Probabilistic Characterization of Wind Diurnal Variability for Wind Resource Assessment

Abstract

As wind energy penetration is expected to grow in the future, wind resource assessment becomes important in modern power grid operations. Selecting an appropriate wind farm site can benefit from understanding nonstationary characteristics of wind speeds. In particular, wind speed exhibits a diurnal pattern and the pattern varies, day-by-day and site-by-site. The goal of this article is to develop a new probabilistic modeling approach for quantifying variation in the wind diurnal pattern for assessing wind resource at unmonitored locations. Specifically, we formulate the coefficient of wind model as a latent random process and incorporate both day-to-day and spatial variability into the latent process. The estimation performance of the proposed approach is validated with actual data collected in west Texas. The results demonstrate that our approach can capture both spatially- and daily-varying patterns and quantify the uncertainty successfully.