|
|
|
|
|
Please use one of the tutorial cases to demonstrate the problem and show how your suggested change provides improvement to the results. |
|
|
Hi Henry, thank you for your attention. For velocity will simple BC, such as fixedValue/fixedGradient/mixed, it is not an issue, as the variable of 'boundaryDiagCmpt' will always be zero (addBoundaryDiag/negate/addCmptAvBoundaryDiag cancelled out). However, if the velocity boundary condition is pressureInletOutlet type (transform,directionMixed), it can not be canceled out. for (direction cmpt=0; cmpt<Type::nComponents; cmpt++) { scalarField psiCmpt(psi_.primitiveField().component(cmpt)); scalarField boundaryDiagCmpt(psi_.size(), Zero); addBoundaryDiag(boundaryDiagCmpt, cmpt); boundaryDiagCmpt.negate(); addCmptAvBoundaryDiag(boundaryDiagCmpt); Hphi.primitiveFieldRef().replace(cmpt, boundaryDiagCmpt*psiCmpt); } A test run on tutorial case of damBreak (via interFoam) along with a modified solver(simply change UEqn.H() to UEqn.myH()) are both attached. An image show result difference is also given. dambreak-comparison.png (235,690 bytes) |
|
|
Do all the tutorial cases run correctly with your proposed change? Even cases with symmetry planes and other transform boundary conditions for which the code you removed is needed? |
|
|
The current formulation has been developed, tested and validated over a long period of time on a wide range of cases and has proved to be robust and reliable. If you are not happy with this implementation for your particular cases you can of course change it for your purpose but we cannot make this change to the core of OpenFOAM based on a single result or the preference of a single user. |
- Anonymous
