Page 1 of 1

XNORMP and RNORMP

Posted: Mon Jan 24, 2011 7:31 pm
by Chebli Rezki
Hello,
Please , why the convergence ( in the pressure system ) is directly related to RNORM ?
Regards.
Rezki

Re: XNORMP and RNORMP

Posted: Thu Feb 03, 2011 1:47 pm
by Chebli Rezki
Hello,
Firstly I am sorry if I have a purely theoretical question.
Indeed, I want to understand why the residual of normalization (RNORM) is introduced as a convergence criteria.  
In the theoretic guide of code Saturne (2.0.0-rc1) (page 171). it is mentioned that if  RESIDU < 10 × ε × RNORMP  then convergence is reached .
Residue of normalization represent the second member of pressure system if the pressure gradient is not taken into account.
I think, but I'm not sure,  that it is the product of the difference between these two equations:
 
[Rho U* -   Rho U(n)]   /  delta t   + .......   =  -  grad P(n) + TS
                                      ( - )
[Rho U*² -   Rho U(n)]  /  delta t   + .......  =   TS                                 
---------------------------------------------------------------------
div [ Rho U* + delta (t) grad P(n) ] = mass source.                              
 
because :  div (Rho U) = mass source (continuous equation)
Rho U*² : mass flow if the pressure gradient is not taken into account in momentum equation.
U*: predicted velocity.
 
Can you affirm (or confirm) that, and explain me why the RNORM is introduced in the convergence test.
Thank you in advance.
 
Rezki