Scotrail Delay-Repay & Performance Data Issues

[Haywain]

but their search didn't show up the initial journey I noticed;

If the recording of the arrival time in their records is incorrect, it's difficult to see how an error could be detected by a search.

there's no cross referencing of human input against available technical data to check it.

If there was other technical data they wouldn't be using manual input.

 

[RodMcMillan]

If the recording of the arrival time in their records is incorrect, it's difficult to see how an error could be detected by a search.

If there was other technical data they wouldn't be using manual input.

I asked them for any records of human error, they found one they had corrected, I know of another they haven't. I was presuming there would be a lot of corrections but apparently not.

There is other data that these can be checked against. Though some will sit with TOC and only some with Network Rail. But - signalling logs, adjustments to departure boards, Comms between signalers and trains, tokens for single tracks, even logs showing driver times. I give one cross reference where the driver log shows the driver stopped driving 55 minutes after the data says the train terminated.

== Doublepost prevention - post automatically merged: 15 Dec 2024 ==

I asked them for any records of human error, they found one they had corrected, I know of another they haven't. I was presuming there would be a lot of corrections but apparently not.

There is other data that these can be checked against. Though some will sit with TOC and only some with Network Rail. But - signalling logs, adjustments to departure boards, Comms between signalers and trains, tokens for single tracks, even logs showing driver times. I give one cross reference where the driver log shows the driver stopped driving 55 minutes after the data says the train terminated.

*Log of corrections
*Have one cross reference. Sorry typing on phone.

 

[Haywain]

There is other data that these can be checked against. Though some will sit with TOC and only some with Network Rail. But - signalling logs, adjustments to departure boards, Comms between signalers and trains, tokens for single tracks, even logs showing driver times.

These pretty much all rely on the same human being.

 

[Dr Hoo]

An aspect that surprises me is that on a ‘modern’ electronic system such as that relating to the ‘virtual tokens’ issued and surrendered on a single line route like the West Highland apparently leaves no recorded trace of train movements but instead relies on a human being in the Banavie control centre remembering to write it down manually at the time.

I could understand that on a traditional route like Stranraer a signaller at Dunragit or wherever might make an occasional error or omission but the whole thing seems very odd. (I do understand that the West Highland doesn’t have traditional signals, track circuits or axle counters other than at Fort William.)

 

[Falcon1200]

I asked them for any records of human error, they found one they had corrected, I know of another they haven't. I was presuming there would be a lot of corrections but apparently not.

Thanks for the additional detail. Where manual inputs are required (at only a handful of locations) human error is always possible, but there should certainly be some means of identifying and correcting such mistakes. I can only say that in my experience, in Network Rail Control in Scotland, these were very rare.

 

[RodMcMillan]

Thanks for the additional detail. Where manual inputs are required (at only a handful of locations) human error is always possible, but there should certainly be some means of identifying and correcting such mistakes. I can only say that in my experience, in Network Rail Control in Scotland, these were very rare.

Thank you, yes, I'd expect them to be rare, but I was more surprised the system allowed them, and there's apparently no system for checking them or monitoring them or even counting them.

 

[Haywain]

I was more surprised the system allowed them,

Do you mean allowing manual entry or allowing errors? Errors happen because there is no available automated method of recording the times.

 

[RodMcMillan]

Do you mean allowing manual entry or allowing errors? Errors happen because there is no available automated method of recording the times.

That there was no available automated method.

I know i'm oversimplifying, but not by much:

1. Set up a Google Workspace corporate account
2. Set up each train with a smartphone with location on
3.Google timeline will record to the minute where that train is, and when it got to the station.

Like Uber but on a train.

Or that there's no sense check on manually recorded data. (Appreciate done here by Network Rail) But ScotRail had logged internally that the train was delayed.

 

[35B]

That there was no available automated method.

I know i'm oversimplifying, but not by much:

1. Set up a Google Workspace corporate account
2. Set up each train with a smartphone with location on
3.Google timeline will record to the minute where that train is, and when it got to the station.

Like Uber but on a train.

Or that there's no sense check on manually recorded data. (Appreciate done here by Network Rail) But ScotRail had logged internally that the train was delayed.

The existence of inaccuracies, not necessarily accidental, has been a matter of public record since at least 1915 (the Quintinshill disaster).

Going to questions of delay repay, these logs are being relied upon for a purpose they were never originally designed for, using systems that sit on top of the operational data.

I’m not sure why adding additional technology, with the associated maintenance and support requirements, is necessary for a system that already works effectively in the vast majority of circumstances.

Edge cases exist (see other threads here about door opening times vs berth times), and errors exist. I’d be more concerned, especially if the operator professes to operate a quality assurance system like ISO9000, at the inconsistency of the back office case handling than the needle in haystack of signallers’ data logging.