We are very excited to announce the release of PreonLab 4.0. This major release expands, improves, and accelerates the core capabilities of PreonLab, highlighted by the following features:
This is just a selection of new features. PreonLab 4.0 is a truly massive update that also introduces improved file management, enhanced rendering capabilities, air pressure force fields, faster MPI, an improved car suspension model, arbitrarily shaped volume sensors and much more. We will publish videos showcasing some of the new features in the coming weeks.
Make sure to follow us on LinkedIn so that you don’t miss them!
PreonLab started as a highly specialized tool for certain fluid simulations, but it has evolved into a more general solution over time, capable of simulating rigid body dynamics, highly viscous fluids and thermodynamics. PreonLab 4.0 continues this evolution with the introduction of a particle-based snow solver, which allows you to simulate snowfall and accumulation of snow. Just like the Preon fluid solver, it can handle any imported geometries without meshing and offers competitive performance. We also extended Preon Renderer so the simulated snow not only behaves like snow but looks like it as well. The results are quite spectacular. Check out the video to see an example!
The new snow solver allows to, for example, simulate car snowing which helps to optimize sensor locations for autonomous vehicles.
PreonLab is currently used across various industries for many water and oil management applications, including the prediction of flow paths, lubrication, and force loads and losses. Until now, this often involved careful tailoring of solver parameters which varied depending on the application.
Our ultimate goal is to eliminate this overhead by providing a solver that performs consistently and reliably with the same parameters regardless of simulation details. (For instance when water is simulated, the same parameters that produce an oscillating droplet in space should also be suitable for simulating a vast ocean filled with turbulent waves.) With the updated solver in PreonLab 4.0, we are taking a big step in this direction. The improved solver has been carefully benchmarked against classical CFD examples as well as various industrial applications. We have already published some of these benchmarks on LinkedIn and will continue to do so in the following weeks.
The video below shows the improved solver accurately capturing the capillary rise effect.
The improved cohesion model in PreonLab 4.0 is crucial to capture the capillary rise effect.
Explicit simulation of thermal diffusion imposed a restriction on the timestep, which could drastically slow down the entire simulation. The new implicit thermodynamics solver fixes this problem, greatly improving performance in many scenarios. The new solver brings the same level of robustness and speed you already know from our pressure and viscosity solvers to thermodynamics.
PreonLab is very efficient at simulating many particles on multiple cores. However, computational efficiency alone is sometimes just not enough to achieve a satisfactory overall turnaround time. A simulation with half a billion particles might be feasible in terms of computation time when using MPI, but post-processing, analyzing, and archiving the resulting dataset can quickly become extremely time-consuming. Some cases might simply require so many particles that, even with MPI, the simulation would take months to complete.
Conveniently, most of these scenarios involve large bodies of fluid that actually only require a high level of detail in certain regions. Consider a wading simulation: Do you really need an accuracy of 5mm 20 meters ahead of the car? With PreonLab 4.0, you can now set two levels of detail in the same simulation by using simulation boundary domains to specify which regions of the simulation require a high resolution. For the exemplary wading case that is depicted below (simulated for 20 physical seconds), this brings down computation time from 11.5 to 6 hours and reduces the size of the generated data from 133 to 57 GB.
We are very excited about how local fluid refinement potentially accelerates almost every application and even open up new ones. However, please note that this feature is considered experimental in 4.0 and just represents the first step. Most notably, some post-processing features are still missing that will be added in 4.1.
Local refinement allows you to use small particles near the region of interest (the car) and coarser particles elsewhere.