## Fluids

#### Issue with mesh independence

• Irving
Subscriber

I'm simulating a kind of counterflow heat exchanger air-air, which is made up of two different sections and sizes: a trapezoidal duct and a rectangular duct. They are separated by an aluminium flat plate, which is generating heat. The flow enters through a section of the trapezoidal duct and the outlet of the trapezoidal duct is connected to the inlet of the rectangular duct (please see the attached pictures). I'm using a hex mesh with wall treatment. Mesh quality is good: skewness max=0.60817, orthogonal quality min=0.264, aspect ratio max=10.8. The issue is that mesh independence was not achieved. I'm measuring the heat transfer coefficient of the flat plate, but it never stops decreasing as the mesh becomes finer, reaching values ​​close to zero. I've tried with different turbulence models(k-e ETW, SST, and SST Transition) and the results are the same. I also tried transient simulation, but the results were insane. Do you have any suggestions?
• Nikhil Narale
Ansys Employee
Hello Irving,

Can you please post the images/screenshots directly in the comments section and not attach it? Ansys employees are not allowed to download any attachments.

Nikhil
• RK
Ansys Employee
Hi Irving,
When you say, you are refining the mesh, do you have a plot of comparison of values with different mesh sizes?
• Irving
Subscriber

Hello, Nikhil
Sure. The first image corresponds to the velocity contour and the second one is the geometry.
• Nikhil Narale
Ansys Employee
Hello Irving,

These images won't help us to understand the problem. As Rahul suggested, can you share the plot of HTC with respect to the different mesh count?
Moreover, which method are you using to calculate the HTC?

Nikhil
• Irving
Subscriber

The first image corresponds to HTC average in function of y+. As y+ decreases, so does HTC. On the seccond one, Htc is plot in function of element number.
which method are you using to calculate the HTC? I obtain HTC from the Fluent across the surfacesintegrals: Area-Weighted Average-> Surface Heat Transfer Coef.
I also tried with points scattered along the flat plate.

• RK
Ansys Employee
Just out of curiosity, can you try with poly mesh? You can use Fluent meshing watertight workflow and that should be easy to generate a poly mesh.