Date of Award


Document Type

Master Thesis

Degree Name

Master of Engineering (Research)


Civil, Structural and Environmental Engineering

First Advisor

Mr. David Cadogan


This paper is focused on the Measure Correlate and Predict (MCP) methodologies used to estimate long term wind resource at a prospective site. The first common step in all MCP methodologies is a successful wind measurement campaign at the prospective site. This campaign involves the erection of a wind measurement mast (ideally to the hub height of the proposed turbine) and careful selection of booms, mountings and anemometers. The wind measurement campaign is a vital part of the process as the correlation and prediction analysis is only as accurate as the measured wind data. Correlation is a measure of the statistical relationship between two comparable data sets, in order to assess if the relationship is strong enough to be used to estimate the future wind resource at the prospective site. The two most established MCP methods are (1) linear regression and (2) error variance ratio method. Linear regression method operates on the assumption that the resulting correlation coefficient correctly describes the long-term relationship between the two random variables and not only during the correlation period, which is limited. However this neglects the potential variance of the predicted wind speed about the mean, leading to the development of the error variance ratio method, which attempts to ensure that the predicted values at the target site have the same overall mean and variance as the observed values. Alternative methods were examined including the matrix and Mortimer methods. In order to make an estimation of the future wind resource the correlation must be compared to a recognised probability density function, the scale and shape parameters must be identified first, once these parameters are estimated using the least squares method the Weibull and Rayleigh probability density functions are used to characterize the wind speed distribution at the target site. The MCP^ module in the WindFarmer software uses two linear techniques (1) linear least squares regression and (2) principal components analysis. A comparison of the leading wind farm design software packages was conducted. It was decided that the best way of assessing the performance of the MCP methods used in wind farm design software, was to conduct an MCP analysis of identical sites using the WindFramer software and a combination of the linear regression and error variance ratio methods using Microsoft Excel spread sheet software. The metrics used to compare the results would be (1) mean wind speed, (2) wind speed distribution and (3) annual energy production.

Access Level