2018.09.62

• Fixed vicous forces calculation in high order solver which were a factor of 2.0 too large
• Upgraded versions of CUDA, Intel MPI and MKL used. CUDA 9.2 is now required for GPU runs
• Added validate_input script to allow checking of control dictionary before run time
• Enforce input parameters that are either True or False to be booleans when validating the dictionary. i.e. 'true' is no longer valid
• Changed colour of timer breakdown output to improve readability

,

2018.07.60

• Optimised GPU execution with 2x performance improvement on V100
• Fixed bug in activate script. zdeactivate didn't restore PYTHONPATH correctly

,

2018.05.58

• Upgraded to Paraview 5.4
• Bug fix to node ordering for DG tetrahedra
• MUSCL reconstruction no longer demotes interface faces between FV and high order cells
• Improved low Mach number preconditioning for turbulent flow for SST and SA models
• DG Performance improvements
• Fixed wavespeed calculation for rusanov scheme for FV solver
• Fixes for P4 hexahedra
• Default parameter values output when solver is run without controldict present are now no longer merged with the supplied control dictionary
• Fix for launching using LSF job scheduler
• Added option to set cell orders in cylindrical regions for DG solver
• Correction to dynamic viscosity output and added output of turbulence kinetic energy
• Added sponge layer option for farfield boundary conditions
• Improved warning message for cell zone definition
• Set velocity to cell velocity on the farfield boundary for ALE cases
• Fix rotating cell zones partition face flux ale velocity

,

2017.2.3

• Corrected DG WALE solver and exposed DG WALE solver to the user
• Correction to BR2 scheme
• Bug fix to velocity gradient vtk output for DG solver

,

2017.2.2

• Print correct release version number, 2017.2.1 release inadvertently reported 2017.2.0
• Improved memory allocation reservations outputting to VTK format
• Always sync gradients from GPU when outputting
• Fixed logic to detect lu-sgs
• Fix to allow P4 hexas
• Efficiency improvements to the BR2 diffusive flux scheme
• Wall functions corrected for HO solver

,

2017.2.1

• Added ability for use to select SST DES variants
• When requested Python dictionary is not available, write default
• Fix issue on Skylake based CPUs
• Improve performance of P0 cases
• Correction to force output extracted in surfaces

,

2017.2.0

• Added ability to override convolution function from controldict
• Python libraries now link against bundled openssl
• Performance improvements
• Added some validation of parameters before running the solver
• Corrected parsing of filtering epsilon and filtering strength values so they are treated as double instead of truncated to int
• Fix to roughness values when running in parallel
• Added BR2 scheme to High Order code
• Ensured halo conditions are correct for strong Boundary Condition formulations
• Added improved interpolation for temperature field input
• Ensured restricted mut is the average of the child cells
• Fixes to restart logic
• Timing output has been improved to report in days. hours, minutes and seconds
• Enabled CUDA optimisations for Pascal architecture GPUS
• Added support for P3 Tris
• Bug fix to prism elements when running in parallel
• Allow user to scale cfl number for viscous timestep
• Added update frequency parameter for turbine zone defintions
• Added a native implementation of the convolution function
• Added specific keywords for specifying the turbine fluid zone length factor and the reference point offset
• Added automatic check to make sure that the reference point is located inside the turbine region
• Added fix for case where only one turbine location is specified as this broke the numpy logic
• Added warning when users specify a thrust coefficient greater than 1 as this invalidates the induction factor calculation later
• Added turbine power curve to turbine fluid zones
• Added keyword to specify the model (induction or simple) to be used for modelling turbines
• Added function 'create_turbine_segments_simple' which uniformly applies thrust and torque to the actuator disc segments, based on the thrust coefficient, tip speed ratio and power curves supplied by the user
• Added thrust summation for simple turbine
• Added use of a search tree to speed up cell searching in turbine regions
• Turbine region function now use line and tube filters rather than cylinder
• Added dissipation to boundary fluxes
• Turned on extrapolation for Farfield Boundary conditions
• Added support for Cray XC systems. Contact us to get a specific version for this platform
• Added relocate option for when the executable on the compute nodes are in a different location to the submit node
• Fixes to WALE model for High Order solver
• Added SA Neg Model for Finite Volume solver
• Added SA Model for High Order solver
• Improved output of high order eddy viscosity
• Robustness fixes for reference plane cell identification
• Added option to output fluidic zone cell sets
• Removed warnings when TRBX file specifies thrust coefficient greater than 1
• Added warning when thrust coefficient > 1 is specified for the 'induction' model. CT is then scaled back to 0.8
• Added turbine power output for Betz limit turbine
• Allow user to set shock sensor variable
• multigrid_cycles variable can be use to turn off Polynomial Multigrid for the High Order solver
• Allow specifying the real time step frequenecy at which the restart H5 file gets written
• Added user option to scale residuals by volume
• Ensured dynamic viscosity and kinematic viscosity output total mu+mu_t
• Allow using Rusanov for turbulence when HLLC is chosen for High Order solver
• Added ABL fluidic zone type
• Added ability to specify the PMG cycle pattern from the python dictionary using the 'multipoly cycle pattern' key
• Ensure that the last solve cycle does a PMG cycle in the High Order Solver
• Added STL file option for turbine offset specification
• Improved parallel load balancing
• Added SAS to LES options
• Changed default ambient turbulence to off so that SA and SST results match
• Improve startup performance for High Order Solver
• Added option to turn on rotation correction for SA
• Fix to memory allocation for SAS
• Added bousinessq buoyancy
• Write turbine power to file on update
• Added libgfortran to tarball for systems where it's not already installed

,

2017.1.1

• Moved cfl logic for coarse grids into the driver layer
• Allow overriding of RungeKutta scheme from users control dictionary
• Fixed bug with incorrect gradient values being calculated at boundaries
• Slip walls now extrapolate at the boundary
• Updated a number of third party dependencies

,

2017.1.0

• run_zcfd script now passes through the tasks per node parameter to the smartlaunch script which is needed when not running through a scheduler
• Made Intel MPI aware of the filesystem it is running on to improve performance on parallel file systems and address issues seen running on GPFS
• Fixed bug with periodic boundary condition
• Removed some code duplication in the python layer
• Fixed bug in create_trbx_zcfd_input that caused turbine regions not to be rotated.
• Bug fix to physical nodal locations in high order code
• Added curvature approximation for high order
• Added porus media support
• Fixed bug detecting actuator disc reference cell when running in parallel
• Ensure the generated turbine region contains the reference point
• Added thrust and tip speed interpolation functions
• Added screen output for actuator disc initialisation
• Performance improvements of file reading during restart
• Added eddy output for boundary faces for high order
• Bug fix to quad nodal location ordering
• Fixed bug in high order filtering
• Added MUSCL reconstruction option for P0
• Changed binding behaviour to bind to sockets if oversubscribing the number of processes on a node
• Added checks for existence of the unix tool numactl
• Added checks for the presence of the mesh and control files
• Bug fix to viscous force calcuation for high order
• Stability improvements to polynomial multigrid
• Upgraded to Python 2.7.13
• Removed requirement for a C++11 compiler for the foamtozCFD converter
• Added missing face area scaling for viscous friction calculation
• Only convect at lower polynomial orders for PMG
• Include pressure and friction forces in reports for High Order
• Fixes to support multiple moving reference frames
• Allow selection of Rusanov scheme for high order
• Added support for multiple cell zones in a single fluid zone entry
• Added translating mesh option
• Fix to ALE HLLC fluxes
• Added moment reporting
• Updated reporting notebook with plotting functionality
• Enable time history vtk output for global timestepping
• Various fixes to global timestepping scheme
• Added vorticity vector output for HO
• Fix support for mixed element CGNS files
• Fix dimensional Q criterion output
• Added helicity output for HO
• Added ability to choose between finite difference and analytic jacobian for LU-SGS FV
• Ensure reported residual is factored by CFL
• Fixes to FV LU-SGS residual reporting
• Added option to use rusanov scheme for FV LU-SGS jacobian
• Added pressure limiter to implicit time marching
• Added generic finite difference boundary jacobian calculator
• Added analytic jacobians for Wall, Wall Function, Farfield, Inflow, Outflow, Canopy and Symmetry boundary conditions
• Added ability to freeze viscous fluxes for FV LU-SGS sweeps
• Added support to restart from LnOmega turbulence model
• Bug fixes to preconditioning
• Added support for P4 Hexes for HO
• Added scaling of turbulent CFL for FV LU-SGS
• Added shock sensor for HO
• Enabled FV LU-SGS with multigrid
• Added output of artificial viscosity for HO
• Added logic to supply a start and end real time cycle for outputting solution in vtk format
• Add momentum source jacobians for FV LU-SGS
• Added capability to identify cells at a plane in actuator disc region
• Added actuator disc plane average reference velocity calculation
• Use diffusive flux on the boundary faces rather than applying penalty parameter
• Only rename the hdf5 file on the master process
• Added new functions for conserved gradient calc
• Fix for implicit solver with farfield subzones
• Added in prolongation relaxation factor for PMG
• Allow user to scale artificial viscosity
• Changed behaviour or restart reporting. Will create a backup of the restarted report file and start a new report file with all of the variable being reported at the top
• Put all *vtu and *vtp in a VISUALISATION sub-directory to help tidy up our file output. *.pvd files remain where they are and point to the files in VISUALISATION
• Put log files in a LOGGING sub-directory
• Added OpenSSL libraries to release distribution

,

2016.3.0

• Initial public release