Posters

Abstract

The contribution of offshore wind to the energy mix has been increasing significantly in the past years with the decrease of its electricity price. This is, amongst others, due to turbines with higher ratings becoming available in the market. Simultaneously, against a background of economic uncertainty and high supply costs, business cases for the development of new farms are becoming increasingly complex. In this development process, finding optimal turbine positions within a farm is key to maximise energy production considering turbine interactions while minimising costs.    At the same time, the surveying of the offshore bathymetry is a very cost intensive task – where it is not possible to survey the whole site, particularly with increasing farm sizes. In addition, due to the relatively long lead times of offshore wind projects, higher-rated turbines may become available before construction, impacting the business case. That is, fewer turbines will be required to achieve the same wind farm capacity and, therefore, only a subset of the surveyed layout positions may be required. A possible solution is to follow a turbine-agnostic approach by using one turbine type with separation distances adjusted to allow for the placement of larger expected future turbine types. Positions in the layout can then be “down-selected” to account for the higher capacity of larger wind turbines. However, this down-selection method is not trivial due to the complexity of wake interactions. The problem is discrete in nature, as the potential turbine locations are pre-defined through a limited set of coordinates of the original layout. To be able to find the best subset of positions, different strategies can be used.   To address this challenge, various heuristic sub-selection strategies were developed and analyzed with the goal of finding a layout that maximizes energy production). These methods are compared based on the relative internal wake loss and the required computational time. The sensitivity of the preferred method to the used wake model and the relative amount of positions being selected with respect to the total amount of initially available positions is also evaluated. RWE’s internal VV turbine interaction model is used as reference for the comparison. Openly available wind farm reference cases, are used to demonstrate the different strategies.   The results of this study identify the best performing method for layout positions sub-selection. In addition, they show the importance for using accurate wake models to  avoid underestimating wake losses and converging on sub-optimal solutions. The potential use of these methods for more robust optimization set ups is further discussed – as these methods could be integrated in the layout optimization process to find the layout that provides the best trade-off for different turbine sizes. In this case, computational efficiency becomes more important when comparing suitable sub-selection methods.