Spanish Rail Crash

[phil-b259]

Though it was my understanding that AWS (which was installed) also stops a train if it passes a danger signal (though I know TPWS also has speed limiters built in). As the driver evidently over-rode his AWS, there's no reason to think he wouldn't have done the same with TPWS (or can't TPWS be over-ridden?).

 

Wrong!

 

AWS is an advisory system which can be cancelled by the driver and when done so has absolutely no effect on the train at all, even if the train is approaching a red signal. This is one of the primary causes of the Purley smash in 1991 where a driver acknowledged the cautionary warnings received for the double yellow, single yellow and then the red with no atempt at slowing down untill he observed a train crossing over in front of him.

 

TPWS is a mandatory system in which the driver plays no part. If the driver is going above the permitted speed the brakes are immediately applied and cannot be released until the train has come to a complete stand and been stationary for 2 minutes.

 

Remember AWS was primary introduced as a updated version of the GWRs ATC system and was designed around the old mechanical signalling system to give drivers notice as to what a distant signal was showing. it was never designed with MAS signalling in mind and suffers acordingly (giving the same indications for red & yellow aspects)

[Norm81]

If you did this by train, with a sleeper from Paris to Barcelona, then you'd be both travelling past the site of the French crash at Bretigny-sur-Orge, and on TALGO stock...

 

Um, thanks for that... but was going to do a daytime run to Barcelona then stay there overnight.

 

RSSB has done some research which concluded that seatbelts on trains would be counterproductive. The UK rail industry has also produced a lot of interesting research on "secondary crashworthiness" measures to minimise damage from flying luggage, fittings detaching, or people being projected out of their seats. Recent train designs in the UK have incorporated many of the results but I don't know whether the Spanish have done anything similar.

 

Most of the coaches of the train are structurally intact, as was the 777 at San Francisco; most people survived in the plane and I think the same will be so for this part of the train. However the coaches towards the rear the train appear to have hit something very solid, probably the start of the retaining wall, and these coaches are severely damaged to the extent that one of them has lost a side and most of its seats. I think most of the casualties will be here - had there been a large concrete obstruction where the plane hit the ground then the results there would have been much more severe.

As in the case of http://en.wikipedia.org/wiki/Singapore_Airlines_Flight_006 which hit a large obstruction on the runway.

[The Stationmaster]

And therein lies the problem - in busy suburban areas the driver can be hitting the "acknowledge" button every signal, because he's running on single or double yellows all the time. It must become a reflex - and then  you acknowledge one too many....

This is another part of the 'design of the machine' problem although in this case it also extends to the design of the timetable.  The SNCF approach is perhaps a telling lesson for others, there Driver training includes a very straightforward element - sight a signal at caution and the power comes off and the brake goes in end of story irrespective of any other factors.  It is a very different approach from Britain where in many areas parsimony over infrastructure provision - as a result, often, of Govt stringency with the purse strings - means much tighter timetabling with the almost inevitable result that trains finish up running on double yellows.

 

Now there's nothing wrong with running hard on double yellows but Drivers need to be doubly alert when doing so and if it becomes an 'every day every trip' thing it can lead to the AWS auto cancelling syndrome, and there danger lurks.  Will we ever get out of I wonder - well 'defensive driving' has had some effect and probably a lot more effect where it has been taught and applied intelligently and  not just as a series of 'you must be doing X speed at location Y when approaching a signal at danger' type of nonsense (which is basically an operator covering their own back and blame shuffling instead of managing) or trying to prevent SPADs with promises of dismissal for those who err.  

[Dazzler Fan]

From the video, it was clearly the first coach that left the tracks, and pulled everything else with it.

The train was nearing the Terminus, and as is the habit, I suspect passengers were walking forward

for a fast exit when the train arrived.

If this is the situation from the inquiry, then it may follow, quite apart from the coach being lighter on

the rails than the locos, the standing passengers also added to a high centre of gravity, and the speed

added to them leaning outwards on the curve - all resulting in the coach tipping off the rails.

[melmerby]

 the standing passengers also added to a high centre of gravity, and the speed

added to them leaning outwards on the curve - all resulting in the coach tipping off the rails.

Not so sure about that.

 

The passengers were not attached to the coach in any way and as far as I can see the only load they would place is on the floor of the vehicle.

I cannot see how standing would increase the centre of gravity.

 

Could somebody verify the ins and outs of the dynamics?

 

Keith

[Ozexpatriate]

... the standing passengers also added to a high centre of gravity, and the speed

added to them leaning outwards on the curve - all resulting in the coach tipping off the rails.

Not so sure about that.

It's hard to imagine. You could just as easily argue that detraining passengers removing heavy luggage from an overhead (if that were the case here) would correspondingly lower the centre of gravity.

 

I haven't done any maths but it's hard to imagine standing versus seated passengers having a material effect on the dynamics of this accident unless the Talgo is very flimsy indeed.

[Edwin_m]

The Talgo appears to be quite light - Wikipedia quotes 315 tonnes for a 200m set.  However 50 people standing up would represent less than five tonnes being raised half a metre or so.  As I suggested further back, I think the generator car (second vehicle in from each end) would have a higher centre of gravity as its roofline is significnatly higher and it has heavy equipment above floor level where the coach is mostly open space. 

[Fat Controller]

From the video, it was clearly the first coach that left the tracks, and pulled everything else with it.

 

The train was nearing the Terminus, and as is the habit, I suspect passengers were walking forward

for a fast exit when the train arrived.

 

If this is the situation from the inquiry, then it may follow, quite apart from the coach being lighter on

the rails than the locos, the standing passengers also added to a high centre of gravity, and the speed

added to them leaning outwards on the curve - all resulting in the coach tipping off the rails.

Santiago was not the terminus, but a (busy) intermediate station. The fact that passengers were standing wouldn't have affected the centre of gravity; however, it would have probably caused the casualty figures to be higher than they would have been. If they had remained seated, their movement would have been constrained by the seat in front.

[meil]

The reason why the train tipped over was because it was going too fast. A few people standing was insignificant.

 

If the train speed was 190kph and the curve at the accident site is 400m rad then the centripetal acceleration at 190kph is 5.5 times higher than at 80kph.

 

So even with 150mm of cant the train didn't stand a chance of staying on the track. The centripetal acceleration would have been at least 60% of g.

 

The stated max for the Talgo is 12.5% of g.

 

[The Stationmaster]

The Talgo appears to be quite light - Wikipedia quotes 315 tonnes for a 200m set.  However 50 people standing up would represent less than five tonnes being raised half a metre or so.  As I suggested further back, I think the generator car (second vehicle in from each end) would have a higher centre of gravity as its roofline is significnatly higher and it has heavy equipment above floor level where the coach is mostly open space. 

And that was where the initial derailment appeared to take place - at the rear of the second vehicle.

[34theletterbetweenB&D]

...The passengers were not attached to the coach in any way and as far as I can see the only load they would place is on the floor of the vehicle.

I cannot see how standing would increase the centre of gravity.

 

Could somebody verify the ins and outs of the dynamics?

All the mass within the vehicle, whether rigidly attached or simply acting on the structure by gravitation, contribute to the effective centre of mass. But as already posted, the movements of passengers and their baggage is likely to produce a very small effect in moving the centre of mass, as the mass of the train is roughly an order of magnitude greater than that of the passengers and baggage.

 

Now in aviation it is a different story. Aircraft have been lost due to passenger movements within the cabin exceeding the flight controls ability to compensate. The last I remember reading about was on a circa 30 seat aircraft, where practically the whole passenger complement went to the rear of the cabin to observe an event on the ground as the aircraft was coming into land. Movement of the centre of mass well aft of the centre of lift induced a stall from which no recovery was possible that close to the ground.

[Dunsignalling]

The Talgo appears to be quite light - Wikipedia quotes 315 tonnes for a 200m set.  However 50 people standing up would represent less than five tonnes being raised half a metre or so.  As I suggested further back, I think the generator car (second vehicle in from each end) would have a higher centre of gravity as its roofline is significnatly higher and it has heavy equipment above floor level where the coach is mostly open space. 

The Talgo design does result in a lightweight train, but most of the weight saving is below the floor, with a notable absence of heavy bogies and associated equipment. However, the way the running gear is attached and the fact that each coach partially supports its neighbour, implies that the bodyshell itself is load-bearing and would have to be pretty rigid.

 

It also seems from photographs that Talgo coach ends are not always open (this presumably varies between different generations of stock).

 

In this case the ends do appear to have been open, or at least had apertures much larger than a typical UK vestibule.

 

Anyone being thrown around would have been free to travel a long way with ejection likely once the coaches began to disconnect. 

 

John

[Miss Prism]

I agree with Meil that a 80km/h speed is right on the limit of what a 15000kg Talgo coach is designed to cope with on a 400m curve, but although the radius of the sharpest part of that Santiago de Compostella curve is in the region of 350m (I estimate), the curve is transitioned, albeit rapidly, and the radius of the curve at the initial point of derailment seems to be far larger. Moreover, the initial improper movement of the interface between the second and third vehicles (the generator car and the first passenger coach) appears to be vertical rather than sideways.
 
An American commentator, in questioning the number of waivers the Talgo units were granted from American railway regulatory requirements, has likened the Talgo units to "a string of soda cans between a pair of bricks".

 

 

 

[meil]

I agree with Meil that a 80km/h speed is right on the limit of what a 15000kg Talgo coach is designed to cope with on a 400m curve, but although the radius of the sharpest part of that Santiago de Compostella curve is in the region of 350m (I estimate), the curve is transitioned, albeit rapidly, and the radius of the curve at the initial point of derailment seems to be far larger. Moreover, the initial improper movement of the interface between the second and third vehicles (the generator car and the first passenger coach) appears to be vertical rather than sideways.

 

An American commentator, in questioning the number of waivers the Talgo units were granted from American railway regulatory requirements, has likened the Talgo units to "a string of soda cans between a pair of bricks".

 

santiago-de-compostella-map.png

It's on the limit without any cant. If there were 150mm cant then it could just survive at 110kph.

 

 

 

The cant plus the flange depth (riding up the rail) will give it a vertical trajectory.

[Glorious NSE]

 An American commentator, in questioning the number of waivers the Talgo units were granted from American railway regulatory requirements, has likened the Talgo units to "a string of soda cans between a pair of bricks".

 

That comment needs to be put into context with a common US implementation of Talgo - where the coaches are generally sandwiched between an F59 (circa 120 tons) and a former F40PH cab car (power unit removed, but I would presume ballasted back up to circa 120 tons)

 

This spanish one was between a pair of similarly lightweight cab cars, so the analogy doesn't translate well.

 

Nothing i've seen so far suggests those coaches were more flimsy in construction than other modern European stock that's had similarly catastrophic impacts (Turbo at Ladbroke Grove, DVT at Heck etc...)

[Miss Prism]

The cant plus the flange depth (riding up the rail) will give it a vertical trajectory.

 

The cant looks to me to be less than 80mm at the sharpest radius point of the curve (which the majority of the train never reaches whilst on the rails). Whilst the curve is severely cant deficient even for its line speed, the cant gradient would appear to be generous. The line east of the Autoestrada bridge is straight-ish, so presumably at zero cant. 100m west of the bridge, which seems to be the initial discernible point of catastophe, the cant is probably no more than 20mm (??). A typical flange depth is say 40mm. The sum of those vertical displacements would be imperceptible on the security cam video. What the video shows is a gross vertical displacement developing immediately west of the bridge.

 

Something doesn't add up to me.

[dibber25]

Probably a stupid question but do the power cars/generator cars tilt? If so, do they use the same suspension system as the passenger cars? The Talgos used by Amtrak clearly have locos (including driving trailer ex-locos) which don't tilt but then the trains probably don't reach enough speed to tilt the coaches, either.

[Adrian Wintle]

 

The cant looks to me to be less than 80mm at the sharpest radius point of the curve (which the majority of the train never reaches whilst on the rails). Whilst the curve is severely cant deficient even for its line speed, the cant gradient would appear to be generous. The line east of the Autoestrada bridge is straight-ish, so presumably at zero cant. 100m west of the bridge, which seems to be the initial discernible point of catastophe, the cant is probably no more than 20mm (??). A typical flange depth is say 40mm. The sum of those vertical displacements would be imperceptible on the security cam video. What the video shows is a gross vertical displacement developing immediately west of the bridge.

 

Something doesn't add up to me.

 

 

What you could be seeing it the inside wheel dropping onto the sleepers, kicking the outside of the coach up. This would assume that the outside wheel climbed the rail - and it may still have had the flange on top of the rail as the inner set dropped off the inside rail.

 

Adrian

[Edwin_m]

From what I remember, derailments at curves can be of two forms:

 

A flange climbing derailment is when the lateral forces are enough for the flange to slide up the rails, so this depends on the ratio of horizontal to vertical force compared to the flange angle.  The train effectively goes off sideways but remains upgright (at least at the time of derailment). 

 

An overturning derailment is where the lateral forces tip the train over, pivoting at where the wheels touch the outer rail.  The train will be on its side and may roll further.  This depends on the ratio of horizontal to vertical force compared to the ratio of centre of gravity height to half the gauge (I think).  In both cases cant reduces the tendency to derail. 

 

As far as I can imagine it, the factor that determines the type of derailment must be the height of the centre of gravity above the rail - a higher centre of gravity means that overturning may happen before flange climbing.  I think the weight of the train cancels out in both cases.  So it is possible that the generator derailed by overturning and the lower-slung passenger coaches derailed by flange climbing. 

[Edwin_m]

I'm not sure why people are commenting about the lack of ends.  As in the photo posted way back up, the gangway of the Talgo is mororless between the wheels, each of which sits underneath some kind of suspension unit that goes up nearly to roof level.  This means the end walls aren't readily visible but they aren't missing, simply inset from the actual end of the vehicle to create a recess for the running gear.  In the majority of the coaches where the sides are intact the end probably is too. 

[Rugd1022]

 

Now there's nothing wrong with running hard on double yellows but Drivers need to be doubly alert when doing so and if it becomes an 'every day every trip' thing it can lead to the AWS auto cancelling syndrome, and there danger lurks.  Will we ever get out of I wonder - well 'defensive driving' has had some effect and probably a lot more effect where it has been taught and applied intelligently and  not just as a series of 'you must be doing X speed at location Y when approaching a signal at danger' type of nonsense (which is basically an operator covering their own back and blame shuffling instead of managing) or trying to prevent SPADs with promises of dismissal for those who err.  

 

I don't think we will get out of it Mike, not with the current capacity problems on our busiest mainlines, especially at the southern end of them. Defensive driving certainly does help, we have it drummed into us as often as possible, but it has the added problem of delaying whatever is behind you all the more, costing more in delay attribution penalties in the long run. I can vouch for the fact that it's no fun at all going from Willesden to Bletchley (about 40 miles) or Cricklewood to Bedford (about 44 miles) on endless double or single yellows. Do this day in, day out and you realise what fatigue is all about. Part of the defensive driving policy we have to adhere to at FLHH is doing no more than 10mph over the AWS magnet on approaching a red 'un, this can have an appalling effect on time keeping and delaying following traffic, a lot of signalmen still don't realise this (not their fault by the way), or even that the speed differs between TOCs / FOCs.

[The Stationmaster]

What you could be seeing it the inside wheel dropping onto the sleepers, kicking the outside of the coach up. This would assume that the outside wheel climbed the rail - and it may still have had the flange on top of the rail as the inner set dropped off the inside rail.

 

Adrian

Judging by the speed with which they got the line back into traffic and the seeming lack of sleeper damage/changing that doesn't seem to fit (unless they don't have much concern for concrete sleepers that have been damaged by passing wheels).  Marks at site - if checked quickly enough after the derailment will tell much of the story of the dynamics of the derailment but as yet we've heard nothing (unless the media consider it to be to geeky to report?).

[Jim]

I suspect with Ladbroke Grove the inexperienced driver of the DMU counted for an awful lot. I still don't think you can train someone off the street to dive a train within the time periods given these days. but then I'm one of the old school.

 

With the Spanish crash, surely the second driver should have warned the man in the chair he was going too fast?

 

As for speed warnings, BR were just as guilty as RENFE in having relatively high speed sections going down to 20. Higham Tunnel into Strood springs to mind, 75 down to 20 at the tunnel exit has seen more than one train likely to go for a dip in the Medway. I speak from experience when I was passing out as driver when the instructor/examiner told me "brake when you hit daylight, there's a 20. He failed to say the 20 started at the end of the tunnel!

It bristles with TPWS now!

 

Jim

[The Stationmaster]

From what I remember, derailments at curves can be of two forms:

 

A flange climbing derailment is when the lateral forces are enough for the flange to slide up the rails, so this depends on the ratio of horizontal to vertical force compared to the flange angle.  The train effectively goes off sideways but remains upgright (at least at the time of derailment). 

 

An overturning derailment is where the lateral forces tip the train over, pivoting at where the wheels touch the outer rail.  The train will be on its side and may roll further.  This depends on the ratio of horizontal to vertical force compared to the ratio of centre of gravity height to half the gauge (I think).  In both cases cant reduces the tendency to derail. 

 

As far as I can imagine it, the factor that determines the type of derailment must be the height of the centre of gravity above the rail - a higher centre of gravity means that overturning may happen before flange climbing.  I think the weight of the train cancels out in both cases.  So it is possible that the generator derailed by overturning and the lower-slung passenger coaches derailed by flange climbing. 

But it's also fairly easy to tell which was which from examining the track approaching and at the point of derailment.

[Kris]

Latest update http://www.bbc.co.uk/news/world-europe-23507348 says he was doing 153kph at the time of derailment and the driver was on the phone.