CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   CFX Natural Convection, please help. (

dbecker October 13, 2010 17:24

CFX Natural Convection, please help.
4 Attachment(s)

I am tasked to perform a natural convection analysis on electrical components. I am using CFX 11.0 and ICEM.

Before I began the analysis I wanted to see if CFX is capable of performing this type of analysis. I made a simple cylinder in a box, the cylinder extends to the walls of the box and I put slip planes on the end walls to simulate infinitely long cylinder. I put no slip planes on the sides and top and bottom.

Cylinder is set at 500C and roof and floor and side walls are at 55C. I turned on bouyancy and I am using shear stress transport because ultimately I am looking for heat transfer coefficients on the cylinder.

I ran the solution and this is the result, heat transfer coefficients are about right according to my hand calcs for a cylinder in air with this deltaT, but the gas temperature distribution is almost all uniformly 55C. This is wrong, there should be a spread of temperature between 55 and 500. This does not make sense and I need to know how to validate and run natrual convection models in CFX.

I tried using the help file and there is about 2 paragraphs on the subject and that's it. Does anyone have any experience using CFX for natural convection? Please help!


- D

michael_owen October 13, 2010 17:30

I'm not sure what model you are looking at, but there is a spread in temperature between 55 and 500C. Look at your second graph.

dbecker October 13, 2010 17:42

It looks almost uniformly 55C to me, maybe a little more, but there isnt much effect on gas temperature due to the rod which is why I ask.

michael_owen October 13, 2010 17:48

Probe it. The bottom two ticks on the legend are 55 and 166 C.

Do you have some expectation of what it should be that is different from what it is?

dbecker October 13, 2010 18:11

3 Attachment(s)
Alright take a look, I think the variation is there but it's just very suttle. I think I was expecting to see larger gradient, gradient exists but it's almost as if this is all conduction with little convection. Something I wasnt expecting with a 500C rod in a 55C room.

Gas is moving up which is a good thing; at least bouyancy is in effect.

By the way gravity is -9.81 along X which is vertical in this case.

I dont know why but my gut is telling me something different.

ghorrocks October 13, 2010 19:07

Is this a steady or transient model?

If transient then don't you just have to run it longer to get more hot gas in there so it is easier to see?

Another point is that over the range 50-500 deg C the boussinesq assumption for buoyancy is quite inaccurate. You will probably need to use an ideal gas approach, also maybe with variable properties (Cp, viscosity mainly).

All times are GMT -4. The time now is 00:22.