[cal25]

I am currently modeling a heat exchanger using chtMultiRegionSimpleFoam and the k-epsilon model. The case is set up and runs fine, the only issue I am having is that I am severely OVERpredicting the heat transfer when compared to experimental data (>100%). It is also important to note that I am UNDERpredicting heat transfer when using the laminar assumption, so the solution lies somewhere between the turbulent and laminar solutions I am currently getting.

I have been testing multiple parameters in an attempt to isolate where the overprediction is occuring, and I believe I have narrowed it down to the effective thermal conductivity (kappa_eff). kappa_eff is fairly large in my simulation compared to the thermal conductivity of water. By reducing kappa_eff, the amount of heat that transfers through the boundary conditions would decrease and alleviate the overprediction I am currently getting.


I believe that effective thermal conductivity is either solved from the turbulent thermal diffusivity (alphat) or the turbulent Prandtl number (Prt)

If it is from Prt:
kappa_eff = cp*mut/Prt
where cp is a constant specific heat capacity and mut is the turbulent viscosity

If it is from alphat:
kappa_eff = alphat*cp*rho
where rho is the density

I have 2 questions for the problem I am having

1) Are one of those options the way that the effective thermal conductivity is calculated?

2) What could I change to lower the effective thermal conductivity? (examples: reducing Prt, altering the turbulence model, reducing alphat)

Thank you in advance for any advice

[kalle]

You are relying quite a bit on the turbulent wall functions for your case. What wall BCs are you using and how does your y+ look?

[cal25]

The wall boundary conditions are as follows:

Temperature - compressible::turbulentTemperatureRadCoupledMixed
U - fixed value (0,0,0)
p_rgh - zeroGradient
k - compressible::kqrWallFunction
epsilon - compressible::epsilonWallFunction


As for my y+ here is the output from the yPlusRAS

Selecting RAS turbulence model kEpsilon
Patch 0 named HXout y+ : min: 661.4 max: 1195.77 average: 1048.74

Patch 1 named HXin y+ : min: 117.625 max: 568.273 average: 270.882

Patch 2 named HX_to_tunnel y+ : min: 5.35252 max: 3280.44 average: 295.612

Patch 3 named HX_to_plug2 y+ : min: 7.61621 max: 65.7355 average: 27.2033

Patch 4 named HX_to_plug3 y+ : min: 7.48487 max: 56.0866 average: 29.122

Patch 5 named HX_to_plug1 y+ : min: 7.22408 max: 59.7763 average: 27.5941

Writing yPlus to field yPlus

End


There are some areas that have significantly large y+ values. I was attempting to minimize the number of nodes I need (currently at 6M) but it looks like I may have to increase the number of nodes near the wall.