The TORCS Racing Board
Username: Password: Remember Me?
Lost Password Register
Author: berniw | Created: 2017-02-07 22:35:53
Subject: Initial temperature model in CVS (only temperature)
Hi all

Just for the curious ones, the initial tire temperature model is in CVS. Influence on grip and wear will follow soon (not yet in). Next steps:
- Plug wear model in
- Plug grip model in
- Plug graining model in
- Parametrize car models
- Pit stop modifications
- Cleanup, make some more stuff configurable
- Dashboard (ingame info), make data accessible in tCarElt

See:
http://torcs.cvs.sourceforge.net/viewvc/torcs/torcs/torcs/src/modules/simu/simuv2/wheel.cpp?revision=1.26.2.15&view=markup&pathrev=r1-3-1
http://torcs.cvs.sourceforge.net/viewvc/torcs/torcs/torcs/src/modules/simu/simuv2/wheel.h?revision=1.7.2.4&view=markup&pathrev=r1-3-1

Kind regards

Bernhard

Last Edited: 2017-02-07 22:38:10 by berniw
    Author: berniw | Created: 2017-02-09 00:42:19
    Subject: Re: Initial temperature model in CVS (only temperature)
    Update, done the following, get it from CVS:
    - Plug wear model in
    - Plug grip model in
    - Plug graining model in

    So you can now run the model actually to get an impression.

    TODO:
    - Pit stop modifications
    - Parametrize car models
    - Cleanup, make some more stuff configurable
    - Dashboard (ingame info), make data accessible in tCarElt

    Regards

    Bernhard
    Last Edited: 2017-02-09 00:42:19 by berniw
      Author: firechief | Created: 2017-02-09 15:18:15
      Subject: Re: Initial temperature model in CVS (only temperature)
      In combination with pro mode, this changes things enormously. The established order of teams from the last few years is now thrown out the window - welcome to the brave new world :)

      All we need now are more competitors...
      Last Edited: 2017-02-09 15:18:15 by firechief
    Author: berniw | Created: 2017-02-12 19:07:25
    Subject: Re: Initial temperature model in CVS (only temperature)
    Further updates (current CVS, r1-3-1):
    - Pit stop does now automatically refresh tires (for now)
    - Dashboard available, data available in tCarElt (hit "1" till you get the dashboard)
    - Race configuration parameter available (wearFactor)

    The model is now ready testing and fine adjustments, TODO:
    - Testing the model, missing test cases:
    1. Does the tire cool down on stillstanding (constant part in convection model)
    2. Burnouts
    3. Does toe in/out have the expected effect
    4. Do suspension changes have the expected impact
    5. Wear for various cars/tracks, get more experience

    - Only consider wear/temp model for pro mode
    - Test factors and adjustment for models
    - Pit stop modifications (use time to change tires, just change on demand)

    Optional:
    - Faster wear/grip dropoff on overheat
    - Initial race temperature adjustment

    What I found fascinating is that the temperature model does do pretty well without adjusting "magic" numbers, amazing:-)

    Regards

    Bernhard
    Last Edited: 2017-02-12 19:16:29 by berniw
    Author: berniw | Created: 2017-02-14 23:26:10
    Subject: Re: Initial temperature model in CVS (only temperature)
    Hi all

    I am now verifying the wear/temperature model, things are looking pretty good (IMHO).

    Kind regards

    Bernhard
    Last Edited: 2017-02-14 23:26:10 by berniw
    Author: berniw | Created: 2017-02-18 15:06:11
    Subject: Re: Initial temperature model in CVS (only temperature)
    Hi all

    Another little update: All verification of the model on my part has passed, so I will go forward with it:
    - Tires do cool down on stillstanding
    - Burnouts are doing as expected
    - Toe in/out do have the expected effect (e.g. with a tita 7 "pro modified bot" with no rear toe max temp is 86.6, lifespan 90 laps, with max rear toe in (1 degree) max temp goes to 91.7, lifespan is 68 laps
    - Manual pushing/nursing tests passed (see video, pushing the tires gives 55 laps of lifetime, nursing it with 1:19er laps gives 87 laps lifetime)

    So the only things left to do for the "core" tire wear model are (for now):
    - Making tire change on pit stop optional
    - Adjusting the cars which need adjustment (trb cars seem to be fine out of the box)

    Kind regards

    Bernhard
    Last Edited: 2017-02-18 15:32:52 by berniw
      Author: alassio | Created: 2017-03-04 20:34:01
      Subject: Re: Initial temperature model in CVS (only temperature)
      Now that we can play around with the new tyre wear model in 1.3.8-test1,
      I would like to better understand the physics/mathematical model
      behind it. Is there a reference to some papers the implementation
      is based on?
      In order to understand impact on driving behaviour,
      racers (robot programmers) should have some understanding of
      the effect onto friction/traction of the car.

      Regards
      Rico
      Last Edited: 2017-03-04 20:34:01 by alassio
        Author: berniw | Created: 2017-03-04 22:18:33
        Subject: Re: Initial temperature model in CVS (only temperature)
        Hi

        It is not based on one paper, because the papers (which I read) just covered a certain aspect, but not all in combination (usually they assumed some variable to be constant, "steady state"). You find the model which emerged here in the code, there are quite a few comments (see SimWheelUpdateTire, ~line 400):
        http://torcs.cvs.sourceforge.net/viewvc/torcs/torcs/torcs/src/modules/simu/simuv2/wheel.cpp?revision=1.26.2.22&view=markup&pathrev=r1-3-1

        Kind regards

        Bernhard
        Last Edited: 2017-03-04 22:34:06 by berniw
          Author: alassio | Created: 2017-03-04 23:05:40
          Subject: Re: Initial temperature model in CVS (only temperature)
          Thanks for the reference in the code, is really well commented.

          From inspecting the code, I understand the following effects:
          - Tire slip and z force cause more heat and pressure in the tire
          - Convection cools it down
          - Overheating causes graining which is cumulative over time (wear)
          - Grip is optimal at ideal temperature of tire and without wear
          - Any difference to ideal temperature (too cold or hot) and wear reduce grip

          Is this description correct?
          Is there any major difference between accelerating and braking
          with regards of heating up the tires?

          Regards, Rico
          Last Edited: 2017-03-04 23:05:40 by alassio
            Author: firechief | Created: 2017-03-04 23:49:35
            Subject: Re: Initial temperature model in CVS (only temperature)
            Hi Rico,

            I didn't write the code but I have tested it quite extensively so I might be able to help answer some of your questions.

            - Tire slip and z force do indeed warm the tires.
            - I'm not sure on the exact cooling process, but driving a little slower definitely cools tires (less slip when cornering and less z forces when braking) and they also cool as they get older, probably because there's less rubber in the tread to be energized to create heat.
            - Graining occurs when the tires are cold, which can be either early in their life before they've warmed up, or as they're aging and the heat can no longer be retained in the tire. Unlike the graining we've seen in Formula 1 tires over the last couple of years, in TORCS it does not necessarily cause a permanent loss of grip assuming the tires can be warmed up enough that the graining process stops, after which wear will remove the grained surface.
            - Grip is optimal at ideal temperature and without graining. Provided these two are maintained, grip will remain constant throughout the life of a tire.
            - Correct, too hot and too cold will reduce grip, plus when its too hot it wears faster.

            Accelerating will tend to increase the weight on the back wheels, plus they're more likely to slip, so they will heat faster than the front. Braking is the opposite, though front tires heat the most when cornering rather than braking.
            Last Edited: 2017-03-04 23:49:35 by firechief