[Mohammad74]

Hello,

I'm trying to simulate an array of 3*3 cubes using k-w sst turbulence model in Ansys Fluent. I've used gambit to generate a fully structured hexahedral mesh and with relatively fine mesh near the walls for boundary layer simulation. As for boundary conditions, I defined inlet boundary as velocity inlet with custom velocity profile and 10% turbulence intensity, outlet boundary as pressure outlet and side walls as symmetric. I posted pictures of the mesh below:

1.jpg
2.jpg
3.jpg

After the simulation, I compared the velocity magnitude along a vertical line located at the middle of the top of central cube with the data I obtained from a journal article. I posted the plot of this comparison below:

4.jpg

As can be seen by the picture, my velocity values in vicinity of the wall is underpredicted by a relatively large margin. What is the problem here? What am I missing?
I'm struggling to find the reason since my domain dimensions are larger and my cell counts are higher than that of the article I mentioned earlier.

Any helps/suggestions would be much appreciated.

[Mohammad74]

Hello again,

So I've been struggling with this problem for over a month now, and wasn't able to solve it. I've tried different mesh resolutions with different strategies and non of them worked. In my latest attempt, I simplified my domain by using an array of 1 building in the span wise direction and 3 buildings in the stream wise direction and used symmetric BC on side walls to account for the removed 6 buildings. Also wall y+ values accross building walls is around 1,here's my mesh::

1.jpg2.jpg3.jpg

But still I'm having trouble validating my results near the 2nd building wall. I've noticed the separation zone above the 1st building is extended all over to the 2nd building!? Which gives me wrong results. I'm posting pictures of the velocity contour and velocity profile at the point I'm having problem with:

4.jpg5.jpg

Any Ideas why this is happening?

[AlexanderZ]

for me it seems, that this separation zone should be bigger, isn't it?

you may try to play with inlet BC, of example turbulence intensity, and the other one, don't remember the name


you may try to use other turbulence model.

other idea is to extend zone with refined mesh higher above "buildings" as that separation zone from the first building could interact with course mesh and could be influence by course mesh. Its hard to say from pictures is it the case or not.

the last "expensive" approach is to simulate this in transient with relatively small timestep.