Ian Batten wrote:
On Friday, 22 September 2017 14:59:16 UTC+1, Recliner wrote:
Peter Able wrote:
Fascinating report on what could have been a deadly incident at Bank DLR.

An alignment fault on a CCTV camera was noted but, by the split of
responsibilities, the maintainer sent wasn't allowed to re-align the camera.

An intending passenger wedged himself into the door of a train. This caused
the doors to re-open.

None of this was seen by the driver as for unproven reasons he decided not
to follow normal departure procedure but relied on CCTV images alone to
judge departure.

A late-arriving passenger saw the doors opening to release the wedging and
thought that she had plenty of time to board - but quickly found that she
couldn't and backed off (and this is all at the CCTV-invisible door,
remember)

A draw-string on her coat happened to get caught in the door and she was
dragged forward with the departing train.

Fortunately she got one arm out of the coat and the train dragged the coat
off her.

Oh, and the door obstacle detection specification was that the system must
be able to detect a "test block" 30mm thick. Now how many coats,
draw-strings - or human hands - are at least 30mm thick under compression?

https://www.gov.uk/government/news/r...t-bank-station


'Driver'?

Indeed.

One solution could be for the doors to close to the point at which object
detection ceases, then pause, then close finally. The closure to a small gap
would provide a very, very strong warning, and no-one is going to try
to jump through that gap. The pause need only be a second or to: time to
withdrawn an object that is caught.

The force required to remove objects from power doors is considerable.

Actually, she didn't try to jump through the gap, but stopped just before
the doors. I guess her forward momentum meant that her coat's loose
drawstring swung forward, and got caught in the doors just as they closed.
It was soft enough to be squashed, and wasn't detected.