CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   RE: How to reduce the running time with FLUENT (http://www.cfd-online.com/Forums/fluent/34475-re-how-reduce-running-time-fluent.html)

Hp August 24, 2004 13:41

RE: How to reduce the running time with FLUENT
 
Hi,

I am using the FLUENT to do the long time simulation. Does anyone have some good suggestions on decreasing the running time?

The problem has grid size about 9000 cells. And I tried parallel processing with 2cpus, and it takes more time.

Thanks a lot!

HP

Andrew August 24, 2004 14:13

Re: RE: How to reduce the running time with FLUENT
 
I am assuming you have sufficient RAM

Hp August 24, 2004 17:29

Re: RE: How to reduce the running time with FLUENT
 
Hi,Andrew. Do you mean if I add more RAM, the speed will be increased?

Thanks.

Hp

Colin August 24, 2004 18:11

Re: RE: How to reduce the running time with FLUENT
 
One of my ways is to somewhat sacrifice result accuracy by reducing the grid numbers. This method is very effective. You also could buy Fluent Parallel Process Licenses with more calculation nodes, it is not expensive.

Andrew August 24, 2004 19:01

Re: RE: How to reduce the running time with FLUENT
 
yes, if you don't have sufficient RAM it will slow things down. You need to know how much RAM you have to begin with

anindya August 25, 2004 02:03

Re: RE: How to reduce the running time with FLUENT
 
As you just have 9000 cells, parallel processing will/might actually take more time, due to other overhead charges. parallel processing is not effective for problems with such a small number of cells. Infact you will not be able to appreciate the benefits of parallel fluent with cell count less than atleast an order of magnitude higher.

venugopal August 27, 2004 10:32

Re: RE: How to reduce the running time with FLUENT
 
Haai, You do grid independence study. Means run the cases with less that 9000 cells, for example you can go for 8500,8000 and so on. And in each case you compare the results, and if the error is less that considerable range ( for exampe 5-8 percent) you can go for that coarse grid instead of fine grid. This is the method you have to select appropriate number of cells for a particular problem.


All times are GMT -4. The time now is 15:02.