As I experimented with my carsim I realized that my car''s front tyres loose grip easily and have a large slip angle. What can be done to improve/avoid this? More weight to front? More power to rear?

Better rubber?

(Translated as: higher coefficient of friction)

I believe the problem is the same as that causing your braking problem, ie: you need multi directional friction modelling to account for the fact that tyres slip less easily in their lateral direction than roll direction. Read my latest post in your ''braking'' thread. I believe this will solve both your problems. Hope this helps

Thank you all.

Now the static friction coefficient is 1.1. This is quite good, I think on dry tarmac. Or not? And the kinetic friction coefficient decrease linearly with velocity difference, so that at 40 km/h velocity difference the friction coefficient is still 0.9. I find this a bit much, but without this my car is slippery in corners because of loosing traction.
What coefficient do you suggest for different velocity differences for dry tarmac?

My model consist the followings:
- full rigid body simulation for the body
- wheels treated as masspoints
- vertical spring-damper at each wheel (independent suspension), no anti rollbar
- the weight transfer works correctly
- torque curve based engine
- no clutch yet
- simple gearbox
- open differencials
- brush model based tyre model for each tyre (using Fz, slip angle, velocity, angular velocity, no camer, no toe)
- brake torque at each wheel
- handbrake torque for both rear wheels
- rolling resistance at each wheel
- no downforce, just air resistance opposite the velocity
- simple steering model (not Ackerman, it should be (?))

This is in my model now.

So the main problem is that the front wheels slip more then the rear ones. The car is understeered. This should be compensated somehow.

I played around with your car sim and I''m very impressed. Its really shaping up to be something special. I''m still convinced your problem lies with having isotropic friction modelling however. While playing your game it became evident that when you really throw the car into a corner the amount of slide you get remains fairly constant all throughout the cornering. If you play GT or Collin Mc''Rae you''ll notice that the car initialy slides but after a certain point finds its grip again and shoots off forwards (unless you really push it hard, then you spin out). This is due to the multi directional friction modelling I described in my posts in your ''braking'' thread. Basically the car''s facing is intially thrown right off its direction of travel so you get lots of lateral force and hence sliding. As you continue around the corner the car begins to align itself with the direction of travel and the lateral force is reduced until it no longer overcomes the tyres'' lateral static friction and hence the sliding stops and you shoot forward again. It really isn''t such a difficult concept and I definately recommend you give it a go. Let me know if you do, I''d love to see the results.

In my carsim program you downloaded, the static friction was 1.3
It''s too much. But the inertia of the car was too little. That''s why it rotates and slips so easily. And this is unrealistic.
No static friction 1.1, and the inertia is quite more than earlier. And with these setting the car behaves much different, much more relistically, but it is still very understeered.

I play with the tyre setting yet (longitudinal and lateral stiffness), to make it more realistic, but the tyre model is still unchanged. I will try to introduce this non-isotropic friction. It''s like a friction ellipse instead of friction circle, isn''t it?

Thats one way of looking at it yeh. I quite like that description actually. Easier to visualise. Although I wouldn't try using some kind of ellipse math, mush easier to just decompose the force of motion for each tyre into their parallel and tangential components and work from there. Just remember its the difference between static and sliding friction that is the key. Keep up the good work anyway, its looking fantastic. Make a post when its done, I'd love to see the finished result. Good luck to you.

[edited by - motorherp on June 6, 2004 2:12:46 PM]

yeah. Should not hard to implement, split it into the two axes of the tire.

The tire stiffness can play an important part as well. If you have stuff like tire pressure, and make them more ''springy'', you''ll get a more subtle drift.

downforce is good, but moving the centre of mass forward, I''m not too sure. Car manufacturers go into so much trouble trying to achieve a 50/50 balance, it must be for a good reason

you can have some sort of electronic stabiliser, given how much a tire slips, redistribute the power output through the other wheels. Same for braking.

hmm... all that, and playing the GTR demo makes me want to have a go at it

also, you can have slightly different tire models, from the front and rear, since the tires aren''t matched. The tires at the rear a much wider usually.