Finally got around to trying this...
Wow, lovely piece of work! Thanks
And double-extra brownie points for making the source code available

(Anything special needed to recompile? Haven't tried yet, but I do have the cheapncheerful VS C++ EE already.)

{I know this thread is old, but it's referenced by the current version of LRA, so...}

Nobody seems to have quibbled about whether 1/4 of the unsprung mass is the right correction to use for the lumps of "stuff" hanging off the suspension.
I'm wondering though if it wouldn't be quite different at the front and back? For many cars (FZR and BF1 are just two examples) the front and rear tyres seem very different in size, and probably the hubs and brake units might have rather different mass too. So we might be seriously underestimating the unsprung mass at the rear of the car, and overestimating at the front?

Quote from tristancliffe :

But you cannot plot that graph against engine RPM. You can plot it against wheel RPM, which is the same thing as road speed (ignoring slip).

There is a lot of crap in this thread!

Em, not sure to whom the "crap" remark was directed but I'm aware that you can't plot that graph against RPM in that fashion and still be able to see the crossing points for the gears. However, it was plotted against RPM, making the horizontal axis just plain wrong for all but one of the traces plotted...

Edit: I should add two things... Firstly, I actually agree that a lot of crap ends up in threads like these, and part of this is because people don't read the first pages once the thread gets a few pages long Secondly, a few weeks ago in a thread amazingly like this one, AndroidXP wrote a beautiful summary of the whole "when to shift"/"shift light" situation... As far as I can see he got it all absolutely correct, and answered just about every issue people have raised. It would be cool if lots of people read it now before adding extra posts

Last edited by Neilser, Sat, 2 Jan 2010 18:10. Reason : Addenda...

Quote from AndRand :

And now I am convinced It even shows that gear change should occur at redline as torque is always higher on lower gear.

Nice graph, though the horizontal axis only seems to be RPM for the engine curve, and the others appear to use road speed. And indeed, for this engine, the curves don't cross so I agree that all shifts would be optimal at the redline.

Quote from Iron :

That doesn't make much sense to me. [...]
Perhaps you meant wheel speed.

Oh damn Yes, I meant road speed, argh, can't believe I checked that and still missed it. Argh and double argh Thanks for spotting the nonsensical bit!

Quote from JeffR :

Link to image of rear wheel force (in g's) versus speed for a old ZX11 motorcycle. If shift time was instant, you'd shift where the lines cross, since shifts take some amount of time, you'd shift at slightly higher speed. 1st to 2nd gear shift is at red line, then the rpm for ideal shift point decreases with the gear.
http://rcgldr.net/misc/shiftpoint.jpg

Yup, a nicer case than AndRand's, as the gearbox in his example needed a few extra cogs in it This is the situation in most LFS cars, with most setups, I think.

One thing these curves conceal btw is that the amount of time spent in each gear is dramatically different. So when tuning a box (while you need to think about the track of course) it's much more important to get it right for the higher gears, where you spend much much longer than in 1st or 2nd. So in a long straight run to max speed in a 6-speed box, you might spend a reasonably short time in 3rd, and it doesn't matter so much that you spend quite a big fraction of that time way off the power peak (i.e. big rev range, corresponding to a big fractional increase in speed while in the gear) but in 5th, you want to spend most of your time as close as poss to the power peak (either side of it, naturally) and that will correspond to a much more limited rev range, and a much smaller fractional speed increase while in the gear...
(I've never actually done the sums to compare the average engine power output over a whole acceleration run with different box setups though - must play with it sometime. Anybody recommend a good replay analyser? I think though that the setup which maximises average power will result in both the highest final speed and the fastest time over a set distance...)

Quote from AndRand :

your acceleration comes from torque, not the power which enables to deal with growing opposite forces (ie. air drag), so in proper gearing you should change gears as soon as torque is diminishing as torque on higher gear should be higher.

Well, torque and power are intimately related, but it would be a grotesque understatement to say that they are often misunderstood

Again I'm unsure here if you are talking about engine torque or torque at the wheels.

The peak acceleration in a given gear would happen at peak engine torque if it weren't for air resistance (which increases with road speed). However, the peak acceleration at a given engine speed [edit: argh, no, I meant road speed!] if you have a choice of gears will be in the gear which maximises the engine power.

Since power = torque * angular speed, the engine power peak is always at higher RPM than the torque peak. Depending on the engine, it can be slightly higher, or much much higher - it depends entirely on how steeply the torque curve falls above peak torque.

So if (as I suspect) you meant that you should change gear after the engine torque peaks, then I'm afraid you're wrong - this will be dramatically too early, so early in fact that for most cars you should be trivially able to prove it to yourself on a drag strip by reference to the "Info" tab's engine torque peak RPM value... Only if peak torque and peak power are really close together does this point become moot (don't know for sure if any LFS car has that property).

If you meant wheel torque in what you wrote, then your first statement was correct, that acceleration comes from wheel torque, but you still don't change gear when the wheel torque diminishes, as the wheel torque in any given gear is simply proportional to the engine torque.

Hope that was mostly clear

Last edited by Neilser, Fri, 1 Jan 2010 12:17. Reason : thinko

Quote from RevengeR :

Those 1-2 tenths of a second makes all the diference. Just think of how many upshifts you do on a track, let's say FE4, or AS5. Even 0.05 sec in each upshift makes quite a lot on a whole lap.

Ah, sorry, wasn't clear enough - I meant only a tenth or two over the entire drag run (4 upshifts in a 5-speed box). And a tenth or two is probably similar to the error margin anyway, since I didn't do very many runs; I was simply trying to reassure the OP that the variation would be pretty modest in general terms. But yes, in the right places on certain tracks, I'm not at all sure that the effect would be entirely negligible for a full 500 rpm variation in shift-point.
If I could work out a really easy way to do a more precise test, I'd do it. Maybe a replay analyser (haven't downloaded any yet) would be the right way to do it...

One thing I did notice while replaying various replays carefully to see what the acceleration was in the various gears before and after a shift - if you don't get too close to the redline and shift with your foot kept flat to the board, the engine revs keep rising while the clutch (auto or other) is down, and then when the clutch is released the extra energy the engine has built up while the car was decelerating compensates somewhat for the slight speed reduction during the shift. (Of course, even if you do hit the redline, the revs will still have to drop significantly when the higher gear is hooked up by the clutch, but the gain is a bit smaller then...)
You can clearly see a couple of MPH increase as the clutch engages, with much higher (like 1.5x or even 2x) acceleration than when the revs have fallen back and the clutch is no longer slipping. (Of course, this does warm your clutch up! )

Quote from AndRand :

heh, just the same the car havent changed if you want to be superaccurate get replay analyzer to see max torque vs. rpm. The indicator was nothing else but this rpm.

If you mean torque at the crankshaft then no, the shift point is massively higher than this.

The power peak in the XFG is about 7000 rpm, while the torque peak is more like 5400 rpm. (Straight from the info screen, I don't have the torque/power curves either.)

But the shift point depends on the gear you are changing up to, so an accurate shift light is NOT simply rpm-dependent.

I was driving a close-ratio set earlier, where the 3rd to 4th and 4th to 5th shift points were (in my opinion) not very high at all - maybe 7500 rpm or even less. The basis I use is to try to match the power in the gears (basically the same as the torque to the wheels * wheel speed), so that if I change from 4th gear 7300 rpm to 5th gear 6700 rpm, I'm probably vaguely matching the power I had in 4th to the power I will have in 5th. But without a torque curve, you can't get it spot-on...

I did some tests yesterday, just playing about (in FXO mostly), and it really didn't make much difference to the drag-race timing (a tenth or two) if I used shift-up points differing by 500 rpm or more

In most cars you won't be too far off the optimum by hitting or nearly hitting the red line.
The MRT however, is another thing entirely! (But luckily it does still have a shift light.)

Major intermittent lag problems... Sometimes at peak hours, sometimes not. I've already checked that the problems happen even when nobody else in the house is using the link (ADSL, 4Mbps).

I've had my ISP (Plusnet) on the case, and they have attempted to add LFS to their traffic-management systems, using packet-captures I've supplied. However, they commented that it was the worst game they'd seen in terms of difficulty to recognise the game sessions.
Traffic management is used to make sure that packets for things like streaming video and online gaming get especially well treated, while lower priority things (e.g. ftp) can get held up.

Anyway, things clearly aren't working reliably yet, perhaps partly because LFS servers out there are using many different ports. It seems to be better on servers which use the standard port 63392... (Been studying rather a lot of pcap files :-/ ...)

Anyway, if people have advice, I'd appreciate it.
In particular, since traffic management by ISPs is probably fairly normal these days, how do other ISPs treat LFS? Should I change ISP?

Apologies if this is a dumb question, but what is "lfs message display"? I've not heard of it...

Becky, that old adage comes to mind: "Never attribute to malice that which can be adequately explained by stupidity".

I don't know you so I take at face value that you've had plenty of scrapes, including some seriously extreme ones (SMG fire), but you must realise that this makes you pretty unusual.

Most people can go through their lives and never fall foul of taking the approach that someone driving badly behind them probably isn't trying to attack them, but is in fact just a shit driver.

I guess it's probabilistic. For an average member of the public, taking the extreme "all guns blazing" response to disconcerting situations will in hindsight turn out to be an over-reaction almost all of the time. So the outcome (difficult explanations to the police about it, damage, injuries etc.) will generally be worse than it would have been if you were a more optimistic soul.
But every now and then, an average member of the public gets killed where they might have survived if they had taken the worst-case assumption and gone off the extreme end in their offence=defence reaction.

Everybody makes their own assessment about the relative probabilities of the horror-movie scenario vs. the normal-day scenario... For most people, they will probably have hundreds or thousands of situations which turn out to be harmless and approximately zero of the other kind. So for a quiet/chilled/stress-free life, people mostly assume they aren't being targeted for major grief and nearly all of them get away with it nearly all of the time.

But since you have a somewhat more scary life you may be right to judge that the balance is different for you...

Quote from Bawbag :

Ah I see, wonkers then

Is "wonker" a Scottish insult? Sounds interesting :-P

Quote from Becky Rose :

...A few months later

you mean once they let him out of hospital?

Quote :

he started working at the same place as me and it turns out he was an ok guy, it's an incident I regret.

Every other incident i've ever been in wasn't started by me - but because i've had quite a few of them - with mixed results - I tend to get on the offensive very quickly before I get hurt, as experience has tought me that going on the offense quickly leaves me hurt a lot less often.

You may be on the money there... And I can't argue with how threatened a lone female driver would and should feel when tailgated in that fashion late at night. (You might even be able to explain it successfully to the judge.)
Anyway, glad to hear you're not totally psychotic

Just came across this thread.
Glad you reported it, glad you feel better now, and hopefully it did some good as well.

Most actual traffic officers I've ever spoken to (or observed) drive well and are very professional and patient (god knows they need to be given driving standards these days).
The guys you encountered (young?) clearly hadn't a clue about how to drive, and didn't think about what they were doing. That's training (not) for you
The regular patrol cars and police vans I see around town are almost always driven poorly - lack of lights in poor vis, lack of indications, tailgating, poor positioning, all common faults. I had once assumed that all police drivers allowed out at the wheel of an official vehicle were trained in some way, but since they drive just like normal members of the public, I now believe I'm wrong about that.

(Btw, jesus Becky, I hope I never accidentally walk up behind you in a dark alley, make you nervous and activate your kill-switch )

Last edited by Neilser, Mon, 21 Dec 2009 15:32. Reason : typo

ZOMG, I just realised we don't yet have furry dice either!

Quote from Scawen :

If I try to make the ESP as complicated and as good as the real thing, maybe I'd be on the case for a year. Maybe I'd need to hire a team of engineers and employ them for a few years - obviously not possible. On the other hand if I try to keep it simple then I hope to release the VWS in a reasonable time. I'm thinking, do something that helps stability and traction, using the ability to apply individual wheel brakes. Hope you see what I mean. It's not really something I want to discuss in detail. I'm on the physics, then I want to sort out something for the VWS, more appropriate than LFS's current traction control system, then release it.

Glad you're not trying to make it as good as the real thing.

In fact... I'm certainly not a purist (like "ban ABS/TC/followcam" zealots), but given a choice between a Scirocco in week N with no ESP and one in Week N+several with it, I'll give up the ESP any day
Given what LFS is all about, how many people will even use it? (If switchoffable...) (Does it help or hinder in the real car when racing to the limit?)

(nice job on Z25 btw ;-))

In follow and wheel (right name?) views, the clutch temperature overlaps the rev counter...

A button to turn ABS on/off from the cockpit (like TC) would make testing much simpler and more fun.

(You could even (!) play with turning it on only for certain corners....?)

I know this probably isn't a feature in any real cars btw. (My car doesn't have a reset button though either... anybody? )

Quote from AndroidXP :

Impossible. Reading the speed would be a piece of cake, but there's no way to do a useful speed comparison on an almost full and empty fuel tank. Also there's no track on LFS where you can reach air-resistance limited top speed. The oval comes close, but not quite - even there it depends on how well you exit the last corner. Besides that, the AI absolutely sucks at the oval (it was actually braking there on the KY Nat run :doh

Wow, chalk up another dumbness point for the AI

OK, scratch the terminal velocity approach - maybe it's possible to get at it a little more indirectly by checking the axle torque and car acceleration at the same speed for both tyre states?
You will also know the mass of the car (which will differ due to fuel loadings) and thus I *think* (hey, it's late ) that all the required ingredients will be available to allow us to calculate the power dissipated into the tyres?

The power to the axles (call it axlepower) is proportional to axle torque * axle speed (more precisely, it's 2 * pi * torque * axle-turns-per-second, which will give you a result in Watts if torque is in Nm). If it's the same in both cases (due to the rolling radius being identical, which might be the case) then so much the easier but it's not actually a requirement.

The power soaked up by the rate of change of kinetic energy of the whole vehicle mass (call it accelpower) is easy to calculate from the mass, velocity and acceleration. The kinetic energy is 0.5 * m * v^2, so accelpower will be
m * v * dv/dt or just m * v * a (with kg and m/s and m/s^2 for units, this will also be in Watts).

The power soaked up by the aero drag will be the same for both cases (cos same speed) and thus we don't need to know it (but the RAF file may have enough info to allow it to be calculated directly? I've not yet used a RAF analyser.) The missing piece of power is presumably being used up by the rolling resistance (if I've not forgotten anything), most of which is probably heating up the rubber. (I think??)

So if (axlepower - accelpower) differs between new tyres and worn tyres, we have our smoking gun, provided of course I've not been too hasty and missed something major (!).

Quote from Bob Smith :

Larger tyres generally exhibit lower rolling resistance, so I'm not sure your reduced rolling resistance hypothesis necessarily holds true. Most rolling resistance comes from rubber hysteresis in the sidewalls, which should be unaffected by tread wear.

Goshdarnit Bob, your objection is disgustingly plausible Not sure why a larger tyre should have less resistance, but certainly the sidewall argument seems sound.

But, maybe the sidewall distorts more with a thicker tread layer because of the extra effective leverage? (Reaching a bit here )

OK, recalling that what I'm chasing here is "if and why worn tyres are quicker *in LFS*" (i.e. not in real life), maybe another place to look hard is cornering. (This may be a lost cause! But read on if you want to indulge me...) I've watched a replay where worn tyres *may* have been the secret to carrying a lot of speed through a high-speed complex of corners. If the worn-down tyre is scrubbing less speed off, then a test for this might involve monitoring the axle torque required to maintain steady speed while cornering at a given acceleration. I'll maybe have a go at this...

(I must admit I'm doubting somewhat that worn tyres are actually quicker! But if they do turn out to be quicker, it doesn't have to be physically realistic - it can just be an (accidental) artefact of the way LFS models those rather complex lumps of rubber...)

Quote from AndroidXP :

Looking at both left and right rear tyre data (lateral grip vs. slip angle) and comparing lap 6 to lap 35 shows that there's definitely NO change in tyre behaviour by wear (as I suspected, to be honest). The curves looked pretty much 99% identical and the tiny difference can most likely be attributed to a difference in tyre temperature, air pressure and/or car weight.

Hey - nice one And the negative result isn't a bad thing - it actually simplifies matters, because I just remembered another thing (which we could call number 5 ):
The "less heat while cornering" effect I've been mentioning (or at least hypothesising, but like many people I've personally observed that worn tyres run cooler) almost certainly comes with another effect if physical realism is preserved: lower rolling resistance. Perhaps THIS could be the magic ingredient to explain why worn tyres are "quicker"?
(I should have remembered this sooner, ouch.)

How to verify it? Well, I guess it should mean that the real top speed would be higher - that might be the simplest place to look. So if you review the (actual, not speedo) straight-line speed for your AI at low/high tyre wear...

Quote from AndroidXP :

1) You drove for a quite long time and are more likely to be "in the zone"
2) The tyres have less heat capacity, cooling down from the usual overheat phase thus enabling you to go faster after having to drive careful
3) Your fuel level is lower than on race start
4) The tyres behave differently somehow (???)

FWIW, my money's on number 4, if the "worn tyres are better" hypothesis is correct at all.

AND, I've just realised that your super-cool tyre data-gathering technique could actually help prove what's going on here

All one would have to do would be to test a tyre when fresh and then again when worn down almost to the canvas and compare the grip/slip data.
Now this won't necessarily reveal the effect I mentioned earlier - that a tyre may get heated less for a given cornering force (due to less mechanical work being done in the thinner rubber layer, due to less distortion). But it should certainly show up any "magical" extra grip which is appearing at a given slip angle.
(It might even reveal the reduced heating too though, I guess.)

Quote from AndroidXP :

Errr, the bumping is a direct result of one section having a smaller radius than the others. Of course if all sections are worn the wheel diameter and with that also the circumference decreases.

Er, yup, I know that I'm just saying that LFS isn't necessarily going to model both a height change AND a circumference change due to the thinner piece of rubber. My best guess (but it IS just a guess) is that LFS does not model the circumference change.

Quote :

However, do note that the speedo in LFS works by measuring the speed at the differential / wheels and simply assumes a certain tyre circumference to calculate the actual speed (like real cars do).

A useful reminder - I think I had probably forgotten that LFS uses the axle/diff speed.
However, it wasn't what I was getting at anyway - I didn't say I expected to see a higher top speed come up on the speedo, but that the revs for top speed should be different.
The real top speed (if the limiter doesn't kick in) would be approximately the same with a smaller rolling radius, as it's decided almost entirely by aero drag, with the main thing causing a slight change in speed being the modest change in engine power at the new, higher engine speed.
But in a setup which has max speed sufficiently close to redline revs, I was thinking that the car might actually hit the limiter... But in fact maybe you'd need to use an analyser to see the difference, if it is only a couple of percent or so (2% means only 140 RPM at 7k RPM).
I think I'm going to check this out next time I wear my tyres down a lot

Quote from AndroidXP :

...LFS simulates this (rather poorly if you think about it) by wearing down the affected tyre pads. If you then drive very slowly - preferably in wheel-only view - you'll see the tyre bounce up and down as it rolls over the worn-down pads/section.

Wow, never noticed that. But I guess it doesn't modify the circumference? In other words, what I'd have expected to notice already if it's actually happening is that tyre wear would then raise the RPM for top speed on the straights. Since I've not noticed it (many of my sets would redline with only a modest increase in revs) I had assumed it was absent, but come to think of it I'm notoriously good at *not* noticing things

Quote from AndroidXP :

No idea about Tweak though, I haven't been keeping up with LFS for quite some time now. Heh, two seconds later I find this.

Nice. Look forward to trying that when I am reunited with my wheel... (I'm so bad with a mouse it's pointless!)

Quote from AndroidXP :

This is how a grip/slip curve looks in LFS in general. Note that these are all lateral grip curves - the longitudinal grip curve looks pretty much like the pre-U lateral curve to the right.
Here I found an old tyre pressure grip curve comparison, see that on a high pressure the grip peaks at about 6° slip whereas on the lowest pressure it reaches the maximum at about 10°. The difference visually doesn't look too great here mainly because the line was probably taken with the FOX, which is pretty lightweight so the tyre pressure had less of an effect.

Wow, more lovely data to digest! (Btw, does the RAF file give you access to the tyre wear data?)

I'm most surprised that the lateral grip doesn't drop at high slip angles - I guess this is what Scawen meant about the grip being too high at high slip? Much more pronounced than I had imagined - it really doesn't drop at all. Weird that it was "better" in the older patch (by which I mean that I believe the more physically realistic one to be the one that falls beyond the peak, rather than having a plateau). [Edit: but the longdit. grip does fall at high slip you say? Scratches head....]

And I'm intrigued about the high pressure effect. Modest drop in grip at high pressure, big drop in slip angle required to get it. This should presumably make the car more driveable too? (Not to mention shorter time constant, thus more crisp anyway...)

Quote from Ball Bearing Turbo :

I know you meant cooling. I suggested the open wheeler because there is more exposure to ambient air, and the velocity thereof is not shielded by a wheel well. I suggested the setup/track familiarity because if you are real comfortable with it you'll notice that your heating/cooling cycles are noticably modified by a high wind setting.

Aha. Well, I will give it a whirl, but I would have thought that even a high wind will pale into insignificance compared to a race car's average speed around just about any track. The downforce will be affected as it scales with the square of the total air speed on the wing, but the cooling effect (he says confidently ) should I think just be linear with speed.

Quote from Ball Bearing Turbo :

It's definitely there. More noticeable in open wheelers. If you really want to test it, take one out with a setup & on a trak you're very familiar with & crank the wind to full speed.

Hmm, you sound like you mean it affects downforce (since you mention setup and track familiarity) but I meant tyre cooling. If wind affects tyre cooling wouldn't you prove that by just leaving the car stationary on a windy day and see how long it takes for tyres to cool?
I'm sure wind must affect the handling somehow (or why have it?) but I've actually never tried high wind

Quote from Bob Smith :

I'm not sure if it gets lighter but it should get smaller.

Indeed. I agree worn=lighter tyres should make almost no diff to the car but a smaller rolling radius should be trivial to spot. But I think this effect is definitively absent or I'd (surely!) have noticed it by now...

Quote from AndroidXP :

I know this mainly by messing around with Tweak and the tyre sizes.

Must have a play with Tweak. Where did you find it? Don't suppose source code is available?

Quote :

As far as I know tyre wear does not affect grip at all. It does, however, affect the weight of the tyre (relatively sure) and the heat capacity (100% sure), the former meaning that with worn tyres you drive a lighter car which goes faster through corners, the latter allowing you to punish the tyres more since they can't hold heat as well and therefore won't overheat as readily.

Hmm, that explanation seems a bit fuzzy to me. For a given heat input, thinner tyres should actually overheat *more* quickly (and cool down more quickly), no? So perhaps the point is that the heating *is* less on a thinner tyre (hence we observe it's harder to keep it hot) and therefore you can drive it at a higher slip angle for the same equilibrium tyre temperature and get more grip out of it as a result.... ??

Quote :

The temperature however does have an affect on grip (duh), though the graphs are most likely outdated by now.

It does (click the picture on the right). Just not a numerical one.

Wow. Had NEVER spotted that the tread bars get thinner with time. Sheesh. Could have sworn I'd read that Wiki page properly before; in fact the text mentions "Wear" in the section title but nowhere else... And that pic makes it seem like the Wiki authors are (like me) unaware of the wheel rim temperature being part of the tyre-temp info

And double-wow. Those graphs are superb. Something I've been wanting to find/make for ages now. Thanks a mil. Will have to have a proper look when I get home in a few days. I've been pondering how to try and automate a procedure for generating stuff like this (maybe there's a way to write an AI for LFS to drive a car in controlled circles...?). OR, WAY easier, Scawen could just make the tyre models (and other stuff like torque curves) public... (Big hint there Scawen, if you ever read this )

Quote from AndroidXP :

It is in fact the rim and LFS also models its heating.

Wow, I had never noticed those teeny weeny blips in the corners of the tyre-temp pic.

How do you know for sure it's the rim and that it's modelled? Did you find it mentioned in docs somewhere? Does the whole wheel get modelled (heat reservoir with hugely more mass than the air in the tyre)?

Two things to add to this:
1) I'm interested in the tyre-wear vs. grip (and temperature) effect. LFS doesn't provide a wear-level indicator but wear does seem to influence the grip. I recently saw a guy do a spectacular lap (and some WR HLs do the same) after about fifteen minutes of abusing his tyres and then letting them cool down. And in a recent long race I clearly saw my tyres cooling down (with wear I guess) so I could abuse them more and not have them go red (wasn't 100% sure I had more grip though). [This, despite the fact that slicks have no tread pattern to move around less as the tread-depth drops. But OK, I guess less tyre-thickness might also provide less mechanical deformation and thus less heating.]

SO, I really want to understand how LFS models this tyre-wear effect. Anybody know? Does a high-wear tyre really give more grip? I've searched the forum and found little on this.

2) Tyre cooling: I have sometimes wondered if a car driven slowly (in LFS) cools its tyres more quickly than one left stopped on the track (and if the piece touching the track for the stopped car cools more quickly still ). Some half-hearted checks I made suggested that it made no difference (of course it should, as forced air cooling should be massively more effective at removing heat, no?). This point has come up already in this thread, but it's not clear if people are sure about it being absent. Anyone know for sure if this effect is modelled in LFS?