A Turbulence Intensity Similarity Distribution for Evaluating the Performance of a Small Wind Turbine in Low Wind Speed Regimes

Open Access
Author:
Ward, Nicholas John
Graduate Program:
Energy and Mineral Engineering
Degree:
Master of Science
Document Type:
Master Thesis
Date of Defense:
November 13, 2015
Committee Members:
  • Susan W Stewart, Thesis Advisor
Keywords:
  • Energy engineering
  • small wind turbine
  • turbulence intensity
  • power performance
  • distributions
Abstract:
Field tests from June 2013 to September 2014 on a Skystream wind turbine in a turbulent environment (at the Sustainability Experience Center on the Pennsylvania State University’s main campus) were used to study the impacts of turbulence intensity on power performance. Representations of the power curves for small wind turbines do not currently account for impact of turbulence. The study examined the influence of turbulence intensity on the deviations in power output, commonly experienced by small wind turbines, as compared to published power curves such that more accurate predictions in performance can be made for future small wind projects. The first aspect of this study analyzed the turbine’s performance at its location with respect to manufacturer specifications for the Skystream 3.7. The wind speeds were measured via anemometers on the tower at two different heights in compliance with International Electrotechnical Commission (IEC) standards, and were extrapolated to hub height via the Power Law. The data at each 3 s time step were collected via meteorological packages for temperature, air density, pressure, wind speed and direction as well as power output from the turbine. This data was then averaged and binned for 1 min intervals. One-minute averaged data was used to explore the distribution of turbulence intensities experienced in wind speed bins of 5.00±0.25, 7.50±0.25, and 10.0±0.25 m/s. The resulting turbulence intensity distributions have a similar distribution across these wind speeds and thus it was compared to four common statistical distributions: Weibull (α = 2, β = 0.11), Gamma (k = 12, θ = 0.014), Normal (μ = 0.1889, σ = 0.067), and Exponential (k = 5.2). A goodness of fit study was conducted, finding that a Gamma and a Normal distribution were each the statistically well correlated (with R^2 values ranging from 0.983 to 0.991 and 0.979 to 0.989 respectively). However, based on a visual inspection of the two distributions compared to the tested wind speeds, the Gamma distribution was determined to be the best fit. This finding constitutes a more elegant approach to empirically adjusting power curves for turbulence intensity than has been suggested by previous studies, as well as using a 1 min averaging scale to obtain a better sample of the wind speed and turbulence intensity data experienced at the site and a better representation of the power performance on a more relative scale to the turbulence itself.