Project outline
This project will tackle wind turbine Amplitude Modulation (AM), one of the critical barriers to onshore renewable energy acceptance. Our physics-informed machine learning model will provide the first reliable AM prediction capability, addressing a significant industry need. This enables proactive mitigation, reducing noise complaints and fostering public trust. The data driven part of the model will leverage the existing datasets, that will enable us to build a robust, predictive tool where current methods fall short. The project is interdisciplinary and combines acoustics, aerodynamics, and machine learning disciplines.
Project Partners
The project will be supervised by a cross-disciplinary team that will include academic expertise in acoustics (Dr Anton Krynkin) and aerodynamics (Dr Melika Gul) as well as the industry expertise offered by TNEI Group, Newcastle.
Further project information
Wind turbine amplitude modulation (AM) is referred to “periodic fluctuations in the level of audible noise from a wind turbine” with “the frequency of the fluctuations being related to the blade passing frequency of the turbine rotor” [1]. The audible effect of AM is usually described with ‘‘whoomph’ sound and can sometimes be observed at residential distances from a turbine (or turbines). This differs from well-defined wind turbine noise, often described as blade “swish” when in close proximity to turbines, which is an inherent property of wind turbines as noise is radiated from the trailing edge of the blades as they rotate. It has been reported that AM is widely linked to noise complaints, and the AM phenomenon is mentioned in the guidance on Assessment and Rating of Noise from Wind Farms [2]. The complexity of the phenomenon was associated with the environmental conditions, wind turbine dependency and presence of reflecting surfaces [2].
In this project it is proposed to build a physics-based machine learning model that will account for aerodynamic sound sources, sound propagation in atmosphere and effect of the reflection from the ground. We will investigate and quantify the critical role of turbine blade geometry and blade position with respect to the incoming flow and nacelle, airflow conditions (windspeed and direction, free stream turbulence and shear gradient), wake-turbine interactions. This model will inform understanding of the phenomenon and will include stochastic algorithms such as Markov Chain Monte Carlo enabling inclusion of uncertainty analysis and recovery of the AM onset and its rating [1] as probability distribution.
The results of this project are expected to contribute to the efforts of building a robust approach for measuring and assessing AM. It has the potential to advance planning conditions essential for the successful installation of onshore wind turbines and enhance the management of limiting and controlling noise associated issues [1].
[2] Noise Working Group, Final Report on Assessment and rating of noise from wind farms, ETSU-R-97, September 1996.
Subject Areas
- Wind turbines
- Amplitude modulation
- Aerodynamic noise
- Machine learning
- Stochastic algorithms
- Uncertainties
Required qualifications/skills
Essential
- A minimum of 2.1 or above Bachelor or Master degree in Engineering, Physics, Mathematics or other related areas.
- Experience in acoustics, vibration and numerical simulation with finite element simulation or computational fluid dynamics methods will be desirable.
Student
Commencing in October 2026