Damper setup and bouncy car

Question

I have the bouncing car problem which is related to high damper values, although here are my findings:

• Decreasing fixed timestamp allows for higher values, but I know this isn't optimal.
• The wider the track width the earlier the bouncing occurs.

This is counter-intuitive as cars with wide track with such as a GT3 (Blancpain etc.) racecar have pretty high damper values and are light. If I can believe the values from some games they range from 5k to 23k. On a streetcar track width you can get values above 10k, on the AMG GT3 for example it is about 7k max.

But the question is which values are actually right, for example which value should we use if we have all 4 damper values but can only enter one? Is one preferred or an average of them? The Fast Bump for example seems to work fine without bouncing.

Can Enterprise users configure all 4 values?

Can the issue be resolved with this bugfix that you recently tweeted about? https://twitter.com/VehiclePhysics/status/1176782315575533570

Answer 1

The bouncy car effect on high damper values is caused by the bad implementation of the suspension in nVidia PhysX vehicles. The damper force is computed based exclusively on the contact speed, without considering car mass, inertia and -most important- effective mass.

We can only apply workarounds that mitigate the effect:

• Decreasing fixed timestep. This increases the CPU load as well.
• Increasing the inertia in dimensions X or Y. As inertia is used to compute angular accelerations, more inertia in these dimensions reduces the angular acceleration around the Z axis, thus reducing the chances of the damper to cause bouncing.

The wider track the more contact speed at the wheels, so chances to suffer the problem increase. The inertia must be increased in dimension X so that the vehicle takes full account of the increased track.

But the question is which values are actually right, for example which value should we use if we have all 4 damper values but can only enter one?

I can't really tell. I'd use the initial values as guideline, then I'd find some value that mostly resembles the suspension behavior in the real car.

Can Enterprise users configure all 4 values?

Yes. See: 
https://twitter.com/VehiclePhysics/status/1064446179688308736
https://twitter.com/VehiclePhysics/status/1064533761574887430

Can the issue be resolved with this bugfix that you recently tweeted about? https://twitter.com/VehiclePhysics/status/1176782315575533570

No. That bugfix resolves the car sticking to the ground on high damper values (https://www.youtube.com/watch?v=JbhhT5Zj690), but not the bouncing effects.