Heat transfer coef in Fluent
 

Hi guys,
Im trying to simulate the flow through the reactangular duct with pin-fin
Anyone know about the method fluent use to calculate heat transfer coef ?
I just know Fluent has 2 method to calculate.
First, Fluent use reference value we specified to calculate following q=h(Tw-Tref) and Tref is a constant. I calculate it from Area Weight Average Surface Heat Transfer Coef. So the result Nusselt will depend on Tref which we choose.
Second, Fluent use the first cell nearest the wall to specify Tref. So we have Wall function surface heat transfer coef and it depends on the mesh.
I think the second method is better and correct. But when I have a good mesh with Y+<1 near the wall. Fluent calculates Wall Function Surface HTC is very high, it's make Nu so high, whereas follow Nu=0,08Re^0,723*Pr^0,4 or experiment is approximately 46.925.
Thanks for your time and i'm looking forward to hearing from u.

Honestly, I never understood why the second method is even a thing in CFD solvers. It is ambiguous, prone to misinterpretation, and borderline useless in an engineering context.
My only advice is to use the first one with a fixed reference temperature whenever possible. That's just the opinion I formed having to deal with heat transfer in real-life applications. Maybe htc based on some non-constant reference temperature has other advantages, but I have yet to come across such a case.

So how do u specify the reference temperature? And the reference temperature we input is constant, isn't it?
And I'm reading a article about pin-fin but I can't find the heat flux, they just said that heat flux is constant. I confuse about it. Have u ever meet this case?
This article attach below

This is my article

Same as any other reference quantity: make something up. Ideally something that corresponds to an actual quantity in your simulation. Like e.g. temperature at the inlet. Or if you are trying to recreate an experiment, or another simulation: use the same values they used. And more importantly: when documenting your findings, write down the values you used.
And yes, that reference quantity is usually a constant value.
I can't open the document you linked.
By the way, you can edit your posts, that's usually better than spamming lots of new posts.

Sorry, i did but the link did not appear. Please try to access again. If it still dont work, please respond to me on private message. Thanks for your help

The first method comes from Newton's law of cooling, which is how we define heat transfer coefficient, so that is always the correct method.

Heat transfer coefficient and Nusselt number has a semi-arbitrary reference temperature in it. It plays the same role as the reference length and reference velocity in Reynolds number.

When you use the wall function heat transfer coefficient, you are using the T* as the reference temperature, y* as the reference length, and u* as the reference velocity. You cannot go and compare this Nusselt number with a correlation that uses a Reynolds number using average inlet velocity.

So for example since you want to use Nu=0,08Re^0,723*Pr^0,4

You need to go look up all the reference values (temperature, length, velocity) and use the same ones. That formula is not a formula for wall function heat transfer coefficient.

I already told this person to do this in this thread: https://www.cfd-online.com/Forums/ma...tml#post801433
Guess they didn't like/trust my answer.