transient case diverged all of a sudden after already convergence
Hi everyone,
I am running an internal flow in a duct. Firstly I ran the steady simulation (3500 iterations ) using simpleFoam and everything is OK. Then I started running transient simulation using transient SIMPLE method. In the first 0.9958 s, everything is also OK. However, when the 0.99585 time step starts, the pressure calculation gets suddenly divergence. I tried to modify convergence tolerance control and other linear system solvers and ran from a previous time step, but the divergence exactly recurred at 0.99585 s. Although the pressure field diverged, the velocity field not. It is really strange. The log is as follows. I hope someone could give me a hint. Thank you everyone in advance. Courant Number mean: 0.0002247080246 max: 7.268116999 Time = 3500.99585 Max Inner Iteration: 40 ConvergenceCriterion: 9e08  Inner loop: 0  smoothSolver: Solving for Ux, Initial residual = 3.722149168e06, Final residual = 4.894892661e10, No Iterations 1 smoothSolver: Solving for Uy, Initial residual = 1.756301862e05, Final residual = 1.469450516e09, No Iterations 1 smoothSolver: Solving for Uz, Initial residual = 1.01839746e05, Final residual = 1.132826601e09, No Iterations 1 GAMG: Solving for p, Initial residual = 4.942934403e06, Final residual = 4.204448761e08, No Iterations 7 time step continuity errors : sum local = 1.081550112e13, global = 9.481328711e15, cumulative = 2.379358346e11 smoothSolver: Solving for nuTilda, Initial residual = 9.707065056e06, Final residual = 2.616585371e09, No Iterations 1 > maxResidual(U,p) 4.942934403e06  Inner loop: 1  smoothSolver: Solving for Ux, Initial residual = 1.117159902e06, Final residual = 1.819146731e10, No Iterations 1 smoothSolver: Solving for Uy, Initial residual = 5.270313626e06, Final residual = 4.919923245e10, No Iterations 1 smoothSolver: Solving for Uz, Initial residual = 3.05667921e06, Final residual = 4.055659424e10, No Iterations 1 GAMG: Solving for p, Initial residual = 4.934697876e05, Final residual = 1.730930522e07, No Iterations 4 time step continuity errors : sum local = 4.452616788e13, global = 1.255549445e13, cumulative = 2.39191384e11 smoothSolver: Solving for nuTilda, Initial residual = 2.915621506e06, Final residual = 9.895764204e10, No Iterations 1 > maxResidual(U,p) 4.934697876e05  Inner loop: 2  smoothSolver: Solving for Ux, Initial residual = 3.502918942e07, Final residual = 7.495885771e11, No Iterations 1 smoothSolver: Solving for Uy, Initial residual = 1.675695577e06, Final residual = 1.964754382e10, No Iterations 1 smoothSolver: Solving for Uz, Initial residual = 9.391553624e07, Final residual = 1.568032312e10, No Iterations 1 GAMG: Solving for p, Initial residual = 0.3252932271, Final residual = 0.000827214952, No Iterations 3 time step continuity errors : sum local = 2.924368657e09, global = 1.922521341e10, cumulative = 2.161712725e10 smoothSolver: Solving for nuTilda, Initial residual = 1.225521515e06, Final residual = 4.562680108e10, No Iterations 1 > maxResidual(U,p) 0.3252932271 
your courant number is too large, try using a timestep equal to 1/10 of what you are actually using

Hi,
Thank you for your reply. I know that CFL < 1 is a common recommendation for transient calculation, but I am also told that for URANS, CFL between 1 and 10 can be also acceptable. Is it true ? And even if CFL is too big, but why for the first 0.9958 s does it work 'well' ? 
All times are GMT 4. The time now is 22:59. 