[SystemSafety] Spanish train crash

[Peter Bernard Ladkin]

Folks,

Chick Perrow just sent me the following link to an interview with him about the accident:
http://www.thespainreport.com/2013942/perrow-striking-evidence-for-operator-error/

I haven't checked it out (I don't belong to Facebook) but it suggests that the train driver has a 
Facebook page on which he has talked about behavior which might have some relevance to explaining 
why the train was overspeeding.

What is puzzling me is the overall geometry and control on and around that section of track.

Jens Braband sent me a link to an explanation (in German) of ASFA on Wikipedia 
http://de.wikipedia.org/wiki/ASFA  I imagine people can suitably transform for the page in their 
preferred language. It's a form of (in British) ATP, designed by Westinghouse Air Brake Systems 
(those of the freight braking recently discussed in the wake of the Lac-Megantic accident) in 1972 
to a RENFE specification.

ASFA consists of occasional "balises" (electronic milestones) that work on induction and 
signal-frequency communication with up to 8 "resonant frequencies". The trainside actions consist of 
driver warnings, speed surveillance (here, the German word "Überwachung" is ambiguous between 
surveillance and some amount of control), and emergency automatic braking. It is "point-based" (that 
is, interaction occurs purely at the balise locations) rather than "linear" (continual 
control/sensing, as with ERTMS).

People have said that the accident location is in or near the "transition zone" between ERTMS and 
ASFA. The HS line was finished and has been in operation since late 2011. The pictures show an 
obviously new section of track (for example, the sleepers and gravel are still all light-colored), 
so I imagine that this piece of track follows the older track trace into Santiago, but has been 
rebuilt.

But still - I wonder about a 80kph curve - surely you'd design a transverse gradient into the track 
(German "Querneigung" - I don't know what it's called in English and picked the best-looking word 
from LEO, but it's for roads, not railtrack) in order to enable faster train transition and reduce 
maintenance. What would the reason be here for not going so? Maybe simply because it is close to a 
station at which most or all trains would stop, or because this stretch is also used by local 
stopping trains travelling more slowly, which would make a transverse gradient here uncomfortable 
for their passengers?

I also wonder about the general design of a track which is "transitioning" from high-speed track 
with ERTMS control. If you are designing-in an 80kph curve and you have high-speed 
continuous-control transitioning to low speed and ATP, wouldn't you put a balise and automatic 
emergency brake activation at a suitable point before the curve? After all, it was 2011 and everyone 
in rail knows about Waterfall (New South Wales; the driver is supposed to have had heart failure and 
the "dead man's pedal" didn't work).

Bernd told me the CCTV video of the accident happening was available through the Guardian (reliable, 
worthy British newspaper). I haven't had the discipline to look at it yet.

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