Wind Farm Wake Prediction Using SCADA for Turbine Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional wind farm optimization methods fail to account for real-time variations in ambient conditions and wake effects between turbines, leading to sub-optimal performance due to inaccurate modeling and high computational costs of hi-fidelity models.
Innovation Solution
A data-driven approach using supervisory control and data acquisition (SCADA) data to model wake effects in real-time, leveraging regression models and machine learning to determine optimal control settings for wind turbines based on prevailing conditions and interactions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Power
If conventional engineering wake models are used to optimize wind farm performance, then farm-level power output is improved, but the models become inaccurate when ambient conditions change frequently
Solution Approach 1:
The patent implements dynamic adaptation of wake models by continuously updating model parameters based on real-time ambient condition measurements (wind speed, direction, temperature, humidity). This allows the wake model to transition from static conventional engineering models to dynamic models that adapt to changing atmospheric boundary layer stability and turbulence intensity, resolving the contradiction between maintaining model accuracy and responding to varying conditions.
Solution Approach 2:
The system incorporates feedback mechanisms where actual wake effects measured at downstream turbines are compared with model predictions, and the discrepancy is used to refine and update the wake model parameters. This closed-loop feedback ensures the model remains accurate under varying ambient conditions while continuing to optimize farm-level power output.
2Measurement precision
If hi-fidelity wake models based on computational fluid dynamics are used, then modeling accuracy is improved, but computational cost and system complexity increase significantly
Solution Approach 1:
The patent replaces expensive, complex hi-fidelity computational fluid dynamics models with simpler, computationally efficient wake models that can be rapidly executed. These simplified models use pre-determined wake characteristics and real-time ambient condition data to achieve sufficient accuracy without requiring extensive computational resources or additional complex instrumentation, effectively substituting a 'cheap' solution for the 'expensive' hi-fidelity approach.
Solution Approach 2:
The system changes the parameters used in wake modeling from detailed computational fluid dynamics parameters to simplified parameters based on ambient conditions (wind speed, direction, temperature, humidity) and pre-determined wake characteristics. This parameter transformation maintains adequate modeling accuracy while dramatically reducing computational complexity and making the system practical for real-time operation.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Embodiments of methods 200 and systems for optimizing operation of a wind farm 100 are presented. The method 200 includes receiving 210 new values corresponding to at least some wake parameters for wind turbines 102 in the wind farm 100. The method 200 further includes identifying 212 new sets of interacting wind turbines 102 from the wind turbines 102 based on the new values. Additionally, the method 200 includes developing 214 a farm-level predictive wake model for the new sets of interacting wind turbines 102 based on the new values and historical wake models determined using historical values of the wake parameters corresponding to reference sets of interacting wind turbines 102 in the wind farm 100. Furthermore, the method 200 includes adjusting 216 one or more control settings for at least the new sets of interacting wind turbines 102 based on the farm-level predictive wake model.