• Our new ticketing site is now live! Using either this or the original site (both powered by TrainSplit) helps support the running of the forum with every ticket purchase! Find out more and ask any questions/give us feedback in this thread!

Determining performance allowances

Status
Not open for further replies.

MatthewE707

Member
Joined
17 Apr 2025
Messages
157
Location
Essex
I’m obviously aware of the purpose of performance allowances in timetables, but I was wondering, when planning timetables, how do you know where to put them?

I saw in another thread an example of allowing for a steep gradient on a hill, but surely this would be the same for every train on that line, and I thought performance allowances were specific to each service?
 
Sponsor Post - registered members do not see these adverts; click here to register, or click here to log in
R

RailUK Forums

Joined
9 Dec 2023
Messages
584
Location
High Wycombe
Not an expert but aren’t they sometimes used if the sectional running time has been rounded down to the nearest half minute so it’s just there to actually make it achievable?
 

The Planner

Veteran Member
Joined
15 Apr 2008
Messages
19,658
I’m obviously aware of the purpose of performance allowances in timetables, but I was wondering, when planning timetables, how do you know where to put them?

I saw in another thread an example of allowing for a steep gradient on a hill, but surely this would be the same for every train on that line, and I thought performance allowances were specific to each service?
Depends if you are narrowing it down to actual performance time or diamond <> time. Its normally specified by the operator approaching busy locations, its not normally a NR construct. A steep gradient wouldn't attract any allowance, it should be in the running time unless there is a loop or slower speed junction at the bottom of it, and that would be an adjustment allowance or squiggle time {}. They are in the TPRs.
Not an expert but aren’t they sometimes used if the sectional running time has been rounded down to the nearest half minute so it’s just there to actually make it achievable?
If a SRT is rounded down it will be becuase the previous or next one ia rounded up. SRTs shouldn't be calculated in isolation. If its deficient then it gets fixed for the next timetable unless it really breaks things by altering it. Then it can take a while to remedy.
 

zwk500

Veteran Member
Joined
20 Jan 2020
Messages
18,474
Location
Northampton
I’m obviously aware of the purpose of performance allowances in timetables, but I was wondering, when planning timetables, how do you know where to put them?
The pedantic answer to this is 'where the Planning Rules say'. If you mean 'how did they decide what to put in the rules' it's as @The Planner says.
I saw in another thread an example of allowing for a steep gradient on a hill, but surely this would be the same for every train on that line, and I thought performance allowances were specific to each service?
That would normally be the runtime itself, not performance allowance. Performance allowances are typically applied approaching big junctions or busy stations, particularly after quite long stretches without an allowance.
Not an expert but aren’t they sometimes used if the sectional running time has been rounded down to the nearest half minute so it’s just there to actually make it achievable?
The SRTs should be rounded to account for this. Occasionally SRTs are found to be insufficient but usually Adjustment Allowance {} is added not performance allowance to make it clear to planners the time is there for actual runtime not to protect the timetable. Similarly when terminating on a half-minute, SRTs are rounded UP to the next whole minute with adjustment not performance allowance (usually).

There are also (IIRC) one or two places where the rules specify that performance allowance approaching a terminus can be replaced by advertising the arrival a minute later than booked if required.
 

Spartacus

Established Member
Joined
25 Aug 2009
Messages
3,809
The SRTs should be rounded to account for this. Occasionally SRTs are found to be insufficient but usually Adjustment Allowance {} is added not performance allowance to make it clear to planners the time is there for actual runtime not to protect the timetable. Similarly when terminating on a half-minute, SRTs are rounded UP to the next whole minute with adjustment not performance allowance (usually).

On BIG difference with Adjustment over Engineering, Pathing and Performance allowances is that the latter 3 are all considered recoverable in one way or another (and often perhaps misleadingly are given the name 'recovery time', making people think they're all about performance), whereas Adjustment isn't, as it might say be to allow for a slower speed when approaching a diverging junction, or coming off a branch.

Importantly as a result Adjustment allowances don't show in a TRUST schedule as they're effectively part of the section running time, but the others do. Those 3 are then expected to be recoverable if the train is running late, so for example if a train is 4 late at one timing point, and 4 late at the next but had minutes of various 'recovery times' between the two it would have lost time against what it was expected to do.

That sometimes confuses people, but it's not half as bad as what it does to many computer based signalling systems such as ARS. It'll see the 3 allowances as recoverable even if the train is on time and can trick it into routing a train out at a junction early if it has, most commonly, pathing time approaching it. That's why planners should to be aware of how an area's signalled and use stops at junctions (even a . (dot) stop- effectively zero minutes will do) instead of pathing in ARS areas, though adjustment can sometimes be used to a similar effect.
 

zwk500

Veteran Member
Joined
20 Jan 2020
Messages
18,474
Location
Northampton
That sometimes confuses people, but it's not half as bad as what it does to many computer based signalling systems such as ARS. It'll see the 3 allowances as recoverable even if the train is on time and can trick it into routing a train out at a junction early if it has, most commonly, pathing time approaching it. That's why planners should to be aware of how an area's signalled and use stops at junctions (even a . (dot) stop- effectively zero minutes will do) instead of pathing in ARS areas, though adjustment can sometimes be used to a similar effect.
Agreed, although dot stops are generally discouraged and you can have issues if there's no SRTs in for stopping at a junction, among other problems with the relationship of planning to ARS.
 

Harpo

Established Member
Joined
21 Aug 2024
Messages
3,789
Location
Newport
From days of hand written timetabling/typed timetabling, 40 years plus, ‘Recovery’ was always an engineering allowance intended to cover TSRs, single line working etc.. It was the civil engineers allowance that they were expected to work within on a planned basis. While Rules Of The Route dictated the allowance and its placement, some TPOs allowed it to be moved within the recovery section in lieu of pathing time.

Pathing was not expected to be recoverable as it was simply the time added to maintain correct spacing under restrictive aspects. But then passenger was first on the graph so rarely had pathing time with freight ahead.
 

Class15

Established Member
Joined
30 Dec 2021
Messages
4,111
Location
North London or Mildmay line
The SRTs can be completely broken sometimes. For example, between Gospel Oak and Hampstead Heath, the SRT is 90 seconds, but it always takes at least 2 minutes. Same for Hampstead Heath to Finchley Road & Frognal too. Perhaps there is a speed restriction (although I haven’t noticed any signs)?
 

Spartacus

Established Member
Joined
25 Aug 2009
Messages
3,809
Agreed, although dot stops are generally discouraged and you can have issues if there's no SRTs in for stopping at a junction, among other problems with the relationship of planning to ARS.

Is TPS STILL not able to add the missing ones? I remember the days of TrainPlan when you could add them based on existing ones.
 

The Planner

Veteran Member
Joined
15 Apr 2008
Messages
19,658
The SRTs can be completely broken sometimes. For example, between Gospel Oak and Hampstead Heath, the SRT is 90 seconds, but it always takes at least 2 minutes. Same for Hampstead Heath to Finchley Road & Frognal too. Perhaps there is a speed restriction (although I haven’t noticed any signs)?
As I said above though, sometimes there are issues that are would cause so much ripping up of the timetable that there is little you can do. You may have to wait until a recast until you can do anything. If its 30 seconds out, its likely not ever going to trigger a delay incident. There are also plenty of cases over the years where prodessional driving policies have caused SRTs that were fine to be not achievable anymore.

== Doublepost prevention - post automatically merged: ==

Is TPS STILL not able to add the missing ones? I remember the days of TrainPlan when you could add them based on existing ones.
It can work them out, but the proper process is still to work them out and add to B Plan (A Plan as was). TrainPlan was different, you could add them in to work on a schedule, but if you tried to upload before the SRT was added to A Plan it would fail.
 

zwk500

Veteran Member
Joined
20 Jan 2020
Messages
18,474
Location
Northampton
Is TPS STILL not able to add the missing ones? I remember the days of TrainPlan when you could add them based on existing ones.
I'm no longer working for NR so can't say for sure, but I understand Dynamic Runtimes are still some way from being accepted and therefore the majority of planners must have a full set of SRTs to publish a train. Some planners have elevated rights so can publish trains without every SRT being populated. TPS will give you a 'best guess' based on the data it has, but the planner is responsible for determining the runtime.

It's worth pointing out that TPS has *Always* had the ability to calculate runtimes, and indeed some railways use it for this purpose (AIUI SNCF and IÉ do this), but NR doesn't have a sufficiently accurate dataset of the infrastructure and trains to do this nationally. Dynamic runtime calculation has it's own issues though - if you have a service operated by a mix of stock, and you time to the actual booked stock, you end up with your timings varying from service to service as the stock has slightly different performance characteristics.

TPS can also do full dynamic conflict detection, path identification, and performance simulation, if you set it up to do so (and have the data sufficiently accurate).

== Doublepost prevention - post automatically merged: ==

The SRTs can be completely broken sometimes. For example, between Gospel Oak and Hampstead Heath, the SRT is 90 seconds, but it always takes at least 2 minutes. Same for Hampstead Heath to Finchley Road & Frognal too. Perhaps there is a speed restriction (although I haven’t noticed any signs)?
30 seconds is hardly completely broken. There is also a process to query SRTs if there is data they're too long or too slow.
 

Harpo

Established Member
Joined
21 Aug 2024
Messages
3,789
Location
Newport
There are also plenty of cases over the years where prodessional driving policies have caused SRTs that were fine to be not achievable anymore.
Ditto dwell vs door opening/despatch policies.

Ian Walmsley (Modern Railways) seems to be a lone voice amongst industry commentators in expressing concern about these twin impacts on journey times.
 

The Planner

Veteran Member
Joined
15 Apr 2008
Messages
19,658
Ditto dwell vs door opening/despatch policies.

Ian Walmsley (Modern Railways) seems to be a lone voice amongst industry commentators in expressing concern about these twin impacts on journey times.
Agree, dwells are probably the biggest contributor now, what used to be a 30 second dwell needing 45 or a minute.
 

Cheshire Scot

Established Member
Joined
24 Jul 2020
Messages
1,676
Location
North East Cheshire
Agree, dwells are probably the biggest contributor now, what used to be a 30 second dwell needing 45 or a minute.
My cousin travelled from Warrington to Edinburgh yesterday and most of the time loast was in dwells.

When class 185s were used in the North West by TPE the smaller stations were timed alternately at one minute and a half minute, i.e. average 45 seconds, despite the door cycle from train stopping to completion of the door closing process taking 35 seconds, leaving just 10 seconds for passengers to alight and join. This discrepancy was masked by the power of the 185s which typically bettered the SRTs.
 

Horizon22

Established Member
Associate Staff
Jobs & Careers
Joined
8 Sep 2019
Messages
10,964
Location
London
Agree, dwells are probably the biggest contributor now, what used to be a 30 second dwell needing 45 or a minute.

Certainly notieable on metro operators in and around London in the peak; that extra 15-30 seconds adds up to 2-3 minutes within a few stops.
 

43074

Established Member
Joined
10 Oct 2012
Messages
2,253
Ditto dwell vs door opening/despatch policies.

Ian Walmsley (Modern Railways) seems to be a lone voice amongst industry commentators in expressing concern about these twin impacts on journey times.
Ian Walmsley is mostly the same unhinged rubbish every few months though, he writes as though the railway is still in 1985 so it's no wonder his voice is ignored.
 

zwk500

Veteran Member
Joined
20 Jan 2020
Messages
18,474
Location
Northampton
Certainly notieable on metro operators in and around London in the peak; that extra 15-30 seconds adds up to 2-3 minutes within a few stops.
I don't know if it's more vibes and anecdotal than anything with proper evidence, but particularly post-covid passenger behaviour in crowding doors seems to have slowed dwell times a lot as passengers struggle to disembark through a narrow channel, and then people won't move away from vestibules making it harder for people behind them to get on.
 

Spartacus

Established Member
Joined
25 Aug 2009
Messages
3,809
I don't know if it's more vibes and anecdotal than anything with proper evidence, but particularly post-covid passenger behaviour in crowding doors seems to have slowed dwell times a lot as passengers struggle to disembark through a narrow channel, and then people won't move away from vestibules making it harder for people behind them to get on.

It wouldn't surprise me, although I found it to be becoming a big problem long before Covid. A goose-step off the train usually cleared things pretty quickly :E First place I ever noticed it being a regular issue was at New Street maybe 25 years ago, but it happens everywhere now.
 

Harpo

Established Member
Joined
21 Aug 2024
Messages
3,789
Location
Newport
Ian Walmsley is mostly the same unhinged rubbish every few months though, he writes as though the railway is still in 1985 so it's no wonder his voice is ignored.
As someone with a lifetime career across all areas of operations, Walmsley’s writing chimes very strongly with my perception of a railway that has lost sight of delivering fast efficient services.
 

43074

Established Member
Joined
10 Oct 2012
Messages
2,253
Agree, dwells are probably the biggest contributor now, what used to be a 30 second dwell needing 45 or a minute.
That's Network Rail perceived wisdom though, there was a big drive on dwells a few years back which was mostly rebutted by train operating company colleagues as the data was flawed and they suggested solutions which would have "broken the timetable" on key bits of the network to resolve it.

The difference between DOO and a guard who is on the ball is measured in seconds. Of course the railway doesn't help itself by ordering new stock with glacial door cycles, but daft regulation policies and defensive driving probably contribute to more sub-threshold delays then dwells I would suggest.

== Doublepost prevention - post automatically merged: ==

As someone with a lifetime career across all areas of operations, Walmsley’s writing chimes very strongly with my perception of a railway that has lost sight of delivering fast efficient services.
Safety and things like T-3 performance levels tend to be prioritised more than speed these days.

It's also a railway which is carrying more passengers each year so more passengers are boarding and alighting at stations which in the past would have been quieter. Plus with e.g. a slam door EMU most of those disembarking would have been alighting before the train had stopped, no need for door release by either driver or guard on one of those!
 
Last edited:

The Planner

Veteran Member
Joined
15 Apr 2008
Messages
19,658
That's Network Rail perceived wisdom though, there was a big drive on dwells a few years back which was mostly rebutted by train operating company colleagues as the data was flawed and they suggested solutions which would have "broken the timetable" on key bits of the network to resolve it.

The difference between DOO and a guard who is on the ball is measured in seconds. Of course the railway doesn't help itself by ordering new stock with glacial door cycles, but daft regulation policies and defensive driving probably contribute to more sub-threshold delays then dwells I would suggest.

== Doublepost prevention - post automatically merged: ==


Safety and things like T-3 performance levels tend to be prioritised more than speed these days.

It's also a railway which is carrying more passengers each year so more passengers are boarding and alighting at stations which in the past would have been quieter. Plus with e.g. a slam door EMU most of those disembarking would have been alighting before the train had stopped, no need for door release by either driver or guard on one of those!
Which is exactly what I said previously, some issues cannot be solved easily, or at all. I'd like to know how any data was flawed as I expect it was collected either by people being on trains or on platforms, usually in conjunction with the TOCs.
 

43074

Established Member
Joined
10 Oct 2012
Messages
2,253
Which is exactly what I said previously, some issues cannot be solved easily, or at all. I'd like to know how any data was flawed as I expect it was collected either by people being on trains or on platforms, usually in conjunction with the TOCs.
That particular exercise was from data derived from the berth offsets for the track circuits, which aren't a particularly useful method for measuring dwells as they don't usually reflect the exact location of platforms and stopping positions. It was proven to be flawed by recording on trains and stations.
 

The Planner

Veteran Member
Joined
15 Apr 2008
Messages
19,658
That particular exercise was from data derived from the berth offsets for the track circuits, which aren't a particularly useful method for measuring dwells as they don't usually reflect the exact location of platforms and stopping positions. It was proven to be flawed by recording on trains and stations.
That was daft then and was proposed by someone that didn't understand how that would work in reality.
 

Wilts Wanderer

Established Member
Joined
21 Nov 2016
Messages
3,199
That was daft then and was proposed by someone that didn't understand how that would work in reality.

Well, yes!!! Unfortunately there are quite a lot of those types floating around within the industry. My recollection of that little episode is exactly per @43074 ’s comments and what it actually demonstrated was how inaccurate many of the berth offsets actually are/were. And the lack of appreciation that at any one signal/platform, there is only one berth offset expected to report the arrival/departure of multiple types of rolling stock, be it diesel, electric, steam or pump-trolley, each with different acceleration/braking characteristics. Of course the extrapolated dwell data was going to be nonsensical, it goes almost without saying :D

(Sorry, going for a lie down with a damp cloth on my forehead…)
 

Bald Rick

Veteran Member
Joined
28 Sep 2010
Messages
35,687
Ian Walmsley (Modern Railways) seems to be a lone voice amongst industry commentators in expressing concern about these twin impacts on journey times.

He is the lone voice amongst people who have a column in a railway magazine. Be assured he is not a lone voice.
 

Harpo

Established Member
Joined
21 Aug 2024
Messages
3,789
Location
Newport
Safety and things like T-3 performance levels tend to be prioritised more than speed these days.

It's also a railway which is carrying more passengers each year so more passengers are boarding and alighting at stations which in the past would have been quieter. Plus with e.g. a slam door EMU most of those disembarking would have been alighting before the train had stopped, no need for door release by either driver or guard on one of those!
It’s over 30 years since anyone could alight from an InterCity service before it came to a stand.

In that time the main line network’s dwell times have moved even further away from that of operators such as TfL while increasingly wanting to move towards their high capacity capabilities. TfL has hardly ignored safety along the way.
 

43074

Established Member
Joined
10 Oct 2012
Messages
2,253
It’s over 30 years since anyone could alight from an InterCity service before it came to a stand.

In that time the main line network’s dwell times have moved even further away from that of operators such as TfL while increasingly wanting to move towards their high capacity capabilities. TfL has hardly ignored safety along the way.
I've never found the dwells on the Elizabeth line or London Overground to be any quicker than anywhere else that is DOO.
 

Harpo

Established Member
Joined
21 Aug 2024
Messages
3,789
Location
Newport
I've never found the dwells on the Elizabeth line or London Overground to be any quicker than anywhere else that is DOO.
And there’s the peculiarity. Where there’s a member of staff on board dedicated to despatch (and often doing nothing else) it’s become slower than DOO by design.
 
Status
Not open for further replies.

Top