依極值統計理論預測風場之長期最大風速

Abstract

近年來臺灣離岸風電產業蓬勃發展，但缺乏氣象局觀測到的離岸風速資料，即使有少數風場應政府要求於這幾年開始展開風速觀測，也只有一兩年的資料，無法因應一架風機於二十或二十五年生命週期內結構或機構設計所需考量到的最大風速資訊，只能被動採用國外既有風機設計，令人好奇該等設計是否仍然符合國內風場需求。本研究以苗栗竹南外海離岸風速塔觀測到的兩年風速資料為基礎，考量西門子SWT-6.0-154風機設計壽命為25年，依Gumbel所提出的極值統計理論(Statistical Theory of Extreme Values)預測風機未來25年所會遇到的最大風速。其間，為講求保守並克服數值分析上的困難，本研究分別以每日最大逐時風速與每月最大逐時風速作為極值統計理論所需之初始分布。經分析探討，本研究發現以每日最大逐時風速預測未來最大風速較為精準。而因最大風速及其對應重現週期也是風機設計上另一個重要考量因素，本研究也依極值統計理論分析所得每年最大風速預測該風場內不同最大風速的重現期，並將結果與文獻所記載香港、墨西哥、印度等地不同風場之結果進行比較。

In recent years, Taiwan’s offshore wind power industry has been booming, but there is a lack of information on offshore wind speeds observed by the Central Weather Bureau; and even though a few wind farms have started to observe wind speeds in the past few years at the request of the government, there is only one or two years of data available. It is curious whether the turbines imported and designed primarily based on foreign wind data can still endure domestic wind environments. In this study, the two-year wind speed data collected by an offshore wind tower off Zhunan are used as the basis for predicting the maximum wind speed a turbine established in that area may face in its lifecycle of 25 years based on Gumbel’s statistical theory of extreme values. To be conservative and to overcome difficulties in the numerical analysis, probability distributions of the daily and the monthly maximum wind speeds (measured hourly) are used, respectively, as the initial distribution required in applying the theory. After the analysis, it is found that using the daily maximum wind speed as the initial distribution is more accurate for predicting the maximum wind speed than the other. Since the maximum wind speed corresponding to a prescribed recurrence period is another important factor in the design of a wind turbine, this study also carries out related analysis, also based on the Gumbel’s theory. The result for the studied local wind farm is compared with those reported in literature for wind farms in Hong Kong, Mexico, and India.