CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Problems with additional variable (http://www.cfd-online.com/Forums/cfx/24744-problems-additional-variable.html)

Krishna Premi October 29, 2007 01:34

Problems with additional variable
 
Hi

I have an additional variable (size of a particle) the value of which is extremely small and ranges between 10^-10 metres and 10^-9 metres. Due to some constraints, I cannot use the inherent particle tracking model in CFX. I am solving a transport equation for the additional variable. For this, I have written a Fortran code that takes values of Temperature, Pressure and other variables and computes the necessary source terms for this variable. The global convergence tolerance that I have set is 10^-5, as I know that it is difficult to get variables like momentum and P-mass go below this. Since the values of my additonal variable (range 10^-10 and 10^-9) are very low and less than the convergence limit (10^-5), I think I will have problems with convergence and conservation of the variable. Can somebody advise what I should do to overcome this problem.

1. Can I reduce the value of my additonal variable by a factor of 10^-6 so that its range now lies in 10^-4 and 10^- 3.What could be the effects of doing this?

2.Can I specify different convergence criterion for different variables like in Fluent, say 10^-5 for all variables and 10^-13 for the additional variable. Will I have to use a double precision solver for this?

Kindly help me out.

Regards,

Krishna Premi.


opaque October 29, 2007 09:19

Re: Problems with additional variable
 
Dear Krishna,

For

"2 - Can I specify different convergence criterion for different variables like in Fluent, say 10^-5 for all variables and 10^-13 for the additional variable. Will I have to use a double precision solver for this?"

Have you looked into the Solver Control panel for ANSYS CFX? If you select the Equation Class tab, you could set controls per equation. Will that help?

Very likely you will need to use the double precision solver. Using variable scaling might help, but it is case dependent and you must be careful on which parts of the model are dependent on the scale you select.

Opaque



All times are GMT -4. The time now is 18:09.