What's New

View new features for nanoFluidX 2022.2.

Altair nanoFluidX 2022.2 Release Notes

The 2022.2 SimLab and nanoFluidX workflows are now a single-window environment. This major update improves accessibility.
Most fluids in reality are non-Newtonian. A new nanoFluidX feature enables exploration of Cross model fluids.

SimLab compatible output: nanoFluidX Companion, nFX[c], now outputs a format which can be read and used in SimLab 2022.2 for post-processing. This feature, alongside SimLab’s ability to run nanoFluidX inside the window, creates a single-window environment.; Refer to the SimLab Release Notes for further new features related to nanoFluidX.
Non-Newtonian fluid physics: This feature allows exploration of Cross model fluids.

Optimization of SPH Kernel performance: The implementation of Quintic Spline SPH Kernel is slightly changed to reduce instruction count. The improvement in performance is up to five percent with no repercussions. In general, there should be no noticeable changes.
Corrections to acceleration terms on solids: nanoFluidX now considers acceleration changes in the treatment of boundary conditions when using variable body force. There should be no noticeable changes.
Accessing Tutorial Model Files: You can click a link within a Tutorial to download the required model file(s).
Important: This download option is only available if you (or any user) are connected to the internet. Users attempting to download a model file will be prompted with further directions on how to access the model file online. A zipped package of the model files can be downloaded from the Altair One Marketplace and extracted to a local machine or directory on your company’s local server.

The following issues will be addressed in a future release, as the software performance is continuously improved.

When using the adhesion coefficient for the SINGLE_PHASE surface tension model, the surface tension coefficient must be calibrated against experimental results or visually estimated. If you use the Tartakovsky model, the physical surface tension value is sufficient.

Wallclock time-based recon output occasionally hangs.
Aeration-viscosity models were deprecated in the 2022 release because the models required a set of parameters which were difficult to obtain and/or required calibration.

Single GPU nanoFluidX is available for Windows for the first time. The full feature set of nanoFluidX can be used on a Windows system with a supported CUDA capable GPU.

Windows: Single GPU nanoFluidX is available for Windows for the first time. The full feature set of nanoFluidX can be used on a Windows system with a supported CUDA capable GPU.
Contact Angle: The Tartakovsky surface tension model allows the contact angle for wall phases to be specified. Enabling the contact angle parameter allows simulation of hydrophilic (<90 degrees) and hydrophobic (>90degrees) wetting behavior in single phase flows.

Library Updates

nanoFluidX libraries have been updated and ship with OpenMPI 4.1.2 and NVIDIA CUDA 11.6.2.
Ensight files created by nanoFluidXc can be created in an unstructured format for the fluid grid.
Bodyforce files can have the latch option enforced so that force is extrapolated if simulation time is outside of the bodyforce file range.

The following known issues will be addressed in a future release as we continuously improve performance of the software:

Inlet regions must always use the temperature set for the phase globally. It is not currently possible to set incoming fluid as a different temperature if that fluid phase is already in the domain.

UTM Files

UTM motion files were discarded if a restart was performed. This meant that UTM files of a restart run executed in the same directory as the clean run would only contain data from the current recon run. This resulted in static intervals in the MOVINGWALL renders in ParaView.

Heat Transfer Coefficient (HTC) calculation and export: One of the main motivations for drivetrain oiling simulations is to understand the oil supply and cooling capacity available for different components. Previously, nanoFluidX relied on coupling with AcuSolve to provide this information. This release adds calculation of time-averaged heat transfer coefficients (HTC) directly on the geometry. Thermal analysis is now more convenient and the turn-around time of the process is faster.

Direct export of heat transfer coefficient (HTC): nanoFluidX does have a built in transient thermal solver, but a direct conjugate heat transfer simulation in nanoFluidX is prohibitively expensive. The different timescales of thermal and fluid dynamics mean it is essentially impossible to reach a steady state condition.; Previously, to avoid this time-scale issue nanoFluidX exported a time-averaged flow field to AcuSolve. A conjugate heat transfer simulation was then executed to determine the quasi-steady-state temperatures.; This approach does have limitations, so an option to export time-averaged heat transfer coefficients (HTC) directly on the geometry has been provided. Such an approach makes the thermal analysis more convenient and speeds up the overall turn-around time of the process.

Add ParaView 5.10 State file support: ParaView State files written from nanoFluidX and nFX[c] are now fully compatible with ParaView 5.10.
Run to run reproducibility: Some discrepancies were observed when running a simulation multiple times. Investigation revealed the root cause to be certain GPU programming libraries used by nanoFluidX. This meant the process of addition was not repeated in the same order from run-to-run, leading to round-off errors being accumulated. The latest version of nanoFluidX can enforce order operation and allow for run to run reproducibility, provided the same hardware and software platform.
Use a wrapper file for solver input: nanoFluidX can now accept a .zip file containing the input files (.cfg, .prtl, .txt) as the input in place of separate files.
Removing SPHERE and CUBOID type impose regions: nanoFluidX 2022 deprecates the SPHERE and CUBOID type impose regions, which are superseded by PARALLELEPIPED and CYLINDER shape setups. Considerable background restructuring of these features has taken place, however the UI has been maintained.
Removing legacy inlets and outlets: For a long time nanoFluidX required inlets and outlets to be specified on the domain boundary. This was changed several versions ago and these features are officially being deprecated and replaced by the inlet and outlet regions.

Adhesion coefficient: Adhesion coefficient for the SINGLE_PHASE surface tension model needs to be calibrated against experimental results or visually estimated. Using the Tartakovsky single phase surface tension model is supposed to get around this by using a physical value of the surface tension.

Fix VTK format output for large datasets: Binary output is appended to the file and the position of each property is managed via an offset variable (in bytes). This is handled internally by a 32bit signed integer. This can break when the amount of data is large (>2GB per file), which can happen with a lot of particles per GPU and/or many output variables. The offset counter was updated to 64bit integer to fix this issue. The VTI writer (for interpolated output) and VTP writer (for surface extractor) have also been updated.