Great Western Coffee Shop

The BBC now have it too

http://www.bbc.co.uk/news/world-europe-31547979

No word of an official explanation yet, but the pictures do support the initial reports of a signal passed at red.

There are two through lines and two platform loops at Rafz, and the express train from Zurich to Schaffhausen can be seen to have struck the suburban train a glancing blow and to have been almost entirely derailed. It crossed the other through line (to Zurich), and struck something (presumably an OLE stanchion) hard enough to tear a hole in the power car and result in serious injuries to the driver.

This train continued over a small bridge, where it just stayed within the parapet railing, onto the bare ground beside the track. The suburban train was left upright - and may have stopped by the time of the impact. It is hard to avoid the conclusion that the suburban train set off before its signal had cleared, unless the signal had cleared in error.

Incidentally, both trains were going to Schaffhausen - in Switzerland - but the next two stations are in Germany. Do they still drive on the left there? It's a bit academic, as the line is single at the border so there is no need for an Alsace-style flyover.

Further news (in French) from 24heures.ch (http://www.24heures.ch/suisse/faits-divers/accident-train-rafz-desempare-experts/story/27552588)

This says that both trains were driven by trainees under instruction, and unlike the last Swiss accident at Granges-Marnand the station has an up to date signalling system (ZUB). This ought to at least stop a train trying to pass a red signal, so the current expert view is that that driver error alone can't explain it.

Having looked at the layout on Google Earth, that's wrong. The pictures are misleading - there's a main island platform, and also a second one, but very low. And they use both sides of the main platform for both ways - many trains terminate here and start back both ways too. So the stopping train (RER) was on the left and the express on the rightmost track, which gives the same geometry.

The report of the investigation of this accident (https://www2.sust.admin.ch/pdfs/BS//pdf/2015022001_SB.pdf) has been published by SUST (aka SESE, SISI, and STSB) (https://www.sust.admin.ch/en/stsb-homepage/). It's only in one language, in this case German which is local to Rafz. I'd expected the summary to be translated into French and Italian, but apparently not.

Since I know almost no German, and this is complicated stuff full of long words (even by German standards), that causes some problems. However, it's full of pretty pictures, so it's possible to follow the narrative parts pretty well.

For those who think the Swiss set the highest standards of "how to run a railway", the causes of the accident may be a bit of a surprise. The S-bahn train did depart against a red signal, and the emergency stop when this was realised left it foul of the points where a through track joined the platform track.

This is my take on what report identified as the main reasons:

• The drivers (trainee and trainer) mistook the green signal for the through track for the one applying to theirs. It was straight ahead, and brighter than the correct one which was to one side. I think the report even says that the design rules for signals don't include sighting confusions like this.
  
• The train and signalling system (ZUB) should stop a train attempting a SPAD, but in this case it was not fully operative - the train had reversed at Rafz, and turning on the new driving position reinitialised the system so this autostop was temporarily inactive.
  

There are several other things, including a lot about the operational procedure taught for departing a station (rather than the way the learner was being supervised). That, and other points, make less sense to me.

If anyone can cope with the German, and wants to correct any errors in the above, please do.

Thanks for posting that STUVING.  I must say that having been involved in signal sighting and scheme layout design in the UK over a considerable number of years, that I do find the conflicting signal aspects somewhat supprising, but then they do tend to rely on the Train Protection system being the safeguard in such circumstances.  If I do manage to decipher anymore of the report (I'm lucky as a signal engineer to be able to understand some of the technical issues even though they are in German) I'll add it here  ::) :P