本文針對散熱風扇的扇葉氣動外型進行參數化模型建模與最佳化之系統化設計研究。論文分為三大部分。首先，以美國國家航空諮詢委員會(NACA) 四位數翼剖面為基礎，描述扇葉的翼剖面、攻角、扭曲角和前掠角，並自動形成三維座標的參數點，建構參數化的風扇扇葉幾何模型。基於競品技術規格，探討不同扇葉幾何設計在競品操作壓力阻抗下，藉流場計算取得相對之穩態質量流率。接續，藉由不同風扇扇葉之流場模擬數據，結合類神經網路建立初步參數化性能與參數關係模型。最後，結合基因演算法估測最大流量之相對幾何形狀，並藉由估測最佳設計參數與分布擴增參數的流場數據來豐富建構之性能模型，並提升參數化性能模型的精確性。重複遞迴步驟直到獲得風扇性能的精確模型時，也一併獲得風扇扇葉幾何形狀最佳解。

研究結果顯示，在相同的容許體積與操作阻抗條件下，本設計方法獲得之幾何設計風扇其空氣體積流率比競品大幅增加70%，足以顯示設計方法帶來的優異效益。本文提出之同步建模與最佳化設計流程結合參數化模型、模擬數據及利用類神經網路與最佳化方法進行交互迭代的系統化設計法，極具產品開發成效，有助於各項產品的開發與最佳化設計。

This paper focuses on the parametric modelling and the optimal design of heat exchanging fan blade’s aerodynamic shape. The study can be divided into three parts. Firstly, describing fan blade’s airfoil, angle of attack, angle of twist and swept forward angle based on the NACA four digits airfoil, forming three-dimensional Cartesian coordinate points, and constructing the geometric model of parametric fan blade. Secondly, by referring to the specification of the competitor fan, finding the steady state flow rate of different fan blade aerodynamic shape design, with respect to the competitor fan’s operating point. Thirdly, by combining CFD (Computational Fluid Dynamics) results of different fan blade design and neural network, creating the initial parameters-performance model. Afterwards, searching the aerodynamic shape of the fan blade corresponding to the maximum flow rate by genetic algorithm, and further enriching the data set by predicting the ideal design parameters and their distribution, so as to increase the preciseness of the parametric model, until acquiring the precise model of fan performance and the optimal solution of fan blade’s geometric shape parameters.

The result indicates that, in the same fan case dimension and operating point, the volumetric flow rate of the fan blade aerodynamic shape gained by this method is 70% more than the competitor fan, which is significant enough to demonstrate the superiority of this design method. The simultaneous modelling and optimal deign procedure proposed by this paper combines parametric model, numerical simulation big data and the systematic approach of iteration between neural network and optimal method. Proved to be very effective and powerful, the design procedure could be valuable to various kinds of product development and optimal design.

Gao, G., You Q., Kou, Z., Zhang, X., Gao, X., (2022) ” Simulation of the Influence of Wing Angle Blades on the Performance of Counter-Rotating Axial Fan”, MDPI
Nouri, H., Ravelet, F., Bakir, F., Sarraf, C., Rey, R.,(2012)”Design and Experimental Validation of a Ducted Counter-Rotating Axial-Flow Fans System”, Journal of Fluids Engineering
Wang,J., Ravelet, F., Bakir, F.,(2023) ” Experimental comparison between a counter-rotating axial-flow fan and a conventional rotor-stator stage”, Science Arts & Métiers
Wu, Y., Huang, D.,(2019)” Optimization design of axial fan blade”, Journal of the Chinese Institute of Engineers
方志翔（2007）以Pro/ENGINEER進行軸流式風扇葉片之快速化開發，碩士論文，國立台灣科技大學機械工程學系
李虹錤 (2002) 小型雙級軸流風扇性能之數值與實驗研究，碩士論文，國立台灣科技大學機械工程學系
李灃哲（2017）提高軸流風扇的風扇效率與較低其氣動噪音之最佳化分析，碩士論文，國立台北科技大學
洪國泰（2007）小型軸流風扇之設計、模擬與實驗整合研究，碩士論文，國立台灣科技大學機械工程學系
莊育霖（2022）同軸兩段式風扇的設計探討，碩士論文，國立雲林科技大學機械工程學系
楊淯仁（2017）小型軸流風扇葉片設計及分析，碩士論文，台北城市科技大學機械工程學系
鄭育政（2018）具襟翼之翼型應用於風扇設計的可行性研究，碩士論文，國立台灣科技大學機械工程學系
鄭楷繽（2022）軸流式環形風扇風量分析與最佳化設計，碩士論文，逢甲大學航太與系統工程學系
蔡明倫（2010）風扇性能評估與設計方法之整合研究，博士論文，國立台灣科技大學機械工程學系
賴昆澤（2012）葉片設計對小型風扇性能影響之探討，碩士論文，國立雲林科技大學機械工程學系
簡嘉宏（2014）田口法在軸流風扇扇葉新穎設計上的應用研究，碩士論文，國立雲林科技大學機械工程學系