Page 1 of 1
| Timetable validation improvements 20/11/2018 at 20:05 #113451 | |
AlexRail575
136 posts |
This is mainly because of my current foray into writing a TT for, among others, Swindid. A feature which would be helpful would be if there was some form of warning in the timetable validator for places which are required by the sim (for ARS purposes, for example, at Swindon, Rushey Platt) but which do not have a time other than the auto-generated one (the sim already seems to hold this information, as auto-generated locations are Capitalised Normally instead of being ALL CAPS). Alternatively, the sim could simply auto-generate the locations but not add in a time (since the time it does put in is, often enough, not quite exact). As a related, but independent issue, it would also be helpful if validation pointed out locations which are on a train's timetable but which do not have a time specified (the only valid case were I'd expect to see this is for run-round loco moves). Last edited: 20/11/2018 at 20:06 by AlexRail575 Reason: spelling/typo Log in to reply |
| Timetable validation improvements 20/11/2018 at 22:09 #113454 | |
Peter Bennet
5402 posts |
AlexRail575 in post 113451 said:This is mainly because of my current foray into writing a TT for, among others, Swindid.In what way is the time that is added not quite exact? If you can give me an example of what you have and what you think it should be I can have a look at the factor in the data. Peter I identify as half man half biscuit - crumbs! Log in to reply |
| Timetable validation improvements 20/11/2018 at 23:17 #113456 | |
|
AlexRail575
136 posts |
Quote:In what way is the time that is added not quite exact?This is often off by up to a minute. It doesn't always seem to account for the stopping pattern of the train. An example (amongst countless) is, in the specific timetable I'm writing, 1L38 (0753 Swansea to Paddington, HST Class 43). It stops at Swindon (1002H) and departs at 1004H, and the auto-generated time has it at Highworth Jn at 1005 (30 seconds later). I'm not even sure that is realistic for a non-stopping train. The time I have (from realtimetrains) is a more reasonable 1006 (90 secs. from a stand). I assume that would have been calculated from the sectional running times, available via the Network Rail data feeds. Log in to reply |
| Timetable validation improvements 20/11/2018 at 23:57 #113457 | |
|
JamesN
1608 posts |
AlexRail575 in post 113456 said:Quote:SimSig interpolates the time based on distance between the various timing points.In what way is the time that is added not quite exact?This is often off by up to a minute. It doesn't always seem to account for the stopping pattern of the train. RealTimeTrains fills in non-scheduled points with guesswork based on the times the author of the website thinks trains should take from A to B. An HST departing Swindon isn't timed at Highworth, the website just puts in a time because..... actually don't know why? Log in to reply |
| Timetable validation improvements 21/11/2018 at 02:12 #113459 | |
|
AlexRail575
136 posts |
It puts in a time because they put in a time for every single point a train passes. As I said, it wouldn't surprise me (given the number of timings points...) if the information was actually based on the SRTs as available from the NR "BPLAN" open data feed (though I don't have access to that so can't confirm). As I also said, the main reason I'm talking of this is because the stopping pattern doesn't seem to be taken into consideration, though I don't know how hard that would be to implement. Log in to reply |
| Timetable validation improvements 21/11/2018 at 05:41 #113462 | |
GeoffM
6376 posts 
|
As James said, it's interpolation so SRTs won't help. Well, you could interpolate those as well, but I'm not sure that would help a great deal when they're known to be optimistic at times. I don't know the exact method RTT uses but they're known to be often wildly out as well. You'll notice OTT doesn't bother with guessing intermediate times because of its unreliability. There are so many factors involved which include speed limits, train lengths (because a long train will take longer to clear a low speed restriction), signal placement, actual recording location, and pathing/performance/engineering allowances. Stopping is considered, especially when it comes to two trains conflicting: there is a "red signal start-up allowance" added, and the running time berth-to-berth is increased for the first pair. There are a number of other factors too but the user doesn't get to see the berth-to-berth timings or conflict resolution calculations. Highworth could be off, but the question is more "will it affect anything?" Given the broadly plain 2-track railway ahead of it, probably not. Note that trains are not scored at auto-inserted timing points - it's just to help you. On a side note, I asked Amtrak for a FoIA list of non-station times for a bunch of long distance trains. Many had none. One had a gap of four hours between two timing points. How do the dispatchers know if the train is on time at some point during that 4 hours? They don't, was the answer. Just a guesstimate - but one which is subject to large fluctuation on single lines when you can be "in the hole" (passing loop/siding) for half an hour or more waiting for the opposing train. Different world. SimSig Boss Last edited: 21/11/2018 at 05:43 by GeoffM Reason: None given Log in to reply The following user said thank you: AlexRail575 |