PhD Research Studentship in Multimodal Machine Learning for CFD Acceleration

This PhD studentship at IDMEC (Instituto de Engenharia Mecânica) focuses on integrating multimodal machine learning techniques into Computational Fluid Dynamics (CFD) to enhance computational efficiency and scalability. The project aims to accelerate traditional CFD algorithms, particularly the SIMPLE method, by leveraging AI models for solving pressure Poisson equations, improving pressure-velocity coupling, and eventually replacing parts of the SIMPLE algorithm. The research will generate high-quality renewable energy datasets and develop AI-driven surrogate models for high-performance computing (HPC). The successful candidate will work under the supervision of Prof. João Miguel da Costa Sousa and contribute to publishing research findings in conferences and journals. The position is based in Lisbon, Portugal, and offers a 3-month initial contract with potential renewal up to 12 months.

• Develop AI-driven surrogate models to accelerate CFD simulations, focusing on solving the pressure Poisson equation using matrix solvers with ML acceleration.
• Improve nonlinear loops of pressure-velocity coupling in CFD by leveraging ML models to enhance classical SIMPLE approximations.
• Attempt a complete replacement of parts of the SIMPLE algorithm using ML methods.
• Generate high-quality datasets for renewable energy applications, specifically for wave, current, and offshore wind energy harvesters.
• Design and develop end-user interfaces for the AI models.
• Document all developed work in written form and prepare research results for submission to conferences and journals.
• Conduct research in high-performance computing environments to optimize AI models for CFD applications.