Author(s) | Wenwen Wang, Xiaobing Kong, Gangqiang Li, Xiangjie Liu, Lele Ma, Wenting Liu, Kwang Y. Lee |
Journal | Energy |
Year | 2024 |
DOI / Link | doi.com/10.1016/j.energy.2024.132902 |
Keywords | Wind farm flow control |
As wind farms (WFs) expand in scale, there is a growing need for active power control to track the reference power benchmark issued by the grid dispatch center and also the imperative to reduce the fatigue load on key components of each wind turbine (WT). The presence of the wake effect causes a decrease in power generation for downstream WTs and an increase in the fatigue load. Consequently, the suppression of the wake effect has emerged as a critical control objective for WFs. For tackling the challenge, this article designs a hierarchical WF control framework. The upper-level controller employs a sequential convex programming (SCP) approach to maximize the WF’s captured wind energy function and determine the optimal induction factors for the WTs. The lower-layer controller uses a distributed economic model predictive control (DEMPC) scheme to control the WT locally to achieve reference power tracking while reducing the fatigue load on each WT. Finally, the effectiveness of the designed algorithm was verified by conducting the simulation on a wind farm containing nine WTs.
This paper designs a hierarchical control structure for WFs affected by wake effects by conducting mechanism modelling of wakes and WTs. The upper controller uses the SCP method to optimize the wind energy captured by the WF, thereby obtaining the optimal induction factor of each WT. The lower-layer controller adopts the DEMPC strategy to achieve APC while reducing the fatigue load of each WT. Simulation experiments on a WF composed of nine WTs verified that the upper controller can significantly improve the wind energy capture efficiency under the wake effect. At the same time, the designed DEMPC controller can significantly reduce the fatigue load of each WT. This proves that the control structure can track the grid dispatch instructions and improve the economic benefits of the WF while taking into account the wake effect. However, the main drawback of the structure is the lack of feedback on the upper-layer optimization. Indeed, the lower layer is only based on the measurement of the free stream wind and the wake model, and it does not exploit real-time information on the turbine state and wind inside the wind farm. As a result, the system behaviour is suboptimal in terms of the real achievable power production. Further research on this critical issue will be conducted in the future