Reducing aerodynamic noise of advanced air mobility by prediction based on machine learning

2 Minute read

High-fidelity simulations of turbulent flows and their application to aerospace systems are integral to gain a physical understanding of flow and noise.

Computationally efficient noise predictions require two key elements: an affordable way to produce accurate mean flow information, and a reliable method to predict the corresponding frequency spectrum.

Machine learning is an enabler for both of those aspects, in that novel turbulence models can be developed with the required predictive accuracy and can also be used to produce better empirical surface pressure spectra or to find the appropriate combinations of modes from resolvent analysis to reproduce the correct spectrum.

The key innovation of this project is strongly coupled with the complementary project ‘Modelling of wall pressure fluctuations with resolvent analysis’ in its use of resolvent analysis; however, while this project is focused on the use of machine learning for turbulence and noise models,
the SJTU-based project is focused on the resolvent-based modelling of high-speed flows.

Project goals

The objectives of this project are:

• Use machine-learning (ML) to model surface pressure spectra from airfoil flows.
• ML-based modifications to RANS modelling of airfoil flows.
• Validation of ML-enabled prediction of mean flow.
• Noise predictions using ML-enabled RANS and resolvent analysis and validation.
• Further research on different airfoil configuration, establishing low noise options with minimal aerodynamic penalty through fast prediction.

Supervision team

The University of Melbourne: Professor Richard Sandberg

Shanghai Jiao Tong University: Professor Weipeng Li

*Click on the researcher's name above to learn more about their publication and grant successes.

Who we are looking for

We are seeking a PhD candidate with the following skills:

• Demonstrated research experience in the field of Aeronautics and Astronautics, and Mechanical Engineering.
• Demonstrated ability to work independently and as part of the team.
• Demonstrated time and project management skills.
• Demonstrated ability to write research reports or other publications to a publishable standard (even if not published to date).

Further details

The project and students will benefit greatly from the complementary expertise and experiences of Professors Sandberg and Li, and will develop broad understanding of fluid dynamics, simulation, ML and modelling techniques.

Professor Richard Sandberg’s group are specialists in high-fidelity simulation of turbulent flows and their application to aerospace systems to gain physical understanding of flow and noise. He is also an expert in the development of lower-fidelity models based on machine-learning techniques that can be employed in an industrial context.

Professor Weipeng Li and his group have been researching the turbulent dynamics and multi-scale structures in high-Reynolds-number turbulent channel flows and hypersonic turbulent boundary layers for years and has developed an in-house high order CFD code to simulate airframe noise, jet noise, cavity noise etc.

Professors Sandberg and Li share the common research interest of developing advanced understanding of high-speed flows and the associated noise generation, yet are using complementary approaches and tools.

The UoM PhD candidate will be enrolled in the PhD program at the School of Mechanical Engineering the University of Melbourne, and the SJTU PhD candidate will be enrolled in the PhD program at the School of Aeronautics and Astronautics at Shanghai Jiao Tong University.

If you are interested in this opportunity, read the application guidelines and eligibility information for both UoM and SJTU before contacting the lead supervisors.