Great Western Coffee Shop

A very cross Mrs GTBE has just phoned me (it's always my fault when trains are late of course, as I used to work on the railway) from Pangbourne station, where she's waiting for the 1017 Pangbourne - Reading, hoping to get 1042 Reading - Waterloo.  It's 8 late.  Reason is a late running freight (464A) put in front of it at Didcot East, a common occurrence. 

There seems to be a huge gap between the aspirations to run more trains on the RL's between Didcot and Reading and the current quality of train operation.   All we need is a realistic timetable that trains (not only freights, but also Turbos which are increasingly being driven by Drivers who have been trained to creep into stations in brake step 1) can and do adhere to - is that too difficult? 


Edited by FT, N! to correct fonts

It is certainly becoming increasingly difficult due to the number of Intermodal freights that now plough up and down between Didcot and Reading - looking at Railcam now there's four freights currently between Didcot North Junction and Oxford Road Junction, Reading.  Their average speed is much greater than a stopper so it makes sense for them to go ahead of them, but I agree there are far too many times when the stopper waits at Didcot and leaves five or so minutes late - especially annoying since the train has been sat at Didcot for upwards of ten minutes anyway!  It is usually back on time from Reading due to the generous allowance at Reading station, but that is of no use to anyone only travelling as far as Reading.

Perhaps running the stopper a few minutes later from Didcot would improve the punctuality?  Though in doing that you'd increase the timetabled dwell time at Didcot even more, unless you ran them later from Oxford after the XC departures rather than before - but then you have what used to happen when they left at xx:21 and xx:51 in that the XC would often be late enough to delay the departure of the stopper from Oxford!  It's a bloody complex jigsaw, which might, ironically, be partly solved if and when EMUs run between Didcot and Reading and the Oxford leg is a shuttle DMU service!

I think part of the answer is to decrease the headways by installing four aspect signalling.  It takes an age for some of the sections to 'clear to greens' after a train has gone in front, especially after leaving Tilehurst in the down direction, and Didcot in the up direction.  Another improvement would be to remodel Didcot East Junction to reduce conflicts and get all of the XC's running on the main lines to leave more room for stoppers and freight.

Anyway, hopefully it ended well as I looks like she would have just made the connection?

Mrs GTBE made the connection thanks, but made the point that the up escalator on P15A at Reading is still closed following the accident on 6th (?) Jan.  However, all trains still stop at P15A, and everyone has to walk back to P15B to get up to the footbridge. It would be good if the trains could stop at P15B until the escalator is fixed.  Is that operationally not possible or is it just not worth ^going the extra mile^ for GWR customers?

On the RL capacity point I accept that four aspect signalling will improve things.  I agree it's a complex jigsaw, but this should be resolved at the timetabling stage and not left to the signallers to sort out.   I know the wish (Western Route Study) is to run longer freights and to avoid looping them, but I would like to see all southbound freights given recovery time at Oxford North (whether in a loop or not) and have them setting off in a precise path to get them though to Basingstoke without messing up passenger trains.

Some of them do have a layover at Oxford North Junction, but even with a precise path it just takes a few minutes delay to a XC service and they're screwed.  Passenger departures from Oxford are generally at xx:01, 07, 16, 31, 37 and 43.  The freights most often get a path through Oxford after the XC trains at around xx:19 or xx:48 (it varies the odd minute) which is pretty much as close behind the XC as they can be.  They then get a clear run to Didcot but are still (if bang on time) not going to reach there until a couple of minutes before the local stopper is due to leave Didcot.  An all that is before you factor in things coming from the east wanting to cross over onto the avoider at Didcot East Junction.  Another problem I didn't mention yesterday is the tortuously slow access into Didcot Yard from the east, where much of the station can be stopped for several minutes whilst long freights crawl in there at 5mph.

You could hold the freights on Didcot East Curve for them to run behind the stopper but that then presents them at Reading at often the wrong time to sensibly get through Southcote Junction on the equally busy stretch to Basingstoke and the SWML to Southampton.

I do wonder what the solution might be!

Get the freights to also pull the passenger carriages ?  ;) ;D

The answer may be that they are indeed timed to sit on Didcot East curve and follow the stopper, and then if necessary for pathing to sit on Reading West curve and wait for a properly-timed path to Basingstoke.  That would improve timetable robustness but would of course add journey time for freights, but is that so bad?  As I^ve said before, an extra journey time of 30 minutes for containers from Trafford Park to Southampton is no big deal when they^ll soon be on a boat doing 5 knots down Southampton Water and then 3 weeks getting to Japan or wherever.

We will not improve train performance on the Didcot ^ Reading RL^s until we stop trying to rush freights through with such tight margins that running just 3 minutes late messes up the passenger service.

There is still room for the long lifted Up Goods Loop between Didcot East Jnc and Moreton Cutting.  The current batch of electrification masts do not foul the old formation. On a permissive block at 1 mi 9 ch, you should be able to stack two or three freights in there awaiting a path. The Up Goods Loop used to extend around Didcot East Curve as well.

That's true, though I would suggest that land would be better used to part provide for the land that a grade separated Didcot East Junction would require.

The digital railway is the answer!

Haven’t had a moan for a while about Freight vs Passengers on the RL’s between Reading and Didcot, so here goes:

Yesterday evening Mrs GTBE caught the 1736 Padd – Oxford, 1D35, an HST.  Left Reading on time, got to Tilehurst 5L having been held at Reading West J to allow 455R 1657 Southampton – Trafford Park FL to precede.

Looking at RTT, 455R managed to delay another Class 1 train, 1K79 Bedwyn – Padd, for 5 minutes at Southcote J a few minutes earlier.  So a load of containers (who aren’t being met, looking forward to getting home etc) caused 10 minutes delay in the space of a few minutes to 2 Class 1 trains (full of passengers who are).

The timings for the FL are just 3 minutes in front of Class 1 trains at both Southcote J and Reading W J.  I would like to ask NR’s timetable planners if they think this is a prudent way to devise a resilient timetable ie one where the effects of out of course running can quickly be smoothed out rather than escalating.

PS The digital railway is NOT the answer!

Isn't that the freight which regularly causes a headache as it is too long for many of the loops enroute and once it gets going it has to plod on as it can't be "put inside" in many places?

I'm sure I remember a signalman telling me it had the potential to be a nightmare every evening peak.

I believe GTBE has had a moan before about this specific service as discussed here (http://www.firstgreatwestern.info/coffeeshop/index.php?topic=11873.0)  :)

455Rs actual code being 4M99

It was 4M99 on 17th January (or possibly 16th). It was "running as" 455R yesterday, and as 489U today. I'm sure there's a good reason for that.

Any code involving 3 numbers followed by a letter is simply randomised/scrambled. I believe the first number is usually correct which helps to identify the class of freight, class 4/6/7 etc.  4M99 is the actual code of this service and has been for many years.

I think the headcodes are scrambled because the FOC's are still worried about commercial sensitivities - on my internal systems it always shows correctly as 4M99, which can be a bit confusing when I
also often refer to realtimetrains, railcam and opentraintimes!

That does sound odd as a reason, not least because it clearly ain't foolin' no-one.

I'd imagined that some part of the computerised stuff needs to distinguish the permanent template path from the one invoked on a given day to run a train. So a new identity is generated each time, but all the systems that can only handle one path ID have to choose which one to keep.

That makes sense to me - not that that's of any relevance, of course. I can also just about believe that the FOCs applied pressure to have that one appear in the data feeds, even if it wasn't the initial reason for having it.

That is the reason but I would have thought they would be more concerned about the origin and destination of the train - which are clearly visible.

The origin and destinations of trains on line has killed the mystery that was available for mystery excursions too - a penalty of the information age!

I'm sure it is possibly to suppress details of some trains (eg the Royal Train and some Flying Scotsman trips), so if such mystery excursions still exist, these could be also ? ? 

Back to the main issue here.  2 points…

1   A “headway” of 3 minutes is misleading.  I don’t know what the exact length of 455R (4M99) is, or the distance to/from the signals involved, but if you calculate the time between the loco arriving at Reading West J, and the time the rear of the train clears the overlap of the next signal down the line, and then add on the time for 1D35 to get to Reading West J from a standing start from the signal it’s been waiting at, I think you’ll be left with something a lot less than 3 minutes 

2   Back in the 1980’s I was at BRB HQ (in 222 Marylebone Road, a fine building to work in).  We had many dealings with a department called Operational Research, who were developing mathematical modelling. (At this time the only computers were mainframes in large cupboards with lots of little water pipes to cool the components).  Anyway, one project that was done was the modelling of a busy junction in the SE Division where late running and regulation were causing problems.  The modelling allowed sensitivity analysis eg the effect of 1 train arriving 30 seconds late, and the time taken for the timetable to return to plan.  This informed tweaking of the timetable to minimise late running and the effects of late running, and also produced guidelines for regulation.   

I wonder if NR do anything similar to-day?

I think you will find find nobody does Operational Research or systems Analysis any more it's straight to designing the website. Hence the many rubbish systems.  You can't beat GIGO.

In terms of software, NR/TOC's use systems like this, which can prove a timetable is robust and can be set to see what impact individual delays can have on other services. 

http://www.trapezegroup.com/pdf/solution_sheets/rail-simulations-solutions-sheet.pdf

Sometimes there are just too many trains and too many potential conflicts.  Freight trains through Reading West junction have also got to be pathed through Oxford Road Junction and Southcote Junction - all three being within a few miles of each other and all three being very busy.  The layout at West Junction is now SO much better than it was though, so whilst not perfect there are far fewer delays.

I have only limited experience of timetable planning (using PROTIM), but I don’t think we ever accepted the notion that there were too many trains and that delays were inevitable.  Paths could be moved and/or retimed and recovery time added etc.

Maybe the May and December timetable changes will be an improvement.  Good that NR/TOC’s have a package like Trapeze – I just hope it actually gets used to simulate train running around Southcote, Oxford Road and Reading West Junctions, and that some time and effort is put in to minimising the probability of Freights arriving late (ie less than 3 minutes) in the first place, and then to minimising the knock on delays when they do. 

Industry Insider, I accept that the remodelling of Reading is a big improvement, but the delays saved west of Reading are mainly on the ML’s and not the RL’s

Looks as if your wish has come true. From May it is routed round Reading West curve two minutes earlier than at present. 4M99 (http://www.realtimetrains.co.uk/train/H02166/2017/05/31/advanced)

Another signal on the relief line between Tilehurst and Pangbourne would help in my opinion.  4-aspects would mean a slow accelerating freight train would allow trains behind to get moving much quicker - currently a train a Tilehurst has to wait until the one in front has got beyond Pangbourne before it gets a green signal.  If it departs on a single yellow, sighting of the next signal at Purley isn't great,so drivers understandably crawl along to find that signal has usually long cleared to green. 

A very inefficient use of capacity, and a shame given the improvements in signal spacing between Reading and Tilehurst.