[modorous]

Hi,
I'm stuck on this for a few weeks now, any help you can give is much appreciated.
As shown below, a liquid coolant is poured onto a structure. Liquid coolant and air are the two phases. Initially, domain is only filled with air. I'm looking at the heat transfer between the liquid and the surface. See attached pictures of the setup "Boundry_conditions" & "geometry".

This is a transient double precision simulation with PISO, Presto!, and Geo-reconstruct as the method. Turbulant model is K omega SST. Multi-phase is solved as a VOF with Implicit formulation, Implicit Body force and bounded second-order implicit transient formulation.
To get an accurate heat transfer coefficient I am using fine boundary layers in my mesh (First height - 0.002mm, layers 7, transition ratio 0.3). Mesh has an orthogonal quality of 0.18 and an aspect ratio of 226 (I'm using Fluent Meshing watertight geometry). See attached "mesh"& "BL_mesh_zoom" to check my mesh.

With these settings, all the parameters converge except for continuity at first. After several iterations, all the parameters started to diverge. I also ran a few simulations after increasing the first height of the boundary layer. With higher "First Height" results converge at first. But after around 100 iterations results start not to converge. I also checked the volume fraction of the domain at this point. This is the point where the secondary phase reaches the boundary layer. Finally, I ran a simulation without any boundary layers. Which ran without any issues, but gave wrong heat transfer results.

Is this due to my boundary layer mesh? as you can see in the above picture, I don't have much room to add a lot of layers. can someone tell me what I am doing wrong here? How can I solve this issue?

Thanks,

Best regards,

Waruna

[CFDKareem]

Originally Posted by modorous

Hi,
I'm stuck on this for a few weeks now, any help you can give is much appreciated.
As shown below, a liquid coolant is poured onto a structure. Liquid coolant and air are the two phases. Initially, domain is only filled with air. I'm looking at the heat transfer between the liquid and the surface. See attached pictures of the setup "Boundry_conditions" & "geometry".

This is a transient double precision simulation with PISO, Presto!, and Geo-reconstruct as the method. Turbulant model is K omega SST. Multi-phase is solved as a VOF with Implicit formulation, Implicit Body force and bounded second-order implicit transient formulation.
To get an accurate heat transfer coefficient I am using fine boundary layers in my mesh (First height - 0.002mm, layers 7, transition ratio 0.3). Mesh has an orthogonal quality of 0.18 and an aspect ratio of 226 (I'm using Fluent Meshing watertight geometry). See attached "mesh"& "BL_mesh_zoom" to check my mesh.

With these settings, all the parameters converge except for continuity at first. After several iterations, all the parameters started to diverge. I also ran a few simulations after increasing the first height of the boundary layer. With higher "First Height" results converge at first. But after around 100 iterations results start not to converge. I also checked the volume fraction of the domain at this point. This is the point where the secondary phase reaches the boundary layer. Finally, I ran a simulation without any boundary layers. Which ran without any issues, but gave wrong heat transfer results.

Is this due to my boundary layer mesh? as you can see in the above picture, I don't have much room to add a lot of layers. can someone tell me what I am doing wrong here? How can I solve this issue?

Thanks,

Best regards,

Waruna

Your settings look good for the boundary conditions/discretization. It is possible that the quality of your boundary layers are causing the divergence in the calculation.

I am not very familiar with Fluent meshing so I am unsure of the algorithm it uses to create the inflation layers. However, looking at your mesh it looks like there are some mesh faces cutting across the inflation that may be creating some small mesh elements. See the picture attached with marks. These small mesh elements can definitely cause some issue with convergence.

If you use the "Check Mesh" in Fluent does it give any warnings about small mesh elements?

If you can choose the inflation algorithm in Fluent meshing try using a "pre inflation" algorithm. This will create the inflation layers on the wall before creating the interior mesh. It will help avoid the cutting of the hex mesh elements in the inflation layer.

Finally, I would try using workbench meshing to create the same mesh. You won't be able to use ploy mesh elements, but would be a good check to confirm that the rest of your setup is working well.

Let me know how it works out.