[gautami]

Dear Dr. Jacobsen,

Simulation of a 2D wave tank with a floating object case using waveDyMFoam.

The details are:

wave tank length 13 m (500 cells) , wave tank height 2 m (100 cells). Total 50000 cells.

The simulation was completed in 12 hours for an endTime of 35s.

To have an idea of relaxing.correct() on the simulation time, I made a change to waveDyMFoam.C code by not incluidng “relaxing.correct()” in the code. Then I recompiled waveDyMfoam and simulated the same case. The simulation was completed in 4 hours. As we would expect the results show no wave relaxation at the relaxation zones.

Kindly can you please let me know the inclusion of “relaxing.correct()” in waveDyMFoam.C can increase the simulation time by a factor of 3. Please find attached the case file (testCasesb.tar.gz) and details of the case and the waveDyFoam.C codes (waveDyMFoamsb.pdf). Kindly please let me know if I am doing something wrong.

3D simulation case: 3D wave tank with a fixed truncated vertical cylinder, with a mesh size of 2 million cells and a max Courant number 0.25 was completed in 1 day 13 hours for an endTime of 25s using waveFoam solver (parallel run on 24 cores).
The same 3D simulation case with 2 million cells, providing 3 degree of freedom (surge,heave and pitch) to the cylinder using waveDyFoam solver is getting computationally very expensive. A parallel run on 36 cores is taking 1 day and 12 hours to simulates until 2 secs.

Your insights on this will be highly invaluable.Please do let me know if I need to provide any additional information. Sorry for any trouble caused. Thank you for your time and consideration.

Most&More,
Gautami Erukulla

[ngj]

Hi Gautami,

Thank you for the comments, I am not certain, but I can try to answer your question (btw, you do not need to recompile to remove relaxation zones. Simply do not define them in relaxationZoneNames).

What you experience is likely due to the fact that you reevaluate the relaxation zone every time, or more precisely: You find all the cells in the zone again every time step. Try to use the 'frozen' type of zone. This allows for a static set of cell indices. Note, however, that this only works for topologically static meshes, i.e. deformation is allowed, but not a change in number of cells.

Please let me know, whether this helps.

Kind regards

Niels

[gautami]

Dear Dr. Jacobsen,

Thank you very much for your advise. Will work on it and will keep you posted.
Thanks again!

Most&More,
Gautami Erukulla.

[gautami]

Dear Dr. Jacobsen,

As per your advise, I simulated the 2D wave tank with a floating object case (50000 cells), using Frozen relaxation. (version OpenFoam 2.3.0)
The frozen relaxation zone is defined as:

At Inlet:

relaxationZone
{
relaxationScheme Spatial;

relaxationShape Frozen;
actualRelaxationShape Rectangular;

beachType Empty;

relaxType INLET;
startX (0 0.0 -1);
endX (4 0.0 1);
orientation (1.0 0.0 0.0);
}

At outlet:


relaxationZone
{
relaxationScheme Spatial;

relaxationShape Frozen;
actualRelaxationShape Rectangular;

beachType Empty;
relaxType OUTLET;
startX (9 0.0 -1);
endX (13 0.0 1);
orientation (1.0 0.0 0.0);
}

The simulation took the same amount of time 12.8 hours for and endTime of 35s. I am not sure if I am defining the frozen relaxation zone in the correct way. I also tried to define the frozen relaxation zone as:

At inlet:

relaxationZone
{
relaxationScheme Spatial;
relaxationType Frozen;
relaxationShape Rectangular;
beachType Empty;

relaxType INLET;
startX (0 0.0 -1);
endX (4 0.0 1);
orientation (1.0 0.0 0.0);
}

At outlet:

relaxationZone
{
relaxationScheme Spatial;

relaxationType frozen;
relaxationShape Rectangular;
beachType Empty;

relaxType OUTLET;
startX (9 0.0 -1);
endX (13 0.0 1);
orientation (1.0 0.0 0.0);
}

There was no difference in the time (it took 12.8 hours for and endTime of 35s). I am making a mistake in defining the frozen zones. Kindly can you please help me in defining the frozen relaxation zone the right way. Thank you very much for your consideration.

Most&More,
Gautami Erukulla.

[gautami]

Dear Dr. Jacobsen,

This is in regard to my earlier message about frozen relaxation zone. I had defined the frozen relaxation zone as:

AT INLET :
relaxationZone
{
relaxationScheme Spatial;
actualRelaxationShape Rectangular;
relaxationShape Frozen;
beachType Empty;
relaxType INLET;
startX (0 0.0 -1);
endX (4 0.0 1);
orientation (1.0 0.0 0.0);
}

AT OUTLET
relaxationZone
{
relaxationScheme Spatial;
actualRelaxationShape Rectangular;
relaxationShape Frozen;
beachType Empty;

relaxType OUTLET;
startX (9 0.0 -1);
endX (13 0.0 1);
orientation (1.0 0.0 0.0);
}

Under the folder 'constant' the following four files are created:
cells_inlet, cells_outlet, sigma_inlet and sigma_outlet and these folders define the cells in the frozen relaxation zones (15400). I guess the frozen relaxation zone effect is being implemented during the simulation. However defining the frozen relaxation zone did not have any effect of expediting the simulation time.
3D simulation is getting computationally very expensive, with 2 million cells, Courant number 0.2, is taking 3 days to simulate to 2.5s and it looks it will take 25 days to complete the simulation for an endTime of 25s. I was wondering and hoping if there is any way to use the waveDyFoam solver with the relaxation zone, without increasing the computational time drastically.
I apologize for causing any trouble to you with this issue. Any guidance in this regards will be immensely helpful. Thank you

Most&More,
Gautami Erukulla.