[wanghaojie]

Hi everyone,

Im doing a set of simulation with different wind directions through a building. Meanwhile, for each direction I need to rotate the domain(cubic several times larger than the building), does anyone have better idea that dont need to change the mesh and can simulate different wind direction?

Thanks.

My first thought is to use two inlets in xz and yz planes, but it turns out the velocity is larger at the diagonal and decays to the edges(not uniform)

My second thought is add a momentum source, but i dont know if i need to change the direction, how much i need source i need to add.

[SergeAS]

And if you use a domain of cylindrical shape is insensitive to a change in direction of the external flow?

[wanghaojie]

Sorry, I didnt quite understand you. Do you mean to change the whole domain into a cylinder?
If so, I dont know if Fluent or other CFD program can define a inlet and outlet on a cylindrical surface

[SergeAS]

I meant the following. If the accuracy of the solution in the far wake behind the obstacle is not important but you can use a single-stream boundary condition along the entire perimeter (if the outer perimeter is sufficiently far from away from the structure being modeled)

[wanghaojie]

Thank you. Now the problem is that if i only use one inlet, does it mean i have to change the direction of the inlet face everytime i need to change the inlet direction?

[SergeAS]

Changes only the direction of the velocity vector at the input.

In any case, the algorithm depends on your actions the capabilities of your software. I just set out general considerations of which I proceeded in my own software when it was necessary to conduct a series of calculations of the flow near elements of the aircraft for various angles of attack without the new mesh generation

[wanghaojie]

I use Fluent, and I think it can only generate flow normal to the inlet. Can you tell me what kind of software you use and what treatment did you use for various attack angle please?

Thank you

[SergeAS]

I'm using own (used only internally) software. However, as far as I know, OpenFOAM for example allows to specify an arbitrary velocity vector on the boundary.

Concerning Fluent: In Fluent forum you will be able to exact answer to this question. I'm not so familar for Fluent, sorry

[Ford Prefect]

Originally Posted by wanghaojie

I use Fluent, and I think it can only generate flow normal to the inlet. Can you tell me what kind of software you use and what treatment did you use for various attack angle please?

Thank you

No. You can specify the velocity components instead, if you like.

[wanghaojie]

Originally Posted by Ford Prefect

No. You can specify the velocity components instead, if you like.

Do you mean add a momentum source in Fluent?

[wanghaojie]

Originally Posted by Ford Prefect

No. You can specify the velocity components instead, if you like.

Oh, sorry, I　just find there is a way to specify velocity component in Fluent...
Thank you all.