Is the current London-South Wales timetable peculiarly inefficient?

[whole diary]

[jamestheredengine]

Some idle curiosity/creativity.

How many trains would be needed to run London to Swansea half-hourly? Let's work it out. Say it takes 170 minutes up and 165 minutes down (let's have a nice hypothetical world where TfW behave co-operatively in drafting their timetable), with a 35-minute turnaround at Paddington and a 20-minute turnaround at Swansea:

170
165
35
20 +
===
390 minutes
===

Then divide by 60 to get hours:

6.5 hours

And multiply by 2tph to get trains:

13 trains

So how does this compare to what is currently used? Current operations certainly look inefficient, with most trains sitting around at Swansea for 45+ minutes, and the Cardiff short turn being slow and awkward via a siding beyond the station (we're talking roughly half an hour). How many trains do they currently need? Let's count them (note for pedants: these are technically not diagrams, as I've effectively disregarded the churn at Paddington where trains sometimes arrive a few minutes apart from different origins, and then both depart about half an hour later to the other one's destination – this doesn't have any mathematical effect, even though there is presumably some operational convenience in doing that; I've also disregarded those complications of half a train shuffling off early into Maliphant sidings, as it doesn't really make a difference):

1 (5L01 0148 Maliphant-SWA^» (Swansea - next trains) 0152; 1L01 0346 SWA-PAD^» (Paddington (London) - next trains) 0723; 1B05 0748 PAD-SWA 1034; 1L18 1122 SWA-PAD 1420; 1B19 1448 PAD-SWA 1730; 1B22 1822 SWA-PAD 2110; 1B36 2148 PAD-SWA 0046; 5B36 0230 SWA-Maliphant 0234)
2 (5L03 0251 Maliphant-SWA 0255; 1L03 0458 SWA-PAD 0744; (set swap); 1B06 0818 PAD-CDF^» (Cardiff - next trains) 1010; 5L15 1020 CDF-CDF 1037; 1L15 1041 CDF-PAD 1240; 1B16 1318 PAD-CDF 1517; 5L25 1521 CDF-CDF 1548; 1L25 1554 CDF-PAD 1742; 1B29 1818 PAD-SWA 2120; 5B29 2136 SWA-Maliphant 2140)
3 (5L05 0500 Maliphant-SWA 0504; 1L05 0528 SWA-PAD 0816; 1B07 0848 PAD-SWA 1131; 1L20 1223 SWA-PAD 1512; 1B21 1548 PAD-SWA 1834; 1L34 1922 SWA-PAD 2209; 1B38 2248 PAD-SWA 0226; 5B38 0303 SWA-Maliphant 0307)
4 (5L06 0530 Maliphant-SWA 0534; 1L06 0558 SWA-PAD 0844; (set swap); 1B08 0918 PAD-CDF 1107; 5L17 1124 CDF-CDF 1144; 1L17 1150 CDF-PAD 1341; (set swap); 1B18 1418 PAD-CDF 1607; 5L27 1622 CDF-CDF 1649; 1L27 1654 CDF-PAD 1844; 1B31 1918 PAD-SWA 2219; 5B31 2227 SWA-Maliphant 2231)
5 (5L08 0600 Maliphant-SWA 0604; 1L08 0628 SWA-PAD 0914; 1B09 0948 PAD-SWA 1235; 1L22 1322 SWA-PAD 1609; (set swap); 1B24 1648 PAD-SWA 1933; 5B24 1946 SWA-Maliphant 2027)
6 (5L09 0630 Maliphant-SWA 0634; 1L09 0657 SWA-PAD 0944; 1B10 1018 PAD-CDF 1213; 5L19 1222 CDF-CDF 1250; 1L19 1254 CDF-PAD 1444; 1B20 1518 PAD-CDF 1709; 5L29 1716 CDF-CDF 1748; 1L29 1754 CDF-PAD 1944; (set swap); 1B34 2048 PAD-SWA 2347; 5B34 0005 SWA-Maliphant 0009)
7 (5L10 0700 Maliphant-SWA 0704; 1L10 0720 SWA-PAD 1012; 1B11 1048 PAD-SWA 1333; 1L24 1422 SWA-PAD 1714; 1B27 1748 PAD-CMN 2152; 5S27 2207 CMN-Maliphant 2300)
8 (5L11 0724 Maliphant-SWA 0728; 1L11 0743 SWA-PAD 1034; 1B12 1118 PAD-CDF 1317; 5L21 1322 CDF-CDF 1345; 1L21 1353 CDF-PAD 1541; 1B23 1618 PAD-SWA 1925; 1L36 2022 SWA-PAD 2309; 5L36 2338 PAD-North Pole 2349)
9 (3L12 0545 Maliphant-CMN 0659; 1L12 0725 CMN-PAD 1112; 1B13 1148 PAD-SWA 1430; 1L26 1522 SWA-PAD 1809; 1B30 1848 PAD-SWA 2133; 5B30 2150 SWA-Maliphant 2154)
10 (5B04 0636 West Ealing Sidings-PAD 0654; 9B04 0712 PAD-CDF 0908; 5L13 0912 CDF-CDF 0936; 9L13 0951 CDF-PAD 1144; (set swap, replaced by 5B14 1135 North Pole-PAD 1141); 1B14 1218 PAD-CDF 1406; 5L23 1422 CDF-CDF 1448; 1L23 1454 CDF-PAD 1641; (set swap); 1B26 1718 PAD-SWA 2020; 1L38 2122 SWA-BPW^» (Bristol Parkway - next trains) 2302; 5L38 2315 BPW-Stoke Gifford 2318)
11 (5C00 0434 North Pole-PAD 0439; 1C00 0523 PAD-SWA 0858; 1L14 0922 SWA-PAD 1212; 1B15 1248 PAD-SWA 1532; 1L28 1623 SWA-PAD 1914; 1B32 1948 PAD-SWA 2235; 5B32 2250 SWA-Maliphant 2254)
12 (5H02 0337 Stoke Gifford-BRI^» (Bristol Temple Meads - next trains) 0406; 1H02 0453 BRI-PAD 0626; 1B03 0648 PAD-SWA 0933; 1L16 1022 SWA-PAD 1312; 1B17 1348 PAD-SWA 1630; 1L30 1722 SWA-PAD 2014; 5L30 2030 PAD-North Pole 2036)

So the effect of turning alternate trains at Cardiff all day is to save one train. Quite frankly the Welsh Government should offer to buy GWR^▸ (Great Western Railway) an extra IET^▸ (Intercity Express Train) (and compel TfW to recast their timetable in a helpful way) – lot of bang for one's buck there.

[Hafren]

A particular annoyance is that the 'peak' through service to Swansea in the evening is for the London peak and not the local peak. It's nice not to have to wait very long later in the evening (might as well run the service if they would have to run ECS^▸ (Empty Coaching Stock) to get them back to the depot anyway), but trains are generally fairly empty. A half-hourly service all day would help with the evening peak, especially as traffic returns. Connectionally, a half-hourly core service would improve a lot of journey opportunities.

However I'm not convinced 20 minutes is enough turnaround time at Swansea. Any delay and it would lead to a part-cancellation to avoid knock-on delays, with the result of passengers being squeezed into a 2/3 car train, so I think for long-distance journeys a good 40 minutes is the right think to do... but of course if the service is half-hourly it becomes less of a problem. Operational Rules 2022 shows the minimum turnaround times as:
Swansea - 9/10 car 80X from Paddington: 30 minutes
Paddington - 9/10 car 80X from Swansea: 25 minutes

I'd also be concerned that if London service went to 2tph past Cardiff, in practice one each hour would have a very poor path. This often happens in the evening 'peak' – for example the 18:18 PAD^» (Paddington (London) - next trains)-SWA^» (Swansea - next trains) (not currently running, but returning in Jan it appears) [ https://www.realtimetrains.co.uk/service/gb-nr:L86980/2022-01-12/detailed ] ends up following a Maesteg train and gets to Swansea less than a quarter of an hour ahead of the train that leaves London half an hour later; pathing behind freight approaching Margam can also be an issue. (Some of these suboptimal paths might be partly from having to rework things following loss of the super-fasts but I think it would be an issue regardless.)

Of course, if the idea were taken seriously, a decent recast of the TFW timetable could cover some of this, but I'd quite like to see the local services increased as well and I think we'd have an issue with delay propagation with the service mix unless the infrastructure could be improved.

[grahame]

A particular annoyance is that the 'peak' through service to Swansea in the evening is for the London peak and not the local peak. It's nice not to have to wait very long later in the evening (might as well run the service if they would have to run ECS^▸ (Empty Coaching Stock) to get them back to the depot anyway), but trains are generally fairly empty. A half-hourly service all day would help with the evening peak, especially as traffic returns. Connectionally, a half-hourly core service would improve a lot of journey opportunities.

I have ... seen / noted that when I was working is Swansea and often going across during the evening from Wiltshire. Rammed after Cardiff and hourly if I went "early", half hourly and carrying fresh air if I went "late".

Asking a question "left field" and looking at pathing concerns. I wonder on this flow, and on certain others, if a good future approach might be to take 2 an hour from London to Swansea, but to the west of Cardiff alternate trains on the current stopping pattern and all stations (yes, I really mean that!) ... huge ramifications on other services along there.  Using 1 or 2 extra IETs^▸ (Intercity Express Train) over plan, sure, but isn't there a saving from not running the super fasts which, perhaps, aren't needed for a number of years?

[Rhydgaled]

Operational Rules 2022 shows the minimum turnaround times as:
Swansea - 9/10 car 80X from Paddington: 30 minutes
Paddington - 9/10 car 80X from Swansea: 25 minutes

Smeg. My idea doesn't work by those rules.

Currently* there are two 'fast' trains in the 'standard hour' between Cardiff and Swansea calling at Bridgend, Port Talbot Parkway and Neath. One of these is the the hourly GWR^▸ (Great Western Railway) Paddington service, the other a TfW service to/from Manchester.

Eastbound, the two services appear to provide a nice, neat, half-hourly pattern. As a result running a second GWR service would potentially just be duplicating the existing TfW. Westbound however, the TfW service could be following just six minutes behind the GWR, with a 54 minute gap before the next train on a section of route with 2tph.

I believe the timings of the TfW service are constrained in Manchester and by needing to cross the West Coast Main Line on the flat at Crewe. So, I thought the thing to do would be to keep the eastbound timetable but cut the current westbound GWR Swansea trains back to Cardiff and extend the current Cardiff terminaters through to Swansea in their place. Problem being you then only have around 20 minutes to turnaround in Swansea, less than the required 25.

* well, pre-COVID

Asking a question "left field" and looking at pathing concerns. I wonder on this flow, and on certain others, if a good future approach might be to take 2 an hour from London to Swansea, but to the west of Cardiff alternate trains on the current stopping pattern and all stations (yes, I really mean that!) ... huge ramifications on other services along there.  Using 1 or 2 extra IETs^▸ (Intercity Express Train) over plan, sure, but isn't there a saving from not running the super fasts which, perhaps, aren't needed for a number of years?

Short term isn't there a shortage of IETs due to the cracks issue and longer-term perhaps the super fasts will be wanted again (although by then, one hopes, maybe there will be wires to Oxford and Bristol allowing a new fleet of EMUs^▸ (Electric Multiple Unit) to be built to cover that requirement)?

As for your left-field suggestion, the TfW Manchester-Swansea service should in my view be a Regional Express serving 15 stations (Manchester Piccadilly, Stockport, Wilmslow, Crewe, Shrewsbury, Ludlow, Leominster, Hereford, Abergavenny, Newport, Cardiff Central, Bridgend, Port Talbot Parkway, Neath and Swansea). You propose an IET service serving 13 (or 15 if Miskin and Brackla are openned) in south Wales alone (Newport, Cardiff Central, Pencoed, Llanharran, Pontyclun, Pyle, Port Talbot Parkway, Baglan, Briton Ferry, Neath, Skewen, Llansamlet, Swansea). It's an interesting point to raise, but KeolisAmey (who by the way I think were completely wrong in this regard) saw fit to order a suburban train, optimised for short dwell times with wide doors, lots of standing room, few toilets etc. for the former while your suggestion would see an Intercity train performing the local stopper duties.

[Sixty3Closure]

Travelling to Carmarthen I was always surprised that there were two trains late evening from Swansea. The train crew are pretty good at going up and down the train encouraging you to get the TFW train as it got in about 15 mins earlier than the GWR^▸ (Great Western Railway). Interesting to read above why this situation arose as it always seemed a bit pointless to me.

It might make more sense to run of the services non stop Swansea - Carmarthen and the other service becomes a stopper? Always feels strange have the IET^▸ (Intercity Express Train) doing request stops at these tiny stations and I'm often the only person on the train.

[Hafren]

The evening PAD^» (Paddington (London) - next trains)-CMN is a strange one, with a 30 minute wait at SWA^» (Swansea - next trains) during which there's a connection. It has to have long enough at SWA for the split  and reversal, so I assume someone's decided that this makes it too close to the TFW train, so it just waits around for a later slot. It would seem more efficient to run a different PAD-SWA service through to CMN if one of them can be placed in a more efficient path. I wonder if SLC^▸ (Service Level Commitment) or political 'need' dictates that it has to be this one. In terms of the local provision it does help to fill one of the longer SWA-CMN gaps but it makes it very ineffective as a through 'InterCity' service.

A bit of a conundrum for the planners perhaps – it can't leave PAD much earlier as it has to be a journey that doesn't need to run a return journey from SWA, and an earlier journey might not allow for a long enough day in London – but then would a day tripper want to be leaving PAD in the evening peak anyway? (Is this aimed primarily for business trips or "minimise the changes" leisure travel...? For long-distance leisure travel it's probably more suited to non-day trips anyway.) It it runs later it might be too late for someone to want to use it, and means set and crew would get back to the depot very late. If the next one (18:15 PAD-SWA) were extended (when it starts running again) it would fill a bit of a gap in the local service, but that's the aforementioned train with a very inefficient path - long wait at Cardiff and still needing pathing time behind the Maesteg stopper.

The Kidwelly and Ferryside stops are interesting. They've just been added; they were always present in the morning CMN-PAD (which provides a good morning peak journey into SWA and general day trip opportunity beyond) but I don't remember the evening working stopping there before. In the world of request stops they are quite well-used but I find the addition of the stops to this service a bit strange – again perhaps it's about filling a long evening gap or something more political. Although there's no GWR^▸ (Great Western Railway) precedent (and therefore access agreement) for it, perhaps adding Gowerton would have been more useful!

I've sometimes wondered about that possibility of a second tph running non-stop to Swansea. Pathing would again be an issue though, and I suspect it wouldn't be busy enough off-peak to be considered justifiable, unless it's politically useful. 1tph non-stop and 1tph all-stations would mean slowing down the journey from the likes of Neath, and would create short platform issues with 10-car sets as the rear set would have no access to platforms at the smaller stations. Effectively I've just argued for splitting to 5 cars at Cardiff, but I think that would eat up capacity; splitting and joining work well on the trains that run past Swansea, as Swansea has the capacity and layout to do it – in the early days it created delays but they seem to have things working quite well now. However at Cardiff I think it would be a bit of a nightmare! An occasional stop on the Londons at Pyle and Pontyclun might be justified though – it not often enough to create a general 'slowing' of the service! They are quite big places, with railhead potential for surrounding villages, and with short enough journeys to Cardiff & Swansea to create good journey opportunities, but Pyle in particular has a fairly 'rural' off-peak service level. What's really needed, though, is a better local service... but then the capacity issue comes up again. A decent 4-track section would be the obvious solution but a huge engineering & financial burden – and I'm not convinced multiple loops are the answer as they slow things down more for the local services and need things to be running like clockwork which wold be hard to achieve with a big mix of service origins creating more delay sources.

Rather than slow down the Londons, perhaps the Portsmouths could be extended as stoppers...!

[Sixty3Closure]

I've been catching it for a couple of years and pre-pandemic it was generally very crowded to Swindon/Bristol which I guess is commuting traffic. After Cardiff it got a lot emptier and by Swansea I'm generally on my own as anyone in the know got the TFW train. The stops after Swansea always seemed pointless although I did wonder if they were included because the IET^▸ (Intercity Express Train) can't actually run at speed so do you lose a lot of time stopping at Llanelli and the like?

I generally assumed the main reason for the trains existence is to have an IET in place for the early morning run to London which is busy when I've been on it (in the pre apocalypse days anyway). It was a pleasant surprise to discover a direct train from London/Reading when I started making the journey but I've always half expected it to disappear at some point.

You could change the TFW train but then it goes on to Milford Haven (I think) and does look well used the few times I've been on it. You also have the reverse in the morning about 8am with the Manchester train and local service running pretty close together and I'm not sure you need both of them other than it spreads the passenger numbers out a bit. The staff at Carmarthen though do always try encourage me to get the Manchester train for Swansea but it just means a longer wait in the cold at Swansea for the connection.

[Hafren]

In terms of talk of stopping patterns, there are effectively two discussions going on now...!
• CDF^» (Cardiff - next trains)-SWA^» (Swansea - next trains) - where the big capacity issue is that there are a lot of different services (stopper, express from multiple origins which therefore can't be evenly-spaced, expresses covering for stoppers, freight) and a bigger total number of trains.
• SWA-CMN - where there are fewer trains in total, and a big capacity constraint is the long signalling sections, and the smaller total number of trains means everything has to chip in to cover stops.

So the considerations when deciding to run something non-stop or all-stations are different.

In terms of running non-stop, the dominant linespeed west of SWA is 75mph, and OTTOMH reduces to 60 through Llanelli, and something in the 50-60 range through Ferryside, but doesn't reduce through the other stations. So there would be some time-saving, but not as much as missing somewhere like Didcot! Also in the current slot, I think by running non-stop it would get to Ferryside (start of the longest block section) just after the Milford has cleared Carmarthen junction, so if the Milford is at all delayed a non-stop IET^▸ (Intercity Express Train) would also be delayed.

Politically there would also be an issue - "Llanelli loses daily London train"! The other consideration is that the business case for running through to Carmarthen is probably marginal, and losing any potential traffic from the intermediate stops wouldn't be helpful - especially bearing in mind that Llanelli has the biggest population on the route, although Carmarthen has its importance as a railhead. It's also worth bearing in mind the gap from that Milford to the next TFW service - so GWR^▸ (Great Western Railway) at 21:05 ish fills a gap of well over an hour. An alternative pattern (which would probably create its own logistical issues) might be for the IET to run through sooner, terminate the Milford at SWA and then use it for a Carmarthen local at 21:30 or so to plug the gap. Then run a local to Milford to connect with the IET at Carmarthen - not sure off top of head if anything's available at Carmarthen at that point to do so, but as it's gone 9pm and services are starting to thin out it might be possible to juggle.

The evening service isn't needed to balance the morning one as it returns to Swansea anyway! Incidentally, this gives us an idea of how long a non-stop run might take - https://www.realtimetrains.co.uk/service/gb-nr:L89408/2021-12-21/detailed gives us under 40 minutes.

Maybe an earlier PAD^» (Paddington (London) - next trains)-CMN would be more useful as it would complement the morning one in helping TFW with the burden of the Swansea commuters and Cardiff day-trippers, although it wouldn't necessarily be any better for long-distance travel. Maybe a bit late in the day to consider now with TFW looking to go up to 5 car (east of SWA at least) with the new stock! This would leave the issue of how to form the return SWA-PAD journey as this would still be needed that early in the evening - unless the 5 car set left in SWA would be enough at that point.

[grahame]

I was taking a look and "worrying" this issue from left field ...

Trains arrive at Swansea at :30 from Paddington and leave again at :22.   How about extending 5 carriages of a train at :40 every 2 or 3 hours to Carmarthen or Fishguard, with the previous extension arriving back into Swansea at :10 (30 minutes later) to form part of the Paddington train.

The Fishguard extension(s) to arrive and leave Fishguard Harbour while the ship from Rosslare is there ... the Carmarthen extensions to give a service to West Wales thought the day rather than at what used to be peak time for heading to / from work in the capital and business meetings.

There is a herd of elephants in the room - starting with the boat timing which allows an excellent London to Dublin daytime journey but the other direction is awful - there are 4 trains a day from Dublin to Rosslare Europort - over an 8 hour period and then there's a 16 hour gap.  Then the ship <-> train link at Rosslare was "modernised" so that the railway is now on the edge of the port and really difficult and unfriendly especially with luggage.  And the second visit of the ship to Fishguard is at an obnoxious hour that means there's nothing to London for a train to join to at Swansea.