[jgross]

Hi OFers,

I am doing a 2D simulation for a centrifugal pump. I am using MRF properties to perform a frozen rotor simulation.

If I attempt to perform a simulation with rotation around the positive z axis, the simulation crashes and I receive a floating point exception. The solution at the time before it crashes is as follows:

Time = 29

smoothSolver: Solving for Ux, Initial residual = 0.314753, Final residual = 0.00782674, No Iterations 2
smoothSolver: Solving for Uy, Initial residual = 0.27311, Final residual = 0.00721864, No Iterations 2
GAMG: Solving for p, Initial residual = 1, Final residual = 5.34965, No Iterations 1000
GAMG: Solving for p, Initial residual = 0.00054909, Final residual = 1.28724e-05, No Iterations 2
time step continuity errors : sum local = 5.61498e+46, global = -1.65435e+44, cumulative = -1.65435e+44
smoothSolver: Solving for epsilon, Initial residual = 1, Final residual = 0.00849132, No Iterations 4
bounding epsilon, min: -4.81027e+70 max: 9.2432e+71 average: 9.28914e+67
smoothSolver: Solving for k, Initial residual = 1, Final residual = 0.0984708, No Iterations 2
bounding k, min: -1.91665e+69 max: 7.39064e+69 average: 6.54355e+65
ExecutionTime = 3.07 s ClockTime = 4 s

Anyone have any idea as to what could cause this?

J

[jgross]

As a follow on from the previous comment, I am using k-omega SST simulation with cyclic AMI between a rotating and non-rotating region. I am aware the cyclic AMI is not necessary for MRF simulation, however I would like to be able to have non-conformal meshes between the rotating and non-rotating regions, particularly for when I begin to perform transient analysis.

The vast majority of the divSchemes give diverging results, with the lone exceptions being:

div(phi,U) Gauss linearUpwindV grad(U);
div(phi,k) Gauss upwind;
div(phi,omega) Gauss upwind;

and

div(phi,U) Gauss QUICKV cellLimited leastSquares 1.0;
div(phi,k) Gauss QUICK cellLimited leastSquares 1.0;
div(phi,omega) Gauss QUICK cellLimited leastSquares 1.0;

However, both of these schemes give results that are inaccurate, with the QUICK schemes giving significantly better results (as would be expected). The velocity and pressure fields given by the QUICK scheme can be seen in the attached images.

Looking at the log.simpleFoam, I notice that the bounding omega and bounding k values are larger than I would expect. At the latest time, I get:

smoothSolver: Solving for Ux, Initial residual = 0.0047810936, Final residual = 0.00027593667, No Iterations 1
smoothSolver: Solving for Uy, Initial residual = 0.0037438613, Final residual = 0.0002661536, No Iterations 1
GAMG: Solving for p, Initial residual = 0.2253595, Final residual = 0.0065880435, No Iterations 2
GAMG: Solving for p, Initial residual = 0.028159537, Final residual = 0.0012741423, No Iterations 4
time step continuity errors : sum local = 0.1897246, global = 0.029397946, cumulative = 15.106846
smoothSolver: Solving for omega, Initial residual = 0.0053881477, Final residual = 0.00013673379, No Iterations 1
bounding omega, min: -1213.8054 max: 281780.77 average: 4183.0128
smoothSolver: Solving for k, Initial residual = 0.0073032573, Final residual = 0.00026609842, No Iterations 1
bounding k, min: -22.06202 max: 545.59446 average: 38.38937
ExecutionTime = 41.21 s ClockTime = 42 s

I am thinking that possibly my boundary conditions for omega and k are possibly incorrect. The files for omega is as follows:

boundaryField
{
HUB_SHROUD
{
type empty;
}
VOLUTE
{
type empty;
}
inlet
{
type turbulentMixingLengthDissipationRateInlet;
mixingLength 0.001;
value uniform 10.0;
}
outlet
{
type inletOutlet;
inletValue uniform 10.0;
value uniform 10.0;
}
VOLUTE_WALL
{
type omegaWallFunction;
value uniform 10.0;
}
BLADE_ROT
{
type omegaWallFunction;
value uniform 10.0;
}
GGI_INT
{
type cyclicAMI;
}
GGI_EXT
{
type cyclicAMI;
}

#includeEtc "caseDicts/setConstraintTypes"
}

and for k:

internalField uniform 0.02;

boundaryField
{
HUB_SHROUD
{
type empty;
}
VOLUTE
{
type empty;
}
VOLUTE_WALL
{
type kqRWallFunction;
value $internalField;
}
BLADE_ROT
{
type kqRWallFunction;
value $internalField;
}
GGI_INT
{
type cyclicAMI;
}
GGI_EXT
{
type cyclicAMI;
}
inlet
{
type turbulentIntensityKineticEnergyInlet;
intensity 0.05;
value uniform 0.02;
}
outlet
{
type inletOutlet;
inletValue uniform 0.02;
value uniform 0.02;
}

#includeEtc "caseDicts/setConstraintTypes"
}

I was also wondering if maybe my relaxation factors were possibly off. The relaxation factors are as follows:

relaxationFactors
{
equations
{
p 0.8;
U 0.5;
"(k|omega|epsilon)" 0.4;
}
}

Thanks for any help/advice you can offer.

J

[jgross]

Solved the problem by using more stable schemes and a different turbulence model. Also I think there were some issues with BCs.

[cfd_user_pune]

Hi,

I am facing same issue while simulating mixing tank using MRF approach. Solution get diverged after some iterations.

Can you please let me know how you solved. Or post your scheme files?