LTS is selected by the ddt scheme e.g. in the
tutorials/multiphase/interFoam/ras/DTCHull case:
ddtSchemes
{
default localEuler rDeltaT;
}
LTSInterFoam is no longer needed now that interFoam includes LTS
support.
to support laminar simulations in the new framework.
Unfortunately this allows LES models to be instantiated although they
are not compatible with steady-state simulations.
Select LTS via the ddtScheme:
ddtSchemes
{
default localEuler rDeltaT;
}
The LTS algorithm is controlled with the standard settings in
controlDict, e.g.:
maxCo 0.5;
maxDeltaT 2e-8;
with the addition of the optional rDeltaT smoothing coefficient:
rDeltaTSmoothingCoeff 0.02;
which defaults to 0.02.
For cases with reasonably uniform meshes like the forwardStep tutorial
LTS does not provide much benefit but for cases with large variation in
cell-size like the biconic25-55Run35 tutorial LTS provides significant
speed-up to convergence particularly if started from uniform conditions.
Reverted changes proposed in
http://openfoam.org/mantisbt/view.php?id=1548 as it adversely affects
fixed-value BCs and is formulated to fix an issue with an unphysical
case. Further analysis of the handling of fixed pressure outlet
conditions as the Mach number approaches 1 is required.
Multi-species, mass-transfer and reaction support and multi-phase
structure provided by William Bainbridge.
Integration of the latest p-U and face-p_U algorithms with William's
multi-phase structure is not quite complete due to design
incompatibilities which needs further development. However the
integration of the functionality is complete.
The results of the tutorials are not exactly the same for the
twoPhaseEulerFoam and reactingTwoPhaseEulerFoam solvers but are very
similar. Further analysis in needed to ensure these differences are
physical or to resolve them; in the meantime the twoPhaseEulerFoam
solver will be maintained.
Model which applies an analytical solution for heat transfer from the
surface of a sphere to the fluid within the sphere.
Provided by William Bainbridge
fvOptions does not have the appropriate structure to support MRF as it
is based on option selection by user-specified fields whereas MRF MUST
be applied to all velocity fields in the particular solver. A
consequence of the particular design choices in fvOptions made it
difficult to support MRF for multiphase and it is easier to support
frame-related and field related options separately.
Currently the MRF functionality provided supports only rotations but
the structure will be generalized to support other frame motions
including linear acceleration, SRF rotation and 6DoF which will be
run-time selectable.
SIMPLEC (SIMPLE-consistent) is selected by setting "consistent" option true/yes:
SIMPLE
{
nNonOrthogonalCorrectors 0;
consistent yes;
}
which relaxes the pressure in a "consistent" manner and additional
relaxation of the pressure is not generally necessary. In addition
convergence of the p-U system is better and reliable with less
aggressive relaxation of the momentum equation, e.g. for the motorbike
tutorial:
relaxationFactors
{
equations
{
U 0.9;
k 0.7;
omega 0.7;
}
}
The cost per iteration is marginally higher but the convergence rate is
better so the number of iterations can be reduced.
The SIMPLEC algorithm also provides benefit for cases with large
body-forces, e.g. SRF, see tutorials/incompressible/SRFSimpleFoam/mixer
and feature request http://www.openfoam.org/mantisbt/view.php?id=1714
Rather than forcing the dispersed-phase velocity -> the continuous-phase
velocity as the phase-fraction -> 0 the velocity is now calculated from
a balance of pressure, buoyancy and drag forces. The advantage is now
liquid or particles are not carried out of bubble-column of
fluidised-beds by the fictitious drag caused by forcing the
phase-velocities becoming equal in the limit.
nLimiterIter: Number of iterations during limiter construction
3 (default) is sufficient for 3D simulations with a Courant number 0.5 or so
For larger Courant numbers larger values may be needed but this is
only relevant for IMULES and CMULES
smoothLimiter: Coefficient to smooth the limiter to avoid "diamond"
staggering patters seen in regions of low particle phase-fraction in
fluidised-bed simulations.
The default is 0 as it is not needed for all simulations.
A value of 0.1 is appropriate for fluidised-bed simulations.
The useful range is 0 -> 0.5.
Values larger than 0.5 may cause excessive smearing of the solution.
This formulation provides C-grid like pressure-flux staggering on an
unstructured mesh which is hugely beneficial for Euler-Euler multiphase
equations as it allows for all forces to be treated in a consistent
manner on the cell-faces which provides better balance, stability and
accuracy. However, to achieve face-force consistency the momentum
transport terms must be interpolated to the faces reducing accuracy of
this part of the system but this is offset by the increase in accuracy
of the force-balance.
Currently it is not clear if this face-based momentum equation
formulation is preferable for all Euler-Euler simulations so I have
included it on a switch to allow evaluation and comparison with the
previous cell-based formulation. To try the new algorithm simply switch
it on, e.g.:
PIMPLE
{
nOuterCorrectors 3;
nCorrectors 1;
nNonOrthogonalCorrectors 0;
faceMomentum yes;
}
It is proving particularly good for bubbly flows, eliminating the
staggering patterns often seen in the air velocity field with the
previous algorithm, removing other spurious numerical artifacts in the
velocity fields and improving stability and allowing larger time-steps
For particle-gas flows the advantage is noticeable but not nearly as
pronounced as in the bubbly flow cases.
Please test the new algorithm on your cases and provide feedback.
Henry G. Weller
CFD Direct
Improves stability and convergence of systems in which drag dominates
e.g. small particles in high-speed gas flow.
Additionally a new ddtPhiCorr strategy is included in which correction
is applied only where the phases are nearly pure. This reduces
staggering patters near the free-surface of bubble-column simulations.
Allows the specification of a reference height, for example the height
of the free-surface in a VoF simulation, which reduces the range of p_rgh.
hRef is a uniformDimensionedScalarField specified via the constant/hRef
file, equivalent to the way in which g is specified, so that it can be
looked-up from the database. For example see the constant/hRef file in
the DTCHull LTSInterFoam and interDyMFoam cases.
Disadvantage is that the BC values have to be specified in terms of hU
rather than U. The alternative would be to add complex code to map h
and U BCs into the equivalent for hU.
Resolves bug-report http://www.openfoam.org/mantisbt/view.php?id=1566
This is an experimental feature demonstrating the potential of MULES to
create bounded solution which are 2nd-order in time AND space.
Crank-Nicolson may be selected on U and/or alpha but will only be fully
2nd-order if used on both within the PIMPLE-loop to converge the
interaction between the flux and phase-fraction. Note also that
Crank-Nicolson may not be used with sub-cycling but all the features of
semi-implicit MULES are available in particular MULESCorr and
alphaApplyPrevCorr.
Examples of ddt specification:
ddtSchemes
{
default Euler;
}
ddtSchemes
{
default CrankNicolson 0.9;
}
ddtSchemes
{
default none;
ddt(alpha) CrankNicolson 0.9;
ddt(rho,U) CrankNicolson 0.9;
}
ddtSchemes
{
default none;
ddt(alpha) Euler;
ddt(rho,U) CrankNicolson 0.9;
}
ddtSchemes
{
default none;
ddt(alpha) CrankNicolson 0.9;
ddt(rho,U) Euler;
}
In these examples a small amount of off-centering in used to stabilize
the Crank-Nicolson scheme. Also the specification for alpha1 is via the
generic phase-fraction name to ensure in multiphase solvers (when
Crank-Nicolson support is added) the scheme is identical for all phase
fractions.
The Phi field is read if available otherwise created automatically with
boundary conditions obtained automatically from the pressure field if
available (with optional name) otherwise inferred from the velocity
field. Phi Laplacian scheme and solver specification are required. See
tutorials for examples.
The old separate incompressible and compressible libraries have been removed.
Most of the commonly used RANS and LES models have been upgraded to the
new framework but there are a few missing which will be added over the
next few days, in particular the realizable k-epsilon model. Some of
the less common incompressible RANS models have been introduced into the
new library instantiated for incompressible flow only. If they prove to
be generally useful they can be templated for compressible and
multiphase application.
The Spalart-Allmaras DDES and IDDES models have been thoroughly
debugged, removing serious errors concerning the use of S rather than
Omega.
The compressible instances of the models have been augmented by a simple
backward-compatible eddyDiffusivity model for thermal transport based on
alphat and alphaEff. This will be replaced with a separate run-time
selectable thermal transport model framework in a few weeks.
For simplicity and ease of maintenance and further development the
turbulent transport and wall modeling is based on nut/nuEff rather than
mut/muEff for compressible models so that all forms of turbulence models
can use the same wall-functions and other BCs.
All turbulence model selection made in the constant/turbulenceProperties
dictionary with RAS and LES as sub-dictionaries rather than in separate
files which added huge complexity for multiphase.
All tutorials have been updated so study the changes and update your own
cases by comparison with similar cases provided.
Sorry for the inconvenience in the break in backward-compatibility but
this update to the turbulence modeling is an essential step in the
future of OpenFOAM to allow more models to be added and maintained for a
wider range of cases and physics. Over the next weeks and months more
turbulence models will be added of single and multiphase flow, more
additional sub-models and further development and testing of existing
models. I hope this brings benefits to all OpenFOAM users.
Henry G. Weller
Currently these vectors are generated at the same time as the wall-distance field
by the same run-time selected algorithm. This will be changed so that the wall-reflection
vectors are only generated and stored if required.
When using models which require the wallDist e.g. kOmegaSST it will
request the method to be used from the wallDist sub-dictionary in
fvSchemes e.g.
wallDist
{
method meshWave;
}
specifies the mesh-wave method as hard-coded in previous OpenFOAM versions.
During restart correct phi would need the dilatation from the previous time-step.
Alternative is to run potentialFoam on 0 fields to initialise phi.
Resolves bug-report http://www.openfoam.org/mantisbt/view.php?id=1299
