[h0rst]

The problem is, if I do not use a UDF and use a constant heat flux, it is very imprecise since my heating sources are changing a lot.

Besides that, I don't know where to change it in the boundary conditions since my interface between Fluid and Solid is coupled.

[pakk]

You are contradicting yourself.

Q°= h*(T_Surface-T_Fluid)

You said that h is fixed, T_Surface is fixed and T_fluid is fixed. All of them are known by you, and are values that you give to Fluent.

How is that imprecise? What can vary?

And I asked you if you were calculating the wall temperature, and you said to me that you give the wall temperature. But now you talk about a coupled interface between solid and fluid, which suggests that you calculate the wall temperature. So, what is it? Make up your mind!

[h0rst]

My aim is to use the heat transfer coefficient from Literature and own measurements. I would prefer to use.

Q° and T_Surface, T_Fluid

from Fluent but if this doesn't work, I would use mean-values of that.

Of course I cannot state all 4 values since than Fluent cannot calculate.


If you are right and the problem is the difference between cell temperatures and the actual fluid temperatures, how can I use the mean fluid temperature for letting Fluent calculate the heat flux?

[pakk]

Originally Posted by h0rst

If you are right and the problem is the difference between cell temperatures and the actual fluid temperatures, how can I use the mean fluid temperature for letting Fluent calculate the heat flux?

My advice: DON'T use the mean fluid temperature! That one makes no sense!

If you want to have the most detailed model, use local temperatures just as Fluent is doing. This is what physically happens, so it is the best.

If you want to have a simplified model, calculate (with pen and paper) the heat flux by using the averaged temperatures, and then put this heat flux as constant uniform boundary condition in Fluent.

But if you combine these two models in the way that you are trying, you take the worst part of both. You take the complexity of the detailed model, and the inaccuracy of the simplified model. Don't do that!

[h0rst]

Okay, but why is there such a difference if I extract the mean-average wall temperature heat transfer coefficients from Fluent and then put it back in by using

cid[1]=h
cid[2]=h?


Shouldn't the result be nearly the same?


Best regards
h0rst

[pakk]

If by 'nearly the same' you mean a difference smaller than 1%, I don't see any reason to think that that would be true.

[h0rst]

I've checked, it's not the same at all, temperature is very much higher if I apply cid[1] and cid[2] = h.

If i choose cid[1] and cid[2] = 2*h, then the temperature is almost the same (approximately 0,5 % difference).


Best regards
h0rst

[pakk]

So there is no reason to suspect that these two different situations would give a similar result, the indeed your results show that they don't. Accept that, and move on.

I gave you my advice already.

[h0rst]

I find it a bit strange that these are two different situations.

I extract a value and put it as an input variable in without changing so i would expect the same result

[h0rst]

Anyway, pakk, thank you very much for your time and effort

[pakk]

Originally Posted by h0rst

I find it a bit strange that these are two different situations.

I extract a value and put it as an input variable in without changing so i would expect the same result

I have the feeling that we are talking about something different...
The two situations that I call different:

1. Calculate heat transfer using average temperatures.
2. Calculate heat transfer using local temperatures.

You agree that the local temperature is different than the average temperature, right?

You extract the heat transfer from situation 1, and use it in situation 2. I don't know why you would expect the same result.

[h0rst]

I extract the wall function heat transfer that is created by fluent.

When I plot it, the value varies between 11000 and 15000. The average value is 13000. If I use this value for cid[1] and cid[2] I expect that the result is similar because 11000 and 15000 is not a very big range.
Instead, I need to use the value 26000 to get a very close temperature distribution.

Later on, I want to apply own values (about 15500) but as already this previous test show, setting cid[1] and cid[2] to 15500 creates far too high temperatures whereas cid[1] and cid[2] = 2*15500=31000 fits quite well.