ANSYS, Inc. Release Notes
CFX |
ANSYS, Inc. Release Notes
This sections highlights differences in the behavior between Release 12.1 and Release 13.0 of ANSYS CFX and CFD-Post.
The following change has been made to CFX-Pre:
The abbreviation for dram (dm) has been deprecated and now stands for decimeter.
Below is a list of numerics improvements and other changes made for the CFX-Solver in Release 13.0. The changes are believed to be generally helpful and should be reverted only in the event of a problem.
Convergence behavior changes (that do not affect the converged solution):
Multiphase flow:
The density-pressure linearization for compressible
multiphase flows has been modified to improve robustness, but still
recovers the same linearization as single phase flows as the volume
fraction approaches 1. Revert by setting the expert parameter compressible linearisation option = 1.
A problem with the homogeneous nucleation model of
the small droplet condensation model is now fixed. As well, non-clipped
area densities are used for the condensation rates, which improves
conservation in regions of significant re-evaporation. These changes
can be reverted by setting the expert parameters nes nucleation
fix = f and ipmt area density clip nes = t.
Miscellaneous:
A minor change has been
made to the transient term on rothalpy for compressible rotating systems.
Revert by setting the expert parameter transient compressible
rotation option = 2.
Discretization changes (that affect the converged solution):
Boundary conditions/GGI interfaces:
Flux boundary conditions for energy and scalars now
evaluate profiles at integration points rather than face centers.
Revert by setting the expert parameter use bip flux = f.
The treatment of gradient extrapolation at boundaries
has changed so that it is now consistent in serial and parallel. Revert
by setting the expert parameter boundary vertex extrapolation
option = 1.
For moving mesh cases, the numerical details of moving
gaps/overlaps at GGI interfaces have been modified for better conservation
properties. Revert by setting the expert parameter ggi moving
mesh option = 2.
The continuous phase volume fractions used in the
Gidaspow and Wen Yu drag correlation are now clipped to 0.001. Revert
by setting the CCL parameter Minimum Volume Fraction for
Correction = 0.0.
For moving mesh cases with total energy, a problem has been fixed that caused temperature oscillations at the interface. The moving mesh contribution to the pressure work term in the total energy equation was accidentally accounted for twice.
Multiphase:
For cavitation with thermal effects modeled, the interfacial
temperature is now assumed to be the liquid temperature rather than
the bulk temperature. Revert by setting the expert parameter cavitation tint liquid = f.
The discretization of some non-drag forces (the Favre
averaged drag force and the solids pressure force) has been made more
robust. The previous implementation can be recovered by setting the
expert parameter vfr gradient force option = 0.
Reverting the calculation of non-drag forces to the
original source term implementation by using the expert parameter
settings of vfr gradient force option = 0 and virtual mass force option = 0 can improve the accuracy
at GGI interfaces, at the expense of overall robustness and overall
accuracy.
Particle Tracking:
The discretization of the particle turbulent dispersion has been corrected. This improves convergence and robustness for particle cases using this option. It is not possible to revert this change.
A problem has been fixed in the Sommerfeld collision model which can lead to slightly different answers.
In the Elsaesser wall interaction related routines, clipping of model correlations and weighting factors were introduced to help improve convergence for cases with evaporating wall particles.
Miscellaneous:
A problem has been fixed for FSI cases with non-overlapping faces at the coupling interface. It is not possible to revert this change.
Boundary advection on GGI interfaces between solids
with solid motion is now switched off if the materials in the solids
are not identical. This can be reverted by setting the CCL parameter Boundary Advection = On in the “BOUNDARY CONDITION
| SOLID MOTION ADVECTION” sections of the interface boundaries.
Updates between coupling iterations and time steps of quantities derived from solution variables have changed for all transient two-way couplings that utilize multiple coupling iterations per step. The changes, introduced due to a defect correction, are most evident in the convergence history rather than in the actual results. It is not possible to revert this change.
Parallel
A bug has been fixed for the calculation of the RMS Courant number used in the adaptive time step control.
Other changes:
Turbulence: SST SAS turbulence model
Since
Release 12.0 the two model versions, 2005 and 2007, have existed.
When creating a new SST SAS setup in CFX-Pre, the correct default
version, 2007, was used. However, when the parameter Model
version was missing in a setup (for example, a CFX-11 SST-SAS
setup or a setup of another turbulence model switched to SST-SAS by
editing the CCL), the previous model version, 2005, was
used in Release 12.0. This has been corrected in Release 13.0 so
that the new model version, 2007, is used.
The following changes have been made to CFX-Solver Manager:
The monitor data generated
by the solver is now written to the results file using 8 significant
figures by default (previously, the default was 5). This will increase
the size of the monitor data significantly, although for most runs
the increase in size of the results file will be small. To revert
the behavior, set the expert parameter monitor digits =
5.
In addition, the precision
of the monitor data exported from CFX-Solver Manager (exported by, for example,
using the right-mouse button on a specific plot window, and then selecting Export Plot Data), and the precision of the data extracted
and used by the utility cfx5mondata, has
been changed to be consistent with the above change for the solver
(that is, 8 significant figures). Previously the data was exported
using a free format with undefined precision.
This section describe procedural changes (actions that have to be done differently in this release to get an outcome available in previous releases) as well as support changes (functionality that is no longer supported) in 13.0 of CFD-Post.
Procedural Changes
When CFX-Solver Manager is opened from ANSYS Workbench, the Custom Executable and Arguments fields are no longer present on the Solver Tab of the CFX-Solver Manager Define Run dialog (although these fields remain available in the Standalone version of CFX-Solver Manager). You can set those properties via the Solution cell Properties view (this capability is a Beta feature).
Listing Files from a Transient Simulation
In Release 12.0, the files present in the working directory were checked first and all the files with same base name were listed in CFD-Post. If the base name of none of the files matched, CFD-Post listed any appropriate files found in 'autosave/solution-points'.
In Release 13.0, CFD-Post lists any appropriate files found in 'autosave/solution-points'. If no appropriate file is found, the working directory is checked and files with same base name are listed. The new sequence gives higher priority to the files from the chosen DAT/CDAT file.
Chart Legends
A Release 12.0 chart that has its legend on the inside of the chart area may display with the legend in a slightly different position in Release in 13.0.
Display of Mean Molecular Weight Values from FLUENT Files
In Release 12.0, CFD-Post showed incorrect units for Mean Molecular Weight in results files exported from FLUENT; for example, a value that should have appeared as 23.1577 kg/kmol would display as 23157.7 kg/kmol. In Release 13.0, the same case will display a Mean Molecular Weight of 23.1577 kg/mol.
Maximum Temperature Limit of Zero Pressure Polynomials
In Release 12.0, the maximum temperature limit of zero pressure polynomials was 1000 [K] for the CFX-Solver, 3000 [K] in the RULES file, and 5000 [K] for CFX-Pre.
In Release 13.0, maximum temperature limits for zero pressure polynomials are now consistent in RULES, CFX-Pre, and the CFX-Solver (= 1000 K). Temperature limits for table generation are set by CFX-Pre: 5000 K for ideal gases, 1000 K for real gases.
Support Changes
There are no support changes in this release.