How are diagrams "built"?

[paddington]

Thanks for all the posts regarding the factors that need to be taken into account.

I was wondering how it's actually done in practice. Obviously in the past it must all have been worked out by hand, but I assume that computers are used nowadays. Are there proprietary programs that must be used (and perhaps also used in other countries, I imagine planning European train times must be a lot more complex than in GB) or is there a choice? How much of the work still needs to be done manually? Do the programs run simulations of the entire plan to make sure there are no errors and what happens if errors make it past testing?

When timetables are changed in a major way, do you rebuild everything from scratch or just make modifications to the existing timetable?

 

[edwin_m]

There are numerous pieces of rather expensive software, some doing diagramming and others doing timetabling and simulation. Timetables tend to get a major recast every few years (typically a couple of years into a franchise when a new franchisee is ready, maybe, to deliver their promises) and minor tweaks in between.

 

[whoosh]

Really interesting thread, I hadn't considered all of the factors control have to deal with.

Something I didn't mention in the thread I linked to earlier...

Another factor I've just remembered: At the time I worked at Connex, there were still some old slam door trains. If these coupled up into 8 or 12 cars, the electrical connections were via jumper cables. They were duplicated each side of the front gangway, and the ones on the platform side would be connected up when an attachment took place.

This meant that when a detachment took place, it would need to be done at a platform that was the same side of the train as when they'd attached, otherwise the Shunter would need to get a block on the adjacent line if there was one, and a third rail isolation, as he'd have to go non-platform side climbing up above the third rail.

Long Term Planning had to work things like that out, and occasionally it would affect Short Term Planning as well. When planning a timetable (using a computer system to input it called 'Voyagerplan') at main stations you input the platform number, so sometimes doing Short Term Timetables I'd have to change this, giving pathing time to allow for conflicting movements and enable use of a different platform as units had been attached that weren't usually for example.
Tonbridge springs to mind where platform 2 was bi-directional, and I remember diverting Down trains into it instead of platform 1 for this reason.

Day to day, using the the Long Term Planning timetable, in disruption this would be something Control would need to be aware of.

South Central, later Southern, and South West Trains would also have had this issue with the old slam door stock, and it seems that the class 455s and 442s which are still around, still with jumper cables, are kept out in multiple all day these days, rather than messing about constraining things in the above manner!

 

[43074]

Thanks for all the posts regarding the factors that need to be taken into account.

I was wondering how it's actually done in practice. Obviously in the past it must all have been worked out by hand, but I assume that computers are used nowadays. Are there proprietary programs that must be used (and perhaps also used in other countries, I imagine planning European train times must be a lot more complex than in GB) or is there a choice? How much of the work still needs to be done manually? Do the programs run simulations of the entire plan to make sure there are no errors and what happens if errors make it past testing?

When timetables are changed in a major way, do you rebuild everything from scratch or just make modifications to the existing timetable?

There are no diagramming systems where you push the button and watch the magic happen though, so those computer systems which do exist are more of an aid rather than being capable of doing the work for you. Even the ones which are at the more advanced end still rely on human input to specify what can and can't be done and check the output. As always with systems like that, rubbish in = rubbish out.

 

[306024]

There are no diagramming systems where you push the button and watch the magic happen though, so those computer systems which do exist are more of an aid rather than being capable of doing the work for you. Even the ones which are at the more advanced end still rely on human input to specify what can and can't be done and check the output. As always with systems like that, rubbish in = rubbish out.

Quite true. Even the simplest geography requires a whole lot of rules written into the software to stop the computer doing strange, but to a computer, logical things. It takes a lot of work to train the computer first before you start to use the output in any meaningful way.

Diagrammers who build diagrams manually are assimilating a lot of facts in their head as they work through a particular scenario. When a computer generates a set of results it’s metaphorically like being hit in the face with a custard pie. It’s an instant answer but you’ve no idea of the content without studying the result, which can be a rather boring task compared to the stimulation of actually thinking through a solution.

 

[nw1]

What would be interesting would be to find out explanations for apparently 'irregular' diagramming for no obvious reason such as ensuring that given services have appropriate stock or appropriate length.

A case in point is the SWR Waterloo diagramming before this year's timetable change,so (ISTR) the 18/19 timetable. I remember seeing a Waterloo workings pattern (posted on here somewhere) around a year ago, which showed services in and out of Waterloo during the day and there was very little regularity, many units swapping from one service to another but not consistently, e.g there wasn't a consistent pattern such as up Portsmouth slows becoming down Altons in most hours.

Furthermore, there wasn't always an obvious reason such as ensuring that 444s and 450s were on appropriate services, to provide peak capacity or 444 stock on long-distance services.

Sorry it's a bit vague but I do remember the Waterloo diagramming did seem very 'random' up to this year's change. Obviously some irregularity is necessary (to ensure appropriate stock for given peak hour services, for instance) but it seemed to go beyond that.

 

[edwin_m]

Didn't SWT's lease terms for the 444/450 include a big increase in costs if the mileage went beyond certain limits? Was there a desire to equalize mileage on different diagrams or was this addressed by putting units that were getting near the limit onto diagrams that had lower mileage?

 

[306024]

Been there before, having to manipulate unit diagrams to keep within a certain planned mileage threshold to avoid a leap in ROSCO charges. Doesn't make sensible operational planning any easier.

.....When timetables are changed in a major way, do you rebuild everything from scratch or just make modifications to the existing timetable?

The more operators you have on a certain route harder it is to do a 'clean sheet' exercise. At a practical level the operators and NR usually try to find a way through, but some of the contractual requirements, which were probably based on a different timetable, can tie one hand behind your back before you start and can lead to a less optimal timetable.

Even if you can do a clean sheet, your traincrew won't all move overnight so the manpower planning needs to be carefully managed too.