# Convergence: doesn't converged after 800k iterations

 Register Blogs Members List Search Today's Posts Mark Forums Read

 February 19, 2016, 10:22 Convergence: doesn't converged after 800k iterations #1 New Member   Reza Rohani Join Date: Apr 2015 Posts: 7 Rep Power: 11 Hi I try to simulate a dual throat nozzle in 2D. I reached flow pattern but my case doesn't converged after 800k iterations. Which part of my solution is wrong? here is my cas & data files: https://drive.google.com/file/d/0By1...ew?usp=sharing my solution setup: Density based, steady, planar Energy: on material: Air, ideal gas, viscosity; Sutherland viscous model: k-epsilon(2eqn) Realizable wall treatment: scalable wall function solution methods: Implicit, Roe-FDS, Least square cell based, flow;second order upwind, Turbulent kinetic energy;second order upwind, Turbulent dissipation rate;second order upwind, Thanks in advance __________________ Knowledge is power and it can command obedience. Remember that knowledge is a ruler and wealth is its subject. The Peak of Eloquence, Saying 146 Last edited by rohani; February 19, 2016 at 10:40. Reason: adding files links

 February 21, 2016, 00:23 #2 New Member   arshia Join Date: Feb 2016 Posts: 18 Rep Power: 10 hi , what problem you have during convergency?? what accuracy you want?? how is your residuals now??

 February 25, 2016, 15:52 #3 New Member   Reza Rohani Join Date: Apr 2015 Posts: 7 Rep Power: 11 residuals oscillate, they are under e-3 but solution isn't converged __________________ Knowledge is power and it can command obedience. Remember that knowledge is a ruler and wealth is its subject. The Peak of Eloquence, Saying 146

 February 26, 2016, 04:55 #4 Senior Member   Cees Haringa Join Date: May 2013 Location: Delft Posts: 607 Rep Power: 0 On what basis did you conclude lack of convergence? That something doesn't converge doesn't mean it's wrong. It means that the set of equations cannot be properly solved. You can try a denser mesh, might help. Or try to improve on mesh quality (I don't know what the quality is now, so maybe this is irrelevant). The setup you have is quite complex, including quite a lot of physics, so I can very well imagine it is hard to converge. For stability, it may help to run transient instead of steady. If your problem is steady, the transient solver will still converge to that steady state, but it may do so in a more stable way. Also, maybe it helps to start with regular k-e (in my experience, REA is more accurate but less stable), and with first order discretization first - once that converged, switch to 2nd order. rohani likes this.

 Tags 2d flow, convergence, fluent 15.0, nozzle simulation