+- Forums (https://www.vdrift.net/Forum)
+-- Forum: Project (https://www.vdrift.net/Forum/forumdisplay.php?fid=4)
+--- Forum: Development (https://www.vdrift.net/Forum/forumdisplay.php?fid=9)
+--- Thread: clutch mod proposal (/showthread.php?tid=1221)
clutch mod proposal - NaN - 01-27-2010
I had a look at the clutch, and I am not very happy about the way it's simulated in vdrift.
1. Right now it looks like a kind of linear torque-converter(depends on rpm delta).
2. I have no idea where to get realistic parameters from(sliding friction, radius, area, max-pressure).
A (friction)clutch is a torque transfer device. So it's obvious to describe it by its torque capacity. The maximum torque it's able to transfer.
To account for clutch wear/slip the clutch torque capacity should be at least 1.25 of engine peak torque. If you've got a lot of grip(awd, slicks) 1.5. One usually stays below 2.0 to avoid engine/gearbox damages.
My proposal is to use torque capacity as clutch parameter.
For the 911(CS) instead of:
[ clutch ]
sliding = 0.27
radius = 0.1
area = 0.75
max-pressure = 11079.26
We would have:
[ clutch ]
torque-capacity = 350 # engine peak torque * 1.4
- NaN - 01-28-2010
Had a closer look at the clutch code. It is not as bad as I thought. The clutch torque(slip) step function is approximated by a limited linear slope. There is no lockup but a highly viscous coupling. That's why it looked to me like a torque converter first.
The torque capacity is actually the product of the currently used clutch parameters.
torque_capacity = sliding_friction * max_pressure * area * radius
Here a concrete example(CS):
torque_capacity = 0.27 * 0.1 * 0.75 * 11079.26 = 224.35Nm
engine_peak_torque = 243Nm
So the clutch is too weak for this engine of the 911. It is slipping between 4200-6200rpm :o
To fix this the max pressure could be be set to about 15000.
1.25 * engine_peak_torque = 0.27 * 0.1 * 0.75 * 15000 = 303Nm
We should check this for all cars.
edit: Looks like only the CS had this problem.
- joevenzon - 01-29-2010
I checked in a change to the CS for the 15000 value you suggested. Seemed to work fine for me. I rewrote the way the clutch works recently on the recommendation of some forum folks (before, the clutch wasn't actually simulated if it was engaged, which led to all kinds of weird problems), and indeed that exposed some errors in the car setups that didn't get noticed before. I fixed some of them but it looks like I missed the CS. To be honest, enough of the physics engine has changed or had bug fixes that all of the car settings probably need a going over. The most recent in-depth tunes were to the C7 and F1-02, both of which are a blast to drive!