Surrogate modeling a computational fluid dynamics-based wind turbine wake simulation using machine learning

Abstract Estimating reliably and rapidly the losses of wind turbine annual energy production due to blade surface damage is essential for optimizing maintenance planning and, in the frequent case of leading edge erosion, assessing the need for protective coatings. These requirements prompted the development of the prototype system presented herein, using machine learning, wind turbine engineering codes and computational fluid dynamics to estimate wind turbine annual energy production losses due to blade leading edge damage. The power curve of a turbine with nominal or damaged blade surfaces is determined respectively with the open-source FAST and AeroDyn codes of the National Renewable Energy Laboratory, both using the blade element momentum theory for turbine aerodynamics. The loss prediction system is designed to map a given three-dimensional geometry of a damaged blade onto a damaged airfoil database, which, in this study, consists of 2700+ airfoil geometries, each analyzed with Navier-Stokes computational fluid dynamics over the working range of angles of attack. To avoid the need for lengthy aerodynamic analyses to assess losses due to damages monitored during turbine operation, the airfoil force data of a damaged turbine required by AeroDyn are rapidly obtained using a machine learning method trained using the pre-existing airfoil database. Presented results focus on the analysis of a utility-scale offshore wind turbine and demonstrate that realistic estimates of the annual energy production loss due to leading edge surface damage can be obtained in just a few seconds using a standard desktop computer, highlighting the viability and the industrial impact of this new technology for wind farm energy losses due to blade erosion.

Download Full-text

INTEGRATION OF MACHINE LEARNING AND COMPUTATIONAL FLUID DYNAMICS TO DEVELOP TURBULENCE MODELS FOR IMPROVED LOW-PRESSURE TURBINE WAKE MIXING PREDICTION

Journal of Turbomachinery ◽

10.1115/1.4051417 ◽

2021 ◽

pp. 1-12

Author(s):

Harshal D Akolekar ◽

Yaomin Zhao ◽

Richard Sandberg ◽

Roberto Pacciani

Keyword(s):

Machine Learning ◽

Fluid Dynamics ◽

Computational Fluid Dynamics ◽

Model Development ◽

Learning Approach ◽

Near Wake ◽

Low Pressure Turbine ◽

Machine Learning Approach ◽

Turbulence Closures ◽

Turbine Wake

Abstract This paper presents the development of accurate turbulence closures for low-pressure turbine (LPT) wake mixing prediction by integrating a machine-learning approach based on gene expression programming (GEP), with Reynolds Averaged Navier-Stokes (RANS) based computational fluid dynamics (CFD). In order to further improve the performance and robustness of GEP-based data-driven closures, the fitness of models is evaluated by running RANS calculations in an integrated way, instead of an algebraic function. Using a canonical turbine wake with inlet conditions prescribed based on high-fidelity data of the T106A cascade, we demonstrate that the ‘CFD-driven’ machine-learning approach produces physically correct non-linear turbulence closures, i.e., predict the right down-stream wake development and maintain an accurate peak wake loss throughout the domain. We then extend our analysis to full turbine blade cases and show that the model development is sensitive to the training region due to the presence of deterministic unsteadiness in the near-wake. Models developed including the near-wake have artificially large diffusion coefficients to over-compensate for the vortex shedding steady RANS cannot capture. In contrast, excluding the near-wake in the model development produces the correct physical model behavior, but predictive accuracy in the near-wake remains unsatisfactory. This can be remedied by using the physically consistent models in unsteady RANS. Overall, the ‘CFD-driven’ models were found to be robust and capture the correct physical wake mixing behavior across different LPT operating conditions and airfoils such as T106C and PakB.

Download Full-text

Rapid Estimate of Wind Turbine Energy Loss Due to Blade Leading Edge Delamination Using Artificial Neural Networks

Journal of Turbomachinery ◽

10.1115/1.4047186 ◽

2020 ◽

Vol 142 (7) ◽

Cited By ~ 2

Author(s):

M. Sergio Campobasso ◽

Anna Cavazzini ◽

Edmondo Minisci

Keyword(s):

Machine Learning ◽

Fluid Dynamics ◽

Computational Fluid Dynamics ◽

Wind Turbine ◽

Energy Production ◽

Leading Edge ◽

Prototype System ◽

Desktop Computer ◽

Edge Delamination ◽

Blade Leading Edge

Abstract Estimating reliably and rapidly the losses of wind turbine annual energy production due to blade surface damage is essential for optimizing maintenance planning and, in the case of leading edge erosion, assessing the need for protective coatings. These requirements prompted the development of the prototype system presented herein, using machine learning, wind turbine engineering codes, and computational fluid dynamics to estimate annual energy production losses due to blade leading edge delamination. The power curve of a turbine with nominal and damaged blade surfaces is determined, respectively, with the open-source FAST and AeroDyn codes of the National Renewable Energy Laboratory, both using the blade element momentum theory for turbine aerodynamics. The loss prediction system is designed to map a given three-dimensional geometry of a damaged blade onto a damaged airfoil database, which, in this study, features 6000+ airfoil geometries, each analyzed with Navier–Stokes computational fluid dynamics over the working range of angles of attack. To avoid lengthy aerodynamic analyses to assess losses due to damages monitored during turbine operation, the airfoil force data of a damaged turbine required by AeroDyn are rapidly obtained using a machine learning method trained using the pre-existing airfoil database. Presented results demonstrate that realistic estimates of the annual energy production loss of a utility-scale offshore turbine due to leading edge delamination are obtained in just a few seconds using a standard desktop computer. This highlights viability and industrial impact of this new technology for managing wind farm energy losses due to blade erosion.

Download Full-text