Fully Funded PhD: Development of innovative and efficient computational fluid dynamics simulator based on Physics-Informed Neural Networks

Despite significant advances in numerical techniques and computing hardware, the high computational cost of large-scale 3D computational fluid dynamics (CFD) modelling remains a major challenge, limiting its industrial applications. To overcome the limitations of conventional CFD modelling, a new approach has emerged that integrates data and mathematical models through neural networks. This has led to the development of a method for solving partial differential equations (PDEs) known as physics-informed neural networks (PINNs). However, these approaches are still in their early stages of development and have yet to demonstrate their effectiveness for complex real engineering problems. This project proposes the development of a new CFD simulator for offshore renewable energy applications based on physics-informed deep learning that offers greater efficiency and robustness. This is a unique and exciting opportunity to work in an excellent research group known for its long record of accomplishment in delivering outstanding research in marine hydrodynamics and computational fluid dynamics and their applications in both conventional and renewable offshore energy. You will have the chance to collaborate closely with a team of experienced supervisors, work on cutting-edge CFD and AI for science, enhance research capabilities, and expand your professional network and collaborations at both national and international levels.

This project aims to develop a new and efficient CFD simulator using physics-informed neural networks (PINNs) as an improved approach for tackling costly and complex offshore renewable energy challenges. The project objectives are as follows:

• To reduce computational time during the training process, a linear solution based on potential flow theory is used as the training dataset for the neural networks. 
• A PINN model is developed for the potential flow model to obtain up to second-order nonlinear solutions for water wave interactions with marine structures.
• The developed PINN is further integrated into the open-source software package OpenFOAM, ultimately demonstrating its effectiveness in simulating offshore renewable energy systems.