Weird Train Acceleration in Nightlies

[Emperor Jake]

I"ve complained about this before...it prevents high speed trains reaching anywhere near tjeir intended speed... I hope it is fixed somehow.

[Rubidium]

Does the same consist with the same settings work (roughly) the same way in TTDPatch, the defacto validation source w.r.t. "realistic acceleration"? If it does it's a bug in the NewGRF, not OpenTTD. If it doesn't then I'd consider it possibly being a bug in OpenTTD.

Without savegames showing that the behaviour in TTDPatch is significantly different to OpenTTD there's nothing I'm going to do about this.

[Lord Aro]

I"ve complained about this before...it prevents high speed trains reaching anywhere near tjeir intended speed... I hope it is fixed somehow.

The problem of the original vehicles not being able to reach their top speed was fixed by Terkhen a while ago. However, NewGRF authors have to make sure that they tell OTTD what the air drag of their trains/road vehicles is, otherwise a standard is assumed, which is probably wrong

[michael blunck]

Does the same consist with the same settings work (roughly) the same way in TTDPatch, the defacto validation source w.r.t. "realistic acceleration"?

The problem is more subtle, as michi_cc explained a while ago in the German tt-forums, where we had this particular discussion first:

Vielleicht weil man reale Angaben nicht eindeutig in ein Spiel mit beliebigem Maßstab übernehmen kann? Zeit, Entfernung und Geschwindigkeit stehen bei TTD halt einfach in keinem nutzbaren Verhältnis. Ich habe durchaus die von dir gepostete Lasttafel probiert, aber der der einzige Freiheitsgrad (ohne noch viel gravierende Änderungen am Bewegungsmodell) der Fahrwiderstand ist, führt das bei dem dann notwendigem niedrigerem Widerstand dazu, dass so gut wie kein erkennbarer Unterschied bei den stärkeren Loks auftritt. Oder man ändert andere interne Skalenfaktoren und bekommt dann Züge, die die ganze Karte brauchen um irgendwann mal Maximalgeschwindigkeit zu erreichen.

Außerdem ist das DBSetXL ja nun auch nicht das einzige existierende NewGRF und auch die anderen Sets sollen weiterhin vernünftig spielbar sein, auch wenn diese nicht unbedingt Vorbildwerte für die verschiedenen Eigenschaften nehmen.

Und das ist es dann auch, was ich als Spielbalance bezeichne: Bei X verschiedenen Zugfahrzeugen sollte man im Spiel auch tatsächlich Unterschiede merken (so sollte z.B. eine Schnellzuglok deutlich mehr Felder brauchen, um einen schweren Güterzug auf Tempo zu bekommen, als eine dafür ausgelegte Güterzuglok). Gleichzeitig muss das ganze aber auch noch irgendwie z.b. zu den normalen Städte/Industrie-Abständen passen. Und wenn das halt nur dann zu erreichen ist, wenn bei einer Lok die maximale Last bei Höchstgeschwindigkeit anders als beim Vorbild ist, sehe ich das als akzeptablen Kompromiss an.

http://www.tt-ms.de/forum/showthread.ph ... 4#pid68334

In short: It has been changed because of "game balance" reasons.


OTOH, I doubt that, if a 3000hp/300kN engine cannot pull a 1600t train on flat ground.

regards
Michael

[NekoMaster]

Why are you guys talking about TTDpatch problems when this thread is in "OpenTTD Problem". This problem is not a newgrf problem, its a unwanted side effect of the new acceleration code in OpenTTD. I've tried a few different sets (NARS 2.0, 2cc trains, Tropical Rail Set) and all achieve the same results

[dihedral]

perhaps one possibility is that it could have been developed together with some ttdpatch guys? so the question is, do those grf's have the same effect in ttdpatch or not actually not too hard to understand...

[NekoMaster]

perhaps one possibility is that it could have been developed together with some ttdpatch guys? so the question is, do those grf's have the same effect in ttdpatch or not actually not too hard to understand...

I dont play TTDpatch, I've tried but theres some features in openttd that keep me playing with openttd.

[Rubidium]

Why are you guys talking about TTDpatch problems when this thread is in "OpenTTD Problem".

We're talking about how the NewGRFs behave in TTDPatch.

This problem is not a newgrf problem, its a unwanted side effect of the new acceleration code in OpenTTD. I've tried a few different sets (NARS 2.0, 2cc trains, Tropical Rail Set) and all achieve the same results

If the "unwanted side effect" of the new acceleration code in OpenTTD happens for TTDPatch as well, then the NewGRFs are the problem. TTDPatch is the defacto standard for "realistic acceleration" interpretation of the NewGRF specs, as such TTDPatch's interpretation of the specs and thus TTDPatch's realistic acceleration model is what OpenTTD should try to implement as closely as possible.

Thus, if the acceleration with a particular NewGRF misbehaves (according to you) in TTDPatch, then the bug is in the NewGRF. If the acceleration in OpenTTD misbehaves, but does not misbehave in TTDPatch, then the bug is in OpenTTD. Given that the suggestions for OpenTTD's acceleration model changes came from (IIRC) the realistic acceleration developer of TTDPatch, I'm leaning towards blaming the NewGRFs for relying on OpenTTD's previously flawed "realistic" accelation model.

Yes, you read it right: the previous "realistic" acceleration of OpenTTD was flawed.

[dihedral]

I dont play TTDpatch, I've tried but theres some features in openttd that keep me playing with openttd.

it's more of a "Test" than "Play"

[NekoMaster]

I think the new "realistic" acceleration is flawed in that locomotives that have the power and traction still can't reach their max speeds when hauling a reasonable load of cargo or passengers. Like I've said already, I can under stand if a train accelerated much slower but the new code makes trains unable to reach their max speed when they should be able too. It appears though that the new code is based more on horse power then traction alone because even when I tried some high traction locomotives, the power apparently wasn't enough to get full speed.

[dihedral]

Without savegames showing that the behaviour in TTDPatch is significantly different to OpenTTD there's nothing I'm going to do about this.

Edit: as i am not to simply quote if i cannot add to what had been said, following is what i had intended to communicate with quoting the above:

as mentioned above, a savegame of TTDPatch and OpenTTD are required in order to clarify more precisely what is causing this behaviour and in order to find out if it's a bug or not. I would love to see this issue resolved (either way) which is why i am refering to the quoted text, with the hope that Mr. Someone runs a set of Tests in both games and uploads saves to this thread. (Someone does not refer to any specific person but to any follower of this thread on a Windows based OS and keen to help).

[NekoMaster]

Oh my god...

[dihedral]

Oh my god...

people in the OpenTTD communit usually call me 'dih' or 'dihedral' - you are welcome to do the same

[NekoMaster]

What the hell does TTDpatch have to do with this? I've already stated a few sets that exhibit the same behavior in recent nightlies, they wont reach their full speed when the should be able to with the stats of the locomotive(s) pulling the train. They worked fine in Stables and older nightlies before the acceleration code was changed (being able to reach full speed after some time) but not in nightlies where the same set up wouldn't be enough to reach full speed. I don't see what TTDpatch has to do with this since Im a OpenTTD player and this is a OpenTTD problem, not a new grf problem because if it was a new grf problem then why is it that the sets I've tried do the same thing?

[Nite Owl]

@NekoMaster:

What you are failing to understand is that all of the NewGrfs you have mentioned were coded before the new Air Drag Variable was coded into the realistic acceleration aspect of the game. Either that or, given the fact that this variable has existed in TTDP for a longer period of time, the coders simply chose to ignore it. As such they are given a 'default' Air Drag Variable by the game which is, most likely, not ideal for their acceleration, power, and traction set up. When, and if, newer versions of these NewGrfs are released their respective coders will, hopefully, take this into account and code in the proper variables.

As an example: In my current game I am using the US Transition Set with Realistic Acceleration on and a Freight Multiplier of 5. All of my trains so far have been able to reach their top speed or at least come fairly close to it. In previous games, prior to the addition of the new Air Drag Variable, I had always set the Freight Multiplier to 11 and none of the trains in that set every had any problem reaching their top speed rather quickly. Now with the previously mentioned settings in place it actually takes some time to come up to speed which, in my opinion, is a good thing. Granted that I never run trains that are as long or as heavy as yours (10 tile maximum for me) but my point stands. If you fiddle about with your train lengths and the Freight Multiplier (run a test game) you will find a balance that you can live with until such time as the NewGrfs you use are recoded.

[NekoMaster]

My weight multiplier is set to 1, and although NARS 2.0 and the tropical rail set aren't as up to date, the 2cc set is updated often, as well as UKRS2 (Project 1000) Also, this isn't a matter of my trains not getting top speed as fast as they use to, its that they dont reach top speed at all anymore unless I put more locomotives then needed.

Also, I think that air drag would effect the higher speed, lighter express trains like the TGV or the ICE more then a slower and heavy American Freight locomotive

[Michi_cc]

I think the new "realistic" acceleration is flawed in that locomotives that have the power and traction still can't reach their max speeds when hauling a reasonable load of cargo or passengers.

In TTD game scale, a 32-car train is huge, and I don't see that as an exceptional hardness when a single loco struggles with that somewhat.

It appears though that the new code is based more on horse power then traction alone because even when I tried some high traction locomotives, the power apparently wasn't enough to get full speed.

I thought you wanted realistic acceleration? With some simple real-world physics you'll see that a 3000HP/300kN train is power limited as soon as the speed exceeds about 27 km/h. More traction will not change anything at all above this speed, the train will only reach this speed a bit sooner.

-- Michael Lutz

[Alberth]

What the hell does TTDpatch have to do with this?

The goal is to have NewGRFs behave the same in OpenTTD as in TTDPatch under the realistic accelaration setting.
The reason is that TTDPatch is leading in defining how trains should behave in this case, and OpenTTD is simply following that.

Thus if OpenTTD behaves differently from TTDPatch, OpenTTD is wrong.
If OpenTTD behaves the same as TTDPatch, OpenTTD is correct.
(In the latter case, the next question is then how to fix the NewGRFs.)

I've already stated a few sets that exhibit the same behavior in recent nightlies, they wont reach their full speed when the should be able to with the stats of the locomotive(s) pulling the train. They worked fine in Stables and older nightlies before the acceleration code was changed (being able to reach full speed after some time) but not in nightlies where the same set up wouldn't be enough to reach full speed.

That code was buggy. It behaved differently from TTDPatch, and thus was not implementing 'realistic accelaration' (in the way defined by NewGRF and TTDPatch). It has been fixed (hopefully).

If your idea of 'realistically correct' is the same as what TTDPatch says 'realistic' is (ie OpenTTD now behaves differently from TTDPatch), then the OpenTTD fix was wrong, and should be corrected.
If your idea of 'realistically correct' is different from what TTDPatch says 'realistic' is, your idea of how acceleration should be is not 'realistic', but in fact a 3rd form of accelaration (next to 'original', and 'realistic' (as defined by TTDPatch)) that we had in OpenTTD by accident.
Adding a 3rd form of accelaration is then a feature request.

[michael blunck]

Boys, this has nothing to do with "air drag". In fact, there have been quite a lot of modifications to OTTD´s kinematic model in the recent months.

IIRC, this discussion started with Uwe´s post, first in the German tt-forums, reporting the well-known fact that in OTTD trains are accelerating much faster than in TTDPatch. This has been worked upon, and, concurrently, "air drag" has been introduced in the kinematic model as well.

Now, thing is that (some of) the DBXL locomotives always included factors for "air drag", so it shouldn´t be affected by this modification alone.

OTOH, with the latest modifications, some DB Set locomotives suffered in a similar way as Nekomaster did report here. We´ve been discussing those recent changes in the kinematic model and its non-satisfactory outcome in the German tt-forums, and this is what Michael Lutz (OTTD developer) had been written in reply:

[...] Possibly, because one cannot introduce real data into a game with arbitrary scales? Time, distance and speed in TTD are out of all proportion. By all means, I´ve tried your tables, but because the only available degree of freedom (w/o introducing even more serious modifications in the kinematic model) being the tractive resistance, any attempt to decrease it would result in unnoticeable differences for the more powerful locomotives. The only other solution would be a modification of some internal scaling factors, but this will result into trains needing the whole map to reach their maximum speed, eventually.

Moreover, the DBSetXL is not the only existing NewGRF, other sets should also be playable in a reasonable way, even if they´re not using prototypical values for the diverse kinematic properties.

And that´s what I´d call "game balance". You should indeed see differences for the different sets in game. E.g., an express passenger locomotive should clearly need more tiles to accelerate a heavy freight train, in comparison to a freight locomotive, designed for this task.

At the same time, the whole thing must match the normal distances of towns or industries in game. And if this can only be achieved if for a locomotive maximum load at maximum speed would be different than for the prototypical model, I see this as an acceptable compromise.

(transl. mb)

HTH


Some calculations w/r to Nekomasters problem (3000 hp, 300 kN, 1600t train):

F = P/v = 3000hp / 60 km/h[*] -> 2200 kW / 16.66 m/s -> 132 kN

[*] just a number, I have no idea about max speed for that particular engine under load.

E.g., the ancient German E91 electric freight locomotive with 2200 kW (~3000 hp), max TE = 30,000 kg (~294kN), and max v = 55 km/h was able to
haul a 3000t train with 50 km/h, and still 2750t at its max speed of 55 km/h.

@Nekomaster
Report it in http://bugs.openttd.org/

regards
Michael

[Michi_cc]

Thus, if the acceleration with a particular NewGRF misbehaves (according to you) in TTDPatch, then the bug is in the NewGRF. If the acceleration in OpenTTD misbehaves, but does not misbehave in TTDPatch, then the bug is in OpenTTD. Given that the suggestions for OpenTTD's acceleration model changes came from (IIRC) the realistic acceleration developer of TTDPatch, I'm leaning towards blaming the NewGRFs for relying on OpenTTD's previously flawed "realistic" accelation model.

To be fair, neither the old nor the current OTTD "realistic" acceleration is identical to what TTDPatch does. The current one is more similar though. Maybe we should just rename it to improved acceleration.

-- Michael Lutz