View Issue Details
|ID||Project||Category||View Status||Date Submitted||Last Update|
|0001988||OpenFOAM||[All Projects] Bug||public||2016-02-04 18:24||2016-02-04 20:31|
|Fixed in Version|
|Summary||0001988: Not null effect of fvOptions|
|Description||In driftFluxFoam, particularly UEqn.H, commenting fvOptions line is not equal to have this line but without fvOptions set in dictionaries (no fvOptions file).|
|Steps To Reproduce||1. Run dahl example from the distro saving the log|
2. Edit UEqn.H from driftFluxFoam like this:
// Solve the Momentum equation
+ fvm::div(rhoPhi, U)
// fvOptions(rho, U)
3. Compile and run dahl example again saving the log
4. Compare both logs using diff, results will be different. I think that they should be equal, if not, Which is the correct one?
|Tags||No tags attached.|
Both results are correct but slightly different because of the order in which terms are evaluated and boundary conditions updated. However the dependency on the location or presence of the fvOptions term is unsatisfactory.
The trailing fvOptions causes a zero fvMatrix construction which in tern causes the U boundary conditions to be updated before the explicit part of the divDevRhoReff stress term. Note that in C++ expressions are evaluated right-to-left.
I have reordered the terms in divDevRhoReff to ensure that the U boundary conditions to be updated before the explicit part removing the dependency on the presence or location of the fvOptions term in the equation.
Resolved in OpenFOAM-dev by commit 8e04a529d9edcc35a927be97b6995f9c6d776413