[sdfij6354]

Hello,

I want to simulate airflow through a horticultural greenhouse. The greenhouse has openings on both sides which are covered with an anti-insect netting. The airflow is coming from the Inlet. One part is passing through the netting on side 1 of the greenhouse, through the greenhouse itself and leaves the greenhouse on side 2 through another netting. The remaining air is passing over the roof of the greenhouse.

I created the necessary geometry for the solid "walls" and netting of the greenhouse, as well as the domain. I get realistic airflow results when not implementing the netting-geometry (leaving the respective area blank, like a "hole" in the greenhouse).

Now I'm struggling with setting the correct boundary conditions for the anti-insect netting. According to my research, the airflow through such a netting can be calculated as a flow through a porous medium according to the Darcy-Forchheimer-Equation. I have measurements for K, epsilon and Reynold values for the netting I want to integrate into my CFD (obtained from wind tunnel experiments).
I assume a pressure drop to occur when the wind is passing through the netting.

Do you have any help with choosing the right boundary conditions?
The solid parts of the greenhouse are defined as no-slip wall boundaries. The Inlet/Outlet as Patches.
When defining the netting-geometry as a "cell zone" during meshing, I'm able to further define it as a Porous Medium (using the Darcy-Forchheimer-Equation). However, I'm not getting any feasible results for this - but is this the right start?
Reading about other pressure drop problems, I've found the "Cyclic Jump" boundary condition. Would this be more appropriate for my case?

I'm using Helyx-OS as a GUI for OpenFoam. In regards to this, I'm also facing the problem that the netting geometry (after defining it as a "cell zone") appears three times within my case after meshing. Once as a cell zone, once as a geometric object (where I would need to select a patch type which I don't know) and another geometric object with <name>_slave. Is this behaviour intended or do I make some mistake in using Helyx-OS?

Thanks for any help and recommendations,

[piu58]

Did you consider reproducing the netting simply by a smaller area of the openings? This may give some deviations in the very near of the openings itself. But all in all flow through the greenhouse should not be influenced much.

Always use the simples model which gives results of an accuracy you need. Don't overmodel.

[sdfij6354]

Originally Posted by piu58

Did you consider reproducing the netting simply by a smaller area of the openings? This may give some deviations in the very near of the openings itself. But all in all flow through the greenhouse should not be influenced much.

Always use the simples model which gives results of an accuracy you need. Don't overmodel.

Hi Piu48,

thank you for your answer. That would indeed be an option to model the overall airflow. However, I would like to focus on the airflow very near to the openings since that's an important part of my research. I think the deviations would be too strong.
Do you have another idea how to model the netting?

[Swempa]

Originally Posted by sdfij6354

Hi Piu48,

thank you for your answer. That would indeed be an option to model the overall airflow. However, I would like to focus on the airflow very near to the openings since that's an important part of my research. I think the deviations would be too strong.
Do you have another idea how to model the netting?

Hello Alwin,

Your approach with utilizing the porous media definitions could be a good approach if your looking for analysing the macro effects near the opening and gaining the correct pressure drop.

Only issue that i can see with this approach is that "anti-insect netting" most likely is reality a thin domain. Depending on how you model this it will need a suitable mesh with enough resolution in the stream-wise direction. The thin thickness of the net can maybe also induce some errors in the computation of the coefficients.

When you mention that you get unfeasible results you could start by checking your darcy and forchheimer coefficients, and also check your e1 and e2 vector if you have defined is as an orthotropic porosity.

Would also check the mesh, so that you dont have any prism layers in the transverse direction on the "anti-insect netting" domain, and that you fit enough cells in the stream-wise direction.

Good luck with your project!

Cheers