BUG: Limiting min alpha to SMALL to avoid zero transient term of turbulence model

In certain turbulence models such as kEpsilon, kOmegaSST, etc when used
in the multiphase context throw an error in the preconditioner. This
is due to alpha being zero. Fixes #1789.

src/phaseSystemModels/reactingEulerFoam/reactingMultiphaseEulerFoam/multiphaseSystem/multiphaseSystem.C

@@ -6,6 +6,7 @@
      \\/     M anipulation  |
 -------------------------------------------------------------------------------
     Copyright (C) 2013-2019 OpenFOAM Foundation
+    Copyright (C) 2020 OpenCFD Ltd.
 -------------------------------------------------------------------------------
 License
     This file is part of OpenFOAM.
@@ -704,7 +705,7 @@ void Foam::multiphaseSystem::solve()
         phase.alphaRhoPhiRef() =
             fvc::interpolate(phase.rho())*phase.alphaPhi();
 
-        phase.clip(0, 1);
+        phase.clip(SMALL, 1 - SMALL);
     }
 
     calcAlphas();

src/phaseSystemModels/reactingEulerFoam/reactingTwoPhaseEulerFoam/twoPhaseSystem/twoPhaseSystem.C

@@ -6,6 +6,7 @@
      \\/     M anipulation  |
 -------------------------------------------------------------------------------
     Copyright (C) 2013-2018 OpenFOAM Foundation
+    Copyright (C) 2020 OpenCFD Ltd.
 -------------------------------------------------------------------------------
 License
     This file is part of OpenFOAM.
@@ -339,7 +340,7 @@ void Foam::twoPhaseSystem::solve()
             << endl;
 
         // Ensure the phase-fractions are bounded
-        alpha1.clip(0, 1);
+        alpha1.clip(SMALL, 1 - SMALL);
 
         // Update the phase-fraction of the other phase
         alpha2 = scalar(1) - alpha1;

[Alberto Passalacqua]

Hi,

I think this limitation should be moved to the turbulence models themselves and not affect the alpha-U-p loop causing mass loss (see implementation of the kinetic theory, where this is done too). The scheme should enforce positivity and boundedness, so it is better to clip when using the value than accumulating the error in the alpha-U-p loop.

Best, Alberto

[Sergio Ferraris]

Hi, Yes, you are correct , that was my first intention. But I was not convinced to change the behavior of all the turbulence models used in single phase (i.e kEpsilon, etc, etc). Thanks Sergio

[Alberto Passalacqua]

Hi Sergio,

that's a good point indeed, since alpha is, unfortunately, carried over to single-phase models. I guess it's another reason to split multiphase turbulence out of single-phase implementation on the long run...

Thanks, Alberto