Wind Farm Yaw Optimization via Random Search Algorithm

Abstract

One direction in optimizing wind farm production is reducing wake interactions from upstream turbines. This can be done by optimizing turbine layout as well as optimizing turbine yaw and pitch angles. In particular, wake steering by optimizing yaw angles of wind turbines in farms has received significant attention in recent years. One of the challenges in yaw optimization is developing fast optimization algorithms which can find good solutions in real-time. In this work, we developed a random search algorithm to optimize yaw angles. Optimization was performed on a layout of 39 turbines in a 2 km by 2 km domain. Algorithm specific parameters were tuned for highest solution quality and lowest computational cost. Testing showed that this algorithm can find near-optimal (<1% of best known solutions) solutions consistently over multiple runs, and that quality solutions can be found under 200 iterations. Empirical results show that as wind farm density increases, the potential for yaw optimization increases significantly, and that quality solutions are likely to be plentiful and not unique.

Optimizing wind farms layouts for maximum energy production using probabilistic inference: Benchmarking reveals superior computational efficiency and scalability

“…2a), we consider the power generated by each turbine to be proportional to the cube of the effective local wind speeds at the turbine hub height. We adopt this modelling choice to directly compare our results to those published in previous works, which have overwhelmingly calculated the turbine power output in a similar fashion [50,51,52,54]. In the second method (Eq.…”

Providing power reserve for secondary grid frequency regulation of offshore wind farms through yaw control

“…These types of algorithms have the advantages to be accurate and simple to implement. Meanwhile, they become time‐consuming as the size of the WF grows, since finding the optimum becomes harder with the increase of the number of possible yaw angles and the aerodynamic interactions within the WF 19,20,35 Distributed optimization algorithms : Consider the WF as a directed network, where the turbines are the nodes and the aerodynamic interactions are the edges.…”

“…This optimization algorithm is implemented within the fmincon MATLAB function 33 Random search (RS) algorithm (centralized optimization algorithm) : The RS algorithm, proposed by Kuo et al, 35 is based on a random incremental yaw angle selected in a range that tightens as the iteration number grows. The algorithm starts by computing the output power of the WF where all the WTs are facing the wind direction (wind direction perpendicular to the WTs rotor planes) and saves this configuration as the best one.…”

Review of wake management techniques for wind turbines

“…Fortunately, and in contrast to AIC, many of the lessons learned with a column of turbines still apply in larger arrays. Both optimization techniques 3,36,102,103,111,114,119,120,128,137,140,145,148–150,157,158,161,163–165 and trial and error comparisons 108,123,164 of different arrangements of yaw misalignment angles have been tested and have arrived at different conclusions. Several studies have found that total power is optimized when the upstream turbines are misaligned the most and the misalignment angle is decreased as turbines are located farther downstream 3,108,123,146,164 .…”