dumux-repositories issueshttps://git.iws.uni-stuttgart.de/groups/dumux-repositories/-/issues2019-03-12T09:45:21Zhttps://git.iws.uni-stuttgart.de/dumux-repositories/dumux/-/issues/664Staggered higher order bug in face solution2019-03-12T09:45:21ZTimo Kochtimokoch@math.uio.noStaggered higher order bug in face solutionThe following assertion fails for all freeflow tests after merging !1481:
`/data/src/dumux/dumux/discretization/staggered/facesolution.hh:68: const Dumux::StaggeredFaceSolution::FacePrimaryVariables &Dumux::StaggeredFaceSolution, std::a...The following assertion fails for all freeflow tests after merging !1481:
`/data/src/dumux/dumux/discretization/staggered/facesolution.hh:68: const Dumux::StaggeredFaceSolution::FacePrimaryVariables &Dumux::StaggeredFaceSolution, std::allocator > > >::operator[](IndexType) const [FaceSolutionVector = Dune::BlockVector, std::allocator > >, IndexType = int]: Assertion `pos != map_.end()' failed.`3.1Ned ColtmanNed Coltmanhttps://git.iws.uni-stuttgart.de/dumux-repositories/dumux/-/issues/663Staggered higher order memory2019-03-12T09:45:45ZTimo Kochtimokoch@math.uio.noStaggered higher order memoryIt should be investigated if the higher order implementation should be chosen at compile time so we can save memory and runtime.It should be investigated if the higher order implementation should be chosen at compile time so we can save memory and runtime.Ned ColtmanNed Coltmanhttps://git.iws.uni-stuttgart.de/dumux-repositories/dumux/-/issues/662Wrong use of extrusion factors (freeflow)2019-02-27T09:56:07ZKilian WeishauptWrong use of extrusion factors (freeflow)At certain points, the use of the extrusion factor is implemented wrong:
* fluxoversurface.hh lacks the factor for the calculation of the volume fluxes
* stokesdarcy/couplingdata.hh includes the factor in the calculation of diffusive fl...At certain points, the use of the extrusion factor is implemented wrong:
* fluxoversurface.hh lacks the factor for the calculation of the volume fluxes
* stokesdarcy/couplingdata.hh includes the factor in the calculation of diffusive fluxes, however, those values
are then applied as Neuman-BCs where the factor is again applied
__TODO__ @heck What about the Maxwell-Stefan fluxes in the coupling data?3.1Kilian WeishauptKilian Weishaupthttps://git.iws.uni-stuttgart.de/dumux-repositories/dumux/-/issues/661Possibly include update of Box flux variables cache2023-03-18T18:38:31ZDennis GläserPossibly include update of Box flux variables cacheThe flux variable caches for the box scheme are always assumed to be solution-independent. We should think of a way to support user-defined, solution-dependent caches.The flux variable caches for the box scheme are always assumed to be solution-independent. We should think of a way to support user-defined, solution-dependent caches.https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/-/issues/660Pass FluxVarsCache to Neumann Function2019-08-30T15:17:00ZDennis GläserPass FluxVarsCache to Neumann FunctionWe build the flux variable caches for boundary faces, but do not hand them into the Neumann function, where it could be used. For box, we actually never use the object, so it's unnecessary overhead.We build the flux variable caches for boundary faces, but do not hand them into the Neumann function, where it could be used. For box, we actually never use the object, so it's unnecessary overhead.3.1Dennis GläserDennis Gläserhttps://git.iws.uni-stuttgart.de/dumux-repositories/dumux/-/issues/659Why does newtonProceed require two steps minimum?2019-05-29T13:22:23ZTimo Kochtimokoch@math.uio.noWhy does newtonProceed require two steps minimum?3.1Timo Kochtimokoch@math.uio.noTimo Kochtimokoch@math.uio.nohttps://git.iws.uni-stuttgart.de/dumux-repositories/dumux/-/issues/658[2pncmin] precipitation of NaCl from the gas phase2019-03-27T15:43:22ZTheresa Schollenberger[2pncmin] precipitation of NaCl from the gas phaseA value is set for the source term of precipitated salt in gas, which is not physical. The suggestion is that this should balance out a fugacity of NaCl which was original nonequal to zero. However, the fugacity of NaCl was set to zero i...A value is set for the source term of precipitated salt in gas, which is not physical. The suggestion is that this should balance out a fugacity of NaCl which was original nonequal to zero. However, the fugacity of NaCl was set to zero in commit:9925d2d3 so this part of the source term has to be removed now.https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/-/issues/657Turbulent diffusion in shallow water equations2021-06-07T12:47:16ZFrank PlatzekTurbulent diffusion in shallow water equationsAdd diffusion terms to the shallow water equations. One or more turbulence models will follow later. For now a rudimentary implementation is chosen which is only accurate on orthogonal grids. Possible non-orthogonal corrections will be a...Add diffusion terms to the shallow water equations. One or more turbulence models will follow later. For now a rudimentary implementation is chosen which is only accurate on orthogonal grids. Possible non-orthogonal corrections will be added later on.Frank PlatzekFrank Platzekhttps://git.iws.uni-stuttgart.de/dumux-repositories/dumux/-/issues/656XDMF/HDF5 Support in DuMux2019-02-01T09:38:21ZLeopold StadlerXDMF/HDF5 Support in DuMuxWe actually use XDMF/HD5F for input/output of our simulations. There are two files in the swe branch which allow to read and write triangular grids. There are many advantages compared to the vtk-format but also some drawbacks (installing...We actually use XDMF/HD5F for input/output of our simulations. There are two files in the swe branch which allow to read and write triangular grids. There are many advantages compared to the vtk-format but also some drawbacks (installing parallel hdf5, filesystems,...).
Is there any interest from the DuMux side to include XDMF/HDF5 support directly in DuMux?
I suggest to remove the files and ask the dune core developers if they are interested to create a new dune-module for xdmf/hdf5 support. I'm also not sure if there is a simple python way which might simplify the whole thing (parallel reading/writing of XDMF/HDF5 with h5py and mpi4py).https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/-/issues/655Newton uninitialized value for residual criterion2019-02-25T10:31:33ZTimo Kochtimokoch@math.uio.noNewton uninitialized value for residual criterionThe following settings
```ini
[Newton]
EnableResidualCriterion = true
EnableAbsoluteResidualCriterion = true
SatisfyResidualAndShiftCriterion = true
MaxAbsoluteResidual = 1e-25
```
give an error with valgrind for an uninitialized value i...The following settings
```ini
[Newton]
EnableResidualCriterion = true
EnableAbsoluteResidualCriterion = true
SatisfyResidualAndShiftCriterion = true
MaxAbsoluteResidual = 1e-25
```
give an error with valgrind for an uninitialized value in newtonsolver.hh:521.
Might be `residualNorm_` ...
Needs investigation.3.1https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/-/issues/654Allow C++172020-01-29T13:09:44ZTimo Kochtimokoch@math.uio.noAllow C++17I think quite some new features speak for allowing C++17. There are some compilers supporting all features, gcc 8 or gcc 7 (without std::filesystem), clang 8 or 5 (without std::filesystem). https://en.cppreference.com/w/cpp/compiler_supp...I think quite some new features speak for allowing C++17. There are some compilers supporting all features, gcc 8 or gcc 7 (without std::filesystem), clang 8 or 5 (without std::filesystem). https://en.cppreference.com/w/cpp/compiler_support. However, there seems to be no cray compiler.
Who is using the newest Dumux version on a platform without C++17 support?3.2Timo Kochtimokoch@math.uio.noTimo Kochtimokoch@math.uio.nohttps://git.iws.uni-stuttgart.de/dumux-repositories/dumux/-/issues/653Unefficient call for parameter in each flux evaluation2019-01-15T10:24:42ZSamuel Burbullasamuel.burbulla@mathematik.uni-stuttgart.deUnefficient call for parameter in each flux evaluationI think the call
`static const bool enableGravity = getParamFromGroup<bool>(problem.paramGroup(), "Problem.EnableGravity");`
in e.g. flux/cctpfa/darcyslaw.hh:162
executed during each flux evaluation is a performance drawback.
May one wan...I think the call
`static const bool enableGravity = getParamFromGroup<bool>(problem.paramGroup(), "Problem.EnableGravity");`
in e.g. flux/cctpfa/darcyslaw.hh:162
executed during each flux evaluation is a performance drawback.
May one wants to cache the parameter somewhere, but this is not straightforward since the function is static.
![parameterProfile](/uploads/ac1ba4bec5672ffc96fb6628b4cb35c2/parameterProfile.png)https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/-/issues/652Provide interface for more parallel solvers/preconditioners of ISTL2020-03-13T15:11:22ZLeopold StadlerProvide interface for more parallel solvers/preconditioners of ISTLSo far we have only the AMG-backend (parallel linear solver based on the ISTL AMG preconditioner and the ISTL BiCGSTAB solver). It would be great to add support for more preconditioners and solvers of ISTL.
The following combinations sh...So far we have only the AMG-backend (parallel linear solver based on the ISTL AMG preconditioner and the ISTL BiCGSTAB solver). It would be great to add support for more preconditioners and solvers of ISTL.
The following combinations should be available:
* SSOR with BiCGSTAB
* ILU with BiCGSTAB
* ILU with GMRES
Are there any further wishes ?Leopold StadlerLeopold Stadlerhttps://git.iws.uni-stuttgart.de/dumux-repositories/dumux-lecture/-/issues/7heavyoil sagd_cyclic shows unphysical behaviour2020-11-02T10:46:54ZSimon Emmertheavyoil sagd_cyclic shows unphysical behaviour@holle discovered today that the sagd_cyclic and also the sagd_cyclic_hyst show some presumably unphysical behaviour.
This should be a comparison between dumux releases/2.12 and dumux-lecture on releases/2.12 with dumux releases/3.0 and ...@holle discovered today that the sagd_cyclic and also the sagd_cyclic_hyst show some presumably unphysical behaviour.
This should be a comparison between dumux releases/2.12 and dumux-lecture on releases/2.12 with dumux releases/3.0 and dumux-lecture with releases/3.0. We want to use the 3.0 version and regard the 2.12 as a reference.
I already found some obvious things that were missing so that the exercise actually runs and uses the CheckPointTimeLoop and started fixing this on !74.
The general problem however still seems to be that there is maybe (needs to be checked)"too much injection" and that when the injection is stopped the gas phase saturation does not decrease (as on 2.12) but continues increasing for some time even though the injection (and therefor pressure) are stopped
Our guideline would be:
* [ ] run the standard sagd on 2.12 and compare with 3.0 -> are they the same? Hopefully yes.
* [ ] test sagd_cyclic and sagd_cyclic_hyst
Possible things to investigate
* [ ] is injection the same (hopefully verified with sagd-problem)
* [ ] changes in the privar-switch (problem occurs when gas phase disappears)
* [ ] changes to the fluidsystem/components (molarDensity?)
* [ ] changes to constraintsolver/matrix-preconditioning
* [ ] extension of IAPWS region 2 (minimal change)Katharina HeckKatharina Heckhttps://git.iws.uni-stuttgart.de/dumux-repositories/dumux/-/issues/651Add VTK functionality to grid manager2019-03-28T08:49:55ZTimo Kochtimokoch@math.uio.noAdd VTK functionality to grid managerTimo Kochtimokoch@math.uio.noTimo Kochtimokoch@math.uio.nohttps://git.iws.uni-stuttgart.de/dumux-repositories/dumux/-/issues/650[freeflow][rans] update gridvariables in main file necessary for all rans tests2019-02-07T07:13:02ZKatharina Heck[freeflow][rans] update gridvariables in main file necessary for all rans testsWhen gridvolumevariables caching is switched on it is necessary to update the gridvariables again in the main file of the tests after the dynamic wall properties are updated in each time.
This can be done with
`// update wall propert...When gridvolumevariables caching is switched on it is necessary to update the gridvariables again in the main file of the tests after the dynamic wall properties are updated in each time.
This can be done with
`// update wall properties`
`problem->updateDynamicWallProperties(sol);`
`assembler->updateGridVariables(sol);`
There might be also be a better solution for this problem but this was the first fix @kweis found. I did not test yet if that changes any reference solutions for the rans problems but it changed a lot in a coupled test we look at.
@nedc This is the fix for the problem we had before christmas in the coupled komega test.3.1Ned ColtmanNed Coltmanhttps://git.iws.uni-stuttgart.de/dumux-repositories/dumux/-/issues/649[Freeflow][test][angeli] No convergence against analytical solution for first...2021-05-19T08:55:29ZMelanie Lipp[Freeflow][test][angeli] No convergence against analytical solution for first time stepThe solution does not properly converge for the first time step. The pressure solution is really weird for very small time steps (as the 10^(-12) that is currently used). It becomes less and less weird the larger the first time step is t...The solution does not properly converge for the first time step. The pressure solution is really weird for very small time steps (as the 10^(-12) that is currently used). It becomes less and less weird the larger the first time step is taken. But even for an initial time step of e.g. 10^(-2) the solution is an unsymmetrical version of what it is supposed to be.
It is clear that the solution has the chance to recover for later time steps as the pressure solution at previous time steps does, in contrary to the velocity solution, not play a role for the current solution.
__edit__: The problem also persists on the new staggered implementation: !25713.4https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/-/issues/648Put property definitions in tests (examples/applications) in separate header2020-04-09T17:49:48ZTimo Kochtimokoch@math.uio.noPut property definitions in tests (examples/applications) in separate headerTo make another step away from the problem-does-everything-mentality that was common (and a necessary evil) in dumux 2,
I suggest to put the property definitions into a separate header. There are not connected in any way to the `Problem`...To make another step away from the problem-does-everything-mentality that was common (and a necessary evil) in dumux 2,
I suggest to put the property definitions into a separate header. There are not connected in any way to the `Problem` class, so IMO they shouldn't be defined in the problem header. For the purpose of re-usability they also shouldn't be directly coded in the main file.
__Edit: Result from the vote__: We want to advertise putting properties in a separate header.
* [x] Separate properties into `properties.hh` for the examples (milestone 3.2) !1933
* [x] Separate properties into `properties.hh` for the dumux-course (milestone 3.2) [course#24](https://git.iws.uni-stuttgart.de/dumux-repositories/dumux-course/issues/24)
* [ ] Separate properties into `properties.hh` for the dumux-lecture (milestone 3.3) [lecture#15](https://git.iws.uni-stuttgart.de/dumux-repositories/dumux-lecture/issues/15)
* [ ] Separate properties into `properties.hh` for the tests (milestone 3.3)
This doesn't have to be in one go. If someone edits a test anyway feel free to do this change as well.3.2https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/-/issues/647Remove deprecated IOFields::init functions2019-02-28T15:43:43ZTimo Kochtimokoch@math.uio.noRemove deprecated IOFields::init functionshttps://git.iws.uni-stuttgart.de/dumux-repositories/dumux/-/issues/646Tests: Replace problem functions postTimeStep/preTimeStep2019-02-27T13:52:59ZTimo Kochtimokoch@math.uio.noTests: Replace problem functions postTimeStep/preTimeStepFunctions with such names are due to the version 2 way of implementing simulation flow control. In version 3 such functions should not exist because
* the function name doesn't say anything about what the function does
* the functionali...Functions with such names are due to the version 2 way of implementing simulation flow control. In version 3 such functions should not exist because
* the function name doesn't say anything about what the function does
* the functionality in those functions probably doesn't belong in the problem
The function might be removed by moving the functionality into the main file, or renamed to some appropriate name if the functionality really belongs to the problem (connected to BC/IC/Sources or maybe an analytical solution).3.1