[darlopez]

Hi everybody.
I have a dynamic mesh and I want to keep the orthogonality between cells. I use a directional diffusivity, it seems like:

diffusivity directional ( a b c );

in my dynamicMeshDict archive.

a,b and c are numbers, can anybody tell me what mean these numbers?

I found two versions. One of them says that this numbers are the value of the coefficient of diffusion in the 3 directions of the space. The other says that the first number determines the mean cell non-orthogonality, while the second number determines the mean cell skewness (and the third?). But in my experience I can not determine this numbers correctly.
How can I determine the value of a,b and c?

What other types of diffusivity I can use to keep the orthogonality between cells?

Thanks a lot.

[prasanth]

Dear Darlopez,

I am also facing the same issue. I didn't able to get those numbers. Did you able to find out? If so, Please tell me the what are those numbers.

Regards
Prasanth.

[darlopez]

Dear Parasanth

I solve my problem using other type of diffusivity, concretly I use inverseDistance type, but inverse quadratic diffusivity works also ok.
If you should use directional diffusivity, I think that the three numbers are the value of the coefficient of diffusion in the 3 directions of the space (I saw this running different cases). How to determine the value of this numbers is a mystery.
I suggest you to use an inverse distance diffusivity. And you do not forget put the convenient patches type "slip".

Regards
Darío

[prasanth]

Dear lopez,

Thanks for your reply, I will use inverse distance and I will back to you about my experience regarding this.

Regards
Prasanth.

[e_boesso]

Hi,

I think it depends on what you have to do...
for example, if you want to move a boundary patch in the y direction you can put "diffusivity directional (1 1e-6 1)": the b value sets the extension of the moving grid in the y direction. Value 1 for a e c implies that the cells in the x and z directions are "transported" without distortions of the grid.
I think there is no way to set "a priori" these values since they depends on the relative dimensions of the problem.
I use the moveDynamicMesh utility until the mesh moves in the way I want and then I pass to the solver run.
This is only my experience...

[alexB]

It seems to me that a,b,c define a vector.

I tested it with a 2D case (x-z-plane).
-b seemed to make no difference. makes sense cause its not in the plane.
-(0 0 1) gives the same result as (1e-7 0 1). makes also sense, cause the vectors are nearly the same.

greetings
Alex