Wake Effects and Energy Yield Optimization under Realistic Wind Conditions at Ngong Hill Wind Farm

Abstract

This study evaluates wind resource characteristics, wake effects, layout optimization, uncertainty, and operational strategies using long-term hourly wind data (2010 - 2019) and a one-year validation dataset from the year 2022. The power loss is analyzed using the Jensen wake model, and simulation for power output done with simulations in PYTHON®. The most frequent wind speeds occur within the 4 - 6 m/s range, with a mean extrapolated hub-height wind speed of 9.38 m/s at 50 m. The shape and scale parameters were k = 3.29 and c = 9.81 m/s, corresponding to a Betz-adjusted extractable power of 509 W/m², classifying the site as Wind Power Class V at 50 m. Wake modelling showed that the existing layout experiences wake losses of 28.2%, reducing the no-wake Annual Energy Produced (AEP) from 61.86 GWh to 44.41 GWh. Genetic Algorithm-based layout optimization aligned turbine spacing with the dominant wind direction (≥4D along-wind and ≥3D cross-wind), reducing wake losses to 23% and yielding an approximate 5% AEP improvement. Sensitivity analysis demonstrated that ±0.02 variations in the wind shear exponent result in 6–8% changes in AEP, while ±10% perturbations in the Weibull scale parameter produce energy yield variations exceeding ±15%, magnitudes comparable to wake-loss reduction gains. Further, approximately 27% of annual hours occur at wind speeds below 6 m/s, primarily during May-August, as well as during nocturnal and early-morning hours (2100hours - 0300 hours). Maintenance scheduled within these low-wind windows incurs only about 15% of the energy loss associated with maintenance conducted at mean wind conditions, corresponding to an estimated 85% reduction in maintenance-related energy losses. The results demonstrate that maximizing energy yield at Ngong Hills requires a combined strategy integrating aerodynamic layout optimization, uncertainty-aware modelling, and wind-aware maintenance scheduling.