[namCFD]

Hello,

I could not figure out the reasons for the water spray in my simulation becoming unstable when I make the mesh size smaller. I hope the community can point me in a direction for diagnosing this issue. I have included a video for your viewing. If you need the files, I can post them also. The mesh size is 2.5 cm x 2.5cm x 2.5cm.

Following is my set up:
+ dieselFoam for water spray (10 micron min and 50 micron max droplet dia)
+ nozzle dia = 3 mm
+ injection velocity = 10 m/sec
+ mass flow rate = 0.04241 kg/sec
+ injection pressure = ~1.5 bar
+ room size = 4 m long x 2 m high x 2 m wide with ambient air temp (300K ) and ambient pressure

I used VirtualDub (free and portable video editing software) to compress and crop the file.

Thank you for your help.

[schmidt_d]

Without looking at your case in detail, I have a suggestion. You need to greatly increase the number of computational parcels injected into your domain (this doesn't change the injected mass, but rather represents the mass with a greater statistical fidelity).

The reason is that, as a cell gets smaller, the source terms of cells in densely loaded regions get hit with huge noisy source terms from the droplets at are proportionately larger.

At a minimum, you want the number of parcels to go up proportional to the cell count. Even that is not really sufficient.

See:

David P. Schmidt, “Theoretical Analysis for Achieving High-Order Spatial Accuracy in Lagrangian/Eulerian Source Terms,” Int. J. of Numerical Methods in Fluids, 52(8), 2006.

and slightly related:

Sasanka Are, Shuhai Hou, David P. Schmidt, “Second Order Spatial Accuracy in Lagrangian-Eulerian Spray Calculations,” Numerical Heat Transfer, Part B: Fundamentals, 48(1), pp. 25-44, 2005.

-DPS

[namCFD]

Originally Posted by schmidt_d

Without looking at your case in detail, I have a suggestion. You need to greatly increase the number of computational parcels injected into your domain (this doesn't change the injected mass, but rather represents the mass with a greater statistical fidelity).

The reason is that, as a cell gets smaller, the source terms of cells in densely loaded regions get hit with huge noisy source terms from the droplets at are proportionately larger.

At a minimum, you want the number of parcels to go up proportional to the cell count. Even that is not really sufficient.

See:

David P. Schmidt, “Theoretical Analysis for Achieving High-Order Spatial Accuracy in Lagrangian/Eulerian Source Terms,” Int. J. of Numerical Methods in Fluids, 52(8), 2006.

and slightly related:

Sasanka Are, Shuhai Hou, David P. Schmidt, “Second Order Spatial Accuracy in Lagrangian-Eulerian Spray Calculations,” Numerical Heat Transfer, Part B: Fundamentals, 48(1), pp. 25-44, 2005.

-DPS

Thank you very much for your help. I will check out those articles and follow your suggestions. I will also try to post the case files.