Onshore wind turbine availability: A statistical assessment

Abstract

To understand the impact of operational age on availability, the statistical tools are used in the current study to analyze a dataset, leading to the parametrization of availability distribution across different turbine operational ages. The availability distribution, a bimodal distribution, is modeled with a mixed distribution. The study demonstrates that the best-fitting mixed distribution is a combination of two Beta distributions. Fitting the data into a statistical distribution enables parametric investigation of various causal factors, such as long-term and short-term contributors to unavailability. The research examines the contributions of long-term downtime and short-term downtime to overall unavailability. The mixed distribution is formed by combining two Beta distributions-one skewed toward higher availability values to represent short-term downtime, and another mirrored toward lower values to capture short-term downtime effects. This mirroring is achieved through the transformation y =1 - x, which flips the shape of the Beta distribution around the midpoint (0.5), allowing it to peak near 0 while preserving its statistical properties. Together, these two components form a flexible yet stable structure that captures the distinct influences of both short- and long-duration downtime events across all operational years. To understand the impact of operational age on availability, the article uses statistical tools to analyze a large-scale dataset, resulting in the parametrization of availability distribution across different turbine operational ages. The availability distribution, a bimodal distribution, is modeled with a mixed beta distribution. The study demonstrates that the best-fitting mixed distribution is a combination of two Beta distributions. Fitting the data into a statistical distribution enables parametric investigation of various causal factors, such as longterm and short-term contributors to unavailability. The research examines the contributions of long-term downtime and short-term downtime to overall unavailability. The mixed distribution is formed by combining two Beta distributions-one skewed toward higher availability values to represent short-term downtime, and another mirrored toward lower values to capture long-term downtime effects. This mirroring is achieved through the transformation y = 1 - x, which flips the shape of the Beta distribution around the midpoint (0.5), allowing it to peak near 0 while preserving its statistical properties. Together, these two components form a flexible yet stable structure that captures the distinct influences of both short- and long-duration downtime events across all operational years.