Rolling Friction cdt

Hello everyone,

I have a question concerning the rolling friction model cdt. As far as I can see it, the magnitude of the angular velocity is not accurately calculated when torsion_torque is not included. Right now, the magnitude "wrmag " is not updated after subtracting the normal component. Therefore, the normal component of the angular component influences the rolling torque in both tangential directions.
I provide a test case in the attachment. Two particles rolling on a wall. One particle without a normal angular velocity. The second particle with a normal angular velocity (z-direction). If !torsion_torque, both particles should have the same rolling friction torque and therefore the same angular velocity in x-direction because the normal angular velocity should not influence the tangential component. But in the current version, both particles have different tangential angular velocities (x-direction). Am I right?

Cheers,
Johannes

***** At the moment *****

const double wrmag = sqrt(wr1*wr1+wr2*wr2+wr3*wr3);
if(!torsion_torque)
{
double rtorque_dot_delta = r_torque[0]*enx+ r_torque[1]*eny + r_torque[2]*enz;
double r_torque_n[3];
r_torque_n[0] = enx * rtorque_dot_delta;
r_torque_n[1] = eny * rtorque_dot_delta;
r_torque_n[2] = enz * rtorque_dot_delta;
vectorSubtract3D(r_torque,r_torque_n,r_torque);
}

***** Updated version *****

if (!torsion_torque)
{
double wr_n[3];
const double wr_dot_delta = wr1*enx+ wr2*eny + wr3*enz;
wr_n[0] = enx * wr_dot_delta;
wr_n[1] = eny * wr_dot_delta;
wr_n[2] = enz * wr_dot_delta;
wr1 = wr1 - wr_n[0];
wr2 = wr2 - wr_n[1];
wr3 = wr3 - wr_n[2];
}
double wrmag = sqrt(wr1*wr1+wr2*wr2+wr3*wr3);