Germany train crash: Several killed in Bavarian town of Bad Aibling | Hacker News Reader

Germany train crash: Several killed in Bavarian town of Bad Aibling

bbc.com

11 pointskercker13 comments

Note that in this particular case two trains belonging to a small regional short distance company collided.

I have relatives working for "Die Bahn"^, which is the actual national train company. And i can tell you that this kind of thing is a direct result of the privatization of the train industry in germany. First it simply resulted in less quality in transport, with money being saved on the maintenance of non-critical parts, such that i.e. one car per train may have non-functioning air control.

Now small unaffiliated transport companies have sprung up to cover very short distance trips on budgets just as short. The result is what you have in the article.

The only good news is that as of yet, travelling with Die Bahn is still safe, since they at least spend a lot of money on training and maintenance of critical parts.

^pronounced "dee bahn", meaning "The Rail"

As a European I consider trains to be the most comfortable way of traveling available today. Planes have these awkward boarding procedures and trains, even in economy, have nice amounts of moving space.

It is so mind-boggling to me how train-to-train crashes can occur. You can naively assume this must be one of the easier things to prevent. Especially in a highly developed country like Germany.

Does this mean train infrastructure (in Europe) is so outdated? Or why does this seem to happen fairly often?

Edit: According to the Wikipedia list mentioned below there seems to be a train-to-train collision about every 2 years in (Western) Europe. Let's not focus on whether that is a lot or not, but what generalizations can be made about their causes.

Well, does it really happen "fairly often"? Looking at the German train accidents in recent years, a lot of them don't have (human) casualties, so an event like this still seems pretty rare, considering the high amount of daily traffic. And because of that, this gets widely reported.

This is literally my neck of the woods. I just commuted to work on a parallel route. And I actually read this here first. If the same amount of people died on the nearby Autobahn, this wouldn't be international news.

But yeah, I'm very interested in hearing why this could happen at all. I just don't think its endemic, and I don't generally have a high opinion of the current state of the German railroad system (privatization was a bad idea).

A series of very unlucky equipment failures (unlikely, I think), an error in the signalling implementation, or trust in a more passive system (driver must obey signals and speed restrictions) rather than something active (passing red signals stops trains). I'd guess the latter, since this seems to be local trains on a low-traffic line.

In which case, there could be failures like a driver being allowed to do too many night shifts or something like that.

Presumably Germany will produce an official report on the cause.

The last railway fatality in Germany, not involving a road vehicle on a crossing, was in April 2012. So hardly frequent.

I'm guessing this to be the approximate locality: http://www.openstreetmap.org/search?commit=Hledat&query=bad+... - a local branch track, so a passive safety system is likely.

Disclaimer: I've worked on automated transportation systems and have worked on SIL-4, realtime/safety-critical systems for the trains systems of Europe and most of the western world.

Germany has one of the best rail transportation infrastructures in the world. You can travel around the German state on trains in a single day, in much comfort, and hundreds of thousands of people are delivered to their destinations with safety and security, every single hour.

What they also have, is complacency and decadence, and you could say a lot about how the bureaucracy of companies such as Siemens and Thales allows this decadence to flow throughout the infrastructure.

Accidents like this happen today because of human error, through and through. No matter how much we automate safety, in the end it depends on having a conscious human watching the systems and ensuring that things stay, literally, on the rails. It is far too soon to try to understand what went wrong in this case - but I can only say that the fact that an accident occurred, in spite of the extremely impressive infrastructure put in place to keep it safe, means simply: someone screwed up.

> Does this mean train infrastructure (in Europe) is so outdated?

No, European train infrastructure is generally among the most advanced on the planet, probably behind Japan and certain parts of China.

> Or why does this seem to happen fairly often?

This is the first time I've heard of such a crash in a long time. To me, it doesn't seem to happen fairly often.

Beyond that, there are approximately 750m people living in Europe: that's 2.5x as many as in the USA, which I presume is what you're comparing to, so you would expect 2.5x more crashes (assuming all other conditions kept equal). Furthermore, Europe relies on passenger rail as a means of transporting people to a far greater extent than the US, where city-to-city distances are much greater, so it is more advisable to fly. I assume that the density and total use of railway (for passenger transport) is much greater in Europe than in the US, so perhaps it makes sense that you hear more about European train crashes.

(Finally: as a European, have you considered that you mostly hear about European train crashes as a consequence of the media you're exposed to? You tend to hear more about the place in which you live than about other places...)

This seems to happen fairly often (note the emphasis); things that actually happen often are not on the news (because https://en.wikipedia.org/wiki/News_values ).

From https://en.wikipedia.org/wiki/List_of_rail_accidents_(2010%E... : I have counted about 170 fatalities 2010-2016 for the whole of EU, about half of which (81) is from the 24 July 2013, Santiago de Compostela derailment, about another quarter is "train hits car after driver went around the barriers," and there seem to be significant numbers of sabotage (about 20 fatalities). In that, I have counted only 5 train-to-train crashes (Buizingen, Szczekociny, Sloterdijk, Neuhausen am Rheinfall, Bad Aibling).

See also this Eurostat report: even for the deadliest year for a long time (the Santiago derailment again), train collisions accounted for 101 deaths per year. http://ec.europa.eu/eurostat/statistics-explained/index.php/...

You are quite correct that this is a well-understood problem, one that is prevented millions of time per day as the absolute baseline of any railway system. Of course, when it happens, it makes the news - precisely because it's so rare. If two cars collide and people die, that's not news - that happens all the time (at least 27 101 people in 2013, across 23 EU member states).

What is also not on the news (yet it does happen fairly often) are prevented accidents: train passes signal at danger due to driver error, all affected traffic stops immediately and automatically, some delays and a mountain of paperwork appears.

That is an example of the safety system working as designed. No property damage, no injuries, no loss of life - not newsworthy.

re edit: TL;DR: Unable. This is the usual pitfall when troubleshooting a highly reliable system: any uncaught exception is going to be an outlier, almost by definition.

Given the millions of km travelled, one collision in 2 years on the whole continent means "insufficient data to draw conclusions." The only thing that follows is "can't generalize from that." Whether it's a lot or not is completely essential to drawing generalizations: with 5 data points (which I have honestly picked out of Wikipedia under the assumption that such an event is unlikely not to be mentioned - that's a lot of filters right there), what you're saying is equivalent to "there has now been a single confirmed death by meteorite; let's generalize!"

(Tangent follows: Europe (or even the EU) does not have a homogeneous rail network, with the same standards kept everywhere: each country keeps its own set (albeit loosely compatible) of standards and its own overseeing agency; that translates to very different quality of track, service, security etc. The only thing that's common is "the overwhelming majority of track is of the 1435 mm gauge" - there are even countries with four(!) different electric power systems on the same network. (Also translates to a certification nightmare for cross-border trains.)

Moreover, there are various levels of rail: there's the (sparse) TEN-T backbone, national major routes, regional networks, and local lines; although those are interconnected, there are varying levels of safeties, from the (interoperable) ETCS, through various local (automated and semi-automated) block signalling, PTT radio dispatching and manual signal operation, all the way down to 1890s-style train handover by telephone signalling. (No idea which of these is used on this specific line.) )

"Someone screwed up" On today's railways, it is very difficult, deliberate action excepted, for one error to cause an accident of this magnitude. Most modern railway accidents occur following an unfortunate combination of coincidental events and\or procedural breakdowns. Technology and procedures have evolved over the years to ensure that one mistake cannot lead to such an event. Railway authorities are no longer complacent and reactive in their application of safety measures as they once were.

Still, it's quite obviously possible. Nobody has implied that this was specifically due to a single point of failure or a single human error...we'll have to wait for more data.

The JavaScript framework situation is getting out of hand

Reply on news.ycombinator.com