In February 2018, we announced a new multiphase capability in ANSYS Fluent that speeds up accurate spray simulations using a unique hybrid volume of fluid (VOF) to discrete phase model (DPM) method. You can learn more in a blog written by my colleague Muhammad Sami. We ran a webinar on this new capability that garnered great interest. In the past, spray simulations required substantial computing resources to complete with accuracy. This VOF-to-DPM capability requires much less computing time, making it practical to solve a much wider range of problems, including those involving wider length scales. For these problems the focus is to transition large-scale spherical structures in a VOF field to droplets in a Lagrangian field. This helps in optimizing computational resources without compromising the accuracy.
While this approach is great for modeling sprays, there are other problems that start with particles and end with continuous fluids. For example, a spray of droplets impinging on a wall and forming a film or rain falling on a pond. Now we have developed a Fluent user-defined function (UDF) that converts Lagrangian droplets back to a liquid phase in the Eulerian VOF model. Intelligence has been added to the DPM-to-VOF transition algorithm that ensures the transition happens close to the wall or liquid-gas interface, thereby dynamically refining the area occupied by the chosen parcel before conversion and converting only the parcels that are going to hit the wall or liquid-gas interface.
New model speeds simulations, as shown for spray impinging on a wall. The DPM model tracks the droplets until they merge, when they are then tracked by the VOF model.
The DPM-to-VOF transition tool is tested for simple scenarios, such as free-falling droplets, a droplet train moving toward a dry wall, a spray hitting an impingement wall, etc. The tool has been validated for spray impingement on a dry wall and compared with experimental measurements (Arienti et al., 2009) in Figure 1. Compared to the experimental measurements, overall film height predicted falls within the range of experimental uncertainty.
The only drawback of such a multiscale modeling approach could be the overhead on the computational requirement. To perform such a high-fidelity simulation, we still need to work with large element counts and smaller timestep sizes. However, the requirement is not as stringent as the pure VOF simulation, in which the mesh count may grow to billions of computational fluid dynamics (CFD) cells, and overall timestep size may be of the order of nanoseconds. To keep this type of hybrid approach within the limits of an industrial design cycle, ANSYS has also come up with a novel technique to dynamically adapt the mesh of all types — including polyhedral cells — with its polyhedral unstructured mesh adaptation (PUMA) technique. We can start with a coarse background mesh and locally refine or coarsen the mesh based on modeling needs.
PUMA, VOF-to-DPM and DPM-to-VOF can be used together to perform high-fidelity simulations at an affordable computational cost, which will help to eliminate uncertainties that are involved with conventional modeling approaches. Whether a hybrid approach becomes a first-choice tool for prediction of spray and film characteristics, only time will tell.
Learn more about the new multiphase capability in ANSYS Fluent.