open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi all,

I've been running a few test cases with the new AMR_PLAN algorithm implemented in v6p1 and having some difficulty. I started a basic test case with a steady inflow, a fixed downstream water level and a simple stepped temperature profile (i.e. a thermocline) to see how the mesh adapts. However, the free surface seems to become bumpy and after a few timesteps, the model terminates with the message AMR_PLAN : POINT OUTSIDE INTERVAL. The guilty point always seems to be in the top layer, so I assume this is the free surface instability causing problems for the interpolation step in AMR_PLAN. However, the model seems to run okay with conventional vertical discretization (e.g. evenly distributed sigma layers).

Any thoughts? Thanks in advance.

Enda

Hello Enda,

There are a few parameters you can play with.

First the relaxation parameter:
DOUBLE PRECISION :: RR = 0.5D0

Second, the number of horizontal smoothing:
DO JPLAN = 1,20 ! 20 ITERATIONS
(a lower value will save you time, particularly in parallel, a higher value will tend to increase horizontal smoothing)

Third, the coefficient of the shape of the arc-length (based on gradient of your variable):
! ARC LENGTH: MCOEFF = 100/MAX|DSDZ^2|
... Note that a variation based on the curvature would be but has not yet been implemented:
! CURVATURE : MCOEFF = 10/MAX|CURVATURE^2|

Hope this helps.
Sébastien.

Hi Sebastien,

Many thanks for your feedback.

Your last two points are self-explanatory but please could you comment just a little further on the relaxation parameter? Is this just the relaxation factor used in the G-S solution scheme? Could you suggest an appropriate range - less than 1?

Thanks again,

Enda