|
Post by Jimi on Dec 31, 2021 15:59:21 GMT
There have been a number of LUL folks that have given us significant help and support over the many years we've been doing this (MSTS and TS), including Roger, Solidbond, various Control and Signals staff, and of course District Dave. What you may not know is that the accuracy of our train physics and performance is due in large part to Dave's input. This goes back almost 20 years... I was working on our D78 for MSTS at the time, trying to get the physics close to reality. MSTS has a pretty robust physics engine, as does TS, allowing definition of TE, accel and braking force, power, weight, drag coefficient, etc. plus ability to simulate all the various control systems for train management, motoring and braking, so we could simulate the train behavior very well - if we had the right values plugged into the equations. The train seemed to perform pretty good, but as I had no first hand knowledge I asked Dave if he'd give it a try if he had time. Likely the last thing he wanted to do was drive a train, having just done so all day, but he was happy to do so and I sent him the latest train files. He reported back it wasn't quite right. Accel seemed a bit quick, braking a bit too fierce, and various other observations compared to his experience with the real thing. And then he said the key thing that unlocked the solution. "I'm not positive but if I had to guess I'd say overall it feels a bit light". However I had very carefully used the correct weights for the DM/UNDM/T cars on the train so all should have been correct. And then light dawned. I hadn't allowed for any passenger weight! We had empty cars. I did a quick back of an envelope calc of 50% load at a reasonable average person weight and added that to each car, and sent the revised files over again. This time there was a longer wait for response, but worth waiting for. We now had it nailed. He noted he'd driven a few stations to get the feel and then on had simply driven as he usually would - accel and braking notches, braking points wrt scenery, platforms and so on and it all behaved as he expected. We now had a good LUL train physics model - thanks to Dave's testing and feedback.
How he was able to judge that the weight was off, purely by eye - observing the train behavior but without seat of the pants feedback, I have no idea. But he was right, and the resulting physics model was used as the core of all our trains in MSTS and has also been used with minor tweaks in TS. When modeling other stocks or locos we do make adjustments to allow for differences in pinion ratio, wheel size and car weights, but that's about it. The train consist (motored vs. un-motored units) also has en effect as you'd expect. But the core physics model works good - thanks initially to Dave. There was a further tweak to improve accuracy when we were developing the C69. At the time Solidbond was a D78/C69 I/Op and was able to get us yet more data. He mounted a GoPro-type camera so it was focused on the C69 speedo and also picked up audio as heard in the cab. He also added commentary as to what he was doing so we could understand the speedo readings. Later analysis in a video editor allowed us to determine speed against a very accurate time plot, so we could determine acceleration times for 0-5,10,15,...40 etc on level track, plus braking rates in various brake notches. This led to some fine-tuning of the TE vs. Speed physics files and EP/Rheo brake behavior. Later we were able to get some similar 73TS refurb video from Craig so we could model the impact the refurb had on the train (A/C added weight,etc.). If we've done a good job of simulating the train behavior it is due in large part to the LUL professionals who have helped us and kept us honest. As Roger once noted to me on reviewing an item: "I don't see too much wrong with it. Mind you I haven't looked closely yet..." :-)
|
|
|
Post by Roger on Dec 31, 2021 23:06:09 GMT
How he was able to judge that the weight was off, purely by eye - observing the train behavior but without seat of the pants feedback, I have no idea. In principle, quite straightforward. Speed approaching a station is usually fairly constant, if the train is very heavily loaded or exceptionally empty it will vary a smidgen. Most stations have a noticeable feature that drivers use as a braking point and a given brake application will stop the train in the correct place. Therefore, while careful note of what's happening will be needed if the train stops short (light train) or tends to over-run (heavy) and to translate that into useful data. Side note, with experience it becomes possible to watch the platform (think summer, pretty girls) rather than the track ahead and still stop in the correct place.
|
|
|
Post by aussieltfan on Jan 1, 2022 13:51:13 GMT
At the moment supermegaforce the section from Whitechapel to Upminster is in the "under consideration" pigeon hole. I wouldn't want to say "yes" and then have to recant later. The team are now focussed on extending the Piccadilly tracks in various directions. Believe it or not, most of us on the team adore the 73TS and find it one of the most enjoyable stocks to drive, so some more Piccadilly metals would achieve that aim. After that we may revisit the eastbound extension of the District tracks. One problem with that is we would have to model the C2C tracks as well, which will complicate the job, but never say never............... I like everybody else have been thoroughly enjoying driving this route. It is fantastic. I have also been creating my own scenarios using the R, C and D stock in un-refurbished livery, as this is how i remember them before emigrating. As i was looking at the route map, I couldn't help wondering how difficult it would be to complete the circle line, i(t seemed that Algate and Edgeware Rd were not that far apart on the route map) and then that little spur to Hammersmith so then we could drive the district line, circle line, and could even take a jaunt on the Hammersmith and city line from Whitechapel to Hammersmith! I also long to get to Upminster, and the Picadilly idea sounds fantastic too! Gosh with infinite time and resources perhaps we could model the whole network! Thanks for all your outstanding effort. It is much appreciated. David
|
|
|
Post by Jimi on Jan 1, 2022 14:46:27 GMT
Thank you for the very kind comments David! We would be very interested to see/run your scenarios if they're uploaded somewhere?
So far we've met our initial goal re the DR having now placed all the DR metals excepting the east end. We deliberately stopped at ERD and at placing ALD as those are the limits of DR travel. We do have all the detailed charts to extend around the rest of the Circle and the H&C to HAM, and we seriously thought about doing it. However there were rumors of someone else doing the rest of the Circle so we held off as it didn't seem to make sense to duplicate their work - if the rumor proved true. Should that prove false we will certainly reconsider.
So as Ed noted right now we're planning on Picc extensions from ACT to Northfields Dpt, and from HLJ to Rayners Lane. Maybe even extend the deep tube through the city a bit more. Out in the country will be a nice change from working in tunnels, and there's some great scenery in that area, including several main roads and the Grand Union Canal to cross over. Likewise WHI-UPM is on the cards,but as Ed noted the adjacent C2C trackwork will be a significant undertaking. We shall see. Right now we're all taking a brief rest from construction, but planning and discussion continue.
Thanks again for your kind words - much appreciated!
|
|
|
Post by supermegaforcej on Jan 2, 2022 12:58:16 GMT
Maybe even extend the deep tube through the city a bit more. I would love that especially if you would be doing Green park!
|
|
|
Post by Roger on Jan 2, 2022 18:10:24 GMT
Maybe even extend the deep tube through the city a bit more. I would love that especially if you would be doing Green park! The first place to reverse to include Green Park (originally named Dover Street) would have been Covent Garden, again a trailing crossover east of the station, however, I think the crossover has been removed. That means the next place to reverse is King's Cross. The arrangement is different here as there's a connection from the east bound Picc to the northbound Northern Line south of Euston. The Picc crossover is west of that connection and hence the station. That's an awful lot of modelling.
|
|
|
Post by Jimi on Apr 8, 2022 18:18:45 GMT
I feel the need to ask you all, especially those who are developing scenarios, a question.
We're in process of working on a few changes and tweaks to Phase III, including fixes to a few scenarios. One other thing we've been asked for is to add more portals, and we're now doing that.
For those unsure, a portal can be used to either add trains to the network from a pseudo-hidden location, and also to remove trains that are no longer required. This keeps the number of AI trains running at one time to a reasonable number so we don't over-consume resources and the scenario runs smoothly.
For those curious, there's a list of the current portals in the back of the Route Guide which documents the portal name, location and the exact train spawn point (e.g. in rear of signal Axxx). Any portal can be used for either adding or removing a train, and we've taken care in our placement so trains can be added (spawn) successfully within a track section protected by appropriate signals. You don't want another AI train (or the player) running into the back of a just-spawned train, so the signal protection is essential and tends to influence the exact portal placement.
An example of where we could usefully add portals (which has already been asked) is the Outer (upper) Circle between HSK and NHG. This would, for example, be useful to remove any Outer Circle trains that are no longer going to be seen by the player. Right now, the closest such portal is in the tunnel east of ERD. So the intent is to add portals that break up the various route sections or roads (e.g. radiating from ECT to ERD, RMD, WIM, EBY, etc.) so trains can be removed efficiently.
We're aware that a portal that easily works to remove a train can also be used to add them, with the attendant need to carefully place and test it so the spawn is successful. We can handle that and determine the specific location wrt signals, etc.
So what we'd like is suggestions as to locations (be as specific as you can - e.g. between TGN and GUN, or XXX and YYY) that you feel would be useful. Feel free to ask questions and debate the idea here too.
Thanks!
|
|