CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   natural convection problem in cfx (

NVSD BABU November 11, 2008 07:05

natural convection problem in cfx
hi to all

i am the beginner of CFX, for practice i have taken a simple copper plate, and enclosed it in a steel enclosure which is opened at bootom and closed at top,i passed a current of 500amps through the copper bar, same set up made in cfx and i solved but the results are about 50% higher than the experimental values and air inside the enclosure not increasing more than 2degrees but experimentally there was an increse of 8-14degrees.

Can any body help me to resolve this problem.


pratik mehta November 11, 2008 10:36

Re: natural convection problem in cfx

Well, make sure your boundary conditions are close to that of the reality. Like the bottom , I think it would be good to set an opening . Check the direction of the gravity , find out which option of model is best. Also use the material properties as in the experiment since in the CFX library they are very general . try out these , I hope you a good solution.

Rogerio Fernandes Brito November 11, 2008 19:45

Re: natural convection problem in cfx
Take care about the material properties, if they vary with the variation of the temperature or not. Try to multiply the Cp number by 1.7 (or 1.5) for example, and give a look on the results.

NVSD BABU November 12, 2008 02:03

Re: natural convection problem in cfx
thank you mehta.

i wanted to give a clear description of the problem.copper bar dimensions are 1250*80*50mm and steel enclosure dimensions 1630*510*150mm with sheet thick modeling i have created inside air volume and out side air volume, i have given an interface to inside air and out side air at the steel cabin opening.i have given opening boundary condition to the all the outer faces of the outer air volume.In opening boundary condition boundary details mass and momentum option set as opening pressure and drn,relative pressure as zero,flow direction normal to boundary condition.

In my model +Z-direction showing downwards and i have given g in Z-drn is 9.81.In meshing,i have given inflation for inside air volume(i.e.air in steel cabin)and out side air volume.

plaese help me to solve this problem,and tell whether my approach is correct or not,and how to decide whether the flow is laminar or tarbulent. i have calculated raleigh no it is 4.5E+9 at trailing edge of the copper plate.

Thank you

pratik mehta November 12, 2008 04:24

Re: natural convection problem in cfx
Hi, thanks for the description of your problem, if your domain is vetically standing on say +Y axis , so then mention the gravity as -9.81 which means that you will get help to develop a natural convection flow field inside the enclosure and also on the outside . But what is the temperature diff in your domain .

I think Ra number is good for a turbulence to exist. Unfortunately I am not aware how to switch off turbulent quantities calculation in places of your laminar region in CFX , but in fluent one can get this option.

Anyways try to get good y plus as well and fill in the material properties as accurate as possible .

Amod November 12, 2008 10:50

Re: natural convection problem in cfx
,------------------, | | | ,--, | | | | | | | | | | | | | | '--' |Steel Cabin | |


I guess this is the domain. There is not need to model air outside steel cabin volume. Please check the following: 1. The heat source is correct: The unit should be W/m^3 in case you are modeling volumetric heat source. You can also model total heat source in that case CFX will calculate heat generation/volume from geometry of the Cu plate. What is the volumetric heat source you have defined in Cu domain?

2. The choice of approximation, Ideal Gas or Boussinesq alos makes lot of difference. Please follow the solver recommendation for defining Reference Density?

3.Why are you asking basic question whther flow is turbulent or not? You should 1st understand the physics behind natural convection. Any book on H.T. will tell you the range over which flow becomes turbulent! Caution: While calculating Ra no., do take care of the temperature at which properties of the fluid are evaluated. Typically, it is the wall temperature (which is not known in your case). Hence you will have to make a wise guess.

Hope this will help you!


NVSD BABU November 13, 2008 02:24

Re: natural convection problem in cfx
Thanks amod,

Before modeling in CFX i conducted experiment, based on that i calculated Ra No at trailing edge, that is >1e9,so i have given turbulence for CFX model.

i have created a sub domain in copper region,and i given heat source as 33911.44 w/m^3. for air i have selected air at 25 degC. Please tell me on what basis, we have to give Ref buoyancy temp or Ref density.


NVSD BABU November 13, 2008 11:56

Re: natural convection problem in cfx
Hi all

In this problem i have meshed air,copper and steel separtely, and these GTM files imported in to the CFX-pre. can i do it in this way or should all volumes mesh together?


All times are GMT -4. The time now is 19:29.