BUG: Fixing sign of qr in externalWallHeatFluxTemperature

1) The sign of radiative flux (qr) for modes Power and HeatFlux was
corrected.

2) Relaxation factor implementation was incomplete. Now caching
valueFraction and refValue from  the previous time step.

src/TurbulenceModels/compressible/turbulentFluidThermoModels/derivedFvPatchFields/externalWallHeatFluxTemperature/externalWallHeatFluxTemperatureFvPatchScalarField.C

@@ -6,7 +6,7 @@
      \\/     M anipulation  |
 -------------------------------------------------------------------------------
     Copyright (C) 2011-2017 OpenFOAM Foundation
-    Copyright (C) 2015-2019 OpenCFD Ltd.
+    Copyright (C) 2015-2020 OpenCFD Ltd.
 -------------------------------------------------------------------------------
 License
     This file is part of OpenFOAM.
@@ -360,6 +360,9 @@ void Foam::externalWallHeatFluxTemperatureFvPatchScalarField::updateCoeffs()
 
     const scalarField& Tp(*this);
 
+    const scalarField valueFraction0(valueFraction());
+    const scalarField refValue0(refValue());
+
     scalarField qr(Tp.size(), Zero);
     if (qrName_ != "none")
     {
@@ -375,7 +378,7 @@ void Foam::externalWallHeatFluxTemperatureFvPatchScalarField::updateCoeffs()
     {
         case fixedPower:
         {
-            refGrad() = (Q_/gSum(patch().magSf()) + qr)/kappa(Tp);
+            refGrad() = (Q_/gSum(patch().magSf()) - qr)/kappa(Tp);
             refValue() = 0;
             valueFraction() = 0;
 
@@ -383,7 +386,7 @@ void Foam::externalWallHeatFluxTemperatureFvPatchScalarField::updateCoeffs()
         }
         case fixedHeatFlux:
         {
-            refGrad() = (q_ + qr)/kappa(Tp);
+            refGrad() = (q_ - qr)/kappa(Tp);
             refValue() = 0;
             valueFraction() = 0;
 
@@ -459,8 +462,9 @@ void Foam::externalWallHeatFluxTemperatureFvPatchScalarField::updateCoeffs()
         }
     }
 
-    valueFraction() = relaxation_*valueFraction() + (1 - relaxation_);
-    refValue() = relaxation_*refValue() + (1 - relaxation_)*Tp;
+    valueFraction() =
+        relaxation_*valueFraction() + (1 - relaxation_)*valueFraction0;
+    refValue() = relaxation_*refValue() + (1 - relaxation_)*refValue0;
 
     mixedFvPatchScalarField::updateCoeffs();
 

[Simon]

Hi!

Can I ask why is it q_-qr instead of q_+qr? I think it was correct. Let's create an adiabatic patch, thus q_ = 0. In this case the heat flux is qr (you "put back" the radiative flux) Then calculate the resulted wall heat flux: It will be qr, since you prescribed it with the BC, and substract the radiative flux which is also qr, thus the heat flux is zero, which is expected since you made an adiabatic wall.

If you use q_-qr, you will end up with a heat flux of -2*qr. Maybe I'm wrong somewhere so feel free to correct me.

Thanks!

[Lieven Vervecken]

Hi,

Same question as Simon. I also have the impression that the signs were correct. When we apply this 'fix', our simulations start to show nonphysical results which disappear when we turn of radiation. Can you re-evaluate this commit?

Thanks!

[Andrew Heather]

@Sergio

[Simon]

Hi!

It is corrected in commit 712512d4