This major release offers new or enhanced physics and better HPC performances
New LBM collision model (named HRR for Hybrid Recursive Regularized), developed in partnership with M2P2 lab (Aix-Marseille University):
Available for aerodynamics, aeroacoustics and thermal calculations
More stable and cleaner flow calculations
Allowing for larger timesteps and reduced computing time
and much more !
Feature Highlights
Discover the main features of this release
Better backward compatibility
Older versions of LBPre .labs files can be opened in this version and will be automatically updated using our lbrecover tool. When necessary, a backup copy of the original file will be done for safety so that you never loose your setup.
Saving time with templates
You can now define a folder as a “templates folder”, using an environment variable. Any .labs file placed in this folder can be used as a template for LBPre 3.0.
Predefined .labs files
Efficient workflow
Repetitive setup
Read only
Unified multi-selection for boundary conditions
The linked items setup is no longer boundary specific:
Every BC can be associated with a list of geometries
Every BC can be associated with an absorbing region
Better run management
We extended LBPre capabilities, adding features and making it smarter to ease your setup:
Extension of the mathematical functions library
New integrated quantities available for data recording
Specific setup for aeroacoustics simulation available in one tick
Unified user manual for web broswers
The user manual is now part of the ProLB documentation, together with LBsolver manual. This documentation is available in html format and takes advantage of browser functionality :
Less scrolling
Better layout
Easy to update/grow/correct
Web links for fast broswing
Updated 3D viewport
New shortcuts are available for better 3D navigation
New viewport functions were added :
clip planes
geometry statistics
surface normal components
Brand new collision model
The previous in-house Double Relaxation Time (DRT) collision model was replaced by the new Hybrid Recursive Regularized (HRR) collision model [1]. This innovative collision model is better than the DRT on several aspects. It relies on the modification of previously existing recursive regularised BGK models, which consists of hybridising the computation of velocity gradients. The resulting effect is the introduction of a stabilising hyperviscosity term, whose amplitude can be explicitly tuned via a control parameter. This well-controlled hyperviscosity term allows to improve the overall stability of the ProLB simulations. One direct benefit is to be able to increase the time-step value and thus to reduce the simulation time.
Finite volume discretization of the enthalpy equation for energy
Two coupling model between the LBM and FV solvers :
Boussinesq coupling
Full perfect gas model (this model is still in alpha)
Improved Optimal Surface Resolution
Our Optimal Surface Resolution algorithm was improved to be more robust. The OSR allows to re-mesh the output surfaces in order to have nearly the same spatial resolution as the volume mesh resolution. It does not impact physical algorithms themselves: it is only applied on the output mesh.
Updated default schemes
3 schemes are now available in ProLB 3.0 :
D3Q19HRR : default for athermal calculations
D3Q19HRRTherm : default scheme for thermal simulations (Boussinesq)
D3Q19uHRR : alpha-version of the perfect gas scheme
The user can also access previous schemes (D3Q19DRT and D3Q19DRTTherm) but they are now considered as legacy.
New mesh refinement algorithm
We developed with our partners a new mesh transition algorithm to handle the reconstruction of distribution functions, called Direct Coupling (DC) algorithm [1]. In some configurations, the DC algorithm reduces the small spurious errors on the small-scale turbulence vortices due to sudden change in mesh resolution.
We have developed a unique and advanced surface mesh algorithm that resolves intersections between some predefined surfaces of the model, hence its name : surface sewing algorithm. It generates proper wet surfaces of the model, including in very concave regions, and is useful to get accurate integrated quantities.
Unlike well-known wrapping techniques, our sewing algorithm locally calculates the intersection between surfaces and adds some triangles along the intersection line. The initial user topology is then rigorously maintained.
Improved CPU performance
Cache alignment has been optimized in the solver loop
Data transfer (through MPI) has been reduced using data filtering/selection that depends on mesh (case) configuration
Wall law calculation is accelerated by pre-computing and storing the available interpolating (donor) nodes
Which resulted in a 20% CPU time reduction compared to v2.8 HRR beta version
... toward high Mach model
We now focus on the integration of the fully compressible high Mach scheme. This integration is made using a hybrid coupled solver approach :
HRR LBM for mass and momentum
Finite volume discretization (MUSCL-based stencil) of the energy equation
The beta version is planned for this year, stay tuned!