disruption. Of course there are also human errors. Go back to the pilotless flight argument-how many plane crashes are down to human error? Huge numbers have been and no one seriously considers pilotless planes. Both humans and technology have a place working together in these kind of industries but it's very unlikely to go completely automated as technology has its own faults and limitations, not to mention the cost v benefits argument.

Automatic speed controls are unlikely to come about as a result of this incident alone as its a very rare type of incident indeed. They arnt gonna spend many millions on some automatic system when thousands of trains traverse low speed points every single day in the UK and even just looking over the past 12 months only 1 has been considerably overspend and derailed (I can't think off the top of my head the last time something like this happened but I would bet it was a fair few years ago at the very least). All the rest of the drivers safely reduced speed without the need for a computer helping out.

Let's see what the outcome of the investigation is and what factors influenced this incident. There are generally a great number of factors which lead up to any kind of human error. Simply pushing for automation certainly isn't the answer!

 

[CosherB]

Unfortunately ERTMS does not cope with the different braking characteristics of different trains, locations and rail head conditions very well at all. When wheelslip occurs the system doesn't respond very well to it and rolls over resulting in total loss of signalling over the entire area covered by the controlling SCC. It is a system that needs more attention.

I think you'd need something a lot more sophisticated than ERTMS to automate train ops. You'd have to have a 'clean sheet' design that incorporates the infrastructure as well as the trains. You don't automate a nuclear power station by concentrating only on the reactor; the entire station has to be one integrated control system.
--- old post above --- --- new post below ---

Let's face it technology is responsible for as many incidents as humans are. Look at the number of signal failures, cat B spads etc.

Again, don't judge the capability of today's all-encompassing sopisticated control systems by some old and unintegrtaed rail technology that's out there now, nailed onto a non-automated railway, and which regularly fails.

 

[tirphil]

I think you'd need something a lot more sophisticated than ERTMS to automate train ops. You'd have to have a 'clean sheet' design that incorporates the infrastructure as well as the trains. You don't automate a nuclear power station by concentrating only on the reactor; the entire station has to be one integrated control system.

A well made point. And other countries that have implemented ERTMS have usually gone down this route. Should have had you in charge of the ERTMS roll out Captain. Might have been less problematic!

 

[A-driver]

Absolutely right that complete automation would need a complete rebuild of the entire infrastructure which is why it will not be any more than a fantasy. The cost would be unimaginable, look how long it's taken to get a go ahead on HS2, how much protest there has been, how much it is costing, and that's just a very simple stretch from London to Birmingham. Imagen doing that throughout the entire network. There is neither a cost nor safety case for it.

 

[GB]

I think you'd need something a lot more sophisticated than ERTMS to automate train ops. You'd have to have a 'clean sheet' design that incorporates the infrastructure as well as the trains. You don't automate a nuclear power station by concentrating only on the reactor; the entire station has to be one integrated control system.

You will never get a clean sheet on the national network which is why we are where we are.

 

[the sniper]

RAIB findings have been released today.

http://www.raib.gov.uk/publications/current_investigations_register/120203_Bletchley_Junction.cfm

See you in 2015 for the final conclusions...

 

[S19]

You don't automate a nuclear power station by concentrating only on the reactor; the entire station has to be one integrated control system.

And even then you still need an abundance of human input to monitor and operate these 'automated' control systems.

 

[34D]

How does your "automation kit" cope with the different braking characteristics of different trains, locations and rail head conditions? Braking is not a black and white task.

Surely it would be something like: external input says to break from 110mph to 15Mph over course of next two miles. In that stretch there could be 200 check points which check whether the retardation is behind/ahead of target (each time) then alter the level of brake application to suit?

 

[Whistler40145]

Would a Class 90 have different braking modes for Passenger & Freight as older classes of locomotives did?

If yes, could this have had some indication of braking problems whilst running light engine if the wrong mode had been selected?


Sent from my iPhone using Tapatalk

 

[Hydro]

Would a Class 90 have different braking modes for Passenger & Freight as older classes of locomotives did?

If yes, could this have had some indication of braking problems whilst running light engine if the wrong mode had been selected?


Sent from my iPhone using Tapatalk

Doubtful, different brake timings apply to the train brake not the loco brake which would be used on a light loco.

 

[Tomnick]

We still don't know that the accident was down to Driver error, so talk of automation seems slightly premature. Unlikely I know, but if the route indicator had failed to illuminate (but the interlocking still allowed the signal to clear - as I say, unlikely!), then he'd have had no idea he was going over the crossover until it was too late. Given that the loco was brought to a stand by the approach control (has this been stated?), brake failure seems unlikely too. The RAIB hasn't ruled any of those possibilities out yet though!

 

[Whistler40145]

Could be that the route indicator didn't illuminate, therefore the driver probably wasn't expecting to switch lines, so the brakes weren't applied?


Sent from my iPhone using Tapatalk

 

[GB]

I haven't seen anywhere mention that the loco was brought to a stand by the approach control...indeed as you know, the signal's approach control/approach release doesn't affect the brakes anyway, the TPWS does.

 

[Tomnick]

I haven't seen anywhere mention that the loco was brought to a stand by the approach control...indeed as you know, the signal's approach control/approach release doesn't affect the brakes anyway, the TPWS does.

I definitely read somewhere that the loco stopped in Bletchley station (might have been for reasons other than waiting for the approach control to release), but I don't know whether that's come from a reliable source or speculation! My point was, if that's true, the brakes must have been working properly to bring the loco to a stand in the station.

TimSYoung - even if the route indicator didn't illuminate, approach control should still have checked the train right down before allowing the signal to clear. Indeed, the signal shouldn't be allowed to clear until the route indicator is proved alight. As I said, it's unlikely - but still possible.

 

[Whistler40145]

Having thought about it, if the route indicator hadn't illuminated, the driver would have stopped at the signal and contacted the signalman for instruction to proceed.


Sent from my iPhone using Tapatalk

 

[hairyhandedfool]

That would depend on if the driver expected to switch lines or not. Looking at it from the drivers pov, they are brought down to a red at a junction, the driver could see this as approach control, or maybe a train has just gone over the junction and now his path is clear or maybe the signaller forgot to clear the signal. If the route indicator lights up they are going over the junction, if not, they carry on on the same line. Unless the driver knows they can't get where they need to go on that route, or they don't sign the route, there is no reason to stop and question the signal.

As a side note, in my experience, approach control clears the route before the train stops at the signal, so if the loco came to a halt, there is no reason to believe that it was under approach control.

 

[Whistler40145]

I now fully understand.


Sent from my iPhone using Tapatalk

 

[notadriver]

If I could just ask a question of the Captain : you surely trust in the abilities and skills of the flight deck crew as a pilot yourself. Do you not trust in the skills a train driver has too?

 

[Whistler40145]

As far as I am concerned, I trust a Pilot as much as a Train Driver, even more a Train Driver as they do their job without any assistance, unless they have a Traction Inspector or Pilotman.


Sent from my iPhone using Tapatalk

 

[CosherB]

If I could just ask a question of the Captain : you surely trust in the abilities and skills of the flight deck crew as a pilot yourself. Do you not trust in the skills a train driver has too?

Modern airliners warn the crew if they are doing something wrong, and many even prevent the crew from doing something wrong! Especially stuff like exceeding speed limits with flap down, or flying so slowly that a stall is possible.

In the case of Bletchley, there was nothing to stop this locomotive exceeding the speed it should have been at.

It's not a question of trust, it's a question of accepting that humans are fallible, WILL make mistakes, and building in protection systems against that.

 

[notadriver]

I agree with you Captain. But I must ask : how many trains last year had accidents due to excessive speed caused by possible driver error? I don't know but I'd say it was very low. With the exception of ATP on high speed sections of Great Western and Chiltern, train speeds are not continuously supervised.

 

[Nym]

*coughs* ERTMS Cambrian Line, High Speed One TVM430, NXEA/GA Cl.90s and Mk3 DVTs...

 

[notadriver]

*coughs* ERTMS Cambrian Line, High Speed One TVM430, NXEA/GA Cl.90s and Mk3 DVTs...

Yep missed out Cambrian and high speed lines. NXEA cl 90s? How so?

 

[Nym]

On board monitoring equipment installed by depot engineers some time ago, covered in Rail once and then forgoten about...

 

[Tringonometry]

That would depend on if the driver expected to switch lines or not. Looking at it from the drivers pov, they are brought down to a red at a junction, the driver could see this as approach control, or maybe a train has just gone over the junction and now his path is clear or maybe the signaller forgot to clear the signal. If the route indicator lights up they are going over the junction, if not, they carry on on the same line. Unless the driver knows they can't get where they need to go on that route, or they don't sign the route, there is no reason to stop and question the signal.

Have we considered the possibility that the signal cleared to green with the route indication ...

... at the exact moment that the driver's view of the route indication was obscured by the canopy of Platform 4?

Just a thought.

 

[hairyhandedfool]

How close is the canopy to the signal?