Learning droplet dynamics via AI-assisted modeling

• Grant number: 24/15904-7
• Support Opportunities: Scholarships in Brazil - Doctorate
• Start date: January 01, 2025
• End date: September 30, 2028
• Field of knowledge: Physical Sciences and Mathematics - Mathematics - Applied Mathematics
• Principal Investigator: Cassio Machiaveli Oishi
• Grantee: Diego Alecsander de Aguiar
• Host Institution: Faculdade de Ciências e Tecnologia (FCT). Universidade Estadual Paulista (UNESP). Campus de Presidente Prudente. Presidente Prudente , SP, Brazil
• Associated research grant: 13/07375-0 - CeMEAI - Center for Mathematical Sciences Applied to Industry., AP.CEPID

Abstract

The combination of computational simulations and artificial intelligence tools has been explored in various areas of fluid mechanics, but multiphase flows still require further investigation. In this work, artificial intelligence techniques will be applied to solve problems observed in various industrial processes involving complex multiphase flows, such as oil transportation and processing in the petrochemical industry and the control of substance mixtures in the food industry. Focusing on droplet dynamics, the objectives will be to predict properties such as energy transfers and to predefine whether there will be coalescence, bouncing, or atomization. To achieve this, the Vision Transformer (ViT) architecture will be explored in different droplet dynamics, such as spreading and collision. One of the limitations of such applications, however, is that training the models requires a large volume of data. In this regard, aiming to reduce the costs associated with expanding data sets, a generative artificial intelligence technique called Denoising Diffusion Probabilistic Model (DDPM) will also be applied for generating artificial data. At the end of the project, a new methodology will be presented to assist in the challenges of modeling droplet flow.