fvMesh::update() now executes at the beginning of the time-step, before time is
incremented and handles topology change, mesh to mesh mapping and redistribution
without point motion. Following each of these mesh changes fields are mapped
from the previous mesh state to new mesh state in a conservative manner. These
mesh changes not occur at most once per time-step.
fvMesh::move() is executed after time is incremented and handles point motion
mesh morphing during the time-step in an Arbitrary Lagrangian Eulerian approach
requiring the mesh motion flux to match the cell volume change. fvMesh::move()
can be called any number of times during the time-step to allow iterative update
of the coupling between the mesh motion and field solution.
fvMesh is no longer derived from fvSchemes and fvSolution, these are now
demand-driven and accessed by the member functions schemes() and solution()
respectively. This means that the system/fvSchemes and system/fvSolution files
are no longer required during fvMesh constructions simplifying the mesh
generation and manipulation phase; theses files are read on the first call of
their access functions.
The fvSchemes member function names have also been simplified taking advantage
of the context in which they are called, for example
mesh.ddtScheme(fieldName) -> mesh.schemes().ddt(fieldName)
The handling of the div(phid,p) term for transonic support in the pressure
equation is now consistent such that conservation is achieved at convergence of
the pressure system irrespective of the scheme chosen for div(phid,p) and the
relaxation of the pressure equation.
The rhoSimpleFoam tutorials have been updated and improved.
In rhoPimpleFoam, rhoSimpleFoam, buoyantPimpleFoam and buoyantSimpleFoam the
density prediction step at the start of pEqn.H is now consistent between these
solvers and the other compressible solvers. If the density is relaxed in the
corrector it is now also relaxed following the predictor which improves
consistency, stability and convergence.
For some cases, in particular those with very small cells created by snapping in
corners for example, it may be beneficial to convergence rate to limit the
minimum LTS time-step, the new minDeltaT control provides this.
replacing the virtual functions overridden in engineTime.
Now the userTime conversion function in Time is specified in system/controlDict
such that the solver as well as all pre- and post-processing tools also operate
correctly with the chosen user-time.
For example the user-time and rpm in the tutorials/combustion/XiEngineFoam/kivaTest case are
now specified in system/controlDict:
userTime
{
type engine;
rpm 1500;
}
The default specification is real-time:
userTime
{
type real;
}
but this entry can be omitted as the real-time class is instantiated
automatically if the userTime entry is not present in system/controlDict.
Mesh motion and topology change are now combinable run-time selectable options
within fvMesh, replacing the restrictive dynamicFvMesh which supported only
motion OR topology change.
All solvers which instantiated a dynamicFvMesh now instantiate an fvMesh which
reads the optional constant/dynamicFvMeshDict to construct an fvMeshMover and/or
an fvMeshTopoChanger. These two are specified within the optional mover and
topoChanger sub-dictionaries of dynamicFvMeshDict.
When the fvMesh is updated the fvMeshTopoChanger is first executed which can
change the mesh topology in anyway, adding or removing points as required, for
example for automatic mesh refinement/unrefinement, and all registered fields
are mapped onto the updated mesh. The fvMeshMover is then executed which moved
the points only and calculates the cell volume change and corresponding
mesh-fluxes for conservative moving mesh transport. If multiple topological
changes or movements are required these would be combined into special
fvMeshMovers and fvMeshTopoChangers which handle the processing of a list of
changes, e.g. solidBodyMotionFunctions:multiMotion.
The tutorials/multiphase/interFoam/laminar/sloshingTank3D3DoF case has been
updated to demonstrate this new functionality by combining solid-body motion
with mesh refinement/unrefinement:
/*--------------------------------*- C++ -*----------------------------------*\
========= |
\\ / F ield | OpenFOAM: The Open Source CFD Toolbox
\\ / O peration | Website: https://openfoam.org
\\ / A nd | Version: dev
\\/ M anipulation |
\*---------------------------------------------------------------------------*/
FoamFile
{
format ascii;
class dictionary;
location "constant";
object dynamicMeshDict;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //
mover
{
type motionSolver;
libs ("libfvMeshMovers.so" "libfvMotionSolvers.so");
motionSolver solidBody;
solidBodyMotionFunction SDA;
CofG (0 0 0);
lamda 50;
rollAmax 0.2;
rollAmin 0.1;
heaveA 4;
swayA 2.4;
Q 2;
Tp 14;
Tpn 12;
dTi 0.06;
dTp -0.001;
}
topoChanger
{
type refiner;
libs ("libfvMeshTopoChangers.so");
// How often to refine
refineInterval 1;
// Field to be refinement on
field alpha.water;
// Refine field in between lower..upper
lowerRefineLevel 0.001;
upperRefineLevel 0.999;
// Have slower than 2:1 refinement
nBufferLayers 1;
// Refine cells only up to maxRefinement levels
maxRefinement 1;
// Stop refinement if maxCells reached
maxCells 200000;
// Flux field and corresponding velocity field. Fluxes on changed
// faces get recalculated by interpolating the velocity. Use 'none'
// on surfaceScalarFields that do not need to be reinterpolated.
correctFluxes
(
(phi none)
(nHatf none)
(rhoPhi none)
(alphaPhi.water none)
(meshPhi none)
(meshPhi_0 none)
(ghf none)
);
// Write the refinement level as a volScalarField
dumpLevel true;
}
// ************************************************************************* //
Note that currently this is the only working combination of mesh-motion with
topology change within the new framework and further development is required to
update the set of topology changers so that topology changes with mapping are
separated from the mesh-motion so that they can be combined with any of the
other movements or topology changes in any manner.
All of the solvers and tutorials have been updated to use the new form of
dynamicMeshDict but backward-compatibility was not practical due to the complete
reorganisation of the mesh change structure.
to provide a single consistent code and user interface to the specification of
physical properties in both single-phase and multi-phase solvers. This redesign
simplifies usage and reduces code duplication in run-time selectable solver
options such as 'functionObjects' and 'fvModels'.
* physicalProperties
Single abstract base-class for all fluid and solid physical property classes.
Physical properties for a single fluid or solid within a region are now read
from the 'constant/<region>/physicalProperties' dictionary.
Physical properties for a phase fluid or solid within a region are now read
from the 'constant/<region>/physicalProperties.<phase>' dictionary.
This replaces the previous inconsistent naming convention of
'transportProperties' for incompressible solvers and
'thermophysicalProperties' for compressible solvers.
Backward-compatibility is provided by the solvers reading
'thermophysicalProperties' or 'transportProperties' if the
'physicalProperties' dictionary does not exist.
* phaseProperties
All multi-phase solvers (VoF and Euler-Euler) now read the list of phases and
interfacial models and coefficients from the
'constant/<region>/phaseProperties' dictionary.
Backward-compatibility is provided by the solvers reading
'thermophysicalProperties' or 'transportProperties' if the 'phaseProperties'
dictionary does not exist. For incompressible VoF solvers the
'transportProperties' is automatically upgraded to 'phaseProperties' and the
two 'physicalProperties.<phase>' dictionary for the phase properties.
* viscosity
Abstract base-class (interface) for all fluids.
Having a single interface for the viscosity of all types of fluids facilitated
a substantial simplification of the 'momentumTransport' library, avoiding the
need for a layer of templating and providing total consistency between
incompressible/compressible and single-phase/multi-phase laminar, RAS and LES
momentum transport models. This allows the generalised Newtonian viscosity
models to be used in the same form within laminar as well as RAS and LES
momentum transport closures in any solver. Strain-rate dependent viscosity
modelling is particularly useful with low-Reynolds number turbulence closures
for non-Newtonian fluids where the effect of bulk shear near the walls on the
viscosity is a dominant effect. Within this framework it would also be
possible to implement generalised Newtonian models dependent on turbulent as
well as mean strain-rate if suitable model formulations are available.
* visosityModel
Run-time selectable Newtonian viscosity model for incompressible fluids
providing the 'viscosity' interface for 'momentumTransport' models.
Currently a 'constant' Newtonian viscosity model is provided but the structure
supports more complex functions of time, space and fields registered to the
region database.
Strain-rate dependent non-Newtonian viscosity models have been removed from
this level and handled in a more general way within the 'momentumTransport'
library, see section 'viscosity' above.
The 'constant' viscosity model is selected in the 'physicalProperties'
dictionary by
viscosityModel constant;
which is equivalent to the previous entry in the 'transportProperties'
dictionary
transportModel Newtonian;
but backward-compatibility is provided for both the keyword and model
type.
* thermophysicalModels
To avoid propagating the unnecessary constructors from 'dictionary' into the
new 'physicalProperties' abstract base-class this entire structure has been
removed from the 'thermophysicalModels' library. The only use for this
constructor was in 'thermalBaffle' which now reads the 'physicalProperties'
dictionary from the baffle region directory which is far simpler and more
consistent and significantly reduces the amount of constructor code in the
'thermophysicalModels' library.
* compressibleInterFoam
The creation of the 'viscosity' interface for the 'momentumTransport' models
allows the complex 'twoPhaseMixtureThermo' derived from 'rhoThermo' to be
replaced with the much simpler 'compressibleTwoPhaseMixture' derived from the
'viscosity' interface, avoiding the myriad of unused thermodynamic functions
required by 'rhoThermo' to be defined for the mixture.
Same for 'compressibleMultiphaseMixture' in 'compressibleMultiphaseInterFoam'.
This is a significant improvement in code and input consistency, simplifying
maintenance and further development as well as enhancing usability.
Henry G. Weller
CFD Direct Ltd.
The pressure work term for total internal energy is div(U p) which can be
discretised is various ways, given a mass flux field phi it seems logical to
implement it in the form div(phi/interpolate(rho), p) but this is not exactly
consistent with the relationship between enthalpy and internal energy (h = e +
p/rho) and the transport of enthalpy, it would be more consistent to implement
it in the form div(phi, p/rho). A further improvement in consistency can be
gained by using the same convection scheme for this work term and the convection
term div(phi, e) and for reacting solvers this is easily achieved by using the
multi-variate limiter mvConvection provided for energy and specie convection.
This more consistent total internal energy work term has now been implemented in
all the compressible and reacting flow solvers and provides more accurate
solutions when running with internal energy, particularly for variable density
mixing cases with small pressure variation.
For non-reacting compressible solvers this improvement requires a change to the
corresponding divScheme in fvSchemes:
div(phiv,p) -> div(phi,(p|rho))
and all the tutorials have been updated accordingly.
With the new fvModels framework it is now possible to implement complex models
and wrappers around existing complex models which can then be optionally
selected in any general solver which provides compatible fields and
thermophysical properties. This simplifies code development and maintenance by
significantly reducing complex code duplication and also provide the opportunity
of running these models in other solvers without the need for code duplication
and alteration.
The immediate advantage of this development is the replacement of the
specialised Lagrangian solvers with their general counterparts:
reactingParticleFoam -> reactingFoam
reactingParcelFoam -> reactingFoam
sprayFoam -> reactingFoam
simpleReactingParticleFoam -> reactingFoam
buoyantReactingParticleFoam -> buoyantReactingFoam
For example to run a reactingParticleFoam case in reactingFoam add the following
entries in constant/fvModels:
buoyancyForce
{
type buoyancyForce;
}
clouds
{
type clouds;
libs ("liblagrangianParcel.so");
}
which add the acceleration due to gravity needed by Lagrangian clouds and the
clouds themselves.
See the following cases for examples converted from reactingParticleFoam:
$FOAM_TUTORIALS/combustion/reactingFoam/Lagrangian
and to run a buoyantReactingParticleFoam case in buoyantReactingFoam add the
following entry constant/fvModels:
clouds
{
type clouds;
libs ("liblagrangianParcel.so");
}
to add support for Lagrangian clouds and/or
surfaceFilm
{
type surfaceFilm;
libs ("libsurfaceFilmModels.so");
}
to add support for surface film. The buoyancyForce fvModel is not required in
this case as the buoyantReactingFoam solver has built-in support for buoyancy
utilising the p_rgh formulation to provide better numerical handling for this
force for strongly buoyancy-driven flows.
See the following cases for examples converted from buoyantReactingParticleFoam:
$FOAM_TUTORIALS/combustion/buoyantReactingFoam/Lagrangian
All the tutorial cases for the redundant solvers have been updated and converted
into their new equivalents and redirection scripts replace these solvers to
provide users with prompts on which solvers have been replaced by which and
information on how to upgrade their cases.
To support this change and allow all previous Lagrangian tutorials to run as
before the special Lagrangian solver fvSolution/PIMPLE control
solvePrimaryRegion has been replaced by the more general and useful controls:
models : Enable the fvModels
thermophysics : Enable thermophysics (energy and optional composition)
flow : Enable flow (pressure/velocity system)
which also replace the fvSolution/PIMPLE control frozenFlow present in some
solvers. These three controls can be used in various combinations to allow for
example only the fvModels to be evaluated, e.g. in
$FOAM_TUTORIALS/combustion/buoyantReactingFoam/Lagrangian/rivuletPanel
PIMPLE
{
models yes;
thermophysics no;
flow no;
.
.
.
so that only the film is solved. Or during the start-up of a case it might be
beneficial to run the pressure-velocity system for a while without updating
temperature which can be achieved by switching-off thermophysics. Also the
behaviour of the previous frozenFlow switch can be reproduced by switching flow
off with the other two switches on, allowing for example reactions, temperature
and composition update without flow.
To provide more flexibility, extensibility, run-time modifiability and
consistency the handling of optional pressure limits has been moved from
pressureControl (settings in system/fvSolution) to the new limitPressure
fvConstraint (settings in system/fvConstraints).
All tutorials have been updated which provides guidance when upgrading cases but
also helpful error messages are generated for cases using the old settings
providing specific details as to how the case should be updated, e.g. for the
tutorials/compressible/rhoSimpleFoam/squareBend case which has the pressure
limit specification:
SIMPLE
{
...
pMinFactor 0.1;
pMaxFactor 2;
...
generates the error message
--> FOAM FATAL IO ERROR:
Pressure limits should now be specified in fvConstraints:
limitp
{
type limitPressure;
minFactor 0.1;
maxFactor 2;
}
file: /home/dm2/henry/OpenFOAM/OpenFOAM-dev/tutorials/compressible/rhoSimpleFoam/squareBend/system/fvSolution/SIMPLE from line 41 to line 54.
The MomentumTransportModels library now builds of a standard set of
phase-incompressible and phase-compressible models. This replaces most
solver-specific builds of these models.
This has been made possible by the addition of a new
"dynamicTransportModel" interface, from which all transport classes used
by the momentum transport models now derive. For the purpose of
disambiguation, the old "transportModel" has also been renamed
"kinematicTransportModel".
This change has been made in order to create a consistent definition of
phase-incompressible and phase-compressible MomentumTransportModels,
which can then be looked up by functionObjects, fvModels, and similar.
Some solvers still build specific momentum transport models, but these
are now in addition to the standard set. The solver does not build all
the models it uses.
There are also corresponding centralised builds of phase dependent
ThermophysicalTransportModels.
The new fvModels is a general interface to optional physical models in the
finite volume framework, providing sources to the governing conservation
equations, thus ensuring consistency and conservation. This structure is used
not only for simple sources and forces but also provides a general run-time
selection interface for more complex models such as radiation and film, in the
future this will be extended to Lagrangian, reaction, combustion etc. For such
complex models the 'correct()' function is provided to update the state of these
models at the beginning of the PIMPLE loop.
fvModels are specified in the optional constant/fvModels dictionary and
backward-compatibility with fvOption is provided by reading the
constant/fvOptions or system/fvOptions dictionary if present.
The new fvConstraints is a general interface to optional numerical constraints
applied to the matrices of the governing equations after construction and/or to
the resulting field after solution. This system allows arbitrary changes to
either the matrix or solution to ensure numerical or other constraints and hence
violates consistency with the governing equations and conservation but it often
useful to ensure numerical stability, particularly during the initial start-up
period of a run. Complex manipulations can be achieved with fvConstraints, for
example 'meanVelocityForce' used to maintain a specified mean velocity in a
cyclic channel by manipulating the momentum matrix and the velocity solution.
fvConstraints are specified in the optional system/fvConstraints dictionary and
backward-compatibility with fvOption is provided by reading the
constant/fvOptions or system/fvOptions dictionary if present.
The separation of fvOptions into fvModels and fvConstraints provides a rational
and consistent separation between physical and numerical models which is easier
to understand and reason about, avoids the confusing issue of location of the
controlling dictionary file, improves maintainability and easier to extend to
handle current and future requirements for optional complex physical models and
numerical constraints.
Field corrections are effectively explicit constraints applied to the field
after solution rather than to the equation and it significantly simplifies the
implementation to treat them as a special case of constraints. To implement
this the fvOption::correct(<field>) function has been renamed
fvOption::constrain(<field>) and uses constrainsField and constrainedFields.
It is better to not select and instantiate a model, fvOption etc. than to create
it and set it inactive as the creation process requires reading of settings,
parameters, fields etc. with all the associated specification and storage
without being used. Also the incomplete implementation added a lot of
complexity in the low-level operation of models introducing a significant
maintenance overhead and development overhead for new models.
ThermophysicalTransportModel is now instantiated on both the
MomentmumTransportModel and also the particular thermo model model rather than
obtaining the fluidThermo from the MomentmumTransportModel. This gives direct
access to the higher-level thermo model used in the solver, for example
rhoReactionThermo so that complex ThermophysicalTransportModels requiring access
to the composition for example are instantiated only for thermo models that
provide it and also avoiding run-time up-casting of the thermo model.
providing the shear-stress term in the momentum equation for incompressible and
compressible Newtonian, non-Newtonian and visco-elastic laminar flow as well as
Reynolds averaged and large-eddy simulation of turbulent flow.
The general deviatoric shear-stress term provided by the MomentumTransportModels
library is named divDevTau for compressible flow and divDevSigma (sigma =
tau/rho) for incompressible flow, the spherical part of the shear-stress is
assumed to be either included in the pressure or handled separately. The
corresponding stress function sigma is also provided which in the case of
Reynolds stress closure returns the effective Reynolds stress (including the
laminar contribution) or for other Reynolds averaged or large-eddy turbulence
closures returns the modelled Reynolds stress or sub-grid stress respectively.
For visco-elastic flow the sigma function returns the effective total stress
including the visco-elastic and Newtonian contributions.
For thermal flow the heat-flux generated by thermal diffusion is now handled by
the separate ThermophysicalTransportModels library allowing independent run-time
selection of the heat-flux model.
During the development of the MomentumTransportModels library significant effort
has been put into rationalising the components and supporting libraries,
removing redundant code, updating names to provide a more logical, consistent
and extensible interface and aid further development and maintenance. All
solvers and tutorials have been updated correspondingly and backward
compatibility of the input dictionaries provided.
Henry G. Weller
CFD Direct Ltd.
The simplistic energy transport support in compressibleTurbulenceModels has been
abstracted and separated into the new ThermophysicalTransportModels library in
order to provide a more general interface to support complex energy and specie
transport models, in particular multi-component diffusion. Currently only the
Fourier for laminar and eddyDiffusivity for RAS and LES turbulent flows are
provided but the interface is general and the set of models will be expanded in
the near future.
The ThermalDiffusivity and EddyDiffusivity modelling layers remain in
compressibleTurbulenceModels but will be removed shortly and the alphat boundary
conditions will be moved to ThermophysicalTransportModels.
rhoPimpleFoam now produces identical results to rhoSimpleFoam when run
with a steady-state time-scheme. The intention is that this solver can
now be used as a reference when adding steady-state support to other
compressible solvers for which no SIMPLE variant exists.
rhoReactingFoam has also been updated to support SIMPLE operation, as it
shares a pressure equation with rhoPimpleFoam.
Changed the interpolation of HbyA from
fvc::flux(rho*HbyA)
to
fvc::interpolate(rho)*fvc::flux(HbyA)
for consistency with the latest compressible p-U algorithm in rhoPimpleFoam.
For most cases this change does not affect the results but test on highly
compressible, transonic and supersonic cases have shown a small but clear
benefit in the new form.
This switch should be on for phi-correction within the time loop, where
the correction simply serves to keep the phi-field up to date before the
U-equation is solved. It should be off for initialisation
phi-correction, as the necessary data to update the conditions may not
yet exist.
Resolves bug report https://bugs.openfoam.org/view.php?id=3198
The writeEntry form is now defined and used consistently throughout OpenFOAM
making it easier to use and extend, particularly to support binary IO of complex
dictionary entries.
The sub-loops of the solution control are now named more consistently,
with ambiguously named methods such as finalIter replaced with ones
like finalPimpleIter, so that it is clear which loop they represent.
In addition, the final logic has been improved so that it restores state
after a sub-iteration, and so that sub-iterations can be used on their
own without an outer iteration in effect. Previously, if the
non-orthogonal loop were used outside of a pimple/piso iteration, the
final iteration would not execute with final settings.
The new patch field mapping class timeVaryingMappedFvPatchField has been
factored out of the timeVaryingMappedFixedValueFvPatchField BC so that it can be
used to map data onto fields stored within other BCs.
In the process the writeEntryIfDifferent function had to be moved from
fvPatchField to dictionary so that it can still be used in the
timeVaryingMappedFvPatchField class and it made good sense to create the
non-conditional variant writeEntry to simplify the patch field write functions.
This rationalisation has been propagated all other patch fields.
Registration occurs when the temporary field is transferred to a non-temporary
field via a constructor or if explicitly transferred to the database via the
regIOobject "store" methods.
