What’s ado at Walthamstow, and will it see 36 trains an hour?

There was a time when one could travel to Walthamstow on the Victoria Line and find the last few stations pretty dead. I used to travel up there on and off not long after it all opened, and off-peak one hardly saw anyone; even in the peaks it wasn’t very busy. The planners anticipated this and the signalling between Seven Sisters and Walthamstow was somewhat cut-down (2-aspect signals, at only starter and home positions, and no headway posts at which automatic trains could stop). I cannot off-hand recall maximum headways but it must have been in 3-3½ minute range and effectively meant that only one train in two (occasionally two in three) served Walthamstow, so in peaks there was actually only a train every 4½ minutes or so).

When I later worked on the line it was certainly not empty in the peaks, but usually plenty of seats were available somewhere along the train north of Seven Sisters, and in the southbound peak direction there would always have been seats at the rear (favourite destinations had the exits at the front). Well, it’s not like that now.

A recent visit in the evening peak found me on a ‘crowded’ train at Kings Cross that became more ‘crowded’ at Finsbury Park. To my surprise, not only did fewer people alight at Seven Sisters than I recall in days of yore, but those getting off were pretty much matched by an equal number getting on (not something I recall much of, in past times). Tottenham Hale saw little action and there was little more at Blackhorse Road, where I alighted as my business (trying to make some accurate timings) could not be done in conditions where I couldn’t see out of the window. I boarded the slightly less full train following and was astonished on arrival at Walthamstow to find the crowds from the previous train had not cleared the platform. It was apparent that the crowd would only just clear the escalator landing before a further train arrived. I may have arrived at an untypical moment, but I cannot escape the conclusion that Walthamstow is hugely busier than ever before and is another on my list of stations ‘under pressure’. The station still has a fixed stairway between the two escalators (at many other Victoria Line stations a third escalator has been substituted).

On checking the station usage I find annual 2-way usage at 16.7 million in 2013 has gone up by 17.8% in five years (mostly in last year) compared with systemwide increase of 11.8%, so something useful has been going on there.

I mention this in the context of suggestions that Seven Sisters reversers might be done away with, all trains (see my later qualification) to serve Walthamstow. Of course, since the new signalling was installed in recent years, the 1960s signalling capacity constraint has been removed and we already see a substantially better service over this section (I reckon it to be about 26 trains an hour (tph); this leaves roughly one train in four turning at Seven Sisters). For Walthamstow, this is far better than the 14 or so tph of a decade ago, but on the basis of my visit even the enhanced service is pretty busy and could be better.

The problem, it seemed to me, was that there are also active plans to enhance Victoria Line services to 36 tph and I had doubts about this being possible at Walthamstow where the track geometry is quite different from Brixton and there is (or was) always a 25 mph speed restriction over the crossing diamond. I have produced a mathematical model of the working here, confirmed by actual timings, and concluded that it would be hard to exceed 28 tph through Walthamstow. 30 tph might be possible if various recovery and clearance allowances were reduced (increasing unreliability), but 36 was simply not possible with the current arrangement.

However, it is perfectly obvious that LU already knows this and is more than alive to the constraints along the line. The plan is to relay the crossover during this summer, at the same time altering the track geometry to increase the speed limit. Trains approach from the south at 50 mph and I doubt if the speed limit can be lifted entirely, but a much-improved limit over the diamond should be feasible even if this results in a slight reduction into Platform 1, which is unrestricted at the moment. I have remodelled the junction, assuming a run-in speed to Platform 2 (and run-out from Platform 1) of 35 mph by way of example, and find that 32 tph is readily achievable and 34 with some reduction of recovery time, but I really do think that 36 would be a problem, especially if attempted over any sustained period.

The present service requires 39 trains. If these were all extended to Walthamstow at the present intensity of trunk service I think one extra train would be needed. I think that a 36 tph service with no short workings would need 42 trains – the fleet is 47 by the way, so a 5-train float is getting a bit thin, though some will suggest that 12 per cent spares ought to be plenty these days (spares are always calculated with reference to service trains, not fleet size). A 36 tph service with two an hour turning at Seven Sisters might save a train. These suggestions rely heavily on certain assumptions about running times and layover times achievable in practice, and with rising traffic the running times (in particular) could extend.

I would be very interested if there is more detail on all this. Clearly Seven Sisters will remain available for turning late running trains short and reforming the service and so on, but I would not be surprised to find two or three trains an hour scheduled to reverse there to give some flexibility. Before anyone asks, there is (at least at the moment) no staff train service during the peaks (the depot being remote from the railway is linked to it during the day by a 20-minute staff train service to Seven Sisters). This was once unobjectionable as all trains (and operators) were in service in the peaks and the maintenance staff shift times made no demand for a staff train at this time. I’m not sure if this is a little more inconvenient now the control centre is parked there as well.

Following an earlier blog on this subject (16 March 2013) I have also revisited Brixton. There is no doubt that the ambitious 34 tph service has settled down, but even so it is evident that Brixton is the weak link and perhaps running at the edge of what is possible. More often than not, arriving trains do not have a full speed run-in to the platforms as they are checked outside the station by late running departures. Not by much, I should add, but enough to halt the train briefly and add perhaps a couple of minutes to the running time (which has the same impact as taking a train out of the schedule). It would be risky to enhance the timetable any more without fixing this glitch as it otherwise means more trains backing up outside Brixton, creating late running rather than improving intervals in real life.

One of my visits coincided with a problem requiring a train to be held briefly at Stockwell (with no information at all made over the PA during the seven minutes I was stuck, incidentally). Watching events unfold at Brixton a few minutes later I was struck by how long trains were taking to reverse – in one case a train waited for over a minute and a half after the signal cleared, and another waited over a minute. Meanwhile, trains blocked back to Pimlico. The problem appeared related to sorting crews out as a result of a train being turned at Victoria but if this cannot be remedied very much more quickly then this then a 36 tph service will be impossible. I have anecdotal evidence that targets for prompt departures (I think 10 seconds of signal clearing) are being seriously exceeded as a matter of course. Seconds really do count.

Even so, to run a 34 tph service (or very nearly so) is very creditable. What I think we are all seeing is an organization learning effectively from first principles how to run train services at a level not seen for half a century, and it seems to be working (with persistence and enormous technological assistance which I wish we’d had when I was there). A significant feature is the service management system that can supervise the dwell time at every station, evening out gaps and maintaining timetable. At present the timetable is still quite granular. It used to be good for timings to the half-minute, but quarter minute timings came in with the new signalling. New software is available now such that the next Victoria Line timetable could be compiled to the second (I understand that were this to be the case the service control centre equipment could regulate the service to the second as well). If this were all to come to pass then my concerns about 36 tph might be ameliorated a little, and my model shows that at Brixton one might conceive a 38 tph service, but this would be very tight.

To give credit to the present organization, I suggest that running this intensity of service now is actually very much more difficult than half a century ago. I don’t want to divert onto what is a subject in its own right beyond observing that in days gone by it was only necessary to work the magic for about four hours a day and one had huge periods outside the peaks, with very light services, in which to recover anything that had gone amiss. Today, it is virtually ‘peak’ all day, the pressure is relentless and there is no quiet time in which to repair things (the off-peak service now is as good as the peak Victoria Line service was when I first knew the line!).

The weak links remain though. The main weak link is the station dwell time, a factor heavily dependent on the initiative of the station staff and the behaviour of passengers (it only takes one idiot in a hurry to delay a whole train and with 36 tph the service won’t be very forgiving). The other, it appears, is the operator changeover at Brixton where a train needs to be ready to depart as soon as the signal clears, but on the strength of three visits this is not happening. Neither of these are new problems!

Will we ever see the Victoria Line at 40 tph (echoing my earlier blog)? Well not without more trains, I suggest. Could it be done then? For a given level of traffic there may be a benefit if the additional service quantum reduces train loads at the point where loading time begins to become excessive at busy stations as that might actually reduce run times. Even so I think it would be pushing things just a tad too far at Brixton and in any case loadings will probably still increase and overwhelm any benefit, introducing delays into the bargain. More useful will be Victoria reconstruction that will shift the horrendous south-end loading of the trains more evenly and hopefully reduce excessive dwell times.

Since the original post, London Underground has announced that the work at Walthamstow will take place during August this year, requiring the line north of Seven Sisters to be closed for three weeks, a discovery that has not gone down entirely well for people who are regular users of that section. Three weeks sounds a bit excessive, but probably reflects the difficulties involved in replacing the crossover to completely different geometry. Nor is it feasible in the deep tube to bring in sections of completed track and pointwork in the same way that would be attempted in open-air sections. It would be interesting to know if this requires any alterations to the tunnels.

So there we are. Next quest is to see what I’m missing at Walthamstow that has created such an increase in demand. I had better go back and have a look upstairs next time, but I’ll still have my stop watch with me.