buoyantKEpsilon: Changed the additional buoyancy generation/dissipation

term to the more commonly used form of Henkes, R.A.W.M., Van Der Vlugt, F.F. & Hoogendoorn, C.J. (1991).

src/TurbulenceModels/compressible/RAS/buoyantKEpsilon/buoyantKEpsilon.C

@@ -24,9 +24,9 @@ License
 \*---------------------------------------------------------------------------*/
 
 #include "buoyantKEpsilon.H"
-#include "addToRunTimeSelectionTable.H"
 #include "uniformDimensionedFields.H"
 #include "fvcGrad.H"
+#include "addToRunTimeSelectionTable.H"
 
 // * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //
 
@@ -68,7 +68,7 @@ buoyantKEpsilon<BasicTurbulenceModel>::buoyantKEpsilon
         (
             "Cg",
             this->coeffDict_,
-            0.85
+            1.0
         )
     )
 {
@@ -107,8 +107,8 @@ buoyantKEpsilon<BasicTurbulenceModel>::Gcoef() const
         lookupObject<uniformDimensionedVectorField>("g");
 
     return
-        this->alpha_*(this->Cmu_/this->sigmak_)*this->k_
+        (Cg_*this->Cmu_)*this->alpha_*this->k_*(g & fvc::grad(this->rho_))
-       *(g & fvc::grad(this->rho_))/(this->epsilon_ + this->epsilonMin_);
+       /(this->epsilon_ + this->epsilonMin_);
 }
 
 
@@ -124,7 +124,19 @@ template<class BasicTurbulenceModel>
 tmp<fvScalarMatrix>
 buoyantKEpsilon<BasicTurbulenceModel>::epsilonSource() const
 {
-    return -fvm::SuSp(this->C1_*(1.0 - this->Cg_)*Gcoef(), this->epsilon_);
+    const uniformDimensionedVectorField& g =
+        this->mesh_.objectRegistry::template
+        lookupObject<uniformDimensionedVectorField>("g");
+
+    vector gHat(g.value()/mag(g.value()));
+
+    volScalarField v(gHat & this->U_);
+    volScalarField u
+    (
+        mag(this->U_ - gHat*v) + dimensionedScalar("SMALL", dimVelocity, SMALL)
+    );
+
+    return -fvm::SuSp(this->C1_*tanh(mag(v)/u)*Gcoef(), this->epsilon_);
 }
 
 

src/TurbulenceModels/compressible/RAS/buoyantKEpsilon/buoyantKEpsilon.H

@@ -28,33 +28,34 @@ Group
     grpRASTurbulence
 
 Description
-    k-epsilon model for the gas-phase in a two-phase system
+    Additional buoyancy generation/dissipation term applied to the
-    supporting phase-inversion.
+    k and epsilon equations of the standard k-epsilon model.
 
-    In the limit that the gas-phase fraction approaches zero a contribution from
+    Reference:
-    the other phase is blended into the k and epsilon equations up to the
+    \verbatim
-    phase-fraction of alphaInversion at which point phase-inversion is
+        Henkes, R.A.W.M., Van Der Vlugt, F.F. & Hoogendoorn, C.J. (1991).
-    considered to have occurred and the model reverts to the pure single-phase
+        Natural Convection Flow in a Square Cavity Calculated with
-    form.
+        Low-Reynolds-Number Turbulence Models.
+        Int. J. Heat Mass Transfer, 34, 1543-1557.
+    \endverbatim
 
-    This model is unpublished and is provided as a stable numerical framework
+    This implementation is based on the density rather than temperature gradient
-    on which a more physical model may be built.
+    extending the applicability to systems in which the density gradient may be
+    generated by variation of composition rather than temperature.  Further, the
+    1/Prt coefficient is replaced by Cg to provide more general control of
+    model.
 
-    The default model coefficients correspond to the following:
+    The default model coefficients are
     \verbatim
         buoyantKEpsilonCoeffs
         {
-            Cmu             0.09;
+            Cg              1.0;
-            C1              1.44;
-            C2              1.92;
-            C3              -0.33;
-            Cg              0.85;
-            sigmak          1.0;
-            sigmaEps        1.3;
-            alphaInversion  0.7;
         }
     \endverbatim
 
+SeeAlso
+    Foam::RASModels::kEpsilon
+
 SourceFiles
     buoyantKEpsilon.C