Weird Train Acceleration in Nightlies

[Rubidium]

The new acceleration model was basically introduced due to http://www.tt-forums.net/viewtopic.php?f=33&t=50739 which complained that OpenTTD's acceleration was too fast. However, the graphs as plotted by petern show that OpenTTD's realistic acceleration has always had a lower top speed and more resistance while the force was roughly equal.

OpenTTD calls UpdateTrainSpeed (which basically applies the acceleration) twice per tick. Some historical digging in OpenTTD shows this has basically been always the case, so I'm wondering whether TTDPatch might be calling UpdateTrainSpeed only once per tick. Or does TTDPatch half the result of the acceleration function before applying it to the consist?

On the other hand, TTDPatch's formulae are (as far as I can see from the comments) based on Imperial units instead of SI units. So maybe that's causing the difference between the used acceleration value from both acceleration formulae?

In any case, michi_cc has changed some constants so please test the binaries at http://devs.openttd.org/~rubidium/cf/h37d5ee14/ to let us know whether this is better, but still not too fast.

[Uwe]

I just checked this binary, using the DBSetXL with freight multiplier=5 and slope steepness=5%. To me, it looked quite good. Years ago I made a weight table for DBSetXL and TTDPatch (Download here), and the trains behaved quite similar. That is, going over the limits given in those tables resulted in trains not being able to reach their top speed, staying well below the limits gave good acceleration, and running at about the maximum loads for full speed resulted in slow acceleration, but the trains reached top speed eventually. I tested this with some of the early steamers and electrics, but also with the fast BR 05, which is only capable of hauling light express trains at top speed. I did not test the other versions where problems were reported, though, so I cannot compare them.

Regards,
Uwe

[NekoMaster]

The new acceleration model was basically introduced due to http://www.tt-forums.net/viewtopic.php?f=33&t=50739 which complained that OpenTTD's acceleration was too fast. However, the graphs as plotted by petern show that OpenTTD's realistic acceleration has always had a lower top speed and more resistance while the force was roughly equal.

OpenTTD calls UpdateTrainSpeed (which basically applies the acceleration) twice per tick. Some historical digging in OpenTTD shows this has basically been always the case, so I'm wondering whether TTDPatch might be calling UpdateTrainSpeed only once per tick. Or does TTDPatch half the result of the acceleration function before applying it to the consist?

On the other hand, TTDPatch's formulae are (as far as I can see from the comments) based on Imperial units instead of SI units. So maybe that's causing the difference between the used acceleration value from both acceleration formulae?

In any case, michi_cc has changed some constants so please test the binaries at http://devs.openttd.org/~rubidium/cf/h37d5ee14/ to let us know whether this is better, but still not too fast.

Hey I just noticed now that you posted a patched binary. Using the 2cc set I tested it with a very low power high speed MU (the DB Class 601) with 6 passenger cars and so far it looks alright. I also tried a Budd RDC which also seems to be working fine, last, a SDp40F was tested with 11 5 Gen Bi Levels (each weighing 42 tonnes) and it slowly got up to its max speed of 95 MPH. Im not entirely sure if this is fixed but from what I've seen so far it looks good. I'd say go for it if your going to include this into trunk or at least make it an option in the advanced settings window

[Arie-]

I've got no computer with TTDPatch near, and am not fond of messing with computers of others, so cannot test/compare atm.

[Arie-]

Ok, small bump, first results with h37d5ee14. I've now looked at a top speed, but we want also the overall acceleration to be comparable. So I'm going to do some additional testing with the speed at several points.

greatly_improved-topspeed.png (13.65 KiB) Viewed 492 times

edit: Ok, a bit more work is needed and I have to leave now, but apparently depot speeds differ in TTDP and OTTD so results may need a bit of work.

[Arie-]

Triple posting, sorry, but this post contains three attachments:
- a screen shot with some markers with respect to achieved speeds.
- the ttdp save game.
- the ottd save game.
Apparently the acceleration in TTDP is now greater, however, top speeds are better in OTTD, that is, if my findings are correct. This can be seen in the lower line where, in TTDP acceleration was faster until 90mph, then at 100mph the train did not accelerate anymore. In OTTD the train kept accelerating to 115mph.

[SirkoZ]

The formulas are different, of course the acceleration is of different rate; play-testing the latest nigtly I was quite satisfied with michi_cc's latest high speed fix. Watched the 12-wagon-ed Millenniums zip by at not very hot (read: cool) 337km/h.

[DJ Nekkid]

...

See calculations here.

regards
Michael

We are currently looking at building HST lines in norway, and im probably winning an argument with a guy But there is one thing im wondering about, how the max TE affects the needed power. I.e. i might _need_ 50kn to go on a saied hill on a saied speed, but it might have, say, 150kn available at the saied speed. This so i can make a somewhat ok guesstimate on powerconsuption and similars

[NekoMaster]

...

See calculations here.

regards
Michael

We are currently looking at building HST lines in norway, and im probably winning an argument with a guy But there is one thing im wondering about, how the max TE affects the needed power. I.e. i might _need_ 50kn to go on a saied hill on a saied speed, but it might have, say, 150kn available at the saied speed. This so i can make a somewhat ok guesstimate on powerconsuption and similars

Well power isn't everything, the higher the TE the faster it will accelerate, and IRL, the more TE the less chance of wheel spin