Hello,
So let's go:
MASS-LUMPING ON H : this changes the matrix stemming from the derivative in time of the depth into a diagonal. In some sense moves from finite elements to finite volumes. Effects : stabilises, smoothes. In steady state just stabilises, no effect on results.
OPTION FOR THE DIFFUSION OF VELOCITIES : in non conservative shallow water equations the diffusion term for a function f should be in the form 1/h * div (h * NUT grad(f)). Option 2 is this form, in option 1 we discard the terms 1/h and h. Not too much effect, and prevents divisions by 0.
TURBULENCE MODEL FOR SOLID BOUNDARIES : yes just for k-epsilon, other models do not need this "detail".
TREATMENT OF NEGATIVE DEPTHS : the fastest algorithms in Telemac allow some slightly negative depths. To get strictly positive depths TREATMENT OF NEGATIVE DEPTHS = 2 recomputes the continuity equation and yields an equation solved with machine accuracy and positive depths.
FREE SURFACE GRADIENT COMPATIBILITY : a parameter to suppress inf-sup oscillations (that appear when you discretise depth and velocities in the same way). 1 does nothing, 0 suppresses wiggles but may smooth a bit, any value in between accepted. Look for spurious oscillations of the free surface.
IMPLICITATION FOR DEPTH (or VELOCITY) : takes depth or velocity explicitely (time n) or implicitely (time n+1) in various terms of the equations. Implicit is more stable but smoothes, semi-implicit (0.5) is more accurate but the limit for stability. Important only for waves, has no effect on stedy state flows.
That's it,
Jean-Michel Hervouet
P.S. I have a number of Powerpoint presentations on this that should be put on this site soon.
