- Feb 10, 2017
-
-
* remove offset * use common function
-
* Make discretization test compile again
-
-
* No longer needed
-
* Get rid of fullFace stencils, not required anymore * Use new map in assembler and localJacobian
-
-
* Replace setRowSizes_ and addIndices_ functions with occupationPattern concept
-
* Allows usage of range-based for-loop
-
-
* Make use of [] operator of staggeredPriVars class
-
* Make base staggered discretization more general
-
* Set fluxVars in model-specific propertyDefaults * Remove unused code from staggered VolVars
-
* Move all flux calculations form localResidual to fluxVars * Move stencil computation from stencils to fluxVars FluxVars will 4 provide different specilizations: - single component, isothermal - single component, non-isothermal - multi component, isothermal - multi component, non-isothermal LocalResidual has to be adapted accordingly. Can be extended for turbulence models.
-
* Can be accessed with regular priVar indices * In the problem, a helper type has to be used in order to specifiy the velocities as vectors
-
-
*Greatly improves performance because only the local residual of the dof of interest gets calculated (not the whole element)
-
* Cell dof and and individual face dofs can now be treated separately
-
* Remove unused code * Use chached values of the residua for calculation of derivatives
-
*Test may fail due to tiny velocity discrepancies (no flow involved in this test)
-
* Stokes (Re=1) and Navier-Stokes (Re=1000) flow regimes
-
* Affects normal advective momentum flux and pressure term
-
*For Stokes/Navier-Stokes
-
*Tests now use a Dirichlet BC for the pressure to get unique results
-
* Treat boundaries on model-specific localResidual * Add support for Dirichlet BCs for velocity and presure
-
*Remove unused ones *introduce velocityIdx
-
*Same executable used for two different tests with different input files
-
-
*Lid-driven cavity *Hydrostatic pressure distribution
-
*Multiply by density *Use correct extrusion factor in volVars
-
* Use CCPriVars as type for ccResidual * Use FaceSolutionVector as type for daceResidual
-
*Hard coded values, has to be improved
-
* Include no-slip Dirichlet BC * Remove unused code
-
-
*Always take the velocity from the facevars (no Dirichlet values) to prevent zero entries in Jacobian *Account for up-wound velocity in normal momentum term
-
*Now also provides a vector
-
*For umfpack
-
* Do not write drs files * Pressure Dirichlet BC only on right side
-
* use Dirichlet velocity provided by problem. Might me done better in faceVars class ... * account for momentum in/outflux at boundary
-
-
-