open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello everyone,

When I using the 3D module with salinity, always occur the problem about the iteration of MURD3D. I try to change the number of vertical layers and time steps, which will not work. The parameters I use are mainly based on the t3d_estu_gir case.(see attached file) What could be the possible cause for the problem? and what can I do for amending?
BTW, I looked up previous posts, JMH mentioned that the existance of dry zones is important, actually there are dry zones in my modelling area, if this may cause the problem, what should I pay attention to this issue?
Below is the listing information of the last time step when the run stopped.
================================================================================
ITERATION 205 TIME 0 D 1 H 42 MN 30.0000 S ( 6150.0000 S)
================================================================================

---

ADVECTION STEP
DIFFUSION OF WN
GRACJG (BIEF) : 2 ITERATIONS, RELATIVE PRECISION: 0.1710904E-06

---

PROPAGATION AND DIFFUSION WITH WAVE EQUATION
RESCJG (BIEF) : 32 ITERATIONS, RELATIVE PRECISION: 0.8029932E-06
POSITIVE DEPTHS OBTAINED IN 3 ITERATIONS

---

DYNAMIC PRESSURE STAGE
GRACJG (BIEF) : 4 ITERATIONS, RELATIVE PRECISION: 0.1550866E-06

---

VELOCITY PROJECTION STEP

---

ADVECTION-DIFFUSION OF TRACERS
MURD3D OPTION: 5 2 ITERATIONS
DIFFUSION OF TRN1
GRACJG (BIEF) : 3 ITERATIONS, RELATIVE PRECISION: 0.5600978E-06

---

MASS BALANCE

WATER
MASS AT THE PREVIOUS TIME STEP : 0.2789634E+13
MASS AT THE PRESENT TIME STEP : 0.2789809E+13
MASS LEAVING THE DOMAIN DURING THIS TIME STEP : -0.1749809E+09
ERROR ON THE MASS DURING THIS TIME STEP : 0.4569036E-01
FLUX BOUNDARY 1: 5818195. M3/S ( >0 : ENTERING <0 : EXITING )
FLUX BOUNDARY 2: 14500.00 M3/S ( >0 : ENTERING <0 : EXITING )

TRACER 1
ADVECTIVE FLUX THROUGH BOUNDARIES OR SOURCES : -0.1926560E+09
DIFFUSIVE FLUX THROUGH THE BOUNDARIES : 0.000000
MASS AT THE PREVIOUS TIME STEP : 0.5956511E+13
MASS AT THE PRESENT TIME STEP : 0.5962355E+13
MASS EXITING (BOUNDARIES OR SOURCE) : -0.5779681E+10
ERROR ON THE MASS DURING THIS TIME STEP : -0.6518787E+08
MURD3D: ITERATION NO. REACHED 100 , STOP.
ALFA = 0.960590047676882
GUILTY POINT = 289637

PLANTE: PROGRAM STOPPED AFTER AN ERROR

Hello,

Yes, distributive schemes N and PSI (4 and 5) will not work on dry zones, when some nodes have a volume equal to zero, for CFL condition reasons. For tracers on dry zones, prefer the scheme called NERD : scheme 14, it should work much better.

You may ask then why we do not strictly forbid to use N and PSI schemes with tidal flats. It is because we have important applications that do work, when volumes of points are either less than 1.D-3 or much larger, so that using a threshold we can remove dry points from the treatment.

Another distributive scheme able to work on dry zones is on the way for the next version, it is already done in 2D...

With best regards,

Jean-Michel Hervouet

Dear Jean-Michel Hervouet,

I am glad to hear from you, you reply really make me much more clear about the scheme issues, the another try would be make on them. Thank you very much.

Viola