CFD-based optimization for wind turbine locations in a wind park
Avramenko, Anna (2019-05-10)
Väitöskirja
Avramenko, Anna
10.05.2019
Lappeenranta-Lahti University of Technology LUT
Acta Universitatis Lappeenrantaensis
School of Engineering Science
School of Engineering Science, Laskennallinen tekniikka
Kaikki oikeudet pidätetään.
Julkaisun pysyvä osoite on
https://urn.fi/URN:ISBN:978-952-335-367-1
https://urn.fi/URN:ISBN:978-952-335-367-1
Tiivistelmä
Searching for the optimal positions of wind turbines in a wind farm for their maximal power generation has an important and difficult role in the wind energy industry. Several factors have to be considered while designing a wind farm. These include geographical constraints, the maximum desired installed capacity of the wind farm, wake effects, noise assessment, the total cost and visual impact (Mittal and Taylor (2012)). The fundamental aim is to reduce the total costs associated with the wind farm while maximizing power production. ’Micro-siting’ is the process of optimizing the layout of the wind farm.
The determination of wind turbine positions can be treated as an optimization problem that can be solved by various methods. In this thesis, an optimization tool based on evolutionary algorithms is developed together with a fast computational fluid dynamics (CFD) model to improve solutions for the wind turbine placement problem.
This work presents the development of a wind farm optimization tool with CFD based on the Reynolds-averaged Navier-Stokes (RANS) approach. First, the use of CFD in wind farm modelling is validated with laboratory measurements. This is the starting point for this study. Since accurate three-dimensional (3D) modelling is very time-consuming, a fast depth-averaged modelling method was developed for the wind farm optimization process. The developed fast modelling method was validated with accurate 3D modelling. Even though some of the flow characteristics are lost, the depth-averaged model predicts the velocity and power sufficiently well.
In wind farm optimization, the monitored values and objective function have nonsmooth behaviour when CFD simulations are used. Therefore, optimization of the wind farm has been performed with evolutionary algorithms to avoid difficulties in the gradient calculations of the objective function. In general, evolutionary algorithms are shown to be very convenient for this optimization problem, although they require a lot of CFD simulations during the optimization process. Power maximization was chosen as an objective function for optimization in the presented case examples. The CFD-based optimization of the wind farm introduced in this thesis allows researchers to find a more effective wind turbine layout than before. Even if the optimization algorithm is only implemented for one real hill in this work, it can be extended to different geographical geometries as well.
The determination of wind turbine positions can be treated as an optimization problem that can be solved by various methods. In this thesis, an optimization tool based on evolutionary algorithms is developed together with a fast computational fluid dynamics (CFD) model to improve solutions for the wind turbine placement problem.
This work presents the development of a wind farm optimization tool with CFD based on the Reynolds-averaged Navier-Stokes (RANS) approach. First, the use of CFD in wind farm modelling is validated with laboratory measurements. This is the starting point for this study. Since accurate three-dimensional (3D) modelling is very time-consuming, a fast depth-averaged modelling method was developed for the wind farm optimization process. The developed fast modelling method was validated with accurate 3D modelling. Even though some of the flow characteristics are lost, the depth-averaged model predicts the velocity and power sufficiently well.
In wind farm optimization, the monitored values and objective function have nonsmooth behaviour when CFD simulations are used. Therefore, optimization of the wind farm has been performed with evolutionary algorithms to avoid difficulties in the gradient calculations of the objective function. In general, evolutionary algorithms are shown to be very convenient for this optimization problem, although they require a lot of CFD simulations during the optimization process. Power maximization was chosen as an objective function for optimization in the presented case examples. The CFD-based optimization of the wind farm introduced in this thesis allows researchers to find a more effective wind turbine layout than before. Even if the optimization algorithm is only implemented for one real hill in this work, it can be extended to different geographical geometries as well.
Kokoelmat
- Väitöskirjat [1099]