Presentations

Abstract

We present a new class of extremely fast, low-fidelity wake models, primarily designed for the purpose of evaluating farm-to-farm wake impacts between existing and future wind farm lease areas to support national-scale spatial optimisation. We benchmark the performance of the proposed models in terms of speed and accuracy against current engineering wake models. Selecting an approach for assessing the interactions between the atmosphere and turbines typically requires a compromise between accuracy / uncertainty and computational time / expense depending on the complexities of the site in question and the purpose of the analysis.  The Crown Estate (TCE) is responsible for the management of the seabed around England, Wales and Northern Ireland and designs offshore wind lease areas to support the generation of clean and secure energy, whilst minimising impact to the environment and communities. To facilitate this, TCE have developed a suite of digital tools including techno-economic models and a deployment modelling tool that iteratively optimises lease area shapes.  Considering inter-farm turbine interaction effects in this national-scale iterative optimisation context presents a modelling challenge that even the fastest current engineering wake models cannot address. Tens of thousands of potential lease areas and their effect on each other, as well as existing and pipeline projects, must be appraised in minutes, and to a level of accuracy that allows for meaningful comparison of the aerodynamic impacts and wider techno-economic benefits of the various leasing options. In response to this challenge, a new class of ultra-rapid models, derived from trusted existing methods, is introduced. These models work without requiring assumptions to be made about turbine layout whilst maintaining the ability to capture the effect that a site’s shape has on its own performance and on its neighbours. By removing the need to compute individual turbine-to-turbine interactions, these models estimate the wake impacts between lease areas up to 1,000x faster than traditional engineering wake models (Park2, TurbOPark, etc.), while retaining a sufficient level of accuracy for spatial optimisation and early project development. In this session we will: * Explore this unique challenge in more detail, describing TCE’s approach to seabed optimisation and why currently available models do not meet this need. * Introduce the general theory of how engineering models can be accelerated utilising convolutional approaches.  * Describe two examples of ultra-rapid models and their derivation. * Benchmark model accuracy against TurbOPark   * Demonstrate how these models, used inside TCE’s optimisation framework, can help to deliver better leasing outcomes for the UK.