AWS - semaphore signalling

[Amrutha]

Why is AWS provided only on the approach to distant signals in case of semaphore signalling?

[PJW]

Why is AWS provided only on the approach to distant signals in case of semaphore signalling?

Because the whole function of AWS is to act as a back-up to the visual indication gibven by the distant signal. This is the ONLY signal under semaphore signalling at which braking should commence. A signalbox may have many stop signals (homes and starters) along the line within the signalbox's area of control, but unless ALL of them are cleared then the distant must be maintained at caution and the train expecting to stop at the first stop signal; however if this can be cleared then the signaller will clear it once the train has been brought under control and nearly at it, then it would be dropped down to the next signal similarly until the one that can't be cleared.

The question should perhaps be "what is the purpose of fitting AWS to Red/Green signals under colour light signalling?".
I suppose the answer to that is; no real functional reason, mainly because drivers get used to having AWS at all colour lights and it might not be immediately obvious why one were "missing" just because the signal concerned was never a "distant" at which braking to stop at another signal needs to commence.

[Peter]

Before PJW's comments above were posted, I had asked yet another Peter who is far more learned in these matters than me to give some information. Here is his response

In order to answer the question fully, it might be useful to understand how and why AWS developed. In summary, it was invented in Reading by Amos Brooker although patented by Insell and Jacobs. It was introduced in 1906 as the ‘Audible Signal’ with the intention that it should replace the Distant signal. The double line example was installed on the Henley branch, and the single line on the Fairford branch. It used a hump-backed ramp that could have a voltage applied to it, placed between the rails some 200 yards on the approach side of the Distant signal. If the Distant were ‘on’, the ramp was electrically dead and would cause a brake application on a locomotive unless the warning were acknowledged. Thus an un-acknowledged alarm would bring the train to a stand within the ‘allowed’ distance – between the Distant signal and a pre-determined place beyond the Home signal known as the clearing point. If the Distant were ‘off’ the ramp would be electrically live and cause a bell to ring in the locomotive cab, telling the driver where the distant signal was and that all the stop signals to which it applied were clear.

Under the GWR this system was known as ‘Automatic Train Control’. After Nationalisation the contact-less magnetic Hudd version which had been trialled by the LMS on the Fenchurch Street to Shoeburyness line was developed into the standard system, with the ex-Western Region ATC installations being replaced during the 1970s.

Fitting of AWS to semaphore signalling is to signals that display a ‘caution’ aspect only, since (unless all Stop signals to which a Distant signal applies are clear) the signaller is responsible for clearing stop signals only when (s)he is satisfied that a train is under control. Additionally, the spacing of the Stop signals is not prescribed in the same way as for Track Circuit Block (TCB) Multiple Aspect Signalling (MAS). This remains true when semaphore Stop signals are replaced by Red/Green colour light signals. AWS is not fitted to a Red/Yellow/Green signal within a semaphore area where the yellow aspect is only displayed as a Delayed Yellow (when not all the Stop signals ahead of it are clear), since under those circumstances the train will have been brought under control from the Distant signal in rear (which is fitted with AWS). Finally, a Red/Yellow/Green signal in a semaphore area where the Yellow aspect is a free yellow with the track section proved clear up to and including an overlap, would be fitted with AWS since that signal is acting as a ‘Distant’.

Turning to MAS, a major decision point was reached on the ex-WR when Gloucester panel signal G272 was introduced at Standish Junction. This was in the days before TPWS; G272 was the first Red/Green M.A.S. signal and the question was should it be fitted with AWS? The decision was to fit it, and using today’s language it was because it was ‘more ALARP’ to fit than not, given that a driver was to receive an AWS indication at every other main signal on Gloucester panel. It was not strictly required to be fitted since G272 does not display a caution aspect. Since then, all Red/Green signals in TCB MAS areas have been fitted with AWS for consistency.

[asrisaku]

Dear PJW and Peter

May I ask a silly question about AWS? I was wondering why we allow the AWS to be intervened by human? Any benefit for it?
If human cannot intervene AWS, then it would be safer, right? and it will not have a SPAD, unlike TPWS.

Thanks in advance for helping me out.

Best regards
Arnut

[Jerry1237]

AWS is not placed at the minimum braking point from any signal (excepting very low speed scenarios). AWS does not and cannot guarantee a SPAD will not occur. Also, a SPAD is not necessarily dangerous in itself as the TPWS philosophy of stopping the train beyond the signal demonstrates.

What do you mean by intervene? A driver has to interact with AWS at every adverse [non-green aspect] signal. If the question is why are drivers permitted to cancel AWS on approach to a red aspect; the AWS system is not intelligent at all and cannot differentate between the non-green aspects. It was designed in fifty or sixty years ago and selected over the, arguemtatively, better ATC on the Western.

All drivers, for almost every interaction with AWS, acknolwedge it correctly and stop their train in appropriate positions. Sometimes, humans make errors and override or ignore the warning (normally inadvertantly). I'd recommend reading some human factors studies to explain why humans can and do make errors, often totally illogical, especially under even slightly abnormal situations.

You could ask why drivers are still necessary when planes fly without the pilot's assistance!

J

[asrisaku]

Thanks a lot for prompt response. I meant what you explained above. I will try to find human errors to build up my understanding.

[Jerry1237]

You would need to read articles on error rates induced by repetitive actions. The IRSE and/or IET libraries may be able to help. It is not a simple subject.

The importance to us is remembering no system is foolproof and errors will, and do, happen. By careful design and appreciation of factors outside of our control, we are able to mitigate the effects of these errors. For example:

AWS is a reminder. The driver would already have been following a normal aspect sequence and the AWS should be unnecessary. If you look at the RAIB website, there is a good human factors incident just been reported with Bulleid steam loco Tangmere SPADing a pin and overrunning into pointwork. It was purely an HF issue where the crew's actions were inappropriate, against the rules and an incident occurred (fortunately without injury or damage).