Trains forced to wait when clear paths are available

[harz99]

There are exceptions to every 'rule', but being routine is probably not a word that I'd use from my experience.

One consideration for replatforming trains is how many customers will be affected, i.e., those waiting on the station platform and the journey they'll need to take to get to the new platform. Not only that, but also the station staff (where applicable) that need to be informed of the change. Larger stations might only have designated dispatchers on duty at certain times and at certain platforms. Unstaffed stations bring their own problems by ensuring adequate and advance customer information rather than guessing by the fact that a train roughly at the right-time has arrived in a different platform.

Another consideration is what is to be gained by doing so. If I don't want to move customers, shall I replatform another train instead? Does that train then need to crossover using slower speed pointwork and therefore delay it by an extra 2 minutes, when by replatforming the first train I might have lost the same amount of time (if not more) by waiting for the customers to come over.

Depending on the staff hierarchy, an idea to replatform might need to go from the signaller to the signaller supervisor to the Train Running Controller to the TOC control to the station staff. All of this takes time and many decisions regarding train running need to be made right there and right then. A minute later and it's just not worth doing. In this example you were involved in, it might have been a call from the driver to the signaller that made it obvious it was where it was and therefore the chain of communication starts there. But let's remember the signaller might not be able to make that decision.

Nobody is immune to criticism, but it takes a lot to understand the full ramifications and decision making process. Some people have been doing it professionally for years and come away after making a decision to either leave it or do something brave, but it was still wrong. That's life! Nothing is generally done to annoy. In most (if not all) cases it's all done because it was judged as the best thing to do in the time that was available.

So, in summary - I don't know the specifics of your example but I hope this tries to explain a little of the decision making process that goes on.

ARS is generally very simple and sometimes the simplest things work best. It is a little more layered and there are exceptions and priorities that can be programmed in, but essentially it is just trying to do the best job it can by working out whether train 'a' can get to the next ARS area the quickest.

Notwithstanding all you say, at Peterborough 2 and 3 are either side of an island platform which would be a simple walk a few yards across, it seems that DRS/ARS simply did not recognise that 2 was available due to the train that should have occupied it having been cancelled and no train in occupation, and also for whatever reason the signaller did not override the system to route the ex Leeds into 2 and allow the late runner to pass through 3.

I really hope that the signaller not overriding the DRS/ARS in this case, or in other similar cases, is not done by them to prove a point that those systems aren't perfect, as doing that does nothing to protect services or help with public perceptions of the railways reliability.

 

[Steve Harris]

I really hope that the signaller not overriding the DRS/ARS in this case, or in other similar cases, is not done by them to prove a point that those systems aren't perfect, as doing that does nothing to protect services or help with public perceptions of the railways reliability.

That is true, but life has taught me sometimes the only way people in power listen is by proving how rubbish a system is. Besides, what's to say that the signaller wasn't busy dealing with another issue and didn't get time to spot what DRS was doing and we are both barking up the wrong tree !

At the end of the day I think we can both agree that DRS needs improving.

 

[takno]

This would in my experience have been done much the same way at Peterborough before ARS was introduced. It's infuriating to watch, but ultimately the problem was created at Doncaster. Any attempt to fix it at Peterborough is going to take some smart manual working, and has every chance of leaving both trains running late.

Added to this, a non-stop train routed through platform 2 has to slow down very significantly because of the low-speed crossover. This makes it even more likely that the overtaking will delay the Leeds train, and creates a slightly-increased risk of it not slowing down and having an incident, as happened with the Lumo train a couple of years ago.

To be perfectly honest if the fasts from Edinburgh were considered to some kind of special premium train they wouldn't be scheduled to waste 5 minutes stopping at Darlington anyway. They're regulated just like any other train, and the only difference is that on average they'll do the journey around 20 minutes quicker because of having fewer stops.

 

[Belperpete]

The next question I am now curious about is exactly how ARS works? A friend of mine has described it to me as a “simple, first-come-first-served algorithm” but could anyone explain further?

ARS is considerably more complex than first come first served. It takes account of train priorities, connections, crewing and pathing constraints, and a host of other things. BUT all these things need to be correctly programmed in. For example, if someone has entered platform 2 as the only path for that train, then it will not attempt to route it through platform 3. ARS will follow the booked paths unless it has been explicitly programmed otherwise.

ARS essentially sets routes just far in advance that a driver will always have a green aspect in front of him, so as the train passes one signal it will set the route umpteen signals in advance so that the next but one signal ahead of the train clears up from double yellow to green, and so on in a rolling fashion as the train passes each signal. Before it sets each route it will check that doing so won't conflict with a higher priority train, break a connection, or any other constraint that it has been told about. If one train is delayed such that it now conflicts with an on-time train, then there are algorithms that seek to minimise the overall delay. However, it won't generally hold a train running early unless it has been specifically programmed to. So I could envisage a scenario where if train A is running early it could have its route set into the platform before train B has got sufficiently close to be seen as a conflict, if noone has told it that train B has to come in first.

I have no direct experience of DRS, but understand that it is less sophisticated than ARS. ARS needs a lot of information that needs to be constantly reviewed and updated. DRS being a simpler system probably needs less data, and manpower to maintain that data. However, I understand that it is intended to be used with a conflict prediction system which looks further ahead than ARS does, but it is then up to the signaller/regulator to tell the DRS what to do about such conflicts. And if they don't, it will just plough on ...

There are pros and cons to each method. At the end of the day, such systems are never going to be perfect, they are only as good as the information they are programmed with, and are always going to need some signaller input. ARS probably needs the signaller to be more vigilant over what it is doing, whereas DRS probably requires the signaller to intervene more.

 

[harz99]

@takno the ex Leeds was already stopping and could have been routed into Platform 2, taken it's scheduled dwell time and proceeded after the ex Edinburgh had passed through Platform 3, little if any delay to the ex Leeds and unlikely to be any safety implications with the train already slowing for the station stop.

 

[takno]

@takno the ex Leeds was already stopping and could have been routed into Platform 2, taken it's scheduled dwell time and proceeded after the ex Edinburgh had passed through Platform 3, little if any delay to the ex Leeds and unlikely to be any safety implications with the train already slowing for the station stop.

That creates a fair amount of confusion on the platform. I know it's the same island, but the platform staff and people waiting have to have the message communicated to them that the train which just appeared behind them is theirs, and possibly more importantly that the train coming into their platform which they were expecting to stop is actually going to blast through at high speed.

Honestly the basic problem at Peterborough is that it needs two through lines, and ideally a separate up fast line all the way south.

There's always a temptation when you've got something like traksy in front of you to try to think up optimal patterns which could get your train there without delay. The problem is that you aren't seeing the real world impact of that optimization, the casual 5 minute delay from the replatform, the slow crossover, approach activated turnout, or blocking overlap. You are also inevitably optimizing for your train, even if the Leeds train is fuller or has a tighter turnaround at Kings Cross.

I'm not saying ARS and the signallers always get it right, but sometimes getting it right can just look weird on the map.

 

[endofthetunnel]

I’m not an expert on Peterborough but I think platforms 2 and 3 share the same overlap, so the arriving train would in P2 would need to stop and allow the overlap to time out before a route through P3 could be set. Then the train will not be dispatched until the signal is showing proceed after the through train has passed.
This is not achievable in the ‘standard’ dwell time and would probably add 2 minutes.

But as I’ve already said once in this thread, each of these decisions may be (or at least seem) simple when only considering 2 trains at one location in isolation, but a signaller has a lot more to consider than than, and this forum in general is doing this topic to death in the last few months.

 

[Bikeman78]

And the answer (which was given to me in the quoted thread) is DRS, which to me, needs "tweaking" to improve decision making and cut down the added delays that I have seen it cause.

I wouldn't hold your breath. The Cardiff ARS is still rubbish, seven years later.

 

[Steve Harris]

I wouldn't hold your breath. The Cardiff ARS is still rubbish, seven years later.

Well. Perhaps

I think sometimes you literally have to let a system do its own thing and when it all goes pear shaped you actually have evidence to show "the powers that be" how bad "the system"

is. And then the powers that be might actually get off their chair and actually get it sorted !

But as you rightly point out, that probably won't happen. And in the meantime people have stopped using the railways because of delays caused by "the system" (as in DRS/ARS etc).