ClusterDesign Consortium recently completed first round of activities

The project ClusterDesign is building on three ideas:

* To give accurate projections of the energy production and the fatigue-life consumption, the designer of a wind farm must take into account the interaction between wake losses and electrical losses.
* To further optimize a wind farm, the designer should put wake losses to advantage.
* To take the maximum out of a wind farm, the operator of a wind farm needs to find a balance between energy production and fatigue-life consumption of a wind farm. 
To make this possible the consortium has been developing or improving specific models since December 2011, and is ready to integrate these into a Wind Cluster Controller and a ToolBox for Integrated Wind Farm Design. These include:

* Advanced and fast wind farm wake models, and models that give the ambient conditions, and
* Wind turbine load models, electrical collection and grid connection models, and wind farm control strategies.

In the integrated approach developed by the consortium:

* The load model interfaces with the control strategies in order to investigate to what extent these strategies can reduce the mechanical loads on the wind turbine and the foundation structure without reducing the power production.
* The model for the wind farm electrical system and the connection to the grid interfaces with the wind farm wake model. Also coupling with the ToolBox is foreseen in order to dynamically model the electrical issues while keeping the wakes (quasi-) stationary.

Wake Effects within and between Wind Farms

The consortium partners ForWind Oldenburg and ECN compared predictions from various wind farm wake models with different degrees of physical complexity and temporal and spatial resolution to observational data from a real-live wind farm and met mast in the North Sea. Two of these models, FLaP-Ainslie and FarmFlow, were further developed into reliable tools for the estimation of wind farm efficiencies. In addition, with the objective to use model output in load calculations, these models were extended with the ability to predict the turbulence intensity and the shear coefficient of the wind profile. The improved and extended models were benchmarked against state-of-the-art commercially available tools. The FLaP-Ainslie model and the FarmFlow model outperform the other models considered in the benchmark in balance between accuracy and run-time. As to accuracy, FarmFlow slightly outperforms FlaP-Ainslie but for the longer turbine separation distances. Of the two models FLaP-Ainslie is the faster.

To complement or even replace observational data, ForWind Oldenburg used the meteorological model WRF in order to determine the inflow conditions for the wake flow simulations. For the met mast in the North Sea a good agreement was found between the predicted and the observed wind speeds and wind directions, particularly when statistical distributions rather than point values are compared. The agreement was less for other quantities like turbulence intensity and shear coefficient. Significant differences were not found between turbine productions calculated on basis of the predicted or the observed inflow conditions. In a subsequent stage of the project ForWind Oldenburg will employ WRF to create a Wind And Stability Atlas of the North Sea.

As to wake effects between wind farms, ECN demonstrated that FarmFlow can model adjacent wind farms in the form of a big wind farm with internal empty spaces. This functionality replaces the need for separate models to calculate the wakes of wind farms.

Wind Turbine Mechanical Loading

The consortium partner Senvion developed a model for offshore wind turbine load prediction which is the first one to accurately calculate the mechanical loads on multi-megawatt wind turbines in large offshore wind farms. The load modeling is built on the capability of the improved and extended wake models as described above to predict the turbulence intensity and the wind shear at the turbine positions. The type and size of the wind turbine, the configuration of the wind farm and also the interaction with neighboring wind farms are fully taken into account.

In order to calibrate and validate the innovative load prediction tool, the predicted wind turbine loads as stored in a database will in a later stage of the project be validated by coordinated mechanical load measurements in a real-live offshore wind farm in the North Sea. To this end a plan was made to equip some wind turbines in that yet to be build wind farm with load sensors on towers and blades. That plan also foresees wind speed measurements with a LiDAR in order to validate the wind speeds predicted with the wake models. Note the power predictions from these models will be validated by using data from the turbine’s SCADA.

Electrical Collection and Grid Connection

The consortium partner Imperial improved steady-state and dynamic models for the evaluation of the electrical system of wind farms and the connection to the onshore grid. The objective of the improvements was:

* To ensure an efficient connection of the wind turbines and wind farms within clusters, and
* To meet the grid code requirements and supply system services;while at the same time taking into account the interaction of wake losses and electrical losses.

The improved electrical models allow wind farm designers to:

* Evaluate the reliability performance and corresponding wind curtailment associated with alternative topologies with different levels of redundancy and flexibility, and in that way enabling them to consider tradeoffs between benefits and costs across alternative AC and DC designs.
* Develop alternative design solutions to meeting grid code requirements.

In addition Imperial addressed the virtual power plant capability of a wind farm cluster. This involved modeling the contribution the wind farm cluster can make to reactive power and voltage control of the offshore network as well as frequency regulation of the system while considering internal wake effects due to mechanical loads of individual wind turbines. This modeling includes both high-voltage AC and DC connections, and thus is relevant for future offshore grid development, based on high-voltage DC voltage source converter technologies.

Wind Cluster Controller

In the context of wind cluster control, the consortium partner RWE Innogy studied control concepts that at the same time can reduce the load on offshore wind turbines, optimize the overall power output while minimizing mechanical loading, and allow the operation of offshore wind farms similar to conventional power plant.

A range of options for wind farm and wind farm cluster control was investigated, and some will be applied real-live in a yet to be build wind farm in the North Sea. To this end test scenarios were defined and a simulation model of the Wind Cluster Controller was developed. Intelligent wind farm control concepts that combine various objectives - reduce mechanical loads, maximize energy yield, maximize power system support - have not yet been field-tested and demonstrated so far in large-scale offshore wind farms and require careful simulations.

In the process simulations of the wind speed, active power, reactive power and mechanical loads of all turbines in the large wind farm in the North Sea were conducted for normal, power maximization and load minimization modes of operation. In a later stage of the project these data will be used to validate the various control options.

ToolBox for Integrated Wind Farm Design

The consortium partner 3E is combining the various design tool elements into a ToolBox for Integrated Wind Farm Design. These elements are the wake models and the sources of the ambient conditions, the wind turbine mechanical load model, the electrical collection and grid connection models, and the control strategies that maximize the energy yield and reduce design loads on individual wind turbines and allow for operation of offshore clusters as virtual offshore power plants. To this end specific interfaces between the various models are developed.

By using the ToolBox, the designer or a wind farm can:

* Define the optimal distribution within the wind array for a specific wind farm respectively wind cluster, given the available wind turbine types.
* Define an optimal power collection system and internal grid layout for a specific wind farm respectively wind farm cluster that minimizes the costs and is in line with the grid code requirements.
* Take into account wind farm control concepts already in the design phase, to ensure that the proposed wind cluster and grid design will allow the operation as a virtual power plant.
* Predict the energy yields and where relevant the mechanical loading of wind turbines by foreseeing the operation of the wind farm clusters as power plants.