S- Bahn collision Munich

[duesselmartin]

S-Bahn-Unglück in Schäftlarn: Erste Erkenntnisse zur Ursache

In Schäftlarn bei München sind am Montag zwei S-Bahnen zusammengestoßen. Ein Fahrgast starb, zahlreiche Menschen wurden verletzt. Derzeit konzentrieren sich die Ermittler laut Innenminister Herrmann auf die Frage, ob menschliches Versagen vorliegt.

www.br.de

So far only a German language link.
Two S7 commuter trains collided in Schaeftlarn, south of Munich. Several injured, at least one person died. Unconfirmed reports suggest the victim is a driver.

 

[D365]

Was reading that one of the drivers is trapped, not sure if that implies the surviving driver… awful, awful news.

That section of the S7 is predominantly single track. I really don’t want to speculate, but the context sounds eerily familiar.

Worse still - it looks like two S7 trains came within 150 metres of a head-on, six months ago:

Zwei S-Bahnen sind bei Icking unweit des Starnberger Sees auf eingleisiger Strecke fast zusammengestoßen. Die Lokführer erkannten den Kollisionskurs und konnten noch rechtzeitig bremsen. Zwischen den Bahnen lagen nur noch 150 Meter.

 

[duesselmartin]

Also thought of Bad Aibling. However I so not know that line. What signalling is used ect.
I have seen pictures online now. The cab is badly damaged. One does wonder how solid they are on MU's.

 

[Spoorslag '70]

The line to Wolfratshausen is controlled by SpDrS60 relay interlockings, remotely controlled from Wolfratshausen (south of Ebenhausen (inclusive)) and Höllriegelskreuth (on the rest of the line). There was a fault with a level crossing a bit before the incident which could have caused issues (concerning degraded working etc), which iirc was also the situation in last August.

Current official figures are at one dead and ~40 injured.

Condolences to those who lost someone important through this incident and all the best for those who are injured.

 

[duesselmartin]

I assume it is possible to override that system by the Signalman showing Zs1?

 

[Spoorslag '70]

Yes, that's the standard - sadly.

 

[DelW]

S-Bahn-Unglück in Schäftlarn: Erste Erkenntnisse zur Ursache

In Schäftlarn bei München sind am Montag zwei S-Bahnen zusammengestoßen. Ein Fahrgast starb, zahlreiche Menschen wurden verletzt. Derzeit konzentrieren sich die Ermittler laut Innenminister Herrmann auf die Frage, ob menschliches Versagen vorliegt.

www.br.de

So far only a German language link.
Two S7 commuter trains collided in Schaeftlarn, south of Munich. Several injured, at least one person died. Unconfirmed reports suggest the victim is a driver.

There is now an English language translation available via the original link (unless that's being done by software at my end).
It refers to a police statement that the fatality was a passenger.

 

[duesselmartin]

I only see the German text ( from Germany)

For a single track line it has a 20 minute interval.

 

[D365]

For a single track line it has a 20 minute interval.

Quite common in Munich. A lot of the ”branch” lines (S-Bahn only, not mixed used) are single track, with passing loops at stations.

 

[paul_munich]

I guess it happened here, between km 17,8 and km 17,6...

 

[GordonT]

Report in English copied from Sky News (online news version) using following link

https://news.sky.com/story/germany-...er-commuter-trains-crash-near-munich-12542198

One person has been killed and several injured after two commuter trains crashed near Munich.
German Police say the collision happened just before 5pm local time - 4pm in the UK - at Ebenhausen-Schaeftlarn station, 10 miles south of the city.
At least 14 people were injured, and another 80 passengers were being examined by medics.
Police spokesman Andreas Franken told reporters the crash happened on a single-track section of railway.
Helicopters are being used to assist in the rescue operation, and around 130 emergency workers were at the scene.
A replacement bus service is in operation. Police say the cause of the incident is being investigated.

I wonder if this incident has caused the operator to immediately impose any changes to procedures pending the results of the investigations.

 

[MarkyT]

I assume it is possible to override that system by the Signalman showing Zs1?

Oh God, not that again. When is DB Netz finally going to ban the Zs1 on single and bidirectional lines? It's dangerous.

 

[Bletchleyite]

Oh God, not that again. When is DB Netz finally going to ban the Zs1 on single and bidirectional lines? It's dangerous.

Agreed. There is no point in multi level safety systems if one person can just override the lot in a button press.

 

[Gloster]

Bild, that oh-so-trustworthy source, says one dead, a 24-year old originally from Afghanistan, and 18 more or less seriously injured. Reportedly, the late-running train towards Wolfratshausen suddenly stopped on a curve and was hit by the train heading for Munich.

 

[D365]

Sadly, that does sound like what I had suspected. That the northbound train somehow got released onto the single track…

 

[Bletchleyite]

Sadly, that does sound like what I had suspected. That the northbound train somehow got released onto the single track…

Is the line fitted with Indusi? If so, then either there was a serious fault or someone overrode something...

 

[MarkyT]

Is the line fitted with Indusi? If so, then either there was a serious fault or someone overrode something...

According to OpenRailwayMap, yes:

OpenRailwayMap

OpenRailwayMap - An OpenStreetMap-based project for creating a map of the world's railway infrastructure.

www.openrailwaymap.org

The transponders are visible clearly in the cab video posted above

 

[AlexNL]

Many German signals are fitted with what is known as a "replacement signal" or "Zs1" in railway jargon. If a signal can't be cleared in the normal way (for example, because the signalling system thinks that the block is occupied) signallers can use Zs1 to authorise a train to pass the signal at danger without a written order.

When a train departs under Zs1, it is required to run at reduced speed (40 km/h if I'm not mistaken) until it has left station limits. The driver can then accelerate to linespeed, there's no requirement to run on sight. The use of Zs1 is governed by a bunch of protocols and procedures. But Zs1 signals have been involved with multiple serious railway accidents in Germany in the past.

Whether that's what happened here, remains to be seen. But it does sound like it, given what happened in Bad Aibling not too long ago.

 

[BRX]

Surely Germany now has to seriously consider changing their signalling protocols.

 

[Spoorslag '70]

It is actually quite likely that Zs1 was not involved in the accident. Curren speculations in German forums revolve around the misunderstanding that also lead to the incident last August, in which one of the drivers mistakenly took the order to drive on sight and secure the level crossing as an order also allowing passing the starting signal (compare with http://www1.bayern.landtag.de/www/E...cksachen/Schriftliche Anfragen/18_0017773.pdf).

There is quite little slack on the "Isartalbahn" and quite high political pressure to run the service on time. Combine that with single track lines and any kind of operational issues and you will run into problems, as dealing with it corretly costs time.

The protocols are actually fine, it's sticking to the protocols which causes problems. If everything was run to the rules, there would be no incidents (remember that every tragic railway accident in Germany over the last decades was because at least one person was not following the rulebook correctly). The safety culture on the German railways seems to be a bit more lax compared to that in Great Britain.

 

[Bletchleyite]

Many German signals are fitted with what is known as a "replacement signal" or "Zs1" in railway jargon. If a signal can't be cleared in the normal way (for example, because the signalling system thinks that the block is occupied) signallers can use Zs1 to authorise a train to pass the signal at danger without a written order.

When a train departs under Zs1, it is required to run at reduced speed (40 km/h if I'm not mistaken) until it has left station limits. The driver can then accelerate to linespeed, there's no requirement to run on sight. The use of Zs1 is governed by a bunch of protocols and procedures. But Zs1 signals have been involved with multiple serious railway accidents in Germany in the past.

Whether that's what happened here, remains to be seen. But it does sound like it, given what happened in Bad Aibling not too long ago.

I can get the benefits of Zs1 on unidirectional lines - if the rear train drives on sight including checking points positions before crossing them, all is good (bar points shifting under the train, I suppose). But it can't possibly work on a single line or a bidirectional line, because one train driving on sight is all very well, but the other one could be bearing down on it at 200km/h with no idea what is going on.

The mind boggles as to who thought this could even possibly be a good idea.

 

[Spoorslag '70]

The mind boggles as to who thought this could even possibly be a good idea.

It was introduced in the 1920s on the S-Bahn system in Berlin and allowed for all railway lines in 1935. It makes sense if you do have a two-track railway with trains every few minutes (the Berlin case and why it was introduced there), but seems rather out of place elsewhere.

 

[BRX]

It is actually quite likely that Zs1 was not involved in the accident. Curren speculations in German forums revolve around the misunderstanding that also lead to the incident last August, in which one of the drivers mistakenly took the order to drive on sight and secure the level crossing as an order also allowing passing the starting signal (compare with http://www1.bayern.landtag.de/www/ElanTextAblage_WP18/Drucksachen/Schriftliche Anfragen/18_0017773.pdf).

There is quite little slack on the "Isartalbahn" and quite high political pressure to run the service on time. Combine that with single track lines and any kind of operational issues and you will run into problems, as dealing with it corretly costs time.

The protocols are actually fine, it's sticking to the protocols which causes problems. If everything was run to the rules, there would be no incidents (remember that every tragic railway accident in Germany over the last decades was because at least one person was not following the rulebook correctly). The safety culture on the German railways seems to be a bit more lax compared to that in Great Britain.

Surely if a protocol can be bypassed too easily by human error then it's not really fine. If it's the case that this was caused by the driver misinterpreting a signal, it suggests the signalling should be made completely unambiguous, or there needs to be some kind of confirmatory check by the driver.

Unlike things like derailments, shouldn't head on collisions through human error be something that can be made near enough impossible in the 21st century in a wealthy country like Germany.

 

[Bletchleyite]

Unlike things like derailments, shouldn't head on collisions through human error be something that can be made near enough impossible in the 21st century in a wealthy country like Germany.

Completely agreed. If one person's decision could cause such a collision, a protective system should be there to prevent them doing so, and should not be overrideable solely by that same person. At least one other should need to be involved.

Zs1 is absolutely nuts, it makes a mockery of well-established and effective safety systems like Indusi* (a bit like TPWS but slightly more effective). If you're going to have a system where a single signaller can say "well, I don't think there's a train there" you might as well just go back to time-interval working without any form of interlock. It utterly defeats the purpose.

* Induktive Signalsicherung - inductive signal protection - so called because it transmits by way of magnets like AWS does, but is more like TPWS in function, albeit with more advanced speed control.

 

[MarkyT]

Surely if a protocol can be bypassed too easily by human error then it's not really fine. If it's the case that this was caused by the driver misinterpreting a signal, it suggests the signalling should be made completely unambiguous, or there needs to be some kind of confirmatory check by the driver.

Unlike things like derailments, shouldn't head on collisions through human error be something that can be made near enough impossible in the 21st century in a wealthy country like Germany.

I like the PZB Indusi system. It has similar functions to our AWS/TPWS, with distant warning, overspeed on approach to a red, and trainstop at the signal. It's better though, as the onboard equipment enforces a slowdown for a certain distance past a caution and then prevents a reacceleration for a distance after the overspeed. It is also intrinsically failsafe if the trackside equipment power supply fails unlike TPWS (although that's partly covered in UK by proving the kit is powered up and working in the signal to rear like next signal lamp alight proving). Like TPWS however, once you are instructed to override it and pass a red, either verbally or with a special signal like the Zs1 all protection is lost and you are fully reliant on the signaller having performed adequate checks beforehand. Like Bad Aibling, here neither driver would have been aware of the impending collision until the last moment due to curvature severely limiting forward sighting (apparent in the cab video). The recent similar incident of two trains in the block further south on the same line only avoided collision due to better sighting on a straighter section.

 

[Bletchleyite]

I like the PZB Indusi system. It has similar functions to our AWS/TPWS, with distant warning, overspeed on approach to a red, and trainstop at the signal. It's better though, as the onboard equipment enforces a slowdown for a certain distance past a caution and then prevents a reacceleration for a distance after the overspeed. It is also intrinsically failsafe if the trackside equipment power supply fails unlike TPWS (although that's partly covered in UK by proving the kit is powered up and working in the signal to rear like next signal lamp alight proving). Like TPWS however, once you are instructed to override it and pass a red, either verbally or with a special signal like the Zs1 all protection is lost and you are fully reliant on the signaller having performed adequate checks beforehand. Like Bad Aibling, here neither driver would have been aware of the impending collision until the last moment due to curvature severely limiting forward sighting (apparent in the cab video). The recent similar incident of two trains in the block further south on the same line only avoided collision due to better sighting on a straighter section.

In the UK would a train ever be given authority to pass a signal at danger onto a long bidirectional/single line without any other kind of check? Is pilotman working not used in such cases to avoid this exact issue?

 

[MarkyT]

In the UK would a train ever be given authority to pass a signal at danger onto a long bidirectional/single line without any other kind of check? Is pilotman working not used in such cases to avoid this exact issue?

In UK yes pilotman would be used to cover failure on long single lines or for temporary single line working on a multi track route where no bi-di signalling is provided. There are places where it doesn't apply though for conflicts in terminal throats for instance where a single signaller has a good overview of all movements over a fairly short distance and route setting reminder cards are available to manually operate and secure each set of points in the route. Locations where pilotman working is not required in event of signal failure are detailed in sectional appendix I think.

 

[LowLevel]

In the UK would a train ever be given authority to pass a signal at danger onto a long bidirectional/single line without any other kind of check? Is pilotman working not used in such cases to avoid this exact issue?

You can use modified working where authorised where the driver is issued a ticket but it does come with restrictions, including only being able to be used for one train movement, and being authorised in the signal box special instructions/sectional appendix. Google modified working ticket for more details if you're interested.

 

[BRX]

I am interested why there seems to be a more lax approach in Germany - supposedly the home of logic, procedure and precision - than there is in bodge-job UK.

Is it that in Germany, different expectations exist for the fallibility/competence of the individual and there are therefore fewer protections against it?

 

[Bletchleyite]

I am interested why there seems to be a more lax approach in Germany - supposedly the home of logic, procedure and precision - than there is in bodge-job UK.

Is it that in Germany, different expectations exist for the fallibility/competence of the individual and there are therefore fewer protections against it?

I think it is that, yes. That whole culture leads to an expectation that humans will follow procedure perfectly, as everyone does over there when crossing the road. In the UK, a level of fallibility is expected.

Unfortunately, humans, even Germans, make bad automatons.