垂直軸風力抽水機研製與其複合材料葉片最佳化設計

Abstract

本文目的為設計一組裝配阻力型葉片與升力型葉片之垂直軸風力抽水機，並透過葉片元素動量理論(BEM)進行氣動力學分析與效率預估以設計所有葉片之幾何外型，決定風力機所有部件尺寸後，根據葉片尺寸設計模具並使用玻璃纖維布以手積層法製作成型，完成後將風力葉片於自然風場中進行組裝與測試，紀錄結果與BEM理論分析進行比較，結果顯示兩者相符。再利用有限元素分析軟體ANSYS進行應力分析，設計一單點負載實驗驗證應力分析正確性。為減輕風機葉片重量，設計一最佳化方法針對葉片蒙皮正交玻璃纖維疊層進行最佳化設計。最佳化過程使用最佳化軟體SmartDO，由於最佳化問題為離散的設計變數，進行最佳化設計時使用基因演算法，針對各型葉片分別進行最佳化，再將各葉片最佳化結果代入整體葉片有限元素模型中，檢查安全係數是否符合限制條件，若與限制條件不符，則於整體葉片有限元素模型中重新擷取內力作為各葉片之受力條件再次進行最佳化，當各葉片疊層數無法再減少時宣告收斂；透過此疊代迴圈運算方式在第二個最佳化迴圈中即可求得整體風機最佳之玻璃纖維疊層設計。此方法針對本設計案例可有效減重26%，且能簡化變數過多的複雜結構最佳化問題，並減少最佳化運算時間。

In this paper, an optimization technique is proposed to design the layer distributions of the composite blades of a vertical-axis wind turbine (VAWT). The VAWT is a hybrid wind turbine, which is composed of lift- and drag-type blades. The drag-type blades have same semi-circular cross-section. These two types of blades are sandwich structures. Composite materials have been used to manufacture the blades. The finite element code ANSYS is used to perform the stress analysis of the blades during the optimal design process. The accuracy of the finite element model has been verified using the experimental results obtained in a static test. The end forces of each blade extracted from the finite element analysis of the rotor structure are used in the local finite element model of the blade for designing the optimal layer distribution of the blade (termed as the Local Model). In the local optimal design process, the Genetic Algorism which can deal with discrete design variables is used to search for the optimal layer distribution of each blade. The optimal layer distributions of all the blades obtained at the local design stages of the blades are then used in the finite element analysis of the whole wind turbine rotor structure (termed as Global Model) to check whether the safety factors of the blades satisfy the safety factor constraints. If not, the end forces of each blade determined in the Global Model are again treated as the external forces of the blade in the Local Model to repeat the local optimal design process. The optimal design will be performed iteratively until the design variables (number of layers) converge. The proposed optimal design method can make the rotor structure of the VAWT reduce 26% of its initial weight.