[Fole]

Hello,

I am having problems with a CHT simulation. I'll try to explain briefly the necessary details.

Geometry:
A heat exchanger with 1 fluid and fixed heat flux in exterior walls: It is an assembly of cylindrical passes one inside another, and they are connected through holes. The working fluid enters in the outermost cylinder and goes out from the innermost one.

The main problems for automatic meshing with Delaunay tetrahedron with inflation in ANSYS Meshing is that it has borders and corners in the cylinders walls. The walls are thin and they have curvature!

Model:
Flow mainly laminar: Re 1e3. Although it can be transitional an I have tested that kwSST also works.

Custom oil material as working fluid and steel as solid. There is a fixed heat flux of order 1e5 w/m2.

Mesh:
Number of inflation layers 10, the minimal distance of the first cell is aprox 5e-4 m. The mesh skewness is 0.86 caused by inflated cells in corners and borders. Number of cells across gaps is 2. The solid has 1 layer of elements through thickness and it has the same skewness than fluid domain in a handful off cells.

Solver:
Pressure-based Coupled Pseudo-transient. Settings to default.

Simulation:
Temperature limited to lower and upper bounds whatever number they are set, in the fluid domain and later in the solid. In some situations a floating point error appears and fluent exits prematurely.

Things that I have discovered (apart of results from other threads):

1. Changing to air it converges. I think that it is because the air gets hot rapidly and the thermal gradients are acceptable for the mesh.

2. Modelling only the oil domain has also issues. Temperature limited in cells of poorer quality.

3. During the iterations the energy equation is solved very slowly. I have seen it by stopping and watching the temperature contours. Basically the first layers start to get hot. At some number of iterations it diverges.

Conclusion:
- Inflated cells have low quality and they have the issues with the numerical solution of the temperature equation.

Questions:
Which are your experiences in CHT with these types of fluid and skewness? Is it worth to go to ICEM CFD and get a better quality despite the probably time needed, or the problem comes from other issues? I am asking because I am already losing the focus of this problem trying to get it!

Cheers,
Fole.

[Fole]

I have found the problem. The results were ok, the problem was the model that was not considering a heat loss anywhere; that is the reason for high temperatures. I thought that the oil convection was the main energy mechanism, my fault.

Despite of that, I would like to say that I have had to set the energy urf below 1, in this case I set up 0.75 with SIMPLEC 1 skewness correction. The solution is converged by monitoring residuals and energy balances.

Moreover, I've used a coarser mesh with 5 inflation layers, giving max y+ = 2 but with refinement in the walls with high heat fluxes. Despite the apparent coarseness, the flow fields and temperature are being resolved (the streets are relatively thin and no high changes are expected).

The skewness for this mesh was 0.95, in the limit of what Fluent recommends.

Sorry if I have annoyed you! This post could be deleted or anything.
Cheers,
Ovle.