CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   boundary conditions for Natural Convection problem (

Nav June 23, 2008 05:19

boundary conditions for Natural Convection problem
Hi i am dealing with a buoyant flow problem,where there was some heat source and sink attached to it.airvolume was created around the two domains.What boundary conditions best suit the problem if the domains are considered to be placed in open atmosphere...

plz suggest over this issue.. thanks

pratik mehta June 24, 2008 02:10

Re: boundary conditions for Natural Convection pro
ok, as I understand that your system is kept in open atmosphere. Kindly have a sufficiently big domain to aviod backlflow for all the faces of that domain, pressure outlet condition or Outlet as in CFX. One of side of that domain will be wall which is your ground. Later turn on the bouyant flow option. Check the direction of your gravity and set accordingly with proper sign . Try solving it in transient mode.

best of luck

Nav June 30, 2008 00:44

Re: boundary conditions for Natural Convection pro

Pratik Mehta has suggested me regarding Boundary conditions.... Basing on these conditions i tried with 3 cases By having

(1)5 outlets and 1 wall

(2)4 outlets 1 opening 1 wall

(3)same case with different volume

as airvolume domain was having a shape of cuboid

I had simulated according to the boundary conditions suggested by Pratik mehta, but the problem i faced was that--the reversal of flow flow was in the direction of gravity.would it be possible to any other forces would act to divert the path buoyant flow could be dominated ??

plz could any one suggest regarding this aspect..

Thanks in advance

Nav June 30, 2008 03:17

Re: boundary conditions for Natural Convection pro
hi pratik

you had suggested me to solve the problem in transient state --plz could you suggest me total time ie total duration that would best suite for results to be obtained.

i had seen a tutorial but it was of 2D case and in it was taken 2 sec ,but my case here would be of 3D -- a heat source would conduct heat to other solid domain and by then get cooled through air....??ie total of 3 domains.----2 solid and 1 fluid domain.heat from 1 solid domain has to be transferrred to other ,and through natural convection it has to be cooled..

plz kindly help

pratik mehta July 2, 2008 07:47

Re: boundary conditions for Natural Convection pro
Hello Nav,

I can best suggest you the method to solve such flows if you kindly give me the sketch or a geometry of your problem. or explain your study of interest.

Anyways I think the atmospheric boundaries must be kept to outlet and Kindly put your reference pressure 1 atm. Pressure at the outlet boundaries should be 0 bar. set your gravity to -9.81 if your object is resting on flat ground.

Natural convection flows are always transient since there is a big coupling betweeb velocity and temperature. Bounyance has to dominant in your problem since thats the only force you have driving your flow

pratik mehta July 2, 2008 08:05

Re: boundary conditions for Natural Convection pro
ok, I am sure you have to simulate your problem with correct boundary condition at the atmospheric wall . Have you given them Outlet.

I have done natural convection transient simulation but in fluent , I think it should be same method in CFX.

start your simulation with high time steps like 10 , run it for some time till you observe velocity field getting developed , velocity should be monitered. As soon as the flow has kind of developed ,switch on your energy equation and reduce your timestep gradually from 5 , 3 ,1 ,0.1,0.001 , finally the flow should behave as you want , The flow should go from down to top near your heat source . Try this method ,

best of luck

Nav July 4, 2008 05:00

Re: boundary conditions for Natural Convection pro

glad to see ur response.... thanks for guiding regarding intermediate time steps .but had little doubt like what should be the total duration of whole transient case. if 30sec or 60 sec would be suficient for the case to run and plot the results well..

plz kindly respond to this...

thanks in advance

pratik mehta July 7, 2008 07:18

Re: boundary conditions for Natural Convection pro
I have been using CFX but only for steady state simulation so far . So I am not sure how CFX performs transient . But what you must keep in mind is that your time step should be progressive like the one mentioned in my pervious post . To be very frank its just the matter of experience ,which you will get after few trails . Let me know what are your parameters in CFX . I wil see then , donīt forget to keep a watch on, how velocity is evolving in the domain.Try to figure out the Grashofs number to see if your case is at all turbulent or not :)

. See the vector plots well. Study them well ,

Best of luck

Nav July 8, 2008 04:01

Re: boundary conditions for Natural Convection pro
hi Pratik thanks for the response

could you help me in plotting Grashofs number.

initially i was running the case in steady state then by having those initial values i am running it transient case..

i am trying to give heat input as two cases as two different problem 1)heat generation value ie by defining sub domain 2)heat flux value in the interface region

plz suggest me ,thanks in advance

ADI May 10, 2010 15:10

Similar convection case
Hi Pratik Mehta,

I am doing a similar case but with Fluent. I have a cylinder that is 3ft tall and has a 2ft wide diameter. it is filled with water and the top is left open. Of the 3 ft. long side wall, 20% of it near the bottom is being heated so that the temperature is about 10K to 20K more than the rest of the walls. There is no other induced flow rather than the bouyancy driven flow and the flow due to the water being heated and rising from the bottom to top. I am interested in solvong for the behaviour of of the boundary layer as it grows along the wall. my gravity is -1 in y direction. (yeah it is in micro gravity condition)

Can you help regarding how I set my time step size and number and also my boundary conditions. (i am doing 2d to start with)

Rez May 10, 2010 17:54


Outlet boundary conditions might not be the best for you. Depending on your problem, you might have to use symmetry or walls. Generally, you should have the domain as large as possible with respect to your heat sources/sinks. For example in case of heat sources/sinks in open atmosphere, you might want to have a large domain surrounded with walls at the far field fluid temperature (ambient temperature).

Hope this helps,

engahmed June 27, 2010 11:43

Hi people struggling with convection!!
I am studying turbulent convective heat transfer from a heated horizontal pipe surrounded by 4 sides (2D) (Boussinesq approximation-fluent), the sides and bottom as pressure inlets and the top as pressure outlet, I am having difficulties with convergence as I keep getting reversed flow even when I make my domain soooo big (50rx85r)where r is the radius of the pipe.
can anyone pleaaaaaaaaase help me and suggest what I should do to get rid of this problem.
I have been working on this for 1 month now with no results :(
I will be grateful for anyone who can give me any suggestions.
Thanks a lot

rabia June 6, 2011 07:05

Hi Patrik,

Kindly i need your help. I have seen from your posts that you have done the numerical simulation for the natural convection in fluent. I am trying to solve a problem of natural convection over a flat solar panel which is placed under the solar radiations. But i am getting the reversed flow everytime. I made about 50 times wider domain outside the solar panel. Kindly reply me back how i can get rid of this reversed flow. If You have some time kindly look at my mesh and case file.
Eagerly waiting for your reply because i have to submit my results and i have left with very small time.

ghorrocks June 6, 2011 07:37

Asking for help from posts almost three years old will not get you anywhere..... In future please start a new thread for a new question.

If you post some images of your setup and describe what you are doing we might be able to help.

All times are GMT -4. The time now is 09:48.