Optimising new trains on existing signalling

[100andthirty]

In the New Tube for London thread there has been discussion about the performance of 1973 tube stock and how it had the capability to have a much higher acceleration rate which was never exploited. One of the issues identified was that LU trains and signalling systems are matched. This allows the best possible throughput for a given train performance. However it also limits performance improvements if only one part of the system is changed. (For completeness, the power system and track are also part of the system and do affect performance but will be ignored for this post).

Modern trains with solid state traction can, to a limited degree, break out of this. I’ll explain why.

Firstly, the signalling principles on London Underground work something like this – recognise though I’m not a signalling engineer!

1) The overlap for a given signal is determined by the worst case speed at which the train arrives at the signal and
2) The worst credible case distance to stop from that speed plus an allowance.

Point 1 is calculated for an empty train running on the highest voltage with the biggest wheels, and point 2 is calculated for a crush loaded train.
This is safe but standardises the fiction, albeit one erring on the side of safety, that an empty train suddenly becomes full at the signal, and also that older train would go faster on higher voltage and when empty. Most trains don’t achieve that speed, so the best capacity is not delivered

Modern trains are programmed to deliver the same performance irrespective of voltage or load, and this allows trains such as the forthcoming 2024 stock to be programmed to take better advantage of the existing signalling layout than would be allowed with the 1973 tube stock. It's also true that modern trains tend to have a much smaller, load dependent variation, in stopping distance compared with older trains.

T697 applied similar principles to optimise S stock on the existing signalling soon after it was introduced into service.

[Tom]

It might be me getting it wrong (the maths in the overlap calculation were never my strong point, and I haven't got my copy of the Overlap Working Party document to hand) but I always thought that both were calculated on an empty train, on the grounds that a crush loaded train had the better brake force and would stop faster? I appreciate this is probably over-simplifying the physics!

On the rare occasions when an older overlap formula is used (the 1955 formula), there is an assumption that the train is loaded, but not on the 1982 one (and subsequent variations of it).

Is there something in the formula that I'm missing?

[towerman]

When they were testing the overlaps on the Brixton extension the test trains were crush loaded with weights,when the testing was finished had 2 or 3 trains stopped for flatted wheels.

[zbang]

Jul 25, 2022 17:13:05 GMT Tom said:

... but I always thought that both were calculated on an empty train, on the grounds that a crush loaded train had the better brake force and would stop faster?

A loaded train will have better rail adhesion but also more momentum, the question is which increases faster with loading. I assume that the available braking force remains constant (not true if there are track brakes, but do any of the trains have them?).

[Colin]

As far as I'm aware trains are fitted with load sensors that adjust the braking. Whether that means the brake force becomes a constant regardless of passenger loading is something that's outside my scope of knowledge.

[philthetube]

only the more modern ones.

[nig]

What do you class as modern as I'm sure the 73 stock has some sort of passenger load function that effectives the brakes automatically

[t697]

All LUL passenger fleets adjust the brake force available in accordance with passenger load so as to maintain sensibly constant emergency brake rate and adhesion demand. 73TS onwards do this in response to suspension deflection or air suspension pressure thus effectively weighing the passengers. 72TS still uses mercury switch retardation controllers, which dump some brake cylinder pressure air if a preset retardation is exceeded. Thus more pressure is dropped when the train is nearer empty.

[t697]

Jul 25, 2022 17:13:05 GMT Tom said:

It might be me getting it wrong (the maths in the overlap calculation were never my strong point, and I haven't got my copy of the Overlap Working Party document to hand) but I always thought that both were calculated on an empty train, on the grounds that a crush loaded train had the better brake force and would stop faster?

In the formula, any train loading condition is supposed to stop within the calculated distance.  In the rolling stock standard a further distance shortening factor is applied so trains are specified to stop with a margin inside the overlap formula. Those who worked all this out knew to allow margins for unknowns and potential slight errors such as in surveyed gradients etc., etc.

The crush loaded train will have more brake force, but that is offset by the weight increase so that retardation remains approximately the same.

[100andthirty]

I’ve noted all the valid points made by other contributors. It’s the nature of the forum that contributions will be added to and improved as knowledge is pooled.

The key point I was trying to make was that the formula would calculate that a conventional (e.g. 1973 tube stock) train would arrive at a signal at a speed that would rarely be achieved in practice. As the modern train’s traction package is much better controlled when emulating existing stock, its performance can be programmed to arrive at the signal at or closer to the calculated speed. This will give a modest performance increase.

I also note that I neglected to mention that effects of gradient are also taken account of in overlap calculations.

[jimbo]

I presume this is why the new trains will provide 27tph Picc service on signalling that currently seems limited to 24tph.

[t697]

Some localised signalling enhancements as well I think.

[roythebus]

I sometimes wonder if all these modern enhancements are worth it or are more reliable than the seat of the driver's pants? When I was a driver at Waterloo in the early 1980s with an 8 car SUB unit fitted with only a Westinghouse brake, coming into Clapham Junction on morning rush hour, I made the usual brake application just before the road bridge. It was a dry rail, sunny day, it seemed the brake didn't bite. The platform was crowded, I stopped on the 14 car mark. Which considering it was a 12 car platform was a bit embarrassing. what I didn't take into account was the train in front was cancelled and I had a full crush load on! Fortunately I heard no more about the incident.

Fast forward a year, with the introduction of the all-singing, all dancing 508 with Westcode brake, regen brake, wheelslide protection..running into Hampton Court one autumn day, checked the speed to 15 mph for the crossover, train didn't slow down very much. The WSP operated for the length of the platform and the train ended up in the sand drag. I heard no more of that incident either. But it shows that man and machine can make mistakes.

I'm certain there's a limit to how much the system can be optimised to match trains and signalling to the time it takes to load and unload trains. Remember on the Metropolitan District Railway they were running something like 36 TPH through Charing Cross in steam days pre-1906 and that was with mechanical signalling.