I want simulation laminer and Turbulence flow in a cylinder tube . My tube is long and my tube diameter is small. my circular face must have many mesh node(cell),But my tube axial have few cell. 1.What is the best way to mesh my long tube?

2.What is best mesh element in my circular face (Map or multiblock or .....)

3.In my simulation ,my circular face must have many mesh node(cell),But my tube axial have few cell. (I thank fluid field change small in tube axial ) My mesh element ration will very big, May I make mistakes ? What is limited ration ?

(1). Just use H-grid, or hexa grid (bricks), so that it can line up with the flow direction and the wall easily. (2). The mesh distribution is completely up to you, but the result will have to be "mesh independent". That means you will have to increase the mesh size (total number of grid points or cells) until the solution no longer changes with the mesh size. (3). For pipe flow to be fully developed (becomes 1-D flow), it will require more than 40 diameter length. Anything longer than that, the flow will stay fully developed. (assuming that the pipe is straight)

Hi : I wany to use multiblock to create my cylinder tube . multiblock is one square in circular face . How size of square in circular face? big square in circular face ? or small square in circular face ? Thanks!!!

(1). Looks like that you are trying to use a H-grid (bricks) in the center of the tube and a O-grid on the outside so that it can cover the cylindrical surface. (2). You can make a small square mesh in the center, so that the rest of the flow domain is covered with the O-grid. The O-grid is easier to deal with the wall boundary layer. (3). You can also use h-grid to cover the tube, for this, you can take a look at Joe Thompson's book of "Numerical Grid Generation", which include some Fortran code to generate the mesh inside a circle and a square. (4). You may also want to talk to the support engineer of the code vendor, to see if they have some tutorial samples for your simple tube problem.

Hi : Thanks for you suggestion . If I put a small square in center of circle face,I must have many node in small square . If I put big square in center of circle face, It seems can save many node .(because big square can cover 80% circle face) But big square have some problem in square's top point and straight line. Do you have deal with this problem when you use FLUENT? Thanks!!

(1). I don't use Fluent now. (2). You can run some cases to see how the solution is affected by the mesh arrangement. (3). You really need to learn how to generate the mesh on your own before attempting to use a commercial code. (4). And if I were you, I would start with 2-D axisymmetric problem first. Then move on to 3-D cases. Optimization of mesh is something you do after you have the right solution.

Hi Olivia,

I agree with John fot the H-grid that is the most preferable. To do so if you are using Gambit, you can use the Cooper tool to mesh your pipe. You have to make your cross section mappable, then you should imprint a square in the circular cross section (don't make it to big or you will have some skewness problems).

Anyway you can use the Cooper tool without mapping your circular section. Paving it can be nice, as long as the cells will be aligned with the flow in the Cooper direction.

I did so and I got pretty good results. There, once again, John is right: don't forget to check that your results are mesh independant.

Regards

/alice

[Mohammed Raad]

Hi, please what is the best scheme for meshing circular face using gambit and how can I construct it?