Sorry, "less sophisticated" is no option for LFS. The only thing we have is "not yet (fully) simulated".

From my understanding, the "problem" we currently have is that a flatspot is just represented by a section that has just a slightly shorter radius than the rest of the tyre, with no added resistance when rolling "off" the flatspot. Essentially it's just like a short bump down/up, while IRL it would have more severe effects like a short spike in rolling resistance and probably aligning torque, too.

Here's a crude animation of RL vs. (how I think it works in) LFS

Quote from AndroidXP :

Sorry, "less sophisticated" is no option for LFS. The only thing we have is "not yet (fully) simulated".

From my understanding, the "problem" we currently have is that a flatspot is just represented by a section that has just a slightly shorter radius than the rest of the tyre, with no added resistance when rolling "off" the flatspot. Essentially it's just like a short bump down/up, while IRL it would have more severe effects like a short spike in rolling resistance and probably aligning torque, too.

Here's a crude animation of RL vs. (how I think it works in) LFS

The key part of flatspot behaviour is that once you have a flat spot this is the area gets a hammering while under braking as you then tend to lockup easier with the flatspot down. This results in the fatspot getting bigger which meake the situation worse....

Flatspots are something that is just as bad to deal with as cooked brakes or tires.

Going fast is a big part of racing but so is mechanical sympathy

Quote from AndroidXP :

Sorry, "less sophisticated" is no option for LFS. The only thing we have is "not yet (fully) simulated".

My point is that while 1000+ segments may be more realistic than a "less sophisticated" canned effect, < 100 segments is probably less realistic than a canned effect. The segment approach only works well if there are a huge number of very small segments, but this would be a large cpu overhead.

Jeff, you normally make a lot of sense but I have no idea what you're talking about this time TBH. Simulating "something" is inherantly "more" realistic than "not" (canned effect)... Also, having the effects of tire deformation modelled, including vertically, I would really think that the forces that a flatspot creates (as pointed out by Android) would be modelled. Flatspots DO currently tend to cause lockup at the same points, it just doesn't seem sensitive enough (at least to me). Take out a GTR, flatspot the crap of a tire, then take it out in a straight and hammer the brakes. Guess what; 90% of the time the lockup is in the same place and often results in a blowout with the softer tires. Most people only set their brake strength in LFS so that they can stomp on the pedal and not have an incident. Also; if the flatspot did not affect the aligning torque, you would not feel it in the FFB like you can. This is a great thread, but this rubbish about the sectioning of tire heat in LFS is pretty bogus. Perhaps I merely fail to understand the argument, but I'm quite certain that's because it doesn't make any real sense at all

Quote from Ball Bearing Turbo :

Simulating "something" is inherantly "more" realistic than "not" (canned effect)...

Only if the simulation is reasonably close. I'm not sure a 16 segment model is close enough. Which is more accurate, using a table of values with linear interpolation or an equation? Say you're trying to model one cycle of a sine wave. The choices are a table with 1000 values, or a cubic equation. In this case, the table more closely approximates the sine wave. The table eats up more space, but the calculation time is faster. Now a ninth order equation might prove to be closer than a 1000 entry table, but the calculation overhead is much larger.

Maybe it would be more accurate to state my point that the cpu overhead budget might be better spent on other aspects of the simulation (general tire, suspension, chassis, areodynamic aspects) than on hot or flat spots.

I thought the physics core of LFS is actually quite CPU light compared to everthing else. I would have thought that Scawen would have tried various numbers of segments and for some reason has settled on 3x16 segments per tyre.

GPL (yeah.. GPL) has this utility available that shows the tyre surface temnperature. They increase and decrease very very fast indeed. To the point where I'm quite sure that a big powerslide with opposite lock is only sustainable for one or two seconds before tye rears really overheat.

I doubt that is fully realistic but its a VERY big difference to LFS where the temperatures creep up and down very slowly..

From what I know, rubber has poor heat conduction so it would seem to make sense that a short time spinning or locking tyre produces a lot of heat quickly (high friction, low heat conduction) at the surface.

Yes, that's the point: "at the surface". We just don't have a surface layer/display yet.

We're all aware that the tyres are not perfect, and we all know at some point they will get tweaked, either in terms of curves or 'features' (more sections, surfaces, brake heat heating the tyre etc).

But we also know that Scawen won't rely on simple canned effects when he can simulate something like this. Maybe he tried 40 sections, but found little or no improvement worth the extra CPU time.

Personally I'm glad canned effects are banned, otherwise we'll end up with something as crap and unrealistic as rFactor.

But then, JeffR wouldn't be JeffR if he didn't have something to moan about.

Yeah I was thinking.. Perhaps LFS does do quick changing surface temps but doesn't show it to us

Unlikely. Scawen doesn't seem to like calculating something and not showing it. Also we'd be able to feel it through the physics.

Quote :

Yeah I was thinking.. Perhaps LFS does do quick changing surface temps but doesn't show it to us

Quote :

Unlikely. Scawen doesn't seem to like caculating something and not showing it. Also we'd be able to feel it through the physics.

One of the things I noticed when fiddling around with the cmx files is that different parts of the tarmac appear to offer different grip, although I am unsure of the specifics.

cmx? Aren't that the car mesh export files?

Quote from Niels Heusinkveld :

Yeah I was thinking.. Perhaps LFS does do quick changing surface temps but doesn't show it to us

if that was the case drifters would produce smoke a whole lot faster than they currently do

Quote from AndroidXP :

cmx? Aren't that the car mesh export files?

i guess she either means the smx files for her spectator tool or the dds files

afaik the grip change is always clearly visible with different colours/shades of the tarmac

I did mean smx - infact this little bit of dyslexia caused me no end of grief whilst writting various 3d importers...

There seems to be more surface changes than are shown via textures though.

So maybe old and new tarmac have slightly different grip values
Or do you mean even more differentiated?

I'm sure they do, but there are even more divisions and I dont think all of them are shown on textures - i'm just guessing if truth be told, basically what I am really saying is that I suspect the grip of the surface is a lot more involved than most of us realise.

Quote from tristancliffe :

With the flow of air around the wheelarch I don't think that it'll make any difference. Some cars (e.g. the Trevor Sagaris) have big ugly holes in the wheelarches to let some of the excess air out! If you were air, a wheelarch isn't somewhere you'd choose to hide!

they used to, but stones got thrown off the ries through the holes on to the windscreen, so they removed them

Quote from Jakg :

they used to, but stones got thrown off the ries through the holes on to the windscreen, so they removed them

Only a few days late on the reply, and you were beaten by several people.

Hang your head in shame!

Another reason for me to dislike the Sagaris then - pointless frilly bits. They should have just removed them altogether!

Quote from Niels Heusinkveld :

GPL has this utility available that shows the tyre surface temnperature. They increase and decrease very very fast indeed.

That same utility also slows the deeper tread heat and it takes much longer to heat up. When I did Norschleife hot laps, I'd start at the final straight near the end of a lap. I'd nose the car up to a fence then spin the rears for about 2 to 3 seconds, then I'd make a pass up the straight, and lock up the tires 3 times, back down the straight to lock em up 3 more times, then another uturn and it was off to run a hot lap. Obvisouly GPL didn't have a flat spot algorithm, and this method was common practice to avoid having to do a pre 8+ minute warm up lap.

Quote from Becky Rose :

I did mean smx - infact this little bit of dyslexia caused me no end of grief whilst writting various 3d importers...

There seems to be more surface changes than are shown via textures though.

i thought smx files were only the polys and didnt even include the driving durface let alon its parameters ?

Quote from tristancliffe :

Another reason for me to dislike the Sagaris then - pointless frilly bits. They should have just removed them altogether!

its a lot cheaper though to just not cut them out instead of making a completely new form

Quote :

little cpu overhead for the physics

Quote :

We also know that Scawen won't rely on simple canned effects when he can simulate something like this. Maybe he tried 40 sections, but found little or no improvement worth the extra CPU time.

Assuming all this is true, then I sit corrected. I'm just thinking that trying to simulate flat spots would be difficult. I would imagine that trying to include dynamic contact patch area size and deformation in addition to the normal slip / skid formulas used, and now flat spotting seems like it would be pretty cpu intensive. With a 2 foot diameter tire, 150mph translates into 35 revolutions per second, or 1680 tire segments per second (35x16x3). Maybe I'm underestimating the speed of current cpu's in PC's.

So getting back on topic here, it does seem like the hot spots take too long to cool off.

Also in addition to the air, wouldn't the track also be a big contributor to cooling of the tires on straights for the front tires and also for the rears at higher speed where rear wheel torque is less?

Quote from JeffR :

So getting back on topic here, it does seem like the hot spots take too long to cool off.

Also in addition to the air, wouldn't the track also be a big contributor to cooling of the tires on straights for the front tires and also for the rears at higher speed where rear wheel torque is less?

Air & surface temp; sure. There is also the inherant heating that increases with speed due to the tire flex. Lower pressure tires should heat faster even on straights due to the rapid flexing of the sidewall above the contact patch.... @BOB - good point about things being shown... I wonder then how the sim decides that the entire tread should gain heat? Wouldn't there have to be a calculation to add heat to the tread - similar to how the tread affects the air temp over time? I would think then that there would be "one line of pixels" at the top of each "pad" in the F9 display, but maybe he didn't want to show it for some reason.... Hm....

Quote from Shotglass :

if that was the case drifters would produce smoke a whole lot faster than they currently do



i guess she either means the smx files for her spectator tool or the dds files

afaik the grip change is always clearly visible with different colours/shades of the tarmac

Becky = Female ????
Sry to go off topic

Quote from lalathegreat :

Becky = Female ????
Sry to go off topic

One has had one's head in the proverbial toilet bowl, hasn't one?