Abstract
In this paper, we investigate the feasibility of using DNS data and machine learning algorithms to assist RANS turbulence model development. High-fidelity DNS data are generated with the incompressible Navier–Stokes solver implemented in the spectral/hp element software framework Nektar++. Two test cases are considered: a turbulent channel flow and a stationary serpentine passage, representative of internal turbo-machinery cooling flow. The Python framework TensorFlow is chosen to train neural networks in order to address the known limitations of the Boussinesq approximation and a clustering based on flow features is run upfront to enable training on selected areas. The resulting models are implemented in the Rolls-Royce solver HYDRA and a posteriori predictions of velocity field and wall shear stress are compared to baseline RANS. The paper presents the fundamental elements of procedure applied, including a brief description of the tools and methods and improvements achieved.
Highlights
Turbulence modelling is the element of Computational Fluid Dynamics (CFD) that captures the complexity of the physics of turbulent flows using a mathematical model, with the aim of accurately describing the effect of the chaotic behaviour of turbulence on the mean flow
Together with the development of Machine Learning (ML) algorithms, able to make predictions based on large amounts of data, a new opportunity in turbulence modelling is offered: extract data from high fidelity CFD and use ML to relate turbulence behaviour to geometric and mean flow features
As τ12, k and s12 are available from the DNS data, it is possible to infer the distribution of c1 that would make (4) hold to any degree of accuracy. In this simple case, applying the ML process we described is equivalent to look for a non-linear best fit to a specified smoothness of the point-wise mapping c1i ( f 1i ) for all points in the training set, with f 1 being the flow feature chosen to train the ANN
Summary
Turbulence modelling is the element of Computational Fluid Dynamics (CFD) that captures the complexity of the physics of turbulent flows using a mathematical model, with the aim of accurately describing the effect of the chaotic behaviour of turbulence on the mean flow. In recent years an unprecedented growth in computing power has allowed the proliferation of high fidelity CFD calculations. Together with the development of Machine Learning (ML) algorithms, able to make predictions based on large amounts of data, a new opportunity in turbulence modelling is offered: extract data from high fidelity CFD and use ML to relate turbulence behaviour to geometric and mean flow features. The final aim is to develop a process to derive adaptive turbulence models, having as a primary objective not the generality of each of them, but rather high prediction capability for a set of similar cases in a specific application. Weatheritt and Sandberg [3] developed
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: International Journal of Turbomachinery, Propulsion and Power
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
