Machine learning aided model for large-scale gas-particle flow interactions

Project description

Gas-particle flow interactions are significant in many applications including air filters, fluidized bed, pharmaceutical sprays, and blast furnaces. A numerical model with reliable performance, efficiency, and robustness is essential for studying the fundamentals and understanding the physical mechanisms. Coupling computational fluid dynamics (CFD) with discrete element method (DEM) has been proven to be accurate to simulate such circumstances. This Eulerian-Lagrangian method models fluid flow and particles’ motion by solving the local-averaged Navier-Stokes equations and Newton’s laws of motions, respectively, and exchanges the gas-particle interaction forces to achieve coupling. However, it can be extremely computationally expensive when dealing with large-scale systems. To address this, this project aims to develop a numerical method for modelling large-scale gas-particle flow systems by combining the conventional CFD-DEM with emerging machine learning (ML) techniques. An in-house solver for micro-scale simulations has been in development, the results of which will be adopted to train the ML model. Then the trained ML model can be used to draw conclusions on macro-scale gas-particle interactions. This new methodology has the potential to improve the classical gas-particle models, especially for large-scale systems, with greater accuracy and feasible computational cost.

Outcomes

• This project will provide a new methodological framework for solving gas-particle flows problems with accurate predictions but short time consumptions (efficient and effective).
• The framework to be proposed consists of model development, algorithm performance evaluation, results visualization, and decision-making analysis capabilities for any applications.
• This project can improve both individual and collaborative research skills to those who are involved in.
• The project is also expected to produce peer-review journal publications.

Essential capabilities

Critical thinking, computer programming, literature review skills, data analysis, presentation and communication skills.

Desireable capabilities

Industry exposure, knowledge of oil and gas or manufacturing industry, C/C++ programming, Python programming

Expected qualifications (Course/Degrees etc.)

Master’s degree in mechanical engineering/Chemical. Outstanding direct UG performance is also welcome.