Conduction heat not balanced?
 

Hi all,
I'm testing a conduction only case (no convection , no radiation), but the Heat added is not same as the solids gained!
my case is:
(dimension unit is INCH)
A solid(aluminum) cuboid 2x2x2 with 1x1x1 cuboid inside it. The small cuboid served as a heat source ,and now my condition and result shown below,<font color="blue">
1.) all solid initial temp= 0 degF, heat source (Total source) 20 Kg-m^2/S^-3(ie 20 watts), and simulation time=100 s.
2.) after run , I set the heat source as 0 watts and run it for steady state by using the previous result as initial result.
3.) the final solid temp=22.4683 degF </font>
and then I check the Heat added/gained are,
Qin=20 watts x 100 sec= <font color="red">2000 watts</font>
Qgained=M x Cp x dT
where M=Volume x Density=2 x 2 x 2 x (0.0254)^3 x 2702 =0.354223 Kg
Cp=903 [J kg^-1 K^-1]
dT=22.4683-0=22.4683 degF=12.4824 degC
So Qgained=0.354223 x 903 x 12.4824=<font color="red">3992.656 Watts</font> ?

Why I supply 2000 watts into the solid and I gained 3992 watts?
Is there something wrong with my calculation?

Sorry, the last result unit is Joules
 

Sorry, the <font color="red">2000 </font> and<font color="red"> 3992 </font> unit are Joules!

Re: Sorry, the last result unit is Joules
 

I know what's wrong with it! It's due to the transient run with residule target too high. The heat balances as I set the residule target to be 1e-7. <font color="blue"> So what's the criteria for residule target in transient simulation? It will cause so much deviation, so could someone can give me some hint.</font>

Re: Sorry, the last result unit is Joules
 

Hi,

First of all: how the hell can you post with colors ?!?

About your problem: 1e-7 too high??? In the CFX-5.7.1 documentation Solver Modelling, Advice on Flow Modelling, take a look at page 363 and specially page 364 (Judging Convergence).

Are you sure you set eveything right in your simulation? The calculated Qgained is nearly the double of the calculated Qin!

In transient simulations, the Residual Target will only be achieved if you allow the Solver to run enough number of iterations. What's the value of "Maximum Number of Iterations per Time Step"? Take a look at the CFX-Solver Manager, and see if the Residual Target is achieved (a value of 1e-5 will be fairly enough).

A possible cause for your problem may be that, even if obtaining convergence you are not obtaining conservation. So, a possible solution may be to set a Conservation Target.

To check if the problem comes from the transient simulation or from the steady-state simulation, at the end (or at any instant) of the transient simulation, you may calculate the average temperature: volumeAve(Temperature)@Domain . As the solid density and Cp are constants, you may calculate Qgained with this average temperature,and compare with Qin.

Regards,

Rui

Re: Sorry, the last result unit is Joules
 

Hi Rui,
Thanks,
The first time, I used convergence target as 1e-4, so I got so much difference between Qin and Qgained.
As I set convergence target as 1e-7, then Qin and Qgained are almost the same. I didn't setup Conservation Target.
<font color="Green">
About how to post in color:
It's only html code added inside message
for ex.</font><font color=blue>
1.change color to red: You should put < font color="red"> before the text , and put < /font> after it.
2.The use the preformated type: put < pre> and < /pre> to wrap the preformated text(the blue color texts are set by < pre>
3.To change into a new line, just type < br>
</font>

Re: Sorry, the last result unit is Joules
 

Hi,

As Rui says, converging on residuals AND conservation is useful in CHT simulations such as this. I would definitely recommend adding a conservation convergence requirement.

Regards, Glenn

Re: Sorry, the last result unit is Joules
 

Thanks to Glenn Horrocks. I will do it from now on ! Nepal.

Re: Sorry, the last result unit is Joules
 

<font color="red">Hi <big>Nepal</big></font>

<font size=6 color="blue">Thanks for the hint</font>

<font size=3 color="green"><big> Regards,</big><font color="brown" type=italic> Rui</font></font>