chtMultiRegionTwoPhaseEulerFoam, Population Balance
Summary
Population balance (PPB) is not calculated when
solveOnFinalIterOnly true;
is used for the PPB in fvSolution of chtMultiRegionTwoPhaseEulerFoam solver.
Steps to reproduce
Use $FOAM_TUTORIALS/heatTransfer/chtMultiRegionTwoPhaseEulerFoam/solidQuenching2D. Go in constant/water/phaseProperties setup to use predefined PPB:
Change following lines of code from:
type thermalPhaseChangeTwoPhaseSystem;
phases (gas liquid);
// phase change in the bulk of the fluid.
phaseChange off;
//populationBalances (bubbles);
gas {
type purePhaseModel;
diameterModel isothermal;
isothermalCoeffs
{
d0 5e-3;
p0 1e5;
}
Sc 0.7;
velocityGroupCoeffs
{
populationBalance bubbles;
formFactor 0.5235987756;
sizeGroups
(
f0 {d 0.5e-4; value 0 ;}
f1 {d 1.040e-3; value 0 ;}
f2 {d 1.640e-3; value 0 ;}
f3 {d 2.265e-3; value 0 ;}
f4 {d 2.889e-3; value 0 ;}
f5 {d 3.512e-3; value 0 ;}
f6 {d 4.141e-3; value 0 ;}
f7 {d 4.771e-3; value 1 ;}
f8 {d 5.402e-3; value 0 ;}
f9 {d 6.033e-3; value 0 ;}
f10 {d 6.665e-3; value 0 ;}
f11 {d 7.297e-3; value 0 ;}
f12 {d 7.929e-3; value 0 ;}
f13 {d 8.562e-3; value 0 ;}
f14 {d 9.194e-3; value 0 ;}
f15 {d 1.194e-2; value 0 ;}
f16 {d 2.400e-2; value 0 ;}
f17 {d 2.700e-2; value 0 ;}
f18 {d 3.000e-2; value 0 ;}
);
}
residualAlpha 1e-4;
}into
type thermalPhaseChangePopulationBalanceTwoPhaseSystem;
phases (gas liquid);
// phase change in the bulk of the fluid.
phaseChange off;
populationBalances (bubbles);
gas
{
type purePhaseModel;
diameterModel velocityGroup;
velocityGroupCoeffs
{
populationBalance bubbles;
formFactor 0.5235987756;
sizeGroups
(
f0 {d 0.5e-4; value 0 ;}
f1 {d 1.040e-3; value 0 ;}
f2 {d 1.640e-3; value 0 ;}
f3 {d 2.265e-3; value 0 ;}
f4 {d 2.889e-3; value 0 ;}
f5 {d 3.512e-3; value 0 ;}
f6 {d 4.141e-3; value 0 ;}
f7 {d 4.771e-3; value 1 ;}
f8 {d 5.402e-3; value 0 ;}
f9 {d 6.033e-3; value 0 ;}
f10 {d 6.665e-3; value 0 ;}
f11 {d 7.297e-3; value 0 ;}
f12 {d 7.929e-3; value 0 ;}
f13 {d 8.562e-3; value 0 ;}
f14 {d 9.194e-3; value 0 ;}
f15 {d 1.194e-2; value 0 ;}
f16 {d 2.400e-2; value 0 ;}
f17 {d 2.700e-2; value 0 ;}
f18 {d 3.000e-2; value 0 ;}
);
}
residualAlpha 1e-4;
}further change lines
coalescenceModels
(
PrinceBlanch
{
C1 0.05;
h0 1e-4;
hf 1e-8;
turbulentCollisions true;
buoyantCollisions false;
laminarShearCollisions false;
}
);into
coalescenceModels
(
PrinceBlanch
{
C1 0.05;
h0 1e-4;
hf 1e-8;
turbulence true;
buoyancy false;
laminarShear false;
}
); and
driftModels
(
phaseChange
{
pairNames (gasAndLiquid);
}
densityChange{}
); into
driftModels
(
phaseChange
{
pairs ((gas and liquid));
}
densityChange{}
); Then change following line in constant/water/turbulenceProperties.liquid from
simulationType laminar;into
simulationType RAS;Then run
$./AllrunWhen you have a look at the log.chtMultiRegionTwoPhaseEulerFoam you will see that no PPB is calculated during any iteration.
Change following lines in system/water/fvSolution
solveOnFinalIterOnly true;into
solveOnFinalIterOnly false;run
$./Allclean
$./Allrunin the log.chtMultiRegionTwoPhaseEulerFoam, you can see the PPB being calculated each iteration.
Example case
Mentioned in the step to reproduce. Shown at a tutorial case
What is the current bug behaviour?
No PPB is calculated if chosen to run only during the last iteration.
What is the expected correct behavior?
User should be able to choose when the PPB runs. User might think that the PPB runs during the last iteration even though it is not apparent in the log. However, the true is it does not run at all. Explanation in the fix section.
Relevant logs and/or images
Environment information
- OpenFOAM version : v1906
- Operating system : Ubuntu 16.04.6 LTS
- Hardware info :
- Compiler : gcc
Most probably, the same issue will be also with v2006.
Possible fixes
The problem is that chtMultiRegionTwoPhaseEulerFoam does not cunstruct object of pimpleControl class and does not use it. It is rather "hard coded" in the solver to be able to do the iterations based on the user choice. However, populationBalanceModel class uses the pimpleControl class. Then it comes to line populationBalanceModel.C:1236
if (!solveOnFinalIterOnly || pimple_.finalIter())where it decides whether to calculate the PPB. If the solveOnFinalIterOnly true, it means that this if condition should evaluate true only when pimple_.finalIter() = true, but that never happen because the function pimple_.finalIter() gives false
inline bool Foam::pimpleControl::finalIter() const
{
return converged_ || (corr_ == nCorrPIMPLE_);
}because the corr_ is always equal to zero.
Some time ago, I worked on the same solver using the version of OpenFOAM Foundation. There is available class called pimpleMultiRegionControl, which manages the pimple logic for multiregion solvers. Having that, I only made a copy of populationBalanceModel class and changed the costructer to use pimpleMultiRegionControl instead of pimpleControl. Certainly, this leads to code duplication because majority of lines for populationBalanceModel are same. Further more if pimpleMultiRegionControl is used in ESI group version then other multiregion solvers should be adjusted adequatly.
Certainly, there might be some other ways how to pass the information about the last iteration to populationBalanceModel class.