All notable changes to the Lethe project will be documented in this file. The format is based on Keep a Changelog.
-
MINOR The
insertion box minimum/maximum x/y/zparameters have been replaced by one parameter namedinsertion box points coordinates#1074 -
MINOR The
insertion first/second/third directionparameters have been replaced by one parameter namedinsertion direction sequence#1074
- MINOR The mass conservation feature of the VOF model now also monitors the momentum of both phases. This should be useful for some cases and improve the sharpening mechanism. #1073
- MINOR The recently implemented VOF stasis constraint (#1048) was not compatible with adaptive mesh refinements and checkpointing system. When refining the mesh or when restarting a simulation, the filtered phase fraction was reinitialized, but not filled with the appropriate values. This issue is now solved by adding
apply_phase_filter()topost_mesh_adaptation()andread_checkpoint()VOF functions. #1072
-
MINOR In a similar way to "VOF Barycenter", the output of the "VOF Mass Conservation" is now displayed at every time iteration on the terminal when
calculate mass conservationin thepost-processingis enabled (default) and theverbosityis set toverbose. #1062 -
MINOR A new boolean argument in
make_tablefunctions has been added, namelydisplay_scientific_notationto decide if the display of table values is displayed at a fixed decimal or in scientific notation. #1062
-
MINOR Assertion on fields required for physical property models were added to the code. This is done to ensure that if a physical property model necessitating a certain field is employed and that field is not defined, the exception message thrown is comprehensible. #1065
-
MINOR A new
make_table_scalars_vectorsfunction has been added inutilities.h(.cc). #1062
- MINOR The eigenvalue estimation for the multigrid preconditioners in the lethe-fluid-matrix-free solver is now done internally by the PreconditionRelaxation class instead of using a PreconditionChebyshev and an estimate omega function #1064.
- MINOR Temperature-dependent
stasis constraintis now featured in the Melting Cavity heat transfer example. #1061
-
MINOR The new temperature-dependent solid domain constraints for two-phase Volume of Fluid (VOF) simulations has been implemented (
constrain_stasis_with_temperature_vof()). To select cells on which the constraints are applied, we check if the filtered phase fraction at quadrature points in the cell are within a range of accepted values. This range of accepted values is defined with thephase fraction toleranceparameter. This parameter defines an absolute tolerance on filtered phase fraction. Related documentation was updated and a new application test was also added. #1048 -
MINOR Added the capability to locally refine the mesh in the vicinity of the boundary conditions as specified by a list of boundary conditions. #1056
-
MINOR Previously implemented temperature-dependant solid domain constraints for one-fluid simulations (
constrain_solid_domain()) was renamed toconstrain_stasis_with_temperature()and its content changed a bit to avoid copied lines inconstrain_stasis_with_temperature_vof(). #1048 -
MINOR The parameter subsection
constrain solid domainwas renamed toconstrain stasis. Related documentation was also updated. #1048
- MINOR The "file" insertion method has been added to the DEM and CFD-DEM solvers.#1054
- MINOR In LPBF simulations, recoil pressure formulation
$p_\mathrm{rec} = 0.55p_\mathrm{sat} + [1/\rho]m_\mathrm{dot}^2$ was accounting 2 times for the term$[1/\rho]m_\mathrm{dot}^2$ because the latter is included in the term$0.55p_\mathrm{sat}$ . The formulation is now corrected and reads$p_\mathrm{rec} = 0.55p_\mathrm{sat}$ .
- MINOR The previously implemented weak compressibility (#815) was not working with the heat transfer (HT) physic. The missing pressure field was added to scratch data. Two application tests were added to test the compressibility with HT for single-fluid and VOF simulations. #1051
- MINOR Temperature-dependant solid domain constraints were added for one-fluid simulations. In order to improve calculation time and linear system conditioning, homogeneous constraints are imposed on velocity DOFs in a defined solid domain through a temperature range. Additionally, pressure DOFs in the same solid domain that are not connected to fluid cells are also imposed with homogeneous constraints. The temperature range is specified through a new parameter subsection, namely
constrain solid domain. The use of adynamic_zero_constraintsAffineConstraints object, rather than reconstruction of thezero_constraintsone at every time step was determined to be optimal after some profiling tests. Documentation and a new application test have been also added. #1038
- MINOR The
Variableenum class was moved fromparameters.htomultiphysics.h. Using variables from theVariableenum class, the constrained and constraining fields could be parametrized in the future according to user needs. #1038
- MINOR In
establish_solid_domain()the pressure was not constrained in solid cells that were not connected to a fluid due to a small logic mistake. This was corrected. #1038
- MAJOR The first level of the fine search candidates for particle-particle, particle-wall and particle-floating walls was never cleared even when the value of the unordered_map became an empty unordered_map. In simulations with load balancing or particles that were moving significantly, this could potentially lead to a scenario where the size of the first level of the fine search candidate became equal to the number of total particles in the simulation, potentially leading to a crash.
- MAJOR Restarts using lethe-fluid-particles with "void fraction time derivative = true" would be incoherent because the void fraction was reinitialized when restarting, leading to a wrong time derivative of the void fraction. This has been patched.
- MAJOR Robin boundary condition was renamed to incorporate the imposed heat flux (Neumann). The corresponding type parameter was
convection-radiationand the current isconvection-radiation-flux. All application tests were updated accordingly.
- MAJOR All application_tests that use files have now mpirun=1 in their output to ensure that they are run from a mpirun=1 folder. Since dealii/dealii#16551 deal.II 9.6 uses a serial folder for serial tests. However, deal.II 9.5 does not. To maintain portability between the two versions, we manually force a folder called mpirun=1 when running tests and adapt the copy of the files to this folder. This is a temporary fix that can be reverted once we drop support for deal.II 9.5
-
MINOR Darcy-like penalization is now extended to two-fluid VOF simulations with the
PhaseChangeDarcyVOFAssemblerassembler class. Related documentation has been updated and a new application test was added. #990 -
MINOR Extrapolation in time of temperature and temperature gradient values has been added. This is used for source terms in the Navier-Stokes equations requiring them (Marangoni and evaporation recoil force terms). #994
- MINOR Velocity extrapolation has been moved to
include/solvers/vof_scratch_data.h. Solution extrapolations are now located inScratchDatato avoid constraint on assemblers order. #994
- MINOR In the previous implementation of the heat flux postprocessor (#953), there was a confusion related to the
material_idof a cell. As a cell'smaterial_idis defined in the mesh, fluids moving from a cell to another did not have the rightmaterial_idwhen outputted. Furthermore, some deal.II meshes have an intrinsic way of numberingmaterial_ids(e.g. colorized subdivided_hyper_rectangle) that were leading to inadequate comparison of thematerial_ids. As of now, in the heat flux postprocessor, thematerial_idis only used to indentify cell material when solids are present. When simulating with fluids and solids, cells of the mesh in which the fluids lie should have amaterial_idof0. For fluids-only simulations, their heat fluxes are outputted on the entire domain, as the currentDataPostprocessorcan only hold 1dof_handler. During postprocessing, the user could clip the domain of interest of each fluid using thephaseorfiltered_phasefields outputted. #988
- MINOR Phase change thermal expansion now considers the solid thermal expansion within the mushy zone (between Tsolidus and Tliquidus) instead of using a blending rule. This leads to more stable results in the melting cavity benchmark
- MINOR PVD handles for solid DEM objects were written with the wrong text format ("_" separators instead of ".")
- MINOR The "insertion random number range" and "insertion random number seed" parameters got renamed to "insertion maximum offset" and "insertion prn seed" respectively. The old names didn't make sense, as they're not random because they're defined in the parameter file. #970
- MINOR The number of remaining particle to insert of each type is being checkpointed adequately. This means that no modification are required to the "number of particle" parameter after restarting a simulation. #964
- MINOR Solid objects can now be restarted adequately in DEM. They will resume at the position they had at the end of the simulation #959
- MINOR A few tables were omitted from being "checkpointed" causing a lost of valuable information when simulations were restarted (Issue: #916). The missing tables have been added to the
write_checkpoint()andread_checkpoint()of each physic. #938
- MINOR The average diameter of the uniform size distribution with the DEM module was specified with an "average diameter" parameter. It is now specified directly from the diameter parameter. This is correctly documented. #940.
- MINOR The DEM time step verification was outputting the most permissive time step (the biggest) and not the most restrictive (the smallest). This bugfix doesn't affect the uniform particle size simulation. #939.
- MINOR The plane insertion for the DEM was only supporting the uniform diameter distribution. Now it supports all types of distribution.
-
MINOR The maximum number of boundary conditions for all physics was fixed to 14 since the boundary conditions had to be declared before being parsed. A new mechanism is now in place which parses the "number" parameter for each physics and keeps the maximal value. Then, this maximal value is used to pre-declare the boundary conditions. This enables much more robust sanity checking of the input parameter. The major drawback of this (and this is a major one) is that if we ever have another parameter with the name "number" then this parameter would also be parsed and used to establish the maximum number of boundary conditions. In this case, the best approach would be to replace "number" with "number of boundary conditions" in the parameter file. I (B-saurus-rex) did not want to do this at the time of this change to not have a massive PR which breaks every parameter files.
-
MAJOR The "number" parameter within "subsection lagrangian physical properties" and "particle type n" was changed to "number of particles" to prevent confusions with the "number" used for boundary conditions. The "number" for boundary conditions will be changed to "number of boundary conditions" in the near future.
- MINOR The uniform insertion method had been removed. The non-uniform insertion method has been renamed to volume method to remain coherent with the plane method. If you want to use an insertion method equivalent to the uniform insertion method, use the volume method with an "insertion random number range " equal to zero. #926
- MINOR The calculation of the velocity on rotating walls was calculated inadequately #896: The velocity of rotating boundary conditions (e.g. boundaries of the mesh) was calculated inadequately in the case where the normal vector of the wall was aligned with the rotation axis. The whole calculation procedure was slightly messed up and only worked for cylinders. This has been fixed and made general, paving the way for a full refactor of the calculation of the particle-wall contact force.
- MINOR Affine constraints used to bound the void fraction and used as boundary conditions within the heat transfer and the block Navier-stokes solver were corrected #885: There was an error in the application of the affine constraints used to clip the void fraction in the lethe-fluid-vans and lethe-fluid-particles solvers. This led to assertions being thrown in debug. This has been corrected by reiniting the constraints with the appropriate size. For the heat transfer and the block Navier-stokes, the issue was that the constraints were never reinited with the correct size, so they contained ghosted elements. This was caught by a new assert introduced in 2023-09 within deal.II master.
- MINOR The calculation of the source term was enabled using a parameter called "enable". This parameter was used in some physics, not in others and was poorly implemented. We deprecate the usage of this parameter and always enable source term, considering the fact that the default value of the source term is zero anyway. This prevents the false perception that source terms could be enabled or disabled, while the behavior was inconsistent across physics.
- MAJOR All the applications were renamed #882:
gls_navier_stokesis nowlethe-fluid,gd_navier_stokesis nowlethe-fluid-block,nitsche_navier_stokesis nowlethe-fluid-nitsche,gls_sharp_navier_stokesis nowlethe-fluid-sharp,gls_vansis nowlethe-fluid-vans,mf_navier_stokesis nowlethe-fluid-matrix-free,demis nowlethe-particles,cfd_dem_couplingis nowlethe-fluid-particles,rpt3dis nowlethe-rpt-3d,rpt_cell_reconstruction_3dis nowlethe-rpt-cell-reconstruction-3d,rpt_fem_reconstruction_3dis nowlethe-rpt-fem-reconstruction-3d, andrpt_l2_projection_3dis nowlethe-rpt-l2-projection-3d.
- MINOR The rotational vector for the rotational boundary condition in the lethe-particles solver is now define with one line in the parameters file. #920
- MAJOR/MINOR/PATCH Description (#PR).
- MAJOR/MINOR/PATCH Description (#PR).
- MAJOR/MINOR/PATCH Description (#PR).
- MAJOR/MINOR/PATCH Description (#PR).
- MAJOR/MINOR/PATCH Description (#PR).