[SystemSafety] Spanish train crash

[Peter Bernard Ladkin]

Folks,

I have been reading coverage from the Guardian (a reliable British newspaper) as well as looking at 
the WWW, including presentations about ERTMS http://www.ertms.net/ertms/ertms-signalling-levels.aspx 
in Spain from Invensys Rail Dimetronic 
http://www.dimetronic.es/wps/wcm/connect/WebDimetronic_EN/Web/NuestrasReferencias/Todas/LAV+Orense-Santiago 
, which is the consortium leader for signalling and communications on the Ourense-Santiago HS line, 
and from the network operator adif 
http://www.uic.org/cdrom/2012/10_ERTMS-Conference2012/docs/C2-Jorge-IglesiasV1.pdf .

The following things are not clear to me:

1. On the Ourense-Santiago line, where exactly the transition from HS to rebuilt conventional track 
takes place with respect to the A Granderia curve.

2. What level ERTMS is installed on the HS portion of the line. Level 1 or Level 2?

3. Where exactly on the line the transition from ERTMS to ASFA takes place relative to the A 
Grandeira curve. (Reports have said in the vicinity of A Grandeira; they have also said 3 km distant)

4. Whether the trainset was equipped with ERTMS and using it.

Some Guardian reports have called it "Alvia 151" and said it wasn't equipped for ERTMS; 
alternatively that ERTMS "wasn't in use" on Madrid-Ferrol (I don't give references because I think 
this info is suspect; mail me if you want them). There are hardly any references to "Alvia 151" on 
the WWW except for Guardian articles and one YouTube video title. "151" might just have been the 
train number on the timetable (trains get numbers, like flights, in many European countries but I 
don't think in Britain).

The Alvia trainsets are usually known as 730-series trains and the adif presentation says they are 
equipped with ERTMS Level 1 and 2. Of course, as I have personally experienced on the 
Cologne-Brussels line, trainsets of the same type may have some equipped with ERTMS Level 2 and some 
not. If the line is Level 2, then it maybe cannot accept trains not equipped for it, since there 
need be no trackside signals (as there aren't with the German LZB system, which is also partly used 
in Spain as far as I can tell from the adif presentation).

5. The Guardian (Gyles Tremlett) has said in 
http://www.guardian.co.uk/world/2013/jul/25/spain-train-crash-travelling-so-fast that estimates from 
the video timestamp compared with position as estimated from distance between pylons, or between 
"railroad ties" (sleepers to Brits) are
140-190 kph, resp. 150-180 kph (PBL translating mph into kph and rounding conservatively to 5kph). 
Are those right?

6. The driver is reported by El Pais to have gotten an overspeed warning and acked it 
http://www.guardian.co.uk/world/2013/jul/26/spain-train-crash-police-driver  I don't know if 
warning-without-enforcement is possible under ERTMS (either Level 1 or Level 2) but it is possible 
under ASFA. Is there independent confirmation of this indirect report?

7. Nobody has said whether there is any indication that braking was initiated before the derailment.

8. So *if* there is a balise there before the curve, why isn't automatic overspeed protection 
implemented?

9. Why isn't there a transverse gradient on the curve sufficient to allow speeds higher than 80kph? 
(Maybe there is?)

10. What is the status of the 80kph speed limit? Is it purely nominal (for example, on a rule which 
says "all curves with such-and-such geometry, within such-and-such of a station, must be posted 
such-and-such speed limited"), or is it based generally on the curve geometry?

11. I read in the Observer (Sunday paper of the Guardian group) that the driver has been arrested, 
arraigned in handcuffs before a judge, and charged with multiple manslaughter.

That is, of course, an excellent way to ensure that investigators (of any kind) obtain minimal 
useful information about any human factors at play in this accident. Indeed, the article reports 
that he has received legal advice not to make a statement to police.

Let me make some obvious HF points in non-technical language. The guy is an experienced driver, 
middle-aged, far older than the kind of person the police scrape off the roads in Germany every 
weekend of good weather after they come a cropper on a motorcycle. He knows the route well, indeed 
had been driving it for a week. It was daylight - you could see where you are. He has gone round 
that curve in these trainsets many times before, indeed many times recently, so if he has been 
exceeding limits before he knows what it feels like, and he presumably knows what happens to trains 
that take curves way too fast. He may very well be shocked, and grief-stricken, and parading him 
before the legal authorities (and news media) in handcuffs is not going to lighten any feeling of 
culpability and guilt he may have. How are you going to find out, for example, if he had a brief 
loss of consciousness unless you can talk to him sympathetically about anything he remembers? 
(Detecting such a syncope is a matter of *inference* rather than being experienced, in particular 
for the person who underwent it.)

And those are some disadvantages from the point of view of effective accident investigation, leaving 
aside the question, which some might well consider more important, of whether such action is just.

PBL

Prof. Peter Bernard Ladkin, Faculty of Technology, University of Bielefeld, 33594 Bielefeld, Germany
Tel+msg +49 (0)521 880 7319  www.rvs.uni-bielefeld.de