Posters

Abstract

High fatigue loads on wind turbines, caused by ambient climate conditions and wake-induced turbulence, can significantly shorten their lifespan. Limited space available often leads to tightly packed wind farm layouts, resulting in turbine lifetimes below design conditions. It is often favorable to derate turbines when design lifetimes are exceeded and turbine locations are fixed. Curtailment optimization helps determine optimal derating strategies on a wind farm level, per wind speed and wind direction bin, while minimizing losses on the objective, e.g. annual energy production (AEP). Due to non-linear interactions of wake effects that affect both AEP and fatigue lifetime, the optimization problem rapidly demands excessive computational resources as the number of wind turbines increases. We propose a heuristic approach employing a surrogate model for fatigue load assessment to reduce computational demand. We form sub-problems based on wake added turbulence contribution, where the damage equivalent loads (DEL) to AEP ratio is used to identify suitable bins for curtailment. Thereafter, a set of bins with adjustable size is either curtailed or shut down. Moreover, a non-contiguity penalty is introduced to allow steering towards curtailments with less variation between neighboring bins. We performed the proposed curtailment optimization on three projects with different complexities. Respective minimum lifetimes of 15, 14, and 4 years were extended to the lifetime of 20 years with AEP losses of 0.25%, 12%, and 6%. We identified that a high non-contiguity penalty can increase AEP loss up to 7%. The outlined heuristic approach can find solutions that fulfill lifetime constraints with minimum AEP loss and allows to control contiguity of curtailments strategies and the computation time. The controllable computation time allows to embed the presented method in problems where the locations of turbines are also decision variables of the problem.