CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   Relaxation Controls (http://www.cfd-online.com/Forums/main/3006-relaxation-controls.html)

Amber January 14, 2001 21:50

Relaxation Controls
 
Dear all,

I am a student working on my final year project. My work is on the simulations of a residential apartment. The purpose of my CFD modelling is to simulate the air flow throughout the unit. The unit consists of 2 bedrooms, 1 living room, i kitchen and 2 bathrooms. The software that i am using is Phoenics. The model of turbulence used is Lvel.

The problem that i am facing is that i simply cannot get the desired results in the velocity of air flow, at present , the velocity results are simply too high. I have tried changing the relaxation control for velocity to 0.1, with this change, the results i got for interations at 100 gives me the desired results, but the results i got for higher iterations such as 1000 will give me too high a value for the velocity. Can someone please help me out by giving me some pointers. I would really appreciate any forms of assistance because i have to get the results within this week.

A. Rajani Kumar January 14, 2001 23:01

Re: Relaxation Controls
 
Here the important thing is , what type of convergence u r using. when u gave 100 iteration limit ur getting results and if u give 1000 it is showing higher values means (I feel) 1)may be your solution is diverging so u use more under relaxation factor. 2)ur solution is not converged. u allow for more iterations(up to it converge)

Best of luck

John C. Chien January 14, 2001 23:43

Re: Relaxation Controls
 
(1). I would start with just one room first, to see whether converged solution can be obtained or not. (2). After that, add one room and re-calculate the flow field. So, you will have to identify the source of the problem, assuming that someone has already obtained the solution to your problem using the code you mentioned. And the step-by-step isolation method will give you the answer you need. (3). If for certain reasons that the code can not provide the converged solution to your problem, then just document exactly what you have learned in the project. (4). I think, it is not uncommon to stop the iteration at 100, especially when the results look good. But still one would be curious to find out why, assuming that the code can handle it. With nearly six rooms in your simulation, it is hard to know the cause of the problem. (5). So, simplify it first.

Fred Uckfield January 15, 2001 05:18

Re: Relaxation Controls
 
I think Spalding intended LVEL for more cluttered geometries. Have a look at the paper by Spalding and Agonafer (sp?) or check out the CHAM www page.

Fred.

Amber January 17, 2001 05:50

Re: Relaxation Controls
 
Dear all,

Thank a lot for the advices given. Really appreciate it...i think that i have managed to get reasonable results with the advices given. Thank you all so much. However, i am facing a new problem. In my model, there are some internal doors leading to rooms, such doors are rotated to a certain degree usually 290, to create the situation on site. The problem with this is that when i try to run the Earth program, it will stop almost instantaneous with a message "Error Stop". I have checked the results file which lists the problem as data missing from GI3S 3 and grex3 found in GROUND. Can anyone give me some advices on this? I have an average of 5 internal doors, so i shut all of them and left only one open, still the Error Stop pops out! Is it that i missed out something in my modelling process? Thank you all for your kind help.

Fred Uckfield January 17, 2001 06:42

Re: Relaxation Controls
 
I'd take that question to the Phoenix forum if I were you.


All times are GMT -4. The time now is 22:43.