[igor_vidic]

Hi everyone, I' am working on my graduating thesis and I really need help. I am modeling heat transfer between fuel rod and primary water in nuclear reactor, i.e. it's basic cell. So geometry is square cylindar devided in four regions. In center there is cylinder whic represents fuel rod. It is devided in three radial zones (fuel, cladding and helium gap between). Also i divided fuel in 12 parts in axial direction (z-axis) because i had to put differtent volume heat sources in each od them. Around the fuel rod flows water, so final geometry has square chilindar shape. I used four different sketches to draw four regions (water, fuel, gap and cladding). In adition, total geometry is ver long (3.6m) and very thin (only 1.2 cm). I created name selections on the beginning and on the end of the watter region. Furthermore I made input boundary contion as mass-flow input and output as pressure outlet. Other B.C. are set as coupled walls.

My model finnaly converged after about 1500 iterrations and i didn't have any warnings during it, but final results have no any sense. The temperature is everywhere the shame despite the fact that I put heat surce(w/m3) in fuel and velocity is almost everywhere 0, even on the input. I tried laminar and turbelent flow (k, epsilon).
I was told that I had to improve my mash so i put edge sizing and inflation layers but it didn't help. Moreover in this case model didn't converge. What should I do and which mesh method should I use? I need someone who will increas mesh and get real results.
I am working in ANSYS fluent.

[Time4Tea]

My advice would be to go back to the basics. Make a simplified version of your model and get that working with just a momentum solution (i.e. the fluid velocity, no heat transfer), with similar boundary conditions. If you can't get that working, then you need to figure out why before moving on to something more complex with heat transfer.


It might also help if you post some images of your model. Are you able to make use of axisymmetry, to make it simpler and faster to run?

[igor_vidic]

Thank You very much for answering me. Yes I had the same strategy, so first I turned off energy eguation and I analyzed just fluid velocity. I get some result with default mash, but they are incorrect. I increased the mesh but I think the problem is that new mesh has low quality (big number of cells with akewness over 0.5), but I didnt manage to fix the problem. Sorry i am not sure how to use axisymmetry.

[Time4Tea]

My comments on the images:


1. It's not quite axisymmetric, but I think you can definitely use 1/4 symmetry for that case, to reduce the size of the mesh. I.e. just model a 1/4 portion of the model, with symmetry boundary conditions on the cut faces.



2. Your mesh is extremely coarse. I'd be surprised if you get anything even approaching useful with that. For flow through narrow channels, and especially when you include heat-transfer on top, diffusion of momentum and heat in the boundary layer is very important. So, it is vital that you adequately model the boundary layer.

For the meshing, I recommend you enable the 'advanced sizing function' and set the minimum number of cells across a gap to be at least 4-5. I also recommend you apply some inflation layers to the walls of the fluid area where the heat transfer is taking place.


Doing this will increase the size of your mesh and make it take longer to solve; however, adequate mesh resolution is necessary to get accurate results. For that type of constant-cross-section geometry, there should be a way that you can use a 'swept' mesh, so that the element size in the 'longitudinal' dimension is longer. That will help to reduce the cell count (as well as using the 1/4 model).


I also recommend you use the k-epsilon turbulence model, with wall functions.

[igor_vidic]

Thanks a lot for your detailed answer. I ll try all that you advice me.