1. IntroductionMicro Injection Moulding (MIM) is a relatively new technology which is popular in the industry for micromanufacture because of its mass production capability and low component cost. In order to achieve the highest quality components with minimal costs using MIM it is important to understand the process and identify the effects of different independent parameters. One of the methods that can be employed to investigate the overall operation of MIM is Design of Experiments (DoE). In general, DoE can be used to collect data from any process and gain an understanding of the process through data analysis. This procedure can help to optimise the process and eventually lead to quality improvements.This paper is organized as follows. The MIM process is described in Section 2. In Section 3 the DoE is introduced. The collection of experimental data is explained in section 4 followed by results and dataanalysis in section 5. The discussion of results is presented in section 6. Finally the paper ends with conclusions given in section 7.2. Micro Injection Moulding (MIM)Micro-injection moulding 1 is a relatively new technology in the manufacturing world, and as such, it needs to be thoroughly investigated. According to Micro-powder injection moulding, conducted by Liu et.al. 2, micro-system technology were widely used in the new 21st century because of its successful applications in many different fields, e.g. in fluidic, medical, optical and telecommunications. Presented with massproduction capability and low component cost, make the MIM technology to be one of the key production processes for micro manufacturing. The Components of MIM fall into one of the following two categories:Type A: Overall size less than 1mmType B: Micro feature less than 200um.Initial work on DoE and data analysis on MIM, conducted by Sha et. al. 3, primarily focused on the analysis of 5 different factors (the melt and mouldtemperature, injection speed, pressure and flow status) affecting the achievable aspect ratios in three different polymer materials. The aspect ratio is the ratio of a longer dimension to its shorter dimension of a specially designed micro feature for this experiment. Their study concluded that Melt Temperature (Tb) and Injection Speed (Vi) were the key factors affecting the aspect ratios achievable in replicating micro features in all three polymers materials.The effect of tool surface quality in MIM, conducted by Griffiths et. al. 4, primarily focused on the factors affecting the flow behavior and also the interactionbetween the melt flow and the tool surface.The findings of these earlier investigations are taken into consideration in this study.Fig 1 shows a picture of a MIM machine. The planning of DoE and the data analysis was carried out using the statistical software package “Minitab 16”.3. Design of Experiments (DoE)The technique of defining and investigating all possible conditions in an experiment involving multiple factors is known as the Design of Experiments.The two types of DoE that are widely used are the Factorial design and Taguchi Method. According to Minitab design of experiment 6, Factorial design is atype of designed experiment that allows for the simultaneous study of the effects that several factors may have on a response. When performing an experiment, varying the levels of all factors simultaneously rather than one at a time, allows for the study of interactions between the factors.In a full factorial experiment, responses are measured at all combinations of the experimental factor levels. The combinations of factor levels represent the conditions at which responses will be measured. Each experimental condition is called “ run ” and the response measurement an observation. The entire set of runs is the “design”.To minimize time and cost, it is possible to exclude some of the factor level combinations. Factorial designs in which one or more level combinations are excluded are called fractional factorial designs.Fractional factorial designs are useful in factor screening because they reduce the number of runs to a manageable size. The runs that are performed are a selected subset or fraction of the full factorial design. But Roy 7 mentions that using full factorial and fractional factorial DoE may contribute to the following issues: The experiments become unwieldy in cost andtime when the number of variable is large; Two designs for the same experiment may yielddifferent results; The designs normally do not permitdetermination of the contribution of each factor; The interpretation of experiment with a largenumber of factors may be quite difficult.Hence, Taguchi method was developed in order to overcome some of these issues. Taguchi method is the technique of defining and investigating all possible conditions in an experiment involving multiple factors. Taguchi method was first introduced by Dr. Genichi Taguchi after the Second World War 8, 9. He came up with three basic concepts 7:1. Quality should be designed into the product and not inspected into it.2. Quality is best achieved by minimising the deviation from a target. The product should be so designed that it is immune to uncontrollable environmental factors.3. The cost of quality should be measured as a function of deviation from the standard and the losses should be measure system-wide.Dr. Taguchi setup a three stage process to achieve the enhancement of product quality by DoE based upon the concepts above, namely, System design, Parameterdesign, and Tolerance design.For the first stage, system design is to determine the suitable working levels of design factors. It includes design and test of a system based on selected materials,parts and nominal product/process parameters.Parameter design is for finding the factor level that can achieve the best performance of the product/process.The last stage which is the tolerance design is to decrease the tolerance of factors which is significantly affecting the product /process.A special set of arrays called Orthogonal Arrays (OAs) were constructed to lay out the experiment. The OA simplify the experiment design process. It is done byselecting the most suitable OA, assigning the factors to the appropriate columns, and describing the combinations of the individual experiments called the trial conditions.In this study a fractional factorial DoE was conducted in combination with Taguchi design concepts for quality enhancement.4. Collection of Experimental DataThe experiment was designed and set-up as defined by Sha, et. al. 10. This aim of this experiment is to analyse the effects of six factors on the achievable aspect ratios and find the most significant factors in order to reach the optimal settings which would give the highest aspect ratios. Fig. 2 shows the test part with micro features in the form of legs with two level of width (W),200 or 500 um ,and depth(D), 700( ) or 1100 um( ) where the features having the same depth, D1 or D2,were grouped on 1one side of the part.Three different materials, namely, semi-crystalline polymers such as polypropylene (PP), polyoxymethylene (POM) and an amorphous polymer such as acrylonitrilebutadiene-styrene (ABS) were in this study. The parameters investigated were barrel temperature (Tb), mould temperature (Tm), injection speed (Vi), holdingpressure (Ph), the existence of air evacuation (Va) and the width (W) of micro-legs.The aspect ratios, i.e. the ratios between the length of the micro feature and their depths, D1 or D2, are measured during the experiment. The average values of 24 measured responses with the same W and D (two per part) while applying the process setting given in Table 1 are used in this study.5. Results and Data AnalysisA 2-level six factors fractional factorial design (26-2) was applied in this experiment. The DoE was used to identify the factors that were active and significant to study the filling of micro channels. The purpose of this exercise is to look at the results of the DoE responses in order to understand the process and select the significant factors with their appropriate settings which are necessary for optimal performance.5.1. ResultsThe measured experimental responses for the DoE for the ratios between the length of the melt fills and the depth of the channels, D1 or D2 are recorded in Table 2. The value of D1 and D2 shown on the table are the average values of 24 measurements.5.2. Data AnalysisThe statistical software package “Minitab 16” was used to analyse the results obtained from the experiment.The result of the analysis for PP for both the cases of D1 and D2 is given in Table 3.In Table 3 the “Effect” column shows the positive or negative effect of the factor on the measured response. Hence the higher the effect the more significant the factor in consideration will be.The “effect” column determines the factors relative strength,the “p-values” determine which of the factors are statistically significant. In this study the values in the P column of the Estimated Effects and Coefficients table are used to determine which of the effects are significant. To make a decision concerning which factors are significant, further analysis is necessary and this will be discussed in the next section. A typical value for the significance was chosen to be 0.05 throughout this study.6. Discussion of ResultsThe above results were utilised to produce moreevidence to support the claims for strong factors whichmatter the most for the MIM process.Using = 0.05, for PP D1, the p-values found for Tb is 0.038 and Vi is 0.009 indicate that the main effects from these two single factors Tb and Vi are significant, i.e. their p-values are less than 0.05. These two single factors and their effects and other calculated values are highlighted in Table 3. In addition, the above results show that none of the two-way interactions are significant. This is clearly shown by the “Normal Plot of the Standardized Effects” (Fig3) and the “Pareto Chart of the Standardized Effects” (Fig 4).6.1. Normal Effects PlotA normal effects plot is used to compare the relative magnitude and the statistical significance of both main and interaction effects. As shown in Fig 3, Minitab draws a straight line to indicate where the points would be expected to fall if all effects were close to zero. Points that do not fall near the straight line usually signal factors with significant effects. Such effects are larger and generally go further away from the fitted straight line compared to the unimportant effects. By default, Minitab use a=0.05 and labels any effect that is significant. This is shown in Fig 3 by clearly marked labels for factors C and A. The factor C having a much greater weight on the MIM process for PP-D1 compared to factor A can also be seen on this graph.6.2. Pareto ChartA pareto chart of the effects is used to compare the relative magnitude and the statistical significance of both main and interaction effects. As shown in Fig 4, Minitab plots the factor effects in decreasing order of the absolute value of the effects. The reference line on the chart indicates which factor effects are significant. Whenyour model contains an error term, by default, Minitab use a=0.05 to draw the reference line.The results in Fig 3 confirm the results displayed in Fig 4 as factors Cand A are the only two factors that have passed the reference line, and factor C having a much larger effect6.3. Main Effects PlotThe main effects plot shows the basic effect of changing the significant factors. These one-factor effects are called main effects. In this plot bigger main effect isdepicted by a line with steeper slope compared to the effects contributed by less significant factors. To calculate main effects, Minitab procedure subtracts themean response at the low or first level of the factor from the mean response at the high or second level of the factor. It can be seen from Fig 5 that changing Vi fromlevel 1 to 2 has a bigger main effect than changing Tb. This is depicted by a line with steeper slope for Vi.6.4. Interaction EffectsThe next step in the analysis is to look at the significant interactions. The two-way interaction effects calculated in Table 3 can be visually displayed on the interaction plot to see how big these effects are. An interaction plot shows the impact of two suspected interacting factors that changing the settings of one factor has on another factor. Because an interaction can magnify or diminish main effects, i.e. depending onwhether the interaction is positive or negative, evaluating interactions is extremely important. While close to parallel lines indicate little or no interaction between the factors, intersecting lines signal an interaction. The amount of interaction is proportional to the angle of intersection, i.e. close to 90° portrays the strongest possible interaction.The interaction plot in Fig 6 shows that the response, i.e. the aspect ratio for Vi at 100 is higher than for Vi at 50 at both levels of Tb. However, it can be seen that thedifference in aspect ratio between runs using Vi at 100 and runs using Vi at 50 at Tb set to 225 is much greater than the difference in aspect ratio between runs using Viat 100 and runs using Vi at 50 at Tb set to 200. This suggests that to get the highest aspect ratio Tb should be set at 225 while Vi is kept at 100.Similar analysis was carried out for PP D2. Likewise, the experimental results were analysed for POM and ABS for D1 and D2. The significant single factors andinteraction factors for each of these different materials and the recommended settings for the selected significant factors are summarised in Table 4.This study shows that in most cases the aspect ratio is influenced by single factors except in POM-D2, ABS-D1 and ABS-D2 with a two-way interaction. In the case ofPP-D1, Tb and Vi and for PP-D2, Vi only. For POM-D1, Tb, Tm, Vi and W and for POM-D2, Tb, Tm, Vi, W and TbXVi. When ABS was used for D1 the contributingfactors were Tb, Vi, W and TmXPh; for D2 the significant factors were Vi, W and TmXPh. The entries shown in bold in Table 4 indicate the chosen settings for thesignificant factors. The shaded cells in Table 4 show two-way interaction between the factors.Using the process of elimination the critical factors for PP was identified as barrel temperature ( ) and injection speed ( ), for POM as barrel temperature ( ), mould temperature ( ), injection speed ( ) and width (W), and for ABS as barrel fixed at 75. Hence the factors holding pressure (Ph) and the existence of air evacuation (Va) can be ignored in the MIM process. This gives a full factorial of 4 trials for , 16 trials for POM and 8 trials for ABS. Further, as a result of this study, the optimal settings to achieve the highest aspects ratio for different materials used can be summarised as follows: PP-D1: Tb at 225 and Vi at 100; PP-D2: Vi at 100; POM-D1: Tb at 200, Tm at 60, Vi at 100 and W at 500; POM-D2: Same as for D1 except W; ABS-D1: Tb at 258, Vi at 100, W at 500 while Tm is fixed at 75; ABS-D2: Vi at 100, W at 500 while Tm is fixed at 75. Confirmatory trials were conducted to verify the optimal performance for the above settings which have been identified theoretically and repeated 24 times andthe average measured responses gave the best aspect ratios to be found so far. They are as follows: for PP and POM the best aspect ratio of 20 and for ABS it was 21.7. ConclusionsIn this paper an analytical method for understanding the MIM process and optimising the process parameters using DoE has been presented. A fractional factorial experiment with Taguchis quanlity concepts has been conducted in order to save time and effort in performing the trials. The data collected in the form of measured responses has been successfully analysed to identify the significant single factors as well as two-way interactions. Further, the optimal process parameter setting identified through DoE method for different materials used in the study have been validated by running confirmatory trials and the measured responses verified the theoretical results by achieving high aspect ratios for the optimal settings found for the MIM process parameters. The knowledge of MIM gained through this study will help understand and optimise Nano Injection Moulding (NIM) process 11.AcknowledgementsThe authors would like to thank the EC FP7 FlexiTool project for supporting this work.References1 Trotta, G., Surace, R., Modica, F., Spina, R., Fassi, I., 2011. Micro Injection Moulding of Polymeric Components,” AIP Conf. Proc. 2011; 1315:1273-8.2 Liu, ZY, Loh, NH, Tor, SB, Khor, KA, Murakoshi, Y., Maeda, R., Shimizu, T., 2002. Micro-powder injection molding. J Material Processing Technology 2002; 127(2), p. 165.3 Sha, B., Dimov, S., Griffiths, C., Packianather, MS, 2007. Investigation of micro-injection moulding: Factors affecting the replication quality. J Materials Processing Technology 2007; 183, p. 284.4 Griffiths, CA, Dimov, SS, Brousseau, EB, Hoyle, RT, 2007. The effects of tool surface quality in micro-injection moulding. J Materials Processing Technology 2007; 189(1):, p. 418.5 Griffiths, CA, Dimov, SS, Brousseau, EB, Chouquet, C., Gavillet, J., Bigot, S., 2010. Investigation of surface treatment effects in micro-injection-moulding. Int J Advanced Manufacturing Technology 2010; 47(1):, p. 99.6 Minitab Handbook. 5th ed. Canada: Curt Hinrichs; 2005.7 Roy, R., 1990. A Primer on the Taguchi Method. USA: VanNostrand Reinhold; 1990.8 Sudhakar, PR., 1995. An Introduction to Quality Improvement Through Taguchi Methods. Quality 1995, p. 54.9 Taguchi, G., 1996. The Role of D.O.E. For Robust Engineering:A Commentry. Int J Quality and Reliability Eng 1996; 12:, p. 73.10 Sha, B., Dimov, S., Griffiths, C., Packianather, MS, 2007. Microinjection moulding: Factors affecting the achievable aspect ratios. Int J Adv Manuf Technoly 2007; 33, p. 147.11 Zhang, N., Cormac, J., Byrne, CJ, Browne, DJ, Gilchrist, MD, 2012. Towards nano-injection molding. Materials Today 2012; 15(5), p. 216.通過實(shí)驗(yàn)設(shè)計(jì)優(yōu)化微注射成型工藝M. Packianathera*, F. Cha , C. Griffith , S. Dimo , D.T. Phan aIMME 英國卡迪夫 CF243AA 游行皇后大廈卡迪夫大工程學(xué)院英國伯明翰 B152TT 伯明翰大學(xué)機(jī)械工程學(xué)院*通訊作者聯(lián)系電話：+44-29-20875911;傳真：+44-29-20874695；電子郵件地址：packianathermscf.ac.uk摘要本文提出通過試驗(yàn)設(shè)計(jì)（DOE）優(yōu)化微注射成型（MIM）過程。MIM 是一種相對(duì)較新的用于微部件的快速制造的技術(shù)。由于改變工藝參數(shù)，為了滿足質(zhì)量和可靠性的限制，減少操作過程中變異的是非常重要。在這項(xiàng)研究中，對(duì) MIM工藝的理解，它是通過 DOE 的六個(gè)影響表面質(zhì)量的參數(shù)，流動(dòng)長度和長寬比來優(yōu)化的。顯著單一的工藝參數(shù)以及它們之間的相互作用是通過統(tǒng)計(jì)分析確定。為 2 級(jí)的試驗(yàn)中，20：21：20 的縱橫比，分別對(duì)應(yīng)聚丙烯（PP）丙烯腈 - 丁二烯 - 苯乙烯（ABS）和聚甲醛（POM）實(shí)現(xiàn)關(guān)鍵詞：微注射成型（MIM） ，試驗(yàn)設(shè)計(jì)（DOE） ，全因子，部分因子，優(yōu)化設(shè)計(jì)的設(shè)計(jì)第一章 引言因?yàn)樗拇笈可a(chǎn)能力和低元件成本，微注射成型（MIM）是一種在微型制造行業(yè)內(nèi)流行的相對(duì)較新的技術(shù)。為了使 MIM 以最小的成本實(shí)現(xiàn)最高品質(zhì)的元件，理解的過程并確定不同的獨(dú)立參數(shù)的影響是很重要的。一種可以采用的調(diào)查 MIM 的整體操作的方法是試驗(yàn)設(shè)計(jì)（ DOE）的設(shè)計(jì)。在一般情況下，DOE（ DoE）可用于收集從每個(gè)過程，并通過數(shù)據(jù)分析獲得加工工藝的理解。這個(gè)程序可以幫助優(yōu)化過程，并最終使得質(zhì)量的提高。本文的結(jié)構(gòu)如下，在 MIM 工藝在第 2 節(jié)所述，在第 3 節(jié) DOE 的介紹，實(shí)驗(yàn)數(shù)據(jù)的收集之后第 4 節(jié)解釋，結(jié)果和數(shù)據(jù)分析進(jìn)行說明在第 5 節(jié)說明。結(jié)果的討論，在第 6 節(jié)提出，最后在第 7 節(jié)給出結(jié)論的文件結(jié)束。2212-8271©2013 的作者。由 Elsevier BV 公司負(fù)責(zé)出版，羅伯托特提教授同行評(píng)議 DOI：10.1016/j.procir.2013.09.052第二章 微注射成型（MIM）微注射成型1是在制造世界一個(gè)相對(duì)較新的技術(shù)，因此，它需要被深入研究調(diào)查。據(jù) Liu 等人2進(jìn)行微粉末注射成型，因?yàn)樗谠S多不同的領(lǐng)域，例如醫(yī)學(xué)，光學(xué)和電信，成功的應(yīng)用，使得微系統(tǒng)技術(shù)被廣泛使用在新的 21 世紀(jì)，。帶有大批量生產(chǎn)能力和低元件成本，使得 MIM 技術(shù)是進(jìn)行微制造中的一個(gè)關(guān)鍵生產(chǎn)工序。MIM 的組件分為以下兩個(gè)類別之一：A 型：外形尺寸小于 1mm ；B 型：微特征小于 200 。由 Sha 等人3在美國 DOE 進(jìn)行初步工作和 MIM 的數(shù)據(jù)分析，主要集中在 5個(gè)不同的受三個(gè)不同的聚合物材料可達(dá)到的高寬比影響的因素（熔體和模具溫度，注射速度，壓力和流動(dòng)狀態(tài)）的分析。本實(shí)驗(yàn)縱橫比是一個(gè)特殊設(shè)計(jì)的微特征，其為較長尺寸與較短尺寸的的比率。他們的研究結(jié)論是，熔體溫度（TB）和注射速度（六）是受在復(fù)制所有三種聚合物材料的微觀特性中可達(dá)到的長寬比的影響的關(guān)鍵因素。由 Griffiths 等人4進(jìn)行的 MIM 工具的表面質(zhì)量效果主要集中于影響熔體流動(dòng)和模具表面之間的流動(dòng)行為，并相互作用的因素。這些早期的調(diào)查結(jié)果都考慮到了這項(xiàng)研究。圖 1 示出了 MIM 型機(jī)的畫面。DOE 的規(guī)劃和數(shù)據(jù)分析使用的統(tǒng)計(jì)軟件包“Minitab 16”進(jìn)行。圖 1 微型注塑機(jī)5第三章 設(shè)計(jì)實(shí)驗(yàn)（DOE）在實(shí)驗(yàn)中定義和調(diào)查所有可能的條件涉及多重因素的技術(shù)被稱為實(shí)驗(yàn)的設(shè)計(jì)。這兩種 DOE 類型被廣泛采用是析因設(shè)計(jì)與田口方法。根據(jù)實(shí)驗(yàn) Minitab 的設(shè)計(jì)6 ，析因設(shè)計(jì)是一種設(shè)計(jì)的實(shí)驗(yàn)，允許同時(shí)影響研究，一些因素可能對(duì)產(chǎn)生同一個(gè)影響結(jié)果。當(dāng)進(jìn)行實(shí)驗(yàn)，不同的所有因素的水平同步，而不是一次一個(gè)，允許相互作用的因子的研究。在全面析因?qū)嶒?yàn)，響應(yīng)于實(shí)驗(yàn)因子水平的所有組合計(jì)算。因子水平的組合代表了在響應(yīng)將被測(cè)量的條件。每個(gè)實(shí)驗(yàn)條件稱為運(yùn)行和響應(yīng)測(cè)量觀察。整組運(yùn)行的是“設(shè)計(jì)” 。為了最大限度地減少時(shí)間和成本，因此能夠排除一些因子水平的組合