CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   convergence difficulty (

Devin March 2, 2004 00:20

convergence difficulty
hi, all

I got some trouble in using CFX-4.4. How to get good convergence? The value of H(enthalpy) is difficult to converge. The residual of H is from 1E+01 to 1.5E-01, and oscillate between 1E+00 and 1E-01. Is this phenomenon correct ? I've tried to use double precision and relaxtion factor, but it doesn't work. How to improve it ? Thank you for your help!!

Jeff March 2, 2004 19:42

Re: convergence difficulty
Typically, because H = Cp * T (roughly), enthalpy values are on the order of 10^6. Inlet and wall flux values can be even larger. If you're residuals are between 0.1 and 1, that's pretty small compared with the flux values and you may just be hitting the machine round off.

Make sure you're setting your enthalpy reference temperature to just below your expected lowest temperature in the model. This will ensure that H = Cp*[T-Tref] will be as small as it can be without going negative and losing diagonal dominance in the matrix.

Then evaluate your total residual as a fraction of the total heat flux into the model. Something like 1/10^6 or 0.0001% is actually pretty good.


All times are GMT -4. The time now is 22:49.