HS2: Will Old Oak work and what could be done to improve it?

[BayPaul]

A brand-new, multi-billion-pound high speed railway should be aiming for 100% right time and able to have many days of the year - more than not - on which that is achieved. Ask the Japanese. This is not like eking out the last bit of capacity between 1630 and 1830 out of Euston on LNR. This is about designing something to work properly. A railway working properly is one operating precisely to the timetable.



But there will always be the ones that don't, far more so than a commuter railway. There'll be the person needing the ramp. There'll be the one needing help with their bags. There'll be the one that can't read the PIS, either because they're confused or because they have a sight disability. Etc.



The dwell times have precisely nothing to do with how intertwined or not they are with the conventional railway. They simply reflect that it is not feasible to board 500 people onto a train reliably, every time, within 2 minutes.

The timetable is there to make the railway work not vice versa - in your own favourite phrase, there is no need for the tail to wag the dog!. There are places where it needs to be accurately followed, and places where there is benefit in a bit more flexibility. Following it 100% of the time just because you think it looks professional is wasteful and pointless. Just like in Clapham junction, there is no reason that London bound trains need to follow the timetable at all in OOC - arrive when they get there, probably normally a minute or two ahead of the timetable, and then depart as soon as the last passenger disembarks and the (electronic) signal is green. If the train arrives early in Euston, great, it gives a few more minutes recovery time for the next service, and means that a thousand people arrive pleasantly impressed and with an extra minute of life, rather than sitting frustrated at the platform in OOC wondering why the train is delayed.
Northbound, they can't depart early, but if the train is a minute late leaving because this departure had more than its usual share of slow passengers, there is time to catch that up before reaching Birmingham, and much better have 10% of trains very slightly late but with padding to catch up than every train wasting time.

 

[Bletchleyite]

Asians are typically very disciplined and well-behaved compared with British people - in some ways we can only really be compared with other Europeans. However I'd note that most dwells there are 2 minutes except the really quiet stations (Avanti at WFJ might be comparable there, hardly anyone uses it) and the busy ones (as OOC will be) have 3 minutes.

So I maintain my view that 2 minutes at OOC is the wrong choice.

 

[Bald Rick]

brand-new, multi-billion-pound high speed railway should be aiming for 100% right time and able to have many days of the year - more than not - on which that is achieved. Ask the Japanese.

Exactly. And the Japanese use 1-2 minute dwell times, at stations that are busier, and with two way flows.

 

[Ianno87]

Are you proposing it's operated with commuter-layout EMUs like HS1 commuter domestics? Seriously? That's heading into "not fit for purpose" territory.

I'd expect something more like an ICE (i.e. Siemens Velaro) or a souped up 80x, which are both end-doored vehicles, and slightly slower to board than Pendolinos because the vehicles are 26 rather than 24m. And on none of the artist's impressions I've seen have they been anything other than the sort of end-doored high speed EMUs you get, well, everywhere else that does high speed rail.

Another "British exceptionalism" point then.

And that's why rolling stock design and station design and passenger wayfinding and seating allocation and dispatch processes (etc etc) all form part of the solution. Because end door stock struggles when passengers are not distributed evenly across all available doors, but station design etc will help mitigate this. And then, at Old Oak, you'll have the track layout capability to fall back on in the instances that this still isn't sufficient.

If 3 minute dwells were simply accepted now, then there wouldn't be the same focus on getting all these things right.

 

[Bletchleyite]

If 3 minute dwells were simply accepted now, then there wouldn't be the same focus on getting all these things right.

Which is the wrong approach, but again typically British.

ALL those things should be designed properly. We shouldn't rush the timetable to "create focus" on operations. We need to do both. Where are those dastardly Swiss when you need them again?

 

[Ianno87]

Which is the wrong approach, but again typically British.

No it's not. As @Bald Rick notes above, it's how Shinkansen dwell times are achieved.

ALL those things should be designed properly. We shouldn't rush the timetable to "create focus" on operations. We need to do both.

Which is why you set a tight requirement now, as the 2 minute requirement will be baked into the station design from the outset and result in a better solution; 3 minutes can always be fallen back upon for the timetable at a later stage if necessary; the track layout would not prevent this.

Where are those dastardly Swiss when you need them again?

Busy running a completely different system.

 

[Bald Rick]

Which is the wrong approach, but again typically British.

ALL those things should be designed properly. We shouldn't rush the timetable to "create focus" on operations. We need to do both. Where are those dastardly Swiss when you need them again?

Ah, yes, the Swiss with their standard 2 minute dwell times on IC services.

 

[Bletchleyite]

Ah, yes, the Swiss with their standard 2 minute dwell times on IC services.

Deutsche Bahn: bahn.de - Fehler

reiseauskunft.bahn.de

shows between 1 and 11 minutes depending on the station, though admittedly the 11 is a reversal. Bern Hbf has 4 minutes, and I'd consider that more comparable with likely loadings at OOC. OK, some will alight too, but I don't think that requires a doubling in and of itself.

If the infrastructure can handle an increase to 3 that means it won't be disastrous, and this can be trialled in the early phases. But I don't think we should expect 2 to work. And I am not going to agree about MKC.

 

[Bald Rick]

Deutsche Bahn: bahn.de - Fehler

reiseauskunft.bahn.de

shows between 1 and 11 minutes depending on the station, though admittedly the 11 is a reversal. Bern Hbf has 4 minutes, and I'd consider that more comparable with likely loadings at OOC. OK, some will alight too, but I don't think that requires a doubling in and of itself.

If the infrastructure can handle an increase to 3 that means it won't be disastrous, and this can be trialled in the early phases. But I don't think we should expect 2 to work. And I am not going to agree about MKC.

Two can play at that game

Deutsche Bahn: bahn.de - Fehler

reiseauskunft.bahn.de

All 1 or 2 minutes, except for Zurich (reversal allowed for, but actually to match paths) and Bern (to enable connections from Basel and Interlaken, and the matching of the paths either side).

Having used the route a few times, some of these stations are busy for these trains. Geneva, Lausanne and Zurich airport, to name but three. None have level boarding. All have alighting and boarding. And 2 minutes is plenty.

 

[Ianno87]

Deutsche Bahn: bahn.de - Fehler

reiseauskunft.bahn.de

shows between 1 and 11 minutes depending on the station, though admittedly the 11 is a reversal. Bern Hbf has 4 minutes, and I'd consider that more comparable with likely loadings at OOC. OK, some will alight too, but I don't think that requires a doubling in and of itself.

If the infrastructure can handle an increase to 3 that means it won't be disastrous, and this can be trialled in the early phases. But I don't think we should expect 2 to work. And I am not going to agree about MKC.

What's the alternative at MKC?

e.g. the xx20 off Euston *must* call and get away in front of the xx23 (having called Watford), which *must* in turn get away from Watford in front of the xx30 Glasgow?

Seems to have stood the test of over 12 years of operation, accepting that minor over-runs to the dwell times the occur are typically absorbed further north, and in the buffer between the signalling technical headway and the planned headway. Trying to be "perfect" just gives a worse service, if it means taking trains out. Passengers prefer to have more trains, at an attractive frequency.

 

[Bletchleyite]

What's the alternative at MKC?

Probably none, though it doesn't mean I have to like it, and post HS2 it should really be packed in rather than cramming in another Tring stopper or something. But that's the old "cramming lots of trains in on old infrastructure" thing, there's no excuse for that on a completely new railway.

 

[Ianno87]

Probably none, though it doesn't mean I have to like it, and post HS2 it should really be packed in rather than cramming in another Tring stopper or something. But that's the old "cramming lots of trains in on old infrastructure" thing, there's no excuse for that on a completely new railway.

Which is not being done; hence 6 platforms being provided at Old Oak Common when 4 (on paper) would suffice for the normal service. And only 18 paths being timetabled when (on paper) 20 would be theoretically be possible. Etc. etc.

 

[Bletchleyite]

Which is not being done; hence 6 platforms being provided at Old Oak Common when 4 (on paper) would suffice for the normal service. And only 18 paths being timetabled when (on paper) 20 would be theoretically be possible. Etc. etc.

So if HS2 can do 3 minute dwells at OOC, why not do them? One minute is hardly significant in the end to end running times, yet it will all just work that bit better?

 

[Ianno87]

So if HS2 can do 3 minute dwells at OOC, why not do them?

It's about keeping the choice open in future. Old Oak will work with 3 minute dwells if it needs to, but why not try and do better now in terms of what gets baked into the station design and passenger wayfinding (the stuff designed now that can never be changed in the future).

One minute is hardly significant in the end to end running times, yet it will all just work that bit better?

By better, you mean "slower". Which is not "better" for journey times. Getting trains in, loaded, and out again efficiently, F1 pit-stop style is arguably "better".

Or re-occupation of the platform; if trains alternate platforms on a 3 minute dwell time that only gives 3 minutes 'headway' to the next train, rather than a more robust 3-and-a-bit or 4 minutes. The sooner a first train clears the platform rather than hanging around, the sooner it is available for the next train.

 

[BayPaul]

if you're building a multi-billion-pound new railway you shouldn't be cramming it absolutely full on day one

On the other hand, if you gold plate the railway too much, the business case to build it disappears. I'd much rather have an HS2 that works perfectly 95% of the time, and has a few minor delays outside that, than one that is de-scoped to say 15tph to make it up to 99% reliable, but doesn't actually get built as the business case evaporates. And I don't really see an issue with filling it on day 1, so starting to earn money back - there will still be room to grow in the future with double deck trains, and the old main lines will have plenty of released capacity that can also grow.
What would be an issue would be if it was built to a lower-capacity spec, filled on day 1, and then has to be re-fitted at enormous expense to add more paths back in, but that is not currently the case.

 

[Bletchleyite]

It's about keeping the choice open in future. Old Oak will work with 3 minute dwells if it needs to, but why not try and do better now in terms of what gets baked into the station design and passenger wayfinding (the stuff designed now that can never be changed in the future).

I'm arguing to do both. Basically, to set it up for 2 minute dwells, but then operate 3 minute dwells. It will then work exceptionally well.

 

[Ianno87]

I'm arguing to do both. Basically, to set it up for 2 minute dwells, but then operate 3 minute dwells. It will then work exceptionally well.

But then what if 3 minutes dwells are just excessive, if 2 (or even 2 and half) can routinely be achieved reliably, because everything has been designed to deliver that.

E.g. if 2 minute dwells give you 95% reliability, and 3 minutes gives you 99% is that "worth it" for the journey time consequence that affects every single passenger, or results in taking robustness out somewhere else? Surely more "padding" is better placed on the ECML/WCML/MML sections of route, where services are interacting with everything else.

 

[MarkyT]

However I'd note that most dwells there are 2 minutes except the really quiet stations (Avanti at WFJ might be comparable there, hardly anyone uses it) and the busy ones (as OOC will be) have 3 minutes.

3 minutes at only 2 stations, Taipei which is the busiest, and clearly either the only or the by far the most dominant capital city station, and Chunghua, which is mysteriously the quietest, so must be for other reasons, maybe that's just a sensible place to roll up recovery, or there's an element of junction pathing allowance required there; it is next door to the second busiest station of all, Taichung. There are also a number of 1 minute dwells on that chart too, and not always at the quietest places.

 

[Bletchleyite]

E.g. if 2 minute dwells give you 95% reliability, and 3 minutes gives you 99% is that "worth it" for the journey time consequence that affects every single passenger, or results in taking robustness out somewhere else? Surely more "padding" is better placed on the ECML/WCML/MML sections of route, where services are interacting with everything else.

My view would be yes, it is worth it.

 

[BayPaul]

I'm arguing to do both. Basically, to set it up for 2 minute dwells, but then operate 3 minute dwells. It will then work exceptionally well.

It is also very frustrating sitting on a train in a station just waiting for it to leave, when you could be a good few miles on the way to Birmingham

 

[Bletchleyite]

It is also very frustrating sitting on a train in a station just waiting for it to leave, when you could be a good few miles on the way to Birmingham

In one minute?

 

[BayPaul]

My view would be yes, it is worth it.

I would disagree. And following that arguement through - why not go to 4 minute dwells, if it moves you from 99% to 99.8% reliability.... at some point you need to stop!

== Doublepost prevention - post automatically merged: 17 Feb 2021 ==

In one minute?

Yes - 4 miles per minute surely. (and yes, I know that the train takes time to accelerate, but it needs to do that no matter what the dwell time, so effectively you are adding an extra 4 miles of track!)

 

[Ianno87]

In one minute?

At 360km/h you'd be 6km further away.

 

[Bletchleyite]

At 360km/h you'd be 6km further away.

Not exactly significant.

 

[Ianno87]

Not exactly significant.

It is when you consider what proportion of the necessary headway distance between trains that is.

 

[BayPaul]

Not exactly significant.

3% of the journey time (32 minutes to Interchange) sounds pretty significant to me.

 

[The Planner]

It isn't as though there aren't going to be PTT vs WTT differentials anyway to help with this.

 

[WideRanger]

Exactly. And the Japanese use 1-2 minute dwell times, at stations that are busier, and with two way flows.

To back this up, I would take Shinagawa station as the most like the proposed use of Old Oak Common. It's the first stop on the Tokaido Shinkansen after Tokyo, and still within the central area of Tokyo. It has two Shinkansen platforms in each direction, so trains can slightly overlap on the way in and out (a similar arrangement is in place at Ueno station in the opposite direction).

Picking a train at random: http://www.hyperdia.com/cgi/en/trai...t=traintime&diagram=187951&searchday=20210217

This shows a 1 minute dwell, at Shinagawa. JR Easts Shinkansen's are actually timetabled to the nearest 15 seconds (well, they were when I last saw a running timetable). And the published timetable will generally show the rounded-down dwell time. So, I can see how it is possible that the actual dwell time might be as much as 2 minutes. But definitely not more, unless the train is running early (which drivers try to avoid as much as being late). A look at the rest of timetable shows something similar through the day.

How do they achieve it? By having clear signage of where doors will be, having people queue for their trains, and by people generally getting a move on.

Contrast that to commuter lines where it is a bit more of a zoo, and these will invariably become up to 15 minutes late by the end of the morning peak, as trains start stacking up. You know when this is happening, when the departure screens stop showing the timetabled times, and switch to showing 1st, 2nd, etc. Normally takes around an hour after the morning peak for everything to get back on timetable.

 

[Ianno87]

To back this up, I would take Shinagawa station as the most like the proposed use of Old Oak Common. It's the first stop on the Tokaido Shinkansen after Tokyo, and still within the central area of Tokyo. It has two Shinkansen platforms in each direction, so trains can slightly overlap on the way in and out (a similar arrangement is in place at Ueno station in the opposite direction).

Picking a train at random: http://www.hyperdia.com/cgi/en/trai...t=traintime&diagram=187951&searchday=20210217

This shows a 1 minute dwell, at Shinagawa. JR Easts Shinkansen's are actually timetabled to the nearest 15 seconds (well, they were when I last saw a running timetable). And the published timetable will generally show the rounded-down dwell time. So, I can see how it is possible that the actual dwell time might be as much as 2 minutes. But definitely not more, unless the train is running early (which drivers try to avoid as much as being late). A look at the rest of timetable shows something similar through the day.

How do they achieve it? By having clear signage of where doors will be, having people queue for their trains, and by people generally getting a move on.

Contrast that to commuter lines where it is a bit more of a zoo, and these will invariably become up to 15 minutes late by the end of the morning peak, as trains start stacking up. You know when this is happening, when the departure screens stop showing the timetabled times, and switch to showing 1st, 2nd, etc. Normally takes around an hour after the morning peak for everything to get back on timetable.

And the commuter lines (correct me if I'm wrong) are largely "historic" infrastructure that has evolved organically and sub-optimally over time (like the WCML) whereas Shinkansen lines (like HS2) are designed from Day 1 as a "whole system".

 

[CW2]

Quite apart from any matters of railway operational practicality, any extension to the dwell time at OOC translates into a journey time extension. To most of us that one minute extra would be meaningless, but to those people driving a computer model which predicts income, there is a correlation between journey time and patronage. In theory a longer journey time makes the journey less attractive, so fewer people will use it. So sitting for an extra minute on every train at OOC damages the business case quite significantly.