Hello Annalena,
Yes, this is not so far a variable that is computed, I should add it, however we may face the problem that it will be very different if we consider it along the horizontal or on the vertical. To do it simply, I would compute it as : norm of velocity * cubic root of volume around point / DT, and for volume around point I would take the array VOLUPAR. If you use distributive schemes (3, 4 or 5) their number of iterations is about the Courant number.
With best regards,
Jean-Michel Hervouet (just back from Vietnam)