CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   why I can only use SIMPLE of 1 order? (http://www.cfd-online.com/Forums/fluent/29831-why-i-can-only-use-simple-1-order.html)

quain tchew May 29, 2002 04:21

why I can only use SIMPLE of 1 order?
 
I'm computing a combustion chamber of an engine which has two subsonic air inlets, a fuel inlet and a nozzle.

I set 1 the air inlets as pressure inlets with totol pressure of about 950000 Pa; 2 the fuel inlet as mass flow inlet with mass flux of 0.09kg/s; 3 the nozzle as pressure outlet with 26000Pa.

I use ideal-gas model, K-E turbulence model with standard wall function.

Only with SIMPLE method of one order, I got a converged result. But when I switch to two order simple method, it didn't converge.

I don't know why? Any one give some advices?

Jin-Wook LEE May 29, 2002 04:49

Re: why I can only use SIMPLE of 1 order?
 
It is very natural that 1st order is cheaf(CPU time) and easy to converge and 2nd order is expensive and difficult to converge.

Comments :

1. 1st order is easy to converge because of numerical diffusion which has the effect of smoothing the result.

2. To enhance the convergence, there are couple of approaches such as,

- good mesh(low skewness)

- low under-relaxation

- step by step approach(e.g., mixing only --> reaction)

- and so on

Sincerely, Jinwook


Quain Tchew May 30, 2002 05:54

Re: why I can only use SIMPLE of 1 order?
 
Thanks

I have tried low under-relaxation, but....

Yes, maybe my grid is not good.

moiami June 3, 2002 02:34

Re: why I can only use SIMPLE of 1 order?
 
Try this: do 1st order then after it is converged carry on the simulation but replace 1st with 2nd (like in the tutorial). You'll see a rise in peak for the residual plot b4 it peters down to a convergence.

But b4 that you have to be clear on why you're using 2nd order. You may not have to, you know (see message below: Posted By Gavin <gavinbrady85@hotmail.com> Date: Wed, 29 May 2002, 4:12 a.m. in Response To: convergence problems (jeremy))



All times are GMT -4. The time now is 16:39.