Good morning,

i have noticed that CFX11 calculates heat transfer coefficient also when adiabatic walls and a turbulence model (not laminar) are set in CFX-pre (obviously only if a thermal simulation is also set). This capability was not present in CFX10.

What is the analytical formulation used by CFX to compute the heat transfer coefficient in this case?

I know that when a thermal simulation is set and the walls are not adiabatic with a wall temperature fixed, CFX calculates the wall heat flux (Qwall) at wall boundaries and then the HTC as HTC=Qwall/(Twall-Tnear_wall), where Tnear_wall is a Tbulk or the temperature of the first node (if a Tbulk is not specified).

Thank you very much.

Er... isn't the HTC at an adiabatic wall 0?

Hi Alba,

Internally, the turbulence model is actually used to calculate an HTC at the wall, which is then used to determine the fluid temperature and wall temperature or heat flux (depending on the boundary condition), not the other way around. Regardless of the boundary condition, the solver can determind an HTC. If the heat flux is zero, as it is with an adiabatic wall, the HTC will have no effect on the near wall temperature, but it's output anyhow. You can look this up in the near-wall treatment section of turbulence modeling in the documentation.

I don't recommend using this HTC. A heat transfer coefficient is not a real physical quantity, like temperature or heat flux, but more like a linearization of their relationship. The HTC reported by the solver is dependant on the height of the first element; it's not the same as one that you might calculate from experiment.

If you're comparing to experiment, back out the HTC using the same method (i.e. measure the wall temperature and heat flux and compare to a reference temperature).

If you apply an HTC as a boundary condition for a thermal analysis in ANSYS (or another FEA package), compare the temperature and heat flux to that of the CFD solution. The difference gives you an idea of the magnitude of the error between the two. If you can, run a CHT analysis to fully couple the energy solution.

-CycLone

Hi Cyclone,

you say that the HTC reported by the solver is dependant on the height of the first element.

But i made some tests on a straight circular pipe, imposing a mass flow at inlet, a pressure at outlet, k-epsilon turbolence model, adiabatic wall for the tube and thermal energy. The only thing that vary between the different models is the height of the first prism. Plotting the HTC i have notice that for all models its value is always the same and it is near to the value calculated with the Dittus_Boelter correlation: (Kw/D)*0.023*(Re^0.8)*(Pr^0.4).

The simulations shows that (only when adiabatic walls are set) HTC is independet from the height of the first element.

The question is different if the walls are not adiabatic. In this cases the HTC depends on the height of the first prism or on the Tbulk.

The question is: does CFX use only Dittus Boelter correlation to calculate HTC? This correlation has a low precision and should be valid only inside pipe.

Thanks.

Alba

Hi Alba,

CFX isn't using the Dittus Boelter correlation. I recommend you review the documentation to get more information.

-CycLone