CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Heat transfer problem (http://www.cfd-online.com/Forums/cfx/75774-heat-transfer-problem.html)

seojaho May 5, 2010 02:18

Heat transfer problem
 
Hello,

I have a model with three domains (two solid domains, one fluid domain).
One solid domain is for hot plastic melt. (Actually, this domain should be a fluid domain but hot plastic melt does not move and is just contained into this domain. So set this domain as a solid domain for the hot melt)
The other solid domain is for steel into which the heat from the hot plastic melt is transferred.
The remaining fluid domain is for water as a coolant. So the water cools down the steel domain.
When I finished the simulation, it seems that the heat transfer is not conducted from the melt to the steel because the temp. of the steel is increased just little.
Also, the heat transfer from the steel to water is not successful since the temp. difference between inlet and outlet of water is quite small.

(Addtionally, the simulation type is 'transient' and the heat transfer coefficient of ambient air for natural conection of the steel domain is set to 40 W/m2 K.
The initially given hot melt is cooled down by water's cooling without any additional melt during the simulation time. So, I set the initial temp. (204C) of the provided plastic melt an initial condition for the melt domain. )

Please let me know how to solve this unsuccessful heat transfer problem on solid-solid interface and solid-fluid interface.
Thank you.

mvoss May 5, 2010 04:06

hi,

what about convergence? what about the imbalances?

neewbie

seojaho May 5, 2010 15:10

For convergence control and convergence criteria, I gave 100 as max.coeff.loops and RMS (as residual type), respectively.

Convergence results:
--------------------------------------------------------------------
| Equation | Rate | RMS Res | Max Res | Linear Solution |
+----------------------+------+---------+---------+------------------+
| U-Mom | 0.89 | 3.2E-06 | 8.2E-05 | 3.5E-02 OK|
| V-Mom | 0.81 | 7.1E-06 | 3.0E-04 | 3.1E-02 OK|
| W-Mom | 0.81 | 3.0E-06 | 6.0E-05 | 4.2E-02 OK|
| P-Mass | 0.89 | 4.4E-07 | 1.8E-05 | 8.7 3.3E-02 OK|
+----------------------+------+---------+---------+------------------+
| H-Energy | 0.96 | 5.9E-05 | 1.6E-03 | 4.7E-02 OK|
| T-Energy | 0.95 | 1.1E-04 | 1.9E-03 | 5.3 4.7E-02 OK|
+----------------------+------+---------+---------+------------------+
| K-TurbKE | 0.80 | 3.8E-06 | 1.8E-04 | 5.6 1.7E-02 OK|
| E-Diss.K | 0.82 | 5.1E-06 | 1.4E-04 | 9.6 4.1E-03 OK|
+----------------------+------+---------+---------+------------------

The average of each energy imblance is:

water domain: -1.6e-5(H-energy imbalance)
hot melt domain: 0.001 (T-energy imblance)
steel domain: 0.047 (T-energy imblance)

Thank you so much.

ghorrocks May 5, 2010 19:00

Your steel imbalance is large (5%). You need to further convergence to get the steel imbalance down, aim for better than 1% but you might need tighter than that.

seojaho May 5, 2010 23:45

Dear Glenn,

Thank you for kind help everytime.
From http://www.cfd-online.com/Forums/cfx...exchanger.html, you suggested useful tips to solve this kind of problem.

In my case, the simulation is transient. So I cannot control solid domain's timescale which is only valid in steady state simulation. (So, 1) tip is not unavailable). Also I applied GGI interfaces ( 2) tip) but it cannot contribute in my case. For 3) which is convergence parameters, another parameters except the H, T-energy imbalance parameters above are required?
For 4) tip, I gave 100 as max.coeff.loops for convergence control.

So, I wonder how I can further convergence to decrease the steel imbanace...

Thank you so much and have a great day, Glenn.

Jaho

ghorrocks May 5, 2010 23:57

The most important thing in this case is to set a convergence criteria of 0.01 on the imbalances (although you may need tighter - that's up to you to determine) AND the normal residual criteria. You will need to give the solver enough coefficient loops per time step to achieve that convergence tolerance.

seojaho May 6, 2010 00:32

Thank you for quick reply and help.
I will try to do that.
Have a great day, Glenn.

Jaho


All times are GMT -4. The time now is 18:05.