1. 25 Mar, 2015 1 commit
    • Timo Koch's avatar
      [implicit] resize local jacobian if necessay · f70335af
      Timo Koch authored
      The local jocabian is sized 1*2^dim x 1*2^dim for the box scheme and 1 x 2*dim for the cc scheme as default. 
      Some advances approximation schemes, grids with hanging nodes, mixed dimensional grids require different sizes. 
      The actual size was already calculated in implicitlocaljacobian. 
      If the actual size is bigger than the default size the matrix is resized accordingly now.
      
      
      
      git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@14441 2fb0f335-1f38-0410-981e-8018bf24f1b0
      f70335af
  2. 23 Mar, 2015 1 commit
  3. 12 Mar, 2015 1 commit
  4. 20 Feb, 2015 1 commit
    • Bernd Flemisch's avatar
      [implicit] store Geometry if it is re-used · ec5a8c79
      Bernd Flemisch authored
      Since grid implementations like Yasp calculate the Geometries on the
      fly, every call to entity.geometry() may involve computation. This patch
      stores a Geometry when it can be used later on in the same block. Only
      for the implicit models, decoupled contains a lot more.
      
      For simple models this actually pays off: for 1p in 3d on a larger grid,
      the running time could be reduced from 15.5 to 14 s.
      
      Reviewed by Timo.
      
      
      
      git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@14284 2fb0f335-1f38-0410-981e-8018bf24f1b0
      ec5a8c79
  5. 28 Jan, 2015 1 commit
  6. 27 Jan, 2015 1 commit
    • Bernd Flemisch's avatar
      [newton] improve terminology · b5604a68
      Bernd Flemisch authored
      The terminology for the Newton method is improved according to FS#238.
      In particular, what has been referred to as "relative error" is now
      termed "maximum relative shift", while "absolute error" has been renamed
      to "residual reduction". This is particularly important, if
      corresponding parameters or properties are set. See the CHANGELOG for
      details.
      
      Reviewed by gruenich.
      
      
      
      git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@14132 2fb0f335-1f38-0410-981e-8018bf24f1b0
      b5604a68
  7. 23 Jan, 2015 1 commit
    • Bernd Flemisch's avatar
      [implicit] complete and test dim < dimWorld capability · 9f61cb18
      Bernd Flemisch authored
      All implicit porous-media models except 2pdfm are now able to run on
      grids with dim < dimWorld. This still required some replacements of dim
      by dimWorld, for example, in the velocity output of all models. In
      implicit/1p, four new tests are added that run the 1p test problem on
      1d-2d and 1d-3d Alberta grids with box and cell-centered, respectively.
      Compilation has been tested also for all other models, but no runtime
      testing has been performed.
      
      This completes FS#183. Based upon preliminary work and a patch by
      Natalie, thank you.
      
      Reviewed by Natalie.
      
      
      git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@14110 2fb0f335-1f38-0410-981e-8018bf24f1b0
      9f61cb18
  8. 20 Jan, 2015 2 commits
  9. 16 Jan, 2015 2 commits
    • Bernd Flemisch's avatar
      [reverse] reverse unintended part of r14070 · da25ffc8
      Bernd Flemisch authored
      
      
      git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@14071 2fb0f335-1f38-0410-981e-8018bf24f1b0
      da25ffc8
    • Bernd Flemisch's avatar
      [nonisothermal] improve naming and meaning of solid thermal parameters · 8465d28a
      Bernd Flemisch authored
      This implements FS#216.
      
      The "heatCapacity" function of the spatial parameters and the volume
      variables for the implicit nonisothermal models was a misnomer, since it
      returned an effective quantity, namely, 
      heatCapacity*density*(1 - porosity) in [J/(K m^3)].
      Except for mpnc, which resulted in an additional inconsistency.
      
      Corresponding to the decision documented in FS#216, this patch renames
      the function to "solidHeatCapacity" and returns always the "true"
      (non-effective) heat capacity in [J/(kg K)]. This requires an additional
      function "solidDensity" which returns the mass density of the porous
      matrix. Moreover, the functions "thermalConductivitySolid/Fluid" are
      renamed to "solid/fluidThermalConductivity". The decision to prepend
      with "solid/fluid" rather than to append is motivated by consistency
      with components and fluid systems, where "gas" and "liquid" are always
      prepended to the corresponding function names.
      
      _Beware_: this change breaks compatibility. You have to adapt your
      spatial parameters such that they offer functions "solidHeatCapacity",
      "solidDensity" and "solidThermalConductivity".
      
      Reviewed by Alex.
      
      
      
      git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@14070 2fb0f335-1f38-0410-981e-8018bf24f1b0
      8465d28a
  10. 15 Jan, 2015 1 commit
  11. 13 Jan, 2015 1 commit
  12. 10 Dec, 2014 1 commit
  13. 09 Dec, 2014 1 commit
  14. 03 Dec, 2014 1 commit
  15. 28 Nov, 2014 1 commit
    • Bernd Flemisch's avatar
      [cell-centered implicit] fix a bug in the handling of mixed boundary · a78975bb
      Bernd Flemisch authored
      conditions
      
      Mixed boundary conditions here means that a part of the
      equations/primary variables gets Dirichlet conditions, while the rest
      gets Neumann conditions.
      
      While in the cell-centered method pure Dirichlet conditions for all
      equations are handled by calculating the resulting fluxes and adding
      them to the cell residual, such a flux cannot be (easily?) calculated
      for an equation that gets a Dirichlet condition as part of mixed
      conditions. Therefore, such a Dirichlet condition is implemented in a
      strong way by _replacing_ the corresponding cell residual.
      
      It is important that this replacement is done at the very end of the
      residual calculation. However, for corner cells this has not been
      guaranteed so far. Therefore, fluxed resulting from the other boundary
      parts of a corner cell could have been added to the replaced residual,
      obviously leading to a wrong boundary condition treatment.
      
      This patch resolves the issue by guaranteeing that the residual
      replacement is done at the end of the residual calculation.
      
      Brought to attention and reviewed by Thomas. Thanks.
      
      
      
      git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13823 2fb0f335-1f38-0410-981e-8018bf24f1b0
      a78975bb
  16. 25 Nov, 2014 4 commits
  17. 23 Sep, 2014 2 commits
  18. 22 Sep, 2014 1 commit
  19. 16 Sep, 2014 1 commit
  20. 05 Sep, 2014 1 commit
  21. 29 Aug, 2014 1 commit
  22. 29 Jul, 2014 1 commit
  23. 29 Apr, 2014 1 commit
  24. 08 Apr, 2014 1 commit
  25. 31 Mar, 2014 1 commit
  26. 25 Mar, 2014 1 commit
  27. 07 Mar, 2014 1 commit
  28. 14 Feb, 2014 1 commit
  29. 06 Feb, 2014 1 commit
  30. 22 Nov, 2013 1 commit
  31. 13 Nov, 2013 1 commit
  32. 11 Nov, 2013 2 commits
  33. 31 Oct, 2013 1 commit