[KevinZ09]

I'm trying to implement conjugate heat transfer in my simulation, but so far, I've been unsuccessfull.

First, some background info on my problem:
It's a large vessel (7m height) which contains a liquid metal at about 600K. Inside this large vessel, there's an inner loop that contains a heat source, a Volumetric Momentum Source (to force the liquid metal to flow) and a heat sink, as well as some porous media zones. The inner loop is open at two places so that the fluid can flow from the Main Vessel into the inner loop and back out again. Since inside the inner loop there's a heat source, obviously, the temperature of the fluid is higher than in the Main Vessel (about 100K higher).

To simulate buoyance and stratification properly within the Main Vessel, I need to account for heat transfer from the inner loop to the main pool through the walls. Hence my need to implement Conjugate Heat Transfer (CHT). My problem so far has been that whenever I try to implement it, I keep on getting temperature divergence error (Divergece detected in the AMG solver : temperature ). I need to set the CHT up on 13 different wall zones (coupled walls).

I'm using a Realizable k-epsilon model, with scalable wall functions. I've got a y+ > 30, grid, and use mainly 2nd order solvers. Pressure - Velocity Coupling with SIMPLE. I perform Steady - State simulations.

Here's what I have tried so far, with the "results":

- Thin Wall approach: seems OK for one wall zone, but 2-3 it diverges.
- Shell Conduction: same as Thin Wall approach.
- 1st order solvers: same problem.
- Transient runs: same problems.
- Multiple layers in Shell Conduction: same problem.
- Change CHT material/thickness: same problem.
- Constant material properties (I had them T-dependent): same problem.
- Lowered URFs: same problem.
- Body-force weighted pressure discretization: same problem.
- Run in serial mode: same problem.

Now, the interesting things are:
- Make solution isothermal, i.e., turn heat source and sink off: still divergence (temperature goes below 400K and above 900K, my limits)!
- Isothermal, no energy-equation: works fine.
- Isothermal, only energy (no turbulence/flow): slow divergence, i.e., slow increase in temperatures, though not immediate error.
- Improve mesh (had some tets in CHT areas): same problem.
- Use Fluent 17.2 (usually use Fluent 15): still error, though it happens slower. In Fluent 15, it crashes many times right away, while in Fluent 17.2 the temperature first slowly diverges.
- I also tried first spinning up the solution, i.e., no energy-equation, and then turn energy on: still divergence right away.

So that's my current state pretty much. If anyone has any ideas, please let me know cuz I'm running out of any.

Thanks!

Kevin

[attiquejavaid08]

Originally Posted by KevinZ09

I'm trying to implement conjugate heat transfer in my simulation, but so far, I've been unsuccessfull.

First, some background info on my problem:
It's a large vessel (7m height) which contains a liquid metal at about 600K. Inside this large vessel, there's an inner loop that contains a heat source, a Volumetric Momentum Source (to force the liquid metal to flow) and a heat sink, as well as some porous media zones. The inner loop is open at two places so that the fluid can flow from the Main Vessel into the inner loop and back out again. Since inside the inner loop there's a heat source, obviously, the temperature of the fluid is higher than in the Main Vessel (about 100K higher).

To simulate buoyance and stratification properly within the Main Vessel, I need to account for heat transfer from the inner loop to the main pool through the walls. Hence my need to implement Conjugate Heat Transfer (CHT). My problem so far has been that whenever I try to implement it, I keep on getting temperature divergence error (Divergece detected in the AMG solver : temperature ). I need to set the CHT up on 13 different wall zones (coupled walls).

I'm using a Realizable k-epsilon model, with scalable wall functions. I've got a y+ > 30, grid, and use mainly 2nd order solvers. Pressure - Velocity Coupling with SIMPLE. I perform Steady - State simulations.

Here's what I have tried so far, with the "results":

- Thin Wall approach: seems OK for one wall zone, but 2-3 it diverges.
- Shell Conduction: same as Thin Wall approach.
- 1st order solvers: same problem.
- Transient runs: same problems.
- Multiple layers in Shell Conduction: same problem.
- Change CHT material/thickness: same problem.
- Constant material properties (I had them T-dependent): same problem.
- Lowered URFs: same problem.
- Body-force weighted pressure discretization: same problem.
- Run in serial mode: same problem.

Now, the interesting things are:
- Make solution isothermal, i.e., turn heat source and sink off: still divergence (temperature goes below 400K and above 900K, my limits)!
- Isothermal, no energy-equation: works fine.
- Isothermal, only energy (no turbulence/flow): slow divergence, i.e., slow increase in temperatures, though not immediate error.
- Improve mesh (had some tets in CHT areas): same problem.
- Use Fluent 17.2 (usually use Fluent 15): still error, though it happens slower. In Fluent 15, it crashes many times right away, while in Fluent 17.2 the temperature first slowly diverges.
- I also tried first spinning up the solution, i.e., no energy-equation, and then turn energy on: still divergence right away.

So that's my current state pretty much. If anyone has any ideas, please let me know cuz I'm running out of any.

Thanks!

Kevin

are you using material properties constant or as polynomials of temperature? have you checked with piece-wise linear distribution covering temperature bounds of your simulation?

you're just talking about lots of complexities at same time. i'll suggest you start from simple case, adding zones gradually and check if your case works in between. more the variables/parameters to be considered for checking, more it will be harder to exactly know the source of issues

[KevinZ09]

Originally Posted by attiquejavaid08

are you using material properties constant or as polynomials of temperature? have you checked with piece-wise linear distribution covering temperature bounds of your simulation?

you're just talking about lots of complexities at same time. i'll suggest you start from simple case, adding zones gradually and check if your case works in between. more the variables/parameters to be considered for checking, more it will be harder to exactly know the source of issues

Thanks for the reply.

Initially, I had the material properties temperature dependent (through polynomials where possible, otherwise UDF). However, as you rightfully stated, there's many complexities in the model, so that's why I decided to switch to constant material properties.

And to even simplify the model further, I went to isothermal, i.e., T = 573.15K everywhere, and no heat source or sink anymore. Basically, all that's left are the porous medium zones, which are far from the CHT zones that I'm experimenting with, and the momentum source. I'm trying to remove those as well, and see from there.

I don't see how I can strip the simulation down further though. If you've got any suggestions, I'd love to hear them!

[attiquejavaid08]

a sketch/figure would be helpful.
are you sure it's not a meshing issue? run any simulation case which is diverging and just before iteration it diverges and stops, stop solver yourself and look at the temperature contours. examine that specific area/zone.
select one set of conditions and instead of assembling all the zone and solving, just start from few consecutive zones and see if your simulation goes on.
there are lots of do's and don't's which actually varies from case to case.

Sent from my SM-G900P using CFD Online Forum mobile app

[KevinZ09]

Originally Posted by attiquejavaid08

a sketch/figure would be helpful.
are you sure it's not a meshing issue? run any simulation case which is diverging and just before iteration it diverges and stops, stop solver yourself and look at the temperature contours. examine that specific area/zone.
select one set of conditions and instead of assembling all the zone and solving, just start from few consecutive zones and see if your simulation goes on.
there are lots of do's and don't's which actually varies from case to case.

Sent from my SM-G900P using CFD Online Forum mobile app

I'd have liked to add a picture, but unfortunately company policies doesn't allow me to do that....

I'm guessing/fearing it's a meshing issue, but the odd thing is that my model used to work quite well without the CHT. That is, with heat source and sink, the porous medium zones and momentum source, plus some other stuff. The model was working fine, sort of as expected. However, when then trying to add the CHT that's when the problem creeps up. I tried running the simulation with JUST CHT, i.e., none of the other stuff and isothermal, and it still crashes. So I guess it's a meshing problem, but weird it shows up with CHT and not with the other, more complex, stuff.

I know from experience CHT is a delicate process to implement, but I had thought/hoped it'd work in my current model.

As for when and where it crashes, it varies. Sometimes it works fine, nothing weird, meaning in the isothermal the temperature is 573.15K everywhere, when suddenly it goes beyond the limits (400K and 900K). So it's not necessarily a slow divergence, but many times just a sudden blow-up. And I do patch my Shell Conduction zones. So that's why I'm kinda surprised with the errors I'm experiencing...

[Cagatayemre]

Hello Kevin, my advice to you is to look at relaxation factors.

Originally Posted by KevinZ09

I'd have liked to add a picture, but unfortunately company policies doesn't allow me to do that....

I'm guessing/fearing it's a meshing issue, but the odd thing is that my model used to work quite well without the CHT. That is, with heat source and sink, the porous medium zones and momentum source, plus some other stuff. The model was working fine, sort of as expected. However, when then trying to add the CHT that's when the problem creeps up. I tried running the simulation with JUST CHT, i.e., none of the other stuff and isothermal, and it still crashes. So I guess it's a meshing problem, but weird it shows up with CHT and not with the other, more complex, stuff.

I know from experience CHT is a delicate process to implement, but I had thought/hoped it'd work in my current model.

As for when and where it crashes, it varies. Sometimes it works fine, nothing weird, meaning in the isothermal the temperature is 573.15K everywhere, when suddenly it goes beyond the limits (400K and 900K). So it's not necessarily a slow divergence, but many times just a sudden blow-up. And I do patch my Shell Conduction zones. So that's why I'm kinda surprised with the errors I'm experiencing...

[sanjaysajeevan]

hi,
can u please guide me that how u applied conjugate heat transfer in fluent please?.. as part of my project i am running out of time....please