[lentschi]

Hello,

I am currently running an LES WALE case in OF8 using pimpleFoam. An initial solution was adopted from a corresponding RANS simulation. I am encountering a significant increase in initial p residuals during the final iteration of each time step (from the beginning, up to the current time step = 407):

Courant Number mean: 0.008399714967 max: 0.7193162975
  132 Time = 5e-06
  133
  134 PIMPLE: Iteration 1
  135 smoothSolver:  Solving for Ux, Initial residual = 3.798020874e-06, Final residual = 2.178881283e-09, No Iterations  136 smoothSolver:  Solving for Uy, Initial residual = 1.23923313e-06, Final residual = 6.212075878e-10, No Iterations
  137 smoothSolver:  Solving for Uz, Initial residual = 3.326645723e-06, Final residual = 1.500835019e-09, No Iterations
  138 GAMGPCG:  Solving for p, Initial residual = 0.1039002852, Final residual = 7.321548563e-05, No Iterations 3
  139 GAMGPCG:  Solving for p, Initial residual = 0.007223483423, Final residual = 5.231179486e-06, No Iterations 4
  140 GAMGPCG:  Solving for p, Initial residual = 0.001099740598, Final residual = 4.479327689e-07, No Iterations 4
  141 time step continuity errors : sum local = 8.324539494e-12, global = 2.845274255e-13, cumulative = 2.845274255e-13
  142 GAMGPCG:  Solving for p, Initial residual = 0.03220724529, Final residual = 2.08032012e-05, No Iterations 3
  143 GAMGPCG:  Solving for p, Initial residual = 0.002289461202, Final residual = 1.813159765e-06, No Iterations 5
  144 GAMGPCG:  Solving for p, Initial residual = 0.0003657092185, Final residual = 6.333790375e-07, No Iterations 3
  145 time step continuity errors : sum local = 1.177320667e-11, global = -2.500815743e-13, cumulative = 3.444585116e-14
  
  ....
  
  182 PIMPLE: Iteration 5
  183 smoothSolver:  Solving for Ux, Initial residual = 3.571453512e-06, Final residual = 1.389921103e-09, No Iterations
  184 smoothSolver:  Solving for Uy, Initial residual = 5.873638525e-07, Final residual = 9.005931028e-09, No Iterations
  185 smoothSolver:  Solving for Uz, Initial residual = 1.298302045e-06, Final residual = 4.90217827e-10, No Iterations
  186 GAMGPCG:  Solving for p, Initial residual = 0.000810497471, Final residual = 3.461697627e-07, No Iterations 3
  187 GAMGPCG:  Solving for p, Initial residual = 4.458844283e-05, Final residual = 9.773047859e-08, No Iterations 3
  188 GAMGPCG:  Solving for p, Initial residual = 6.643134197e-06, Final residual = 7.799485967e-08, No Iterations 1
  189 time step continuity errors : sum local = 1.893199055e-12, global = -2.463948232e-15, cumulative = -2.505635775e-1
  190 GAMGPCG:  Solving for p, Initial residual = 0.0002424780533, Final residual = 1.0771658e-07, No Iterations 3
  191 GAMGPCG:  Solving for p, Initial residual = 1.31896789e-05, Final residual = 5.020704946e-08, No Iterations 2
  192 GAMGPCG:  Solving for p, Initial residual = 1.971721138e-06, Final residual = 2.429364249e-08, No Iterations 1
  193 time step continuity errors : sum local = 5.912169006e-13, global = -6.088530994e-15, cumulative = -3.114488874e-1
  194 PIMPLE: Converged
  195         Doing final iteration
  196 PIMPLE: Iteration 6
  197 smoothSolver:  Solving for Ux, Initial residual = 1.315012105e-06, Final residual = 1.081018863e-09, No Iterations
  198 smoothSolver:  Solving for Uy, Initial residual = 2.160959541e-07, Final residual = 4.776109417e-09, No Iterations
  199 smoothSolver:  Solving for Uz, Initial residual = 4.772566881e-07, Final residual = 9.535555309e-09, No Iterations
  200 GAMGPCG:  Solving for p, Initial residual = 0.3212578807, Final residual = 0.000273160544, No Iterations 3
  201 GAMGPCG:  Solving for p, Initial residual = 0.02012123337, Final residual = 1.017517052e-05, No Iterations 4
  202 GAMGPCG:  Solving for p, Initial residual = 0.00291566844, Final residual = 1.634454767e-06, No Iterations 4
  203 time step continuity errors : sum local = 3.007467842e-11, global = 1.636254419e-13, cumulative = 1.324805532e-13
  204 GAMGPCG:  Solving for p, Initial residual = 0.1235330451, Final residual = 0.0001013779859, No Iterations 3
  205 GAMGPCG:  Solving for p, Initial residual = 0.007228274664, Final residual = 3.421793683e-06, No Iterations 4
  206 GAMGPCG:  Solving for p, Initial residual = 0.001138691056, Final residual = 6.050579309e-07, No Iterations 4
  207 time step continuity errors : sum local = 1.114932749e-11, global = 2.85878474e-14, cumulative = 1.610684005e-13
  208 PIMPLE: Converged in 6 iterations
  209 ExecutionTime = 220.96 s  ClockTime = 221 s

Excerpt of fvSolution:

 20     p
 21     {
 22         solver          PCG;
 23         preconditioner
 24         {
 25             preconditioner  GAMG;
 26             tolerance       1e-7;
 27             relTol          0;
 28             nVcycles        2;
 29             smoother        GaussSeidel;
 30             nPreSweeps      0;
 31             nPostSweeps     2;
 32             nFinestSweeps   2;
 33             cacheAgglomeration on;
 34             nCellsInCoarsestLevel 10;
 35             agglomerator    faceAreaPair;
 36             mergeLevels     1;
 37         }
 38         tolerance       1e-7;
 39         relTol          0.001;
 40         minIter     1;
 41     }
 42
 43     pFinal
 44     {
 45         $p;
 46         tolerance       1e-06;
 47         relTol          0;
 48     }


 60 PIMPLE
 61 {
 62     nOuterCorrectors    50;
 63     momentumPredictor   true;
 64     nCorrectors         2;
 65     nNonOrthogonalCorrectors 2;
 66
 67     outerCorrectorResidualControl
 68     {
 69         "(U|(p).*)"
 70         {
 71                 tolerance  1e-3;
 72                 relTol     0;
 73         }
 74      }
 75 }


 78 relaxationFactors
 79 {
 80     fields
 81     {
 82           p                     0.7;
 83         "(p)Final"             1;
 84     }
 85     equations
 86     {
 87         U                         0.7;
 88         omega                  0.7;
 89         k                         0.7; 
 90
 91         "(U|k|omega)Final"      1;
 92     }
 93 }

Any hints? I am a bit at a loss.


Many thanks!


//Markus

[lentschi]

No hints in this round?

[Tobermory]

This usually occurs because the intermediate PIMPLE iterations are run with underrelaxation, to help smooth the convergence, whilst the final PIMPLE is run without relaxation to enforce continuity. Looking at your fvSolution file - this is exactly what you have got.

Post edit: see my post below - I meant "time accuracy" and not "continuity"; brain was on holiday!

[dlahaye]

@tobermory

whilst the final PIMPLE is run without relaxation to enforce continuity.

Could you pls. elaborate on how you relate relaxation on p in the final PIMPLE with continuity. Do you encounter cases in which relaxation on p in the final PIMPLE is required to obtain convergence?

Many thanks. All the best.

[Tobermory]

Apologies Domenico - that was careless of me - relaxation of the p field has nothing to do with continuity (that's related to the p equation factor, which you'll note is always 1). What I meant to say was enforce "time accuracy" rather than "continuity". In other words, if you use a value of less than 1 for the p field relaxation factor, then you are artificially holding back the development of the pressure field. That may not be a problem, if you are searching for a steady state solution ... but will introduce some inaccuracy if you are trying for a true transient solution. I'll add a caution to my earlier post.

[lentschi]

Thank you very much for your clarification. I assumed that the relaxation has something to do with this issue but was not sure.