Residuals in CFX-Solver
 

Hello, I got a general question concering the residuals in the CFX-Solver. I had a calculation that did not converge to RMS-resdiual=1e-4 within a certain flow domain. Then I put my article in an adjusted domain which by accident had double the cross section than actually needed (the aim of my adjustment was to reduce the blockage factor from 20 (way too large, I know) to 3 but in the end I had BF=1.5). Within this domain the calculation met the convergence criteria without a problem.

My question is if the explanation can be found in the definition of the RMS in CFX. If the RMS-calculation includes all cells in the same way then I would reduce the RMS-residual just by adding many 'easy' cells that do not experience the complex flow conditions like around my article. If that is right, I could make almost every calculation to converge just by adding a lot of 'easy' cells that is by enlarging my domain. Is there some limiter to this effect except just using the MAX-residual-target?

Regards Andy

Re: Residuals in CFX-Solver
 

Monitoring convergence is an art, as you've seen from your test. You can come up with similar sanity tests to make any single convergence check fail. For most cases, RMS residuals work well, but the best thing is to monitor multiple convergence measures (RMS residual, MAX residual, conservation balance, problem-specific quantities like drag, etc).

Wrong post, sorry.

Monitoring maximum residuals is always a good idea, but as you pointed out, low residuals alone do not guarantee a converged solution.

Since most simulations serve a certain purpose (like calculating pressure drop or certain aerodynamic coefficients) monitoring these quantities can help estimating the state of convergence.