[mostanad]

Hi everybody,
I am using pimpleFoam for solving a multiphase problem. To this end, I have specified the following relaxation factors for converging the simulation through outer iterations:

relaxationFactors
{
    fields
    {
        "(p)"      0.3;
    }
    equations
    {
        "(U)"      0.7;
    }
}

As you know, it imposes 0.3/0.7 until the last outer loop, but considers no relaxation (e.g. 1) for last one. The following shows the residual values for final pimple loops:

PIMPLE: iteration 47
smoothSolver:  Solving for Ux, Initial residual = 0.00348382, Final residual = 3.33273e-09, No Iterations 12
smoothSolver:  Solving for Uy, Initial residual = 0.000400605, Final residual = 8.08511e-09, No Iterations 11
smoothSolver:  Solving for Uz, Initial residual = 0.00348381, Final residual = 7.12709e-09, No Iterations 12
GAMG:  Solving for p, Initial residual = 0.030812, Final residual = 5.75704e-09, No Iterations 14
time step continuity errors : sum local = 2.14038e-21, global = -2.15546e-29, cumulative = -1.77606e-27
GAMG:  Solving for p, Initial residual = 0.0199088, Final residual = 3.64323e-09, No Iterations 16
time step continuity errors : sum local = 1.36398e-21, global = 6.91738e-31, cumulative = -1.77537e-27
PIMPLE: iteration 48
smoothSolver:  Solving for Ux, Initial residual = 0.00344079, Final residual = 3.32801e-09, No Iterations 12
smoothSolver:  Solving for Uy, Initial residual = 0.000403059, Final residual = 8.1732e-09, No Iterations 11
smoothSolver:  Solving for Uz, Initial residual = 0.00344078, Final residual = 7.09306e-09, No Iterations 12
GAMG:  Solving for p, Initial residual = 0.0303855, Final residual = 5.38726e-09, No Iterations 13
time step continuity errors : sum local = 1.91437e-21, global = -3.69477e-30, cumulative = -1.77907e-27
GAMG:  Solving for p, Initial residual = 0.0196223, Final residual = 9.50948e-09, No Iterations 17
time step continuity errors : sum local = 3.38302e-21, global = -1.04939e-29, cumulative = -1.78956e-27
PIMPLE: iteration 49
smoothSolver:  Solving for Ux, Initial residual = 0.0034018, Final residual = 3.32441e-09, No Iterations 12
smoothSolver:  Solving for Uy, Initial residual = 0.000405759, Final residual = 8.26221e-09, No Iterations 11
smoothSolver:  Solving for Uz, Initial residual = 0.00340179, Final residual = 7.06439e-09, No Iterations 12
GAMG:  Solving for p, Initial residual = 0.0301731, Final residual = 9.54839e-09, No Iterations 15
time step continuity errors : sum local = 3.34999e-21, global = -3.40829e-30, cumulative = -1.79297e-27
GAMG:  Solving for p, Initial residual = 0.0194295, Final residual = 7.68963e-09, No Iterations 17
time step continuity errors : sum local = 2.67738e-21, global = -6.22769e-30, cumulative = -1.7992e-27
PIMPLE: iteration 50
smoothSolver:  Solving for Ux, Initial residual = 0.151223, Final residual = 9.88356e-09, No Iterations 596
smoothSolver:  Solving for Uy, Initial residual = 0.00853352, Final residual = 9.95885e-09, No Iterations 640
smoothSolver:  Solving for Uz, Initial residual = 0.151222, Final residual = 9.93596e-09, No Iterations 610
GAMG:  Solving for p, Initial residual = 0.809884, Final residual = 8.12774e-09, No Iterations 25
time step continuity errors : sum local = 7.22642e-21, global = -5.99797e-31, cumulative = -1.7998e-27
GAMG:  Solving for p, Initial residual = 0.718341, Final residual = 7.75604e-09, No Iterations 24
time step continuity errors : sum local = 7.12491e-21, global = -6.79726e-30, cumulative = -1.80659e-27

So as you see, the residuals and number of iterations for final pimple loop suddenly increases. It becomes strange to me as I increase the nOuterCorrectors or decrease the residual tolerance but it still has a very high iterations for final outer iteration. So it seems the solution is not converging through the iterations to provide a good estimation for final outer iteration. So what is your suggestion for this strange behavior?
Cheers
Mo

[piu58]

There is less to say only form this protocol. Mostly, the problem is in your case: Too complicated, to much physics, problems with the mesh.

I recommend to start with the simplest case you can construct and look what happens. Than you might think about take more details in - if that makes sense.

It is alway a good idea splitting the problem and simulate each individual.

[mostanad]

Originally Posted by piu58

There is less to say only form this protocol. Mostly, the problem is in your case: Too complicated, to much physics, problems with the mesh.

I recommend to start with the simplest case you can construct and look what happens. Than you might think about take more details in - if that makes sense.

It is alway a good idea splitting the problem and simulate each individual.

OK. As you said the physics of my multiphase problem is a bit complicated. Its something like this algorithm in every time step:


1- Solve the fluid equations in entire field (pimple or PISO)
2- Find the forces on rigid body, and calculate the rigid body's velocity by Newton's law
3- Impose the calculated rigid body velocity in the domain

So as you see, I need to do the pimple calculation for the first step in entire field, considering the region for rigid body. This requires under relaxation factors because its actually a complicated non-Newtonian problem. Therefore, my problem is that the information in rigid body region shouldn't change drastically in pimple calculation. However, by looking at my outer loops, I see the information inside the solid region is changing every outer iteration because of imposing under relaxation factor. So the question is how I can solve the fluid in rigid particles region with minor effect from outer loops or URF? previously I was using that by PISO for Newtonian, so it didn't have any outer loop and URF. But now I must use pimple for complexity in non-Newtonian. It seems the solvers like PISO suitable for non-Newtonian flows can also be a solution.

If you need any further information, please feel free to tell me. It is a complicated and interesting problem for me.