CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Calculating Wall Temperatures in CFX (http://www.cfd-online.com/Forums/cfx/104346-calculating-wall-temperatures-cfx.html)

Cthames21 July 7, 2012 14:16

Calculating Wall Temperatures in CFX
 
Hello,

I am interested in calculating wall temperatures for the face of a cylinder, that has an axially varying heat flux with turbulent flow at the fluid-wall interface. I have performed this type of simulation using both the k-e turbulent model and SSG turbulent mode. I found that there was no wall temperature variable located in CFX-post, so I created an expression that should calculate the wall temperature directly. Namely,

Wall Temp = (Wall Heat Flux)/(Wall Heat Transfer Coefficient) + Wall Adjacent Temperature

For the k-e model, this method is producing somewhat accurate results. The SSG model is producing results that are up to 100 Kelvin away from the expected answer. This is due to the decrease in the convective heat transfer coefficient axially to such small levels that it causes the first term in my direct calculation to dominate. So this leads to both of my questions:

1. What could be causing the SSG model to predict such low wall heat transfer coefficients knowing that all boundary conditions are identical in both my k-e and SSG runs?

2. Is there another method to calculate wall temperatures in CFX-post?

Thanks,

Cameron Thames

ghorrocks July 8, 2012 07:33

1: Very good question. No idea, I am interested if other people on the forum can help on this one.
2: I seem to recall in a simulation with wall heat sources I did a while back that I could access the wall temperature. Are you sure the Temperature variable is not available on the wall?

Another point - if most of the flow is laminar and turbulence is tripping on the cylinder then the SST turbulence transition model may be a better model to use. You might have to manually set the turbulence inception point as that is well known on a cylinder and the built-in correlations might not be ideal (as they were developed for airfoil flows).

Cthames21 July 8, 2012 10:52

Thank you for the response. My original post may have been misleading with stating that the flow is turbulent at the fluid-wall interface. The flow is completely turbulent throughout the whole system.

Also, I was aware of the temperature variable and had used it to evaluate the average fluid temperature as a function of axial position. If instead of choosing a plane within the fluid for the temperature variable, I can choose the wall itself to obtain the appropriate wall temperatures?

Thanks,

Cameron Thames

ghorrocks July 8, 2012 18:42

I suspect so, but I am not sure. You would have to check it to make sure it is OK before proceeding.

juliom July 9, 2012 11:39

Dear Cameron
It is very weid thaht you can not see the temperature at the wall. In CFX post, you just have to select the wall choose the varaibe you want to look at.. This part I canīt understand.
With your second question, I am quite sure the problem is about the turbulence model, have you chequed how turubulent your flow is?
Because SSG has some limitations as the other models!!
Regards! and good luck!

Cthames21 July 12, 2012 20:13

Update:

It seems that my mesh was the issue. My previous mesh was sitting at about 6 million elements. I have since refined the mesh to approximately 48 million elements. Both the k-epsilon model and SSG model are predicting wall temperatures very well when using the method I described in my first post. The SSG model has produced wall heat transfer coefficients that are nearly identical to the data from which I set as my basis. The k-epsilon heat transfer coefficients are slightly higher than expected, but are still within reason.

Thank you for your help,

Cameron Thames

flotus1 July 13, 2012 00:46

Keep in mind that in CFX, an "expert parameter" "Tbulk for HTC" has to be checked so that heat transfer coefficients can be used in a meaningful way.

Otherwise, the heat transfer coefficient is messed up with the wall adjacent temperature itself, which makes no sense from an engineers point of viev.

VIPIN KAKKAR April 2, 2015 09:54

Quote:

Originally Posted by Cthames21 (Post 370255)
Hello,

I am interested in calculating wall temperatures for the face of a cylinder, that has an axially varying heat flux with turbulent flow at the fluid-wall interface. I have performed this type of simulation using both the k-e turbulent model and SSG turbulent mode. I found that there was no wall temperature variable located in CFX-post, so I created an expression that should calculate the wall temperature directly. Namely,

Wall Temp = (Wall Heat Flux)/(Wall Heat Transfer Coefficient) + Wall Adjacent Temperature

For the k-e model, this method is producing somewhat accurate results. The SSG model is producing results that are up to 100 Kelvin away from the expected answer. This is due to the decrease in the convective heat transfer coefficient axially to such small levels that it causes the first term in my direct calculation to dominate. So this leads to both of my questions:

1. What could be causing the SSG model to predict such low wall heat transfer coefficients knowing that all boundary conditions are identical in both my k-e and SSG runs?

2. Is there another method to calculate wall temperatures in CFX-post?

Thanks,

Cameron Thames

sir, please tell me how to find wall adjacent temperature. what is expression to find this.

Thomas MADELEINE April 2, 2015 10:13

Have a look on the help they explain what the variables are about.

Wall Heat Flux: well it is obviously the heat flux between fluid and wall
Wall Heat Transfer Coefficient: Heat Flux divided by the Temperature difference (with the first cell normally or the specified value for tbulk)
Temperature: is the temperature of the selected item on CFD-Post (Wall for example).

The Turbulence model and the mesh refinement to the wall is here very important (specially with HTC for first cell).


All times are GMT -4. The time now is 22:48.