Queensland Rail and safety interfaces (Part II)


Cleveland street train overrun (Image source: ATSB)

The ATSB has released it’s preliminary report of it’s investigation into the Cleveland street overrun accident which I covered in an earlier post, and it makes interesting reading.

So first off (engage smug mode) I’ll note that my initial educated guess that there were no dedicated anti-climber features was correct. 🙂

The buffer stops located at the end of the platform tracks at Cleveland Station are reinforced concrete structures with rubber fenders. The buffer structure is made of a 2.9 m high concrete block protruding 1.7 m out of the ground attached to a pair of horizontal reinforced concrete beams, one under each rail. The rubber fender is attached to the concrete block and has an adapter plate attached to match the couplers used on the CityTrain fleet. The buffer has been designed for a 200 tonne train with an impact speed of 5 km/h.

ATSB preliminary report RO-2013-005.

Likewise the overriding scenario that I outlined seems to have been right on the money. But what’s really interesting about the design of the buffer stop as described in the report, is that if the designers had actually tried it probably couldn’t have been designed worse for a high speed impact.

First by making the buffer a fabricated cantilevered structure e.g. one attached to the tracks, rather than monolithic, a weakness in shear was built into the design, a weakness that also predisposed the block to rotate under impact. Secondly by using a flexible rubber fender the design allowed for vertical motion of the train relative to the buffer face during impact. Allowing it to mount the barrier. And finally by placing the buffer in front of the platform as a stand-off structure allowed the buffer block to move horizontally while translating rotationally under impact.

Unfortunately the ATSB, like many others before it, has focused upon prevention of the accident rather than considering whether we also need to address the consequences of the accident, .there’s now a working group off chasing issues around rail contamination, and good luck to them.

But focusing on prevention is only a valid approach if you can demonstrate that the subsequent likelihood of a train overrunning a station platform is very, very low. To draw that conclusion we also need to trust the risk analysis, which brings us back to the problem of the myth of mechanical objectivity. For example, ‘fixing’ leaves on the rail is not going to address the risk of a station overshoot due to other reasons, as occurred at Spencer St. Thus any risk assessment that concludes that the resultant residual risk is now acceptable because rail contamination has been addressed is incomplete, for example driver incapacitation could also cause a loss of control.  And that of course does not even start to consider the ‘unknown, unknowns’.

The report also fails to acknowledge the potentially catastrophic nature of the overrun itself, had the station concourse been filled with commuters, a train running into a buffer is one level of disaster, but if it then can over run the platform concourse? That’s a level of magnitude worse.

In not recognising this quantum increase in potential severity it appears that the investigators have fallen afoul of a subtle framing effect in their consideration of the accident, that has led them to view the ‘train in the concourse’ as an aberration, rather than an indication that something may be systemically wrong with the network’s safety management system. Instead they’ve focused on the over-speed as a localised event that can be addressed with specific measures and from which more general lessons need not be drawn.

What I’d still like to know

First I’d like to know what risk assessment (if any) was done by QR on the design of the station, in particular whether the low platform height for a terminus station was considered explicitly in the context of a station overrun. Then I’d like to know a little more about the buffer stop design, if it was really designed, as the report notes for the 5 kmph impact of a 200 tonne train, and if so what verification was actually conducted, test data is acceptable, calculations not so much.

Finally I’d like to know whether the buffer stop as designed met the overrun risk controls, e.g. was the risk of a train overrun only considered as being at 5kmph. Oh, and I’d also like to see the actual calculations, test data and so on, or have faith that the investigators had sighted these document as well as the use of a formal method like Why Because Analysis to lay it all out.

As a final thought, the problem of meaningfully evaluating the risk of a station overshoot could be eliminated by designing a buffer stop capable of preventing a high speed train overshoot. In other words, by adopting a possibilistic rather than probabilistic design approach.

One response to Queensland Rail and safety interfaces (Part II)


    Good stuff Matt.