The Role of High-fidelity Simulations and Machine Learning in Accelerating Development of Gas Turbines

Abstract

CFD predictions are becoming increasingly important in the design of turbomachinery components and have the potential to accelerate and reduce cost of new developments. This presentation will demonstrate that design of experiments using highly accurate first-principles based simulations are now possible at engine scale conditions and including operational aspects such as wear. It will also show how physical insight relevant to designers can be extracted from such data sets that can be exploited for further efficiency gains.
 
The talk will also introduce a novel machine-learning approach that can use both high-fidelity or sparse experimental data for CFD model development. It will be shown that closure models developed using the gene-expression programming approach, which are interpretable and easily implementable into CFD solvers, outperform traditional models both for the cases they were trained on and for cases not seen before. The integration of language-model-based transfer learning into the CFD model development will also be introduced.

Speaker

Prof. Richard Sandberg is the Chair Professor of Computational Mechanics in the Department of Mechanical Engineering at the University of Melbourne. He also is the current Director of the Melbourne Energy Institute. 

His main interests are in (i) high-fidelity simulation of transitional and turbulent flows to gain physical understanding of flow and noise generation mechanisms, and (ii) pursuing novel machine-learning approaches to help assess and improve low-order models that can be employed in an industrial context. 

He received his PhD in 2004 in Aerospace Engineering at the University of Arizona and prior to joining the University of Melbourne, he was a Professor of Fluid Dynamics and Aeroacoustics in the Aerodynamics and Flight Mechanics research group at the University of Southampton and headed the UK Turbulence Consortium (www.turbulence.ac.uk). He was awarded a VESKI Innovation Fellowships in July 2015, entitled: "Impacting Industry by enabling a step-change in simulation fidelity for flow and noise problems" and held an Australian Research Council Future Fellowship for 2020-2024 to work on integrating high-fidelity simulation and machine-learning based turbulence modelling. He was recently elected Fellow of the Australasian Fluid Mechanics Society and, in 2025 has taken on the role as the Director of the Melbourne Energy Institute.