A train expert tells DW that Germany's rail system is highly automated and dependable - but occasionally requires manual signals. Though newer technologies exist, most Germans won't see them any time soon.
At a press conference following Tuesday's train crash in Bad Aibling, investigators said they would not be speculating on the cause of the accident until the trains' black boxes were examined. It is not yet known how the collision occurred - whether the failure was human or technical.
DW conducted this interview with the publisher of the German rail magazine "Privatbahn."
DW: PZB 90 is the name of the system in use on that rail line - and in use throughout Germany. How does it work?
Uwe Höft: The intermittent automatic train control system (Punktförmige Zugbeeinflussung, or PZB), is actually a very dependable and a relatively old system that's deployed by Deutsche Bahn (German railway).
It works as follows: On the tracks, at critical points like signals or pre-signals, for example - and on the train cars - magnets are installed.
That means stop signals are backed by a live, 2,000-hertz magnet, and they would automatically stop the train in the event of a stop signal being driven through,
Then, in area preceding that, there are the other, 1,000-hertz magnets. These are at pre-signals, for example. The conductor has to acknowledge the relevant pre-signal as the train approaches a stop signal, and adjust his or her driving accordingly. That means the train's brakes have to be deployed. If the driver doesn't do so, the train will brake automatically.
And then, in between, there are 500-hertz magnets that monitor the train's velocity. These are usually 150 to 300 meters (500-1,000 feet) before the signal, and if the train has not yet gone below a certain velocity, than the emergency brakes will automatically be deployed.
In that sense, a relatively robust, simple system serves to prevent accidents.
And the fact that two trains collided on a single track? How can that be?
That shouldn't normally happen, since the dispatcher is responsible for the safe operation of trains and the execution of train services. He or she gives the approval for vehicle access and ensures that these things do not happen - supported, of course, by the technology.
But in railway operations there are repeated disruptions. There are different fallback modes, whereby people once again have to act and, above all, have to act properly.
Why this happened in this case ... That's the task of the Federal Railway Authority or the railway accident investigation board, which will clarify the circumstances, the documents at the switch towers, and will evaluate the dispatchers, the relevant calls between the trains and the dispatchers, and, of course, the black box. This memory module in effect shows any action that the conductor has made.
What are some of the conceivable errors that could circumvent this system?
There's the possibility, for example, that if a signal malfunctions, there still has to be a possibility of getting around it. There are some keyboard operations at the switch tower, where I [as a dispatcher] can turn on a so-called surrogate signal, whereby the conductor than recognizes, "OK, I'm not allowed to drive through this signal."
But should you ever drive through a stop signal as a conductor?
No. It's for that reason that they're there: You should stop. But there are sometimes cases where the signal's disrupted and where railway operations must continue to be maintained.
That's not very high up there with regard to security levels, but normally there's the dispatcher and the driver - both people who, through their actions, which are enshrined in regulations, work through it.
The easiest variety is then, actually, that the dispatcher sends a surrogate signal. This signal shows the conductor that he or she is allowed to drive through the signal.
These possibilities are there because there are technical problems in railway operations.
With as far as we've come with autonomous driving, is there any technology that could replace these magnetic signals any time soon?
In principle there are modern security technologies that could theoretically enable autonomous driving. But if you just take a single look at the rail network in Germany with regard to length and needs for investment, then it is probably going to take a very, very long time until such a system can really run safely.
What's an example of some new safety technology?
Well, there's this new European train safety system, the European Train Control System (ETCS). Right now it's in operation on the newly built line between Leipzig and Erfurt. And then there's a line between Berlin and Leipzig where this technology is also installed, although at varying levels. Between Berlin and Leipzig there are also still the conventional signals as a failsafe.
On the Leipzig-Erfurt line, they've completely done away with the signals. It's done with so-called beacons. You can imagine them as small wireless sensors that are laid in the track bed. The information is thereby exchanged between the train and the dispatcher. This is already a very strong move toward automated driving.
A train driver is of course still there, but it's mostly automated.
The investment requirements are gigantic. Germany will certainly not see full coverage in the near future, but rather the PZB system, which also works very reliably.
But as you have already said, PZB may not be entirely reliable. Is there a way to upgrade it?
You'd have to completely replace the safety system. In its basic conception, the PZB system has remedied very many flaws. But, as I said, if an exceptional situation occurs and people act - and people make mistakes - then it can unfortunately lead to an accident. One hundred percent security, even with automated systems such as the ETCS, is unlikely.