[raorjun]

Why

The ClutchLSD stick-slip law (#48) steps the clutch torque from static to slidingFrictionRatio × static at breakaway. With the default 0.85, that step is 15 %. It excites the rear wheel and tire longitudinal relaxation mode (wheel plus halfshaft rotor against the tire's sigma_kappa). That mode's damping ratio is about 0.04 to 0.07. The slip speed swings past zero within half a period, so the clutch sticks again. The cycle repeats at 25 to 40 Hz.

In BobSim's steady-state study on the reference vehicle, 5 of 22 cases log 1000 to 1500 events and do not settle in 420 s wall clock. Two of them hold an ay oscillation the driver cannot settle. The same cases with the Open differential settle in 7.5 to 10.5 s with 6 events.

The 0.85 value has no source. It dates from BobSim's first release, for the old tanh law.

What changes

• DifferentialRecord.slidingFrictionRatio defaults to 1.0.
• The nine VehicleDefn records set 1.0.
• The stick-slip law is unchanged. A caller can still set a ratio below 1, as TestPowertrain does with 0.80.

Evidence

BobSim steady state on the reference vehicle at BobLib 7363fc1, with BobLib's default ay PI gains:

Case (V m/s, ay m/s²)	Ratio 0.85	Ratio 1.0	Open
20, 3	timeout at 32.3 s, 1468 events	settled 8.7 s, 7 events	settled 7.8 s, 6 events
20, 4	settled 12.5 s, 1051 events	settled 9.1 s, 7 events	settled 8.5 s, 6 events
12.5, 2	killed at 1500 s wall, 1864 lock switches	settled 11.1 s, 7 events	settled 10.5 s, 6 events
15, 2	killed at 1500 s wall	settled 9.5 s, 7 events	settled 8.8 s, 6 events
17.5, 2	timeout at 32.2 s, 1011 events	settled 8.9 s, 7 events	settled 7.7 s, 6 events

• With ratio 1.0, ay is within 0.05 m/s² of Open in every case. The handwheel is 0.01 to 0.02 rad higher, which is the intended LSD understeer.
• A 1e-8 solver tolerance does not remove the cycle at 0.85.
• In a 1 Hz sine steer at 15 m/s, ay third-harmonic distortion is 7.6 % with ClutchLSD at 0.85, against 0.5 % with Open.
• make lint, make modelica-lint, make modelica-smoke, make modelica-initialization (39 passed) and make modelica-physics (11 passed) pass.

Limits

• 1.0 removes the static-to-sliding step. It is not a measured calibration. A sourced value below 1 also needs damping on the wheel and tire mode: a positive μ(slip speed) slope, or sourced rear sigma_kappa and wheel inertia (0.02 kg·m² looks low).
• The damping ratio is a hand estimate.
• Measured on the BobSim reference vehicle only.