射出成型製程中機台校正效應對成品多重目標與實驗設計優化法的效能影響之研究

The effect of machine calibration on the multi-objective and the efficiency of optimization method in injection molding process

林采玟(Tsai-Wen Lin)

指導教授 ： 黃招財

淡江大學/工學院/化學工程與材料工程學系碩士班/碩士(2021年)

https://doi.org/10.6846/TKU.2021.00110

摘要

因應工業4.0自動化，自動化技術逐漸應用在射出成型中，以達到快速生產的目標，不過，良好的複合產品常常需要多項目標值同步達成，像是產品外觀、精度、重量等等，此等需求導致品質與量產同步到位非常困難。而產業為了充分使用模具系統，通常喜歡將組合件放在同一付模具內，一次生產為一模多穴的模具系統。雖然，一模多穴模具系統已經在工業製造中用於製造一系列產品多年；但是，針對此等模具系統所生產之產品組合度的分析，科學化資訊並不多。然而，射出後組合件之組合度優劣與否，雖然可見於”面向製造和裝配設計(Design for manufacturing and assembly, DFMA)”相關文獻中討論，但該等產品之組合度評估卻常需要透過手動方式處理，因而非常不容易於設計階段就能有效掌握該等組合度特性。有鑑於此，我們認為射出成型組合件之組合度應該要能被量化，更期盼能有效掌握。為此，本研究將分成三大部份。第一部份將以兩件式之組合件系統，探討如何能提供妥善製造和組裝之設計，我們將針對具有兩個不同組件(後續稱為A件及B件)的一模兩穴模具系統同一次所生產之射出成品的組合度(degree of assembly)進行探討。在此，我們利用數值模擬和實驗驗證方法進行研究。明確而言，我們先利用射出成型之模擬分析所得之保壓壓力作為實務操作參數，藉由其改變來觀察特徵長度之差異變化，進而來量化組合度的變化，其中Xi = (XBi-XAi)，其中XA為A件外部長度，XB為B件內部長度，例如，X1 =（XB1-XA1）是組合後中心部分的特徵長度（以A件為基準），其他部份以此類推。結果顯示，當一模兩穴模具系統之保壓增加時，A件及B件之組裝度變得困難，此部份因較高的保壓壓力將導致B件的內部長度遠遠小於A件的外部長度，進而導致它們的組裝難度增大。另外，我們也利用實體實驗研究此等現象，其中在保壓25％下，A件與B件可以順利組合，但當保壓增加至50％和100％時，組合困難度逐步增加。透過實際實驗，發現實驗與模擬結果中特徵長度變化趨勢相當一致。在第二部分中，我們考慮機台校正效應對組合件系統的影響，透過模擬分析與實際射出實驗結果比對，研究顯示經由機台校正之模擬分析結果與實驗結果差異明顯減小；之後，再透過實際產品組裝測試發現變化之趨勢與模擬分析結果一致，並且確認可組裝之特徵長度範圍為大於-0.250 mm。另一方面，我們也透過實際實驗嘗試定義”密合度”，利用另一個角度掌握射出成品之組合度。實際針對射速30%～70%系統所製作的射出成品進行探討，結果發現當密合度拉力大於50 N之組合件將無法組合，此等50 N應力相當於特徵長度規範為 -0.250 mm。再者，在第三部分將嘗試使用田口法(CAE-DOE)及反應曲面法(CAE-RSM)優化策略，探討實驗設計法應用於組合件組合度之操作參數優化效益探索。顯示在未考慮機台校正效應下進行CAE-DOE優化中，其模擬分析改善率為11%，實際實驗改善率為5%；在考慮機台校正效應後進行CAE-DOE優化中，模擬分析改善率為21%，實際實驗改善率為21%；另外，在未考慮機台校正效應下進行CAE-RSM優化中其模擬分析改善率為25%，實際實驗改善率為29%；在考慮機台校正效應後進行CAE-RSM優化中，模擬分析改善率為29%。整體而言，田口法(CAE-DOE)及反應曲面法(CAE-RSM)優化策略在組合件組合度之操作參數優化有明確成效；另外，機台校正效應確實影響如何妥善製造和組裝之設計。

關鍵字

射出成型 ； CAE模擬分析 ； 多重目標 ； 翹曲變形 ； 實驗設計法(DOE)

並列摘要

In Industry 4.0, automation technology is gradually applied in injection molding to achieve the goal of rapid production. However, good composite products often require multiple target values to be achieved simultaneously, such as product appearance, accuracy, weight, etc. These requirements lead to very difficult to maintain both quality and mass production at the same time. Moreover, in order to enhance the usage of the mold system, people usually like to put the combined parts in the same mold, and produce multiple components at a time. This type of mold is called a multi-cavity mold system. Although this system has been used in industrial manufacturing to fabricate a series of products for many years; however, there is not much scientific information about the degree of assembly of the product which produced by these mold systems. In addition, although the degree of assembly of assemblies can be discussed in the "Design for manufacturing and assembly (DFMA)" related literature, the evaluation of the degree of assembly of these products often needs to be processed manually. Therefore, it is very difficult to effectively grasp the characteristics of the degree of assembly in the design stage. Hence, the degree of assembly of injection molding assemblies should be investigated. In this study, the main contents will be divided into three parts. The first part will use a two-piece assembly system to discuss how to provide a design for proper manufacturing and assembly. We will focus on a two-cavity mold system with two different components (hereinafter referred to as A and B). The degree of assembly will be discussed based on these injection components which are produced at the same time. Here, we have applied both numerical simulation and experimental methods to study the degree of assembly. Specifically, we first utilize the packing pressure obtained from the simulation analysis as a practical operating parameter. Moreover, we have defined the characteristic length as the quantity for evaluation of the degree of assembly. Specifically, the characteristic length is defined as Xi = (XBi-XAi), where XA is the outer length of part A, and XB is the inner length of part B. For example, X1 = (XB1-XA1) is the characteristic length of the central part after the combination (based on part A), and the other parts can be deduced by analogy. The results show that when the packing pressure increases, the assembly of part A and part B becomes more difficult. Due to the higher packing pressure, the inner length of part B will be much smaller than the outer length of part A, which leads to increased difficulty in their assembly. Furthermore, some experimental observation has been performed to verify the numerical prediction. When the packing pressure is 25%, part A and part B can be assembled smoothly, but when the packing pressure is increased to 50% and 100%, the assembly become more difficult gradually. Through these experimental observations, it is found that the variation trend of the characteristic length in the experimental and simulation results is quite consistent. Moreover, in the second part, we have considered the influence of the machine calibration effect on the degree of assembly for this system. By comparing the simulation analysis with the actual injection experiment results, the results show that the difference between the simulation and the experiment through the machine calibration is significantly reduced. Through the actual product assembly test, it is found that the trend of change is consistent with the simulation. Specifically, the specification to guarantee the components with smooth assembly is that the characteristic length ahoulsbe greater than -0.250 mm. On the other hand, we also have tried to define the "fitness" through actual experiments, and use this second real quantity to confirm the degree of assembly for this system. Here, the testing specimens are produced at the injection speed of 30% to 70%. The results show that when the fitness of the assembly is greater than 50 N, the assembly cannot be achieved. The 50 N tensile stresses are equivalent to the characteristic length specification of -0.250 mm. Furthermore, in the third part, we have tried to use the Taguchi method (CAE-DOE) and the response surface method (CAE-RSM) to investigate how to optimize the degree of the assembly. The results show that in the absence of machine calibration effect, comparing to the original design, in CAE-DOE optimization, the deviation by simulation prediction has been improved by 11%, and that of the experiment has been improved by 5%; in the CAE-DOE optimization after considering the machine calibration effect, both the simulation and the experimental results have been improved by 21%. Moreover, in the CAE-RSM optimization without considering the effect of machine calibration the simulation result has been improved by 25%, and that of the experiment has been improved by 29%; In CAE-RSM optimization after machine calibration, the simulation result has improved by 29%. Overall, the Taguchi method (CAE-DOE) and response surface method (CAE-RSM) optimization strategies are effective in optimizing the operating parameters of the degree of assembly; Furthermore, the machine calibration effect does affect the design of assembly significantly.

並列關鍵字

Injection molding ； CAE simulation analysis ； Multi-objective ； Warpage ； Design of experiment

參考文獻

[1] 裴有恆. IoT物聯網無限商機 : 產業概論 X 實務應用. 臺北市: 碁峰資訊股份有限公司. (2017).

[2] Y. Kuo, andM. R. Kamal. The fluid mechanics and heat transfer of injection mold filling of thermoplastic materials. AIChE Journal, 22(4), 661-669. (1976).

[3] B. Pramujati, R. Dubay, andC. Samaan. Cavity pressure control during cooling in plastic injection molding. Advances in Polymer Technology: Journal of the Polymer Processing Institute, 25(3), 170-181. (2006).

[4] H. Hassan. An experimental work on the effect of injection molding parameters on the cavity pressure and product weight. The International Journal of Advanced Manufacturing Technology, 67(1-4), 675-686. (2013).

[5] J. A. Camelio, S. J. Hu, andS. P. Marin. Compliant assembly variation analysis using component geometric covariance. J. Manuf. Sci. Eng., 126(2), 355-360. (2004).