Next Generation Spray Simulation Model

Spray is a core phenomenon in a range of technologies; understanding distribution of droplets in space and how it evolves with the flow and in time is essential for optimising those technologies. Fast and accurate modelling of the spray makes it possible to optimise respiratory drug delivery, conduct more accurate evaluation of airborne transmission of respiratory infections, better control dry powder production in pharmaceutical applications and reduce fuel consumption and exhaust emissions in combustion engines. In the above-named examples, accurate prediction of droplet size distribution is key to achieving the desired outcomes, where distribution of droplets and their deposits, as well as of droplet vapour, are important for end-product quality. The aim of the project is to develop a novel tool to simulate sprays for engineering applications. The novelty lies in the synthesis and development of mathematical and numerical modelling techniques with a view to be applied to engineering applications.
The Fully Lagrangian Approach (FLA), which is being developed in our group, is built around droplet size distribution, making it much more efficient than conventional approaches for droplet modelling, which track individual droplets or small groups of droplets (parcels). We have demonstrated in benchmark cases, our approach is up to 100 times faster than conventional tracking, while preserving acceptable accuracy in reconstructing local droplet distributions. This result was maintained after the corresponding solver had been developed and implemented as a library to open-source and freely available software OpenFOAM, making it accessible to a wide community of researchers and engineering professionals.
The project will push the model to the next level by developing a robust framework for Large Eddy Simulations (LES) of turbulent flows. LES approach has been selected for its efficiency and wide use in the spray community. In LES, only large-scale fluctuations are resolved, while the effects at smaller scale are modelled. The accuracy of simulations essentially depends on the performance of the sub-grid scale (SGS) model. This is particularly important for sprays, where turbulent flow might be generated by injection of liquid and liquid breakup. We will keep the FLA’s core, preserving the droplet distribution formalism, and develop the sub-grid scale (SGS) model based on deep learning. The newly developed model will be validated against experimental data obtained for a flat fan water injection, available in-house. The data includes steady-state and transient injections. After validation, we will adapt and test the model for pressurised-metered dose inhaler and fuel injection simulations. The first study will be done in consultation with Dr Pannala (Biomaterials and Medical Devices and Drug Delivery Research and Enterprise Group). The second will be conducted in collaboration with the UK National Clean Maritime Research Hub.
The main goal of the project is to develop a product, which will be ready to use. We will continue implementation of the model for coupling with OpenFOAM and develop training materials. It is envisaged that adoption of the new solver, which completes calculations within hours rather than days or weeks will boost development of spray-based technologies and support the software users. The final version of the library will be distributed under the MIT license via the online Brighton Research Data Repository. Thus, the outcomes of the project will be accessible to a wide community of researchers and engineers interested in spray phenomena.