PDA

View Full Version : Pressure vs. inner/middle/outer temperatures



dw123
03-07-2015, 11:00
I'm trying to get a handle on how tire pressures and camber affect the temperature spread and general handling of the cars, and I'd been having a hard time seeing enough variation to make any real progress, so I decided to run some extreme-end-of-the-scale tests to get my bearings.

Running a BMW Z4 around Brands GP, I tried firstly, setting all pressures to their absolute lowest (1.50 bar), the running around the track for a few laps, recording the telemetry HUD through OBS so I could concentrate on driving.

I then did exactly the same thing at the absolute highest (2.50 bar).


As I understand it, massively under-inflated tires should show high temperature inside and outside, with a noticably lower temperature in the middle - as the carcass sags... and massively over-inflated tires should show noticably higher temperature in the middle compared to the outer/inner, as it bulges from the excess pressure... right?

Now, under-inflating tended to overheat the tires in general more than over-inflating did, but at most when under-inflated, the middle temperature stayed the same as the outer temperature, while the inner temperature ended up around 2-5 degC higher (nearer 2 on the FR+RL tires, 5 on FL+RR).
Over inflating caused the middle temperatures to pretty much match the inner temperatures, while the outer was slightly cooler... but really, only slightly, and on the FR and RL tires, the temperatures stayed pretty much equal across all three, looking positively well-balanced.


Does this sound right from people's experiences? Are the differences this subtle in reality, or should that level of under/over-inflation cause much larger temperature differences?


The other thing that slightly confounds me is that running the BMW Z4 on *any* track (admittedly most track are clockwise), I always get a hotter inside-temperature on the FL and RR, while their opposites always stay balanced... does that sound expectable? I'd expect it for the FL, seeing as it gets the heaviest pounding, but why would the RR be so noticable while the RL never gets any noticable inside overheating?

danowat
03-07-2015, 11:06
How was the camber set when you ran your tests, as this will have a large affect on how the tyre heats up over the width.

dw123
03-07-2015, 11:08
How was the camber set when you ran your tests, as this will have a large affect on how the tyre heats up over the width.
It was set to the game's stock defaults. -2.4 front, -3.0 rear I believe.

I just ran another quick test with the cambers both set to 0.0, just to see the effects... the FR and RL started to heat the outside more than the middle/inner, but oddly, the RL and RR still kept a slightly higher inner temperature than middle/outer even in that circumstance.

AfterAll14
03-07-2015, 11:09
This game uses SETA tire model. Its dynamical, only God and several programmers inside SMS know how it works.
I couldnt find any logic. No telemetry data avaible about tire behaviour. So we can only guess what is going on there.

dw123
03-07-2015, 11:14
This game uses SETA tire model. Its dynamical, only God and several programmers inside SMS know how it works.
I couldnt find any logic. No telemetry data avaible about tire behaviour. So we can only guess what is going on there.
Yeah, I know it uses the SETA model... but like any literal physics-simulation model, it can model reality well, or it can model reality poorly, depending on how it's been implemented, linked to other physics simulation systems, and preconditioned. Just because it's using a true simulation-derived model doesn't *necessarily* mean it's actually deriving anything close to reality.

(For an example - compare Half Life 2 with Deus Ex: Invisible War - both of them used the exact same Havok rigid body physics engine... but Invisible War would catapult whole tables joyfully around the room if you so much as bumped into them!)

I would hope/presume the SETA model in pCars is implemented well... but that's why I'm asking. I can't get my head around the figures it's showing me, so either it's doing the right thing, and I need to learn to understand it better, or something isn't quite right beneath the bonnet, and spending hours learning will get me nowhere without realising.

danowat
03-07-2015, 11:22
Do you have any real world experience of tyres, pressures and heating to compare the results from PCars with?

dw123
03-07-2015, 11:24
Do you have any real world experience of tyres, pressures and heating to compare the results from PCars with?
Nope, none whatsoever... that's why I'm asking. From what I've read, I got the impression it would be more distinct a difference than I'm seeing, but I acknowledge that I have no actual experience to base that on, and am in all likelihood just not understanding how it works.

danowat
03-07-2015, 11:29
Tyres appear to be a bit of a black art, and can do funny things.

I think that the way the tracks are rendered in PCars (dynamic rubbering, and differing levels of grip and abrasion) would make it tough to come up with any reasoned comparisons.

What are you trying to achieve?

AfterAll14
03-07-2015, 11:40
The only thing you should know about setup tires in pCARS - lowest pressure possible before overheating, max camber on front, min camber on rear. Thats it.

dw123
03-07-2015, 11:54
Yeah, this is a bit what bothers me. I'm methodically minded, and it bugs me that it is such a black-art... I don't really want to just game the system with settings that "just work".

I'd intended to leave everything at default, and test the extreme cases of pressures and camber. What I'm trying to achieve is to push the theory to its logical extremes, so I have a set of boundary cases that show the clear, unequivocal bad results of pushing a particular setup option too far. Logically, if I can demonstrate to myself that each of these settings when pushed to it's limit is having x-amount of effect on the result, I can keep an eye out for proportionally reduced levels of that same effect as an indicator when I'm trying to tune my setup.


If I push it to the absolute extreme, and it still shows me no distinct difference, then I'm not really sure where to go from there, other that to ask whether anyone else thinks it's odd that such extremes should yield such minor differences, or whether that's the kind of tolerances this stuff really operates within, and I should just take it as a given and keep trying.

danowat
03-07-2015, 11:57
I think you might need a bit of a primer on tyre modelling and how things like this affect the model, to get some decent comparative data.

Try googling Pacejka there is a lot of information about tyre models if you look for it.

AfterAll14
03-07-2015, 12:04
I understand you man. I tried the same. But I couldn't find any connection between the knowledge I have about tyres and what I see in pCARS. I only figured that lower pressure -> more grip, more camber on front -> more camber thrust -> more grip, less camber on rear -> more power output -> Vmax rises.
Its still a mystery for me how slip angles work in pCARS... they just work the opposite as they should have in theory...
SMS should reveal something about tyre carcass behaviour, but I doubt they would - it would instantly show flaws in tyre physics.

BTW tyre feels good and "alive", dont get me wrong. But its behaviour under different setup is arguable.

dw123
03-07-2015, 12:13
I think you might need a bit of a primer on tyre modelling and how things like this affect the model, to get some decent comparative data.
Try googling Pacejka there is a lot of information about tyre models if you look for it.
Cheers for the suggestion, but I think the full-blown maths is going to be way beyond me. Mostly I'm just after knowing whether the simplified conventional wisdom:
http://www.wheels-inmotion.co.uk/forum/uploads/monthly_11_2007/post-2-1196429733.jpg
- applies in quite the way it's portrayed - where the extreme cases should hammer the outer-edges or center-line respectively - actually works quite that obviously, or whether I should just accept that it's subtle and basically impossible to clearly discern on its own, even in extreme cases, given all the other complicating factors?

Edit: Another possible way to put it - are the extremes of 1.50 bar and 2.50 bar pressures actually particularly extreme? It may just be that I'm expecting an extreme result when actually the range the game limits me to are actually all within a drivable, fairly reasonable range.

IJerichoI
03-07-2015, 12:14
This comes down to the tire resolution of the STM. It has almost no canned effects like other games. If you double the tire resolution, you would see numbers more exact, but it won't be possible for current hardware to handle @600Hz. Although, the feeling would rather be the same, only a bit more detailed.

So to summarize: The model does what it should, and also these over-under-inflated scenarios are calculated correctly (in the given range). But the view in the HUD is somewhat misleading.

This was the explanation of a dev regarding this topic.

danowat
03-07-2015, 12:18
http://www.wmdportal.com/projectnews/inside-project-cars-seta-tire-model/

I believe that PCars tyre model does things that no other tyre model does?

There was a huge document posted on here the documented much of the development of the game, and the feedback that Nick and Ben were giving on the tyre model makes interesting reading.

danowat
03-07-2015, 12:22
I know it's the brush model, which PCars no longer uses, but it's interesting reading.#

http://www.control.lth.se/documents/2007/jsvenPDH.pdf

https://dspace.lboro.ac.uk/dspace-jspui/bitstream/2134/4822/1/Transient%20analysis%20of%20tyre%20friction%20generation%20using%20....pdf

danowat
03-07-2015, 12:27
Cheers for the suggestion, but I think the full-blown maths is going to be way beyond me. Mostly I'm just after knowing whether the simplified conventional wisdom:
http://www.wheels-inmotion.co.uk/forum/uploads/monthly_11_2007/post-2-1196429733.jpg
- applies in quite the way it's portrayed - where the extreme cases should hammer the outer-edges or center-line respectively - actually works quite that obviously, or whether I should just accept that it's subtle and basically impossible to clearly discern on its own, even in extreme cases, given all the other complicating factors?

Edit: Another possible way to put it - are the extremes of 1.50 bar and 2.50 bar pressures actually particularly extreme? It may just be that I'm expecting an extreme result when actually the range the game limits me to are actually all within a drivable, fairly reasonable range.

Also, you might want to bear in mind the practical differences between racing tyres and road tyres, where conventional wisdom is concerned.

1.5 - 2.5 Bar, try the difference in your road car (real life, on a private road of course!).

dw123
03-07-2015, 12:44
This comes down to the tire resolution of the STM. It has almost no canned effects like other games. If you double the tire resolution, you would see numbers more exact, but it won't be possible for current hardware to handle @600Hz. Although, the feeling would rather be the same, only a bit more detailed.

So to summarize: The model does what it should, and also these over-under-inflated scenarios are calculated correctly (in the given range). But the view in the HUD is somewhat misleading.

This was the explanation of a dev regarding this topic.

Interesting. Yeah, I realise the whole idea is that it's a fully simulated model, with near-zero canned effects. I had wondered - seeing the image of the finite-element model of the tire used, whether the single center division would be enough to actually model major curvature flex in the contact patch... I figured it might just be a straight-up technical limitation. I realise how hard it must be to get any kind of physics system operating at 600 steps per second in realtime.

So, in general, the effect is there, but the limitations of the model mean it's going to be pretty subtle... sound about right?

I wonder if the devs might at some point see fit to give us a "increased tire resolution" option to experiment with, for people with heavily overclocked 6-core+ CPUs or something :-)

dw123
03-07-2015, 17:00
Hmm... odd extra test. I wondered what exactly was causing the inside-tire temperatures to always be noticably higher on the Front-Left, and Rear-Right tire... so I dug around, and noted that Asure Coast would be a good balanced track to test for left-right balance, as it's a straight A-to-B stage, with, if anything, a slight tendency to anti-clockwise turns, compared to most circuits being clockwise.

The result surprised me... I *still* got the FL and RR insides heating up noticably, as though the camber were off, and the FR and RL were *still* evenly balanced across the tire.

I then swapped out the BMW Z4 GT3 for the McLaren 12C GT3, and the result was entirely different. This time, both the right hand tires were balanced temperatures, while both the left hand tires heated up the insides by a good 10 degrees over the outsides/middles. On the same slightly-more-left-hand-turns Asure Coast.


What could be causing that? I've noticed the same two opposite corner tires be unbalanced on all of the 7-8 different tracks I've raced the BMW at so far. This is with stock defaults, sensible tweaked setups, and extreme-case setups, none seem to change the fact that it's always those specific two tires on the Z4.

Edit: I can in fact, set camber to (0.0, -3.5) on the front tires, and (-3.0, 0.0) on the rear tires, and it will *juuust* about balance the temperature spread on all four tires, and even then, even with 0.0 camber the FL tire runs a slightly hotter inside edge.
Is this some quirk of the BMW Z4's weight balance or something? Or is something off here?