Diagramming 101

[Gathursty]

When looking at the timetables at Bere Alston the other day, it occurred to me that I haven't a clue of how such train timetables are made. It must be quite a computational feat.

I thought that the team may start timetabling one line and build up around it to eventually sort out the whole NR network but beyond that I'm not sure. Can anyone explain it?

Thanks.

 

[The Planner]

No, timetables are generally just rolled over unless there is a massive re-cast of a line. Even then it depends on what is being re-cast as to how much it affects. Think of it as a pebble being chucked in a lake, it would make small ripples that need to be dealt with for a normal timetable change, a rock for a re-cast. You just need to decide where you stop the ripples.

 

[Muzer]

I don't think there has ever been a complete recast of every timetable at the same time. So all that needs to happen when a single line is recast is to change the other lines until it all fits together, which is a lot easier than completely redoing everything. Essentially, the current way the timetable fits together has more evolved than having been designed completely from scratch.

 

[route:oxford]

Although it would be interesting to see what a sophisticated piece of software would come up with with a comprehensive dataset such as...

Stock availability, planned maintenance requirements and MTBF.
Fuel efficiency of stock
Current loadings & ticket pricings
Current timetable
Full network map

 

[Eagle]

Essentially, the current way the timetable fits together has more evolved than having been designed completely from scratch.

And this is an evolution that has been constant and ongoing since the 1830s.

 

[The Planner]

Although it would be interesting to see what a sophisticated piece of software would come up with with a comprehensive dataset such as...

Stock availability, planned maintenance requirements and MTBF.
Fuel efficiency of stock
Current loadings & ticket pricings
Current timetable
Full network map

Ha! considering current timetable software is nowhere near being comprehensive enough to do what is required now, you are a long way off that.

 

[Muzer]

Although it would be interesting to see what a sophisticated piece of software would come up with with a comprehensive dataset such as...

Stock availability, planned maintenance requirements and MTBF.
Fuel efficiency of stock
Current loadings & ticket pricings
Current timetable
Full network map

This may take some time, even with a very efficient piece of software and a very fast computer

 

[Rhydgaled]

I suppose there must be some software that checks that the times which have been typed into the timetable are compliant with linespeeds etc. and don't involve two trains occupying the same single line section at once?

 

[Tomnick]

If the software checks for conflicts, it would seem that it's not very good at it. I've not yet worked out how to get a passenger train to overtake a slower freight train in the middle of a long bit of typical double track railway, but it's usually booked to happen several times a week just on one line alone! It's usually STP'd freight or light engines that seem to cause the problems. Whatever happened to looking at the graph to check??

 

[The Planner]

I suppose there must be some software that checks that the times which have been typed into the timetable are compliant with linespeeds etc. and don't involve two trains occupying the same single line section at once?

No, the software imports the SRTs from a separate system (which all the TOCs download from) which are calculated by a variety of methods. It is assumed that the data quality is sufficient.....

It also cannot see clashes on single line or at junctions, that is what planner checks on the graphs.

 

[Tomnick]

Is it meant to be able to detect conflicts between two trains in the same direction then, either overtaking or following on an impossible headway?

 

[The Planner]

No, again, the planner is meant to do that on the graph.

 

[Tomnick]

Thanks, it's always interesting to learn a bit about how it all works, having spent many an hour trying to untangle an impossible set of planned movements . The quality of schedules produced, particularly as I said for STP freights, seems rather variable at times sadly...a lot of which seems to come down to a lack of local knowledge combined with an unwillingness to check the graph for obvious conflicts.

 

[The Planner]

A chunk of it is down to time to validate the schedule properly too unfortunately, STP can often be a case of best in the time you have.

 

[Tomnick]

I wondered whether that might be the case. It's some of the others that are frustrating - sometimes it's obvious that an effort's been made to observe the AB headway, but with only an extra minute or less rather than the two required. My favourites are the freights booked to stand at the entrance to our goods line, then cover the mile-and-a-quarter to the next box (the outlet from the goods line, which is 15mph throughout) in 1H', before immediately picking up the section times for a train running through at line speed.

 

[Rhydgaled]

If the software checks for conflicts, it would seem that it's not very good at it. I've not yet worked out how to get a passenger train to overtake a slower freight train in the middle of a long bit of typical double track railway, but it's usually booked to happen several times a week just on one line alone! It's usually STP'd freight or light engines that seem to cause the problems. Whatever happened to looking at the graph to check??

I can't think what STP is. SPT could be Standard Pattern Timetable, but I doubt that's what you mean.

No, the software imports the SRTs from a separate system (which all the TOCs download from) which are calculated by a variety of methods. It is assumed that the data quality is sufficient.....

It also cannot see clashes on single line or at junctions, that is what planner checks on the graphs.

So, is the process as follows?:

1. For each service
   1. Somebody tells the software what type of train they have, what calling points it has and when they want it to leave the origin point
   2. The software downloads the Sectional Running Times for that type of train with the specified calling pattern and calculates the time it will be at intermdiate stops
2. The software outputs a graph (is that graph as in bar/line/pie or graph as in a collection of nodes each with pointers to other nodes?)
3. A person manually checks this graph for things that are invalid, like two trains on a single track at once or one following too closely behind another

 

[Tomnick]

STP is Short Term Planning - the sort of thing that's done for engineering work or freights responding to customer demand. The graph has time along the x axis, mileage and locations up the y axis, plotting each path as a line - so it's dead easy to spot a conflict on plain line, perhaps less so with headwaters or at junctions where some interpretation might be required.

SPTs are signal post telephones for us .

 

[Eagle]

STP is Short Term Planning - the sort of thing that's done for engineering work or freights responding to customer demand. The graph has time along the x axis, mileage and locations up the y axis, plotting each path as a line - so it's dead easy to spot a conflict on plain line, perhaps less so with headwaters or at junctions where some interpretation might be required.

SPTs are signal post telephones for us .

SPT is Stockport. STP is London St Pancras International. (Witton and Carnoustie can also apply here.)

 

[306024]

Each schedule is created by inputting the traction type, timing points (both passing and stopping), and the time you want to set off. The software will calculate a schedule just like a big adding machine based on the data already loaded for each traction type.

The planner then has to validate that schedule to ensure it doesn't clash with any other. That can be done using a distance / time graph, time on the x axis and the locations are on the y axis (got beaten to that bit ). Alternatively planners may just check against a working timetable print. A graph for the approaches to say Waterloo or Liverpool St can be very difficult to read, but they are good for single lines. The software does have some conflict detection to help the planner, but skill and experience are usually the best tools.

 

[The Planner]

TPS was always sold on conflict detection, never seen it work or used.

 

[306024]

Me neither! The TOC software VoyagerPlan does have some reports that can be used. TPS promised a lot but is still limited to not much more than data processing.

In my experience compared to Network Rail staff turnover is much lower in the former NSE TOC's who tend to have the more experienced staff. Can't speak for 'up north' TOC's.

 

[The Planner]

Always been a case of "dead mans shoes" at the TOC/FOCs which is why when one normally comes up decent NR planners start forming an orderly queue and the TOC can pretty much cherry pick.

 

[Tomnick]

SPT is Stockport. STP is London St Pancras International. (Witton and Carnoustie can also apply here.)

Ah yes, those wretched three letter station codes. They mean very little to me, even BNS and EMP .

 

[High Dyke]

A chunk of it is down to time to validate the schedule properly too unfortunately, STP can often be a case of best in the time you have.

I can agree with that... Earlier this year i started complaining about poor scheduling, so i was invited to make a 'goodwill visit' to my local VSTP planner. I thought this was an office with a group of people working day in, day out. I was wrong as basically it is one person doing a whole area per shift (in my case the ECML and associated branches). A typical request was from DB for a light engine from Doncaster - Leeds, a booked departure time on the request of A.S.A.P. The planner worked hard to achieve a validated plan in less than 10 minutes for the service. Sent it off to Doncaster PSB for them to run the train only to get a phone call back from DB that they "no longer needed the trip."

So hats off to the planners for making the best of a difficult thing.

 

[route:oxford]

Ha! considering current timetable software is nowhere near being comprehensive enough to do what is required now, you are a long way off that.

There is technically no reason why it couldn't be done though.

 

[The Planner]

Probably not, but considering the amount spent on the system there is now I don't think anyone has the wallet big enough.

 

[D6975]

I know someone who used to work at Swindon back in the 80s. His job was to fit extras such as railtours, special freights etc into the existing WTT. He assured me that the job could be an absolute nightmare. It wasn't uncommon for railtours to be retimed from Bristol by up to 2 hours because a pathing hitch had come up somewhere along the route.