Historical wind power in Karnataka differs from predictive models: A granular analysis of performance across climatology, technology, and location.

Wind power constitutes one of the largest forms of renewable energy in India. Many operational models have predicted wind energy in India, but most rely on satellite reanalysis data. In contrast, this paper examines historical generation data at an unprecedented granularity to understand its potential and variance. We find actual wind capacity factor (CF) during the Indian Summer Monsoon to be less than half that predicted in NREL's recent landmark production cost modeling study. Further, we find that the yearly moving average of aggregated CF exhibits a +0.640 correlation with the ENSO SOI, a widely-used climatological parameter. At the feeder level, we confirm an expected exponential relationship between average CF and turbine hub height. However, this effect is 261% more pronounced during the monsoon, a finding with troublesome implications for grid planning. We also test the assumption that higher wind turbines perform proportionately better outside of the monsoon and find no evidence to support it. Finally, we present a new method to measure the geographic smoothing effect that reduces inherent variability by 18% when compared to normalized standard deviation. We find limits to the effect of decreasing variability through geographic diversity; the effect mostly disappears at about 125 km. • Historical wind energy capacity factor is half that predicted by recent models. • Wind energy in Karnataka exhibits strong correlation with El Nino cycles. • Higher hub height leads to higher capacity factor, disproportionately during the monsoon. • New method improves method of measuring geographic smoothing effect. • Geographic smoothing in Karnataka disappears at about 125 km. [ABSTRACT FROM AUTHOR]