本計畫之主要目的，在於發展渦輪葉片鍛造成形問題的最佳化設計技術，其方法乃利用一整合技術，將利用有限元素分析所得到的結果，以程式化方式進行最佳化設計，以完成鍛造葉片之成形尺寸和設計尺寸的誤差為最小化的模具設計。本計畫的另一目的，在於探討目標函數的定義對最佳化效率和結果精確度的關聯性。目標函數可定義成和成形尺寸及設計尺寸之誤差量的函數，且可存在多種定義，例如均方根值，及絕對誤差量的和等。為完成本計畫之目的，本計畫擬採用MARC 有限元素分析軟體（或相當之軟體）和IMSL 數學軟體（或相當之軟體）。因此，除了有限元素解的妥適性必需有賴實驗結果的驗證外，應用前述不同軟體間的相容性亦必需考慮。本計畫的驗證將取自Maccarini 等人[19]的自由擠鍛成形實驗結果。 The main objective of this research proposal is to develop an optimal design technique of die shape in forging of a turbine blade using FEA software. By combining the general non-linear finite element analysis with an optimal software, a programming algorithm is implemented to accomplish the die shape design, by which the deviation between the forged and designated dimensions of turbine blades is minimized. The other objective of this research proposal, however, is to investigate the influence of the cost function on the speed of convergence as well as accuracy of the determined solution. The cost function may be defined as a function of the deviation existing between the forged and the corresponding designated dimensions of the turbine blade. It may be formulated in different ways such as the mean square root of the deviation and sum of the absolute deviation. To arrive at these objectives, in this research proposal the FEA software 「MARC」 (or the equivalent) as well as the IMSL MATH/LIBRARY (or the equivalent) will be employed. Therefore, the capability of using the softwares, which are developed independently, is necessarily accounted for. Besides, it is required to verify the validity of the FEA software by means of the experimental results of Maccarini et al. [19] who performed the test on cold forging operatios. 研究期間：9408 ~ 9507
