OPTIMIZATION STUDY IN MANUFACTURING PROCESS FOR PC/ABS BLENDS

CHINESE JOURNAL OF MECHANICAL ENGINEERINGVol.16, No.3, 2003Huang ChenghungDepartment of Mechanical Engineeing.National Central University,Chungli, Taiwan 320, ChinaOPTIMIZATION STUDY INFung ChinpingMANUFACTURING PROCESSDepartment of System Engineering.Chung Cheng Institute of Technology,FOR PCIABS BLENDSTaoyuan, Taiwan 335, ChinaTB332.063 AChang ShihhsingAbstract: The optimization of injection molding process for polycarbontearlonitrile-butadieneDepartment of Industial Management,styrene (PC/ABS) blends is studied wsing Taguchi method and principal component analysis (PCA).Van Nung Institute of Technology,Four controllable process factors are studied at three levels each in the manufacturing process. The LqChungli, Taiwan 320. Chinaorthogonal array is conducted to determine the optimum process facto/evel combination for singlequality of mechanical properties. In addition, the principal component analysis is employed to trans-Hwang Jiunrenform the correlated mechanical properties to a set of uncorrelated components and to evaluate a com-prehensive index for multi-response cases. Then the optimum process factor/level combination formultiple qualities can be determined. Finally, the analysis of variance is used to find out the mostDoong Jilinginfluential injection molding parameter for single and multiple qualities problems.Key words: PC/ABS Injection molding Optimization Taguchi methodDepartment of Mechanical Engineering,Principal component analysis Mechanical propertiesChungi, Taiwan 320, Chinaperature, mold temperature and packing pressure, each at three0 INTRODUCTIONlevels to find the optimum process factor/level combination forquality of mechanical properties. In addition, the principalCortelateaAs engineering polymers have the advantages of highcomponent analysis is employedransformthe Cstrength to weight ratio, high sifess, and fast productivity, they mechanical properties to a set of uncorrelated components andhave been broadly applied to many different fields in recent years. evaluate a comprebensive index for multi-response cases. ThenIn addition, different materials are blended or alloyed to form the optimum process factor/evel combination for multiple qualiti-composites with a combination of properties that cannot be ob- es can be determined. Finally, the analysis of variance (ANOVA)tained in the original materials. PC/ABS is one of the most fre- is used to fnd out the most influential injection molding parame-quently used engineering composites. Polycarbonate (PC) pos- ter for single and multiple qualities problems.sesses many excellent mechanical properties such as high tough-ness, high impact strength, high heat distortion and dimensional 1 ANALYSIS METHODSstability. However, PC also has some drawbacks such as high meltviscosity, and this cause a problem in injection molding. To solveTaguchi method and signalto-noise ratiois program,n, PC is often blended with acrylonitrile-bncieneTaguchi method is a useful tool for experiment design in re-styrene (ABS)-I. Since PC has the excellent mechanical proper- cent years. In this method the signal-to-noise is used to representties and ABS improves injection molding process, PC/ABS blends quality characteristic and the largest S/N ratio is demanded, Usu-bave been exploited by the industry especially in automotive ally, there are three types of quality charateristics, i.e. the target-appliances.the-better, the-larger-the-better and the smallr-t-e-better.Injection molding is an important manufacturing process to(1) Ter-arget-the-betterpolymers. It provides products with dimensions steadiness andprecision, low manufacture cycle and low cost. However, thauthors and other researchers all found that iection molding(1)nσ司process parameters affect the quality of products very much!4- 7The selection of corret injection molding conditions is, therefore,where o-Standard deviationa major concem in the plastics industry. Traditional experiment- Mcasured propertyresult. Therefore, Taguchi brought up a new experiment design-Number of samples in each test trialthat used orthogonal arrays and signal-to noise ratio to the robust(2) The-large-the-betterdesign of products and processl8. In recent years, Taguchi methodhas become a powerful tool for improving productivity. However,(2)the Taguchi method can only optimize a single quality character-SN -1o却istic.' To overcome this problem, many researchers tried to combi-ne Taguchi method with other methods such as fuzzy method,(3) The-smaller-the bettermutiple liner regression analysis, principal component analysis9-2to deal with mutiple quality caraeteritisi9..SN=-101g(-之)=- 101g(5)(3)The optimum levels of injection molding parameters for themechanical properties of PC/ABS is presented using the Taguchi中国煤化工method and principal component analysis. The experiment design1.2uscs orthogonal array Lq for four factors of flling time, melt tem-The IY片C N M H Gulivariate stiticalmethod that accounts tor the vanance o1 original multi-responseby a small oumber of components. The technique was first intro-Reeved December 30, 202; rceved in evise fom Apil 1s, 203; ac duced by Pearson and then further developed by Holiningl.4.ed May 19, 2003The procedure is described in the following.●234●Huang Chenghung, ct al: Opimization study in manufacturing process for PC/ABS blends(1) The original multi- response arrayTable 1 Basic physlcat properties of PC/ABS blendx,(k);# l,2,.,m;k= 1,2..,n .PropertiesCycoloy 1110FDensity p/(g●cmi”)1.14「x() x(2) . {12)]Mold shrinkage/%Tensile strength p/MPax(1) *x2(2) .. x(n)x=(42379[zod impact srength ax(J●m)xm(1) x(2) . x(n)]Ceoficient of linear thermal expansion a/K-'7.2x 1050.2where m is the number of test trial and n is the number of response.Thermal conductivity W/(m'.K-)The X is the S/N ratio of each response.The experiment was conducted with four controllable 3-level(2) Normalizing the responsefactors and three response variables. It is based on the orthogonalThe response is normalized using the fllowing formula to rray Lg. The four factors of fling time, melt temperature, moldtemperature, and packing pressure with three levels were listed inget rid of the difference in unitsTable 2. The three response variables were fracture toughness,而(k)- xmin(k)_ultimate tensile strength and elongation. The specimens weremolded in an injection molding machine. The dimension of moldcavities, the location and dimension of the pin gate were firstsimulated by C-mold for better flling process.[x() x(2) xi(n)]Table 2 Experimental factors and factor leve' forthe Iiecton mold Condionsof PClABSn mold conditions of PCABS .x*_xi(1) x(2) . x(n)Experimental factorsLevels of[x."(1) x(2) . x(n)]experimentalFillingleltMoldackingtime terperaturetemperaturwhere x (k) is the normalized response.i/s6/CPi%(3) Correlation coficient aray2050The correlation coeffcient array of normalized response ar-40651026080100ray is evaluated as followsThe fracture toughness and tensile test procedures fllowed theCovEx().x()ASTM E399-90 and D638 90 specifcationis10 tons com-(61s using a°xσixputerized MTS (model 810) closed-oop servo lhydraulic system.The procedures were contolld by computer softwarej=1,2,.-,n h= 1,2,.-,.nspecimens were loaded at a speed of 1 mm/min in tensionmonotonically until fracture; in the meantime the load and the cracwhere Co(x(),x(k)一 Covariance of sequences x() length or compliacc dsplacment were continuously rordedand x'(k)The results of fracture toughness test were listed in Table 3.Standard deviation of sequenceEach levelfactor combination of injection molding conditionswas tested three times. The values of plan strain fracture tough-x(k)ness, Kic, are in the range of 5.425~-6.675 MPa.√m . The results(4) Detrmine the eigenvectors and eigeovaluesof ultimate tensile strength and elongation obtained from theThe eigenvectors and eigeavalues are determined from thetensile test were listed in Table 4 and Table 5 respectively. Thecorrelation coefficient array.ultimate tensile strength was 43.403~ 45.573 MPa and the elonga-Eigenvaluestion was 34.173%~ 58.939%.Table3 Experimental results of fracture toughness Kycn;=l, 2,.,nand SAN retiosFracture toughnessEigenvectorsExp.\ BCDKc/(MPa . m啊Average SINNo.1 No.2N.3v=[C。C2。 .. C16.1766.041S.708 5.975 1S.512 .6.042 5.997 15.537where CoCm-- Cofficient of eigenvector5.468 5.316(5) Evaluate the principal components7.156 6.5036.368 6.675 16.456We have the following uncorrelated principal components6.514 6.366.587 6.489 16.2415.9356.4306.298 6.221 15.8626.5806.6796.396 6.555 16.3225.866 6.4346.258 6.186 15.809where y,=[Yr Yz.. Y_.]6.4996.46.4006.4616.186Table 4 Exrperimental results of ultimate tensile strength o。The principal components are created in order to decreaseand S/N ratiosvariance, and therefore the first principal component, Y, accountsEXP.ABCDUltimate tensile strength a MPaAverngc S/Nfor most variance in the data.No.43.61943.620 32.7932 EXPERIMENTAL DESIGN AND RESULTS45.736 44.92245.08545.247 3.1.1144.705 3.000PC/ABS named Cycoloy C1110HF fom general eletric中国煤化工:345.519 3.164corporation was used in this study. It is 60% PC blended with40% ABS and its average density is 1.13 g/cm'. The basic physi-YHCNMHG3445.085 33.07643.403 32.733cal properties are given in Table 1. Before the injection molding45.573 45.736 45.41045.573 33.174process, the material Was heated to 110~120 C for 4~6h to re-45.247 45.085 44.43444.922 33.048move moisture.44.108 43.783 42.969 43.620 32.792●236●Huang Chenghung, et al: Optimization study in manufacturing process for PC/ABS blendsTable 10 Response table and ANOVA for the ftrstbutadiene-stytenelend: [PhD Disetation. Chungli: National Centralprincipal componentTaguchi G Introduction to quality engineering. Asian Productivity Organi-FactcLevel Level Level Degreeof Sumof Mean F Contributionzation.1990...2..3. fredomn_ square squarevlue64- 9 TongL 1, Su C T. Opimizing multiresponse problems in the teguchi0.8341.1201.2350.255 0.1281.53733.8method by fuzzy multiple atribute decision making. Quality and Reliabi-0856100813240.342 0.171 2 06112.6ity Engineering Intermational, 1997, 13(I): 25~340.951 0.984 1.2540.166*0 ReedyP B s, Nishia K, Subash Babu A. Unifcation of robust design andTotal1.404100.0goal progranming for multiresponse optimization-8a case study. Qualityand Rliability Engineering Intermational, 1997, 13(6): 371~-383(Error)_(2) 一(0.16) (0.083)11 SuC T, Tong L I Multi-tesponse robust design by principal componentanalysis. Total Quality Managenent, 1997, 8(6):409-416...4 CONCLUSIONSAutongyJ.Mult-tesponsepopitmtzamton n indusualexpentents usingThe Taguchi method and PCAe used to optimize the in-and Reliabilit Engineering Intermational, 2000, 16(1): 3-8Pearson K. On lines and planes of closest ft to systems of points in spaces.jection molding process of PC/ABS. The results are summarizedPhilosophical Magazine, 1901, 6(2) 559- -572as follows.Hetelling H. Analysis of a complex of statistica! variables into principal(1) The results of the analysis show that different optimumcomponents. Joumal of Educational Psychology, 1933, 24: 417- 441injection molding condition is found for each quality. The fac- 15 Standard test method for plan strain fracture toughness of metalli matern-torlevel combination A2B,CD is recommended for the singleals. Annual Book of ASTM Standards, E399-90quality of fracture toughness, A,B2C;D3 is for the ultimate tensile16 Standard test method for tensile propertics of plastis. Annual Book ofASTM Standards, D638-90strength, and AzB,CD, for the elongation considered only.(2) When fracture toughness, ultimate tensile strength and Blographleal notes: Huang Chenghung is a candidate for doctor's degree inelongation are simultaneously considered, A.B,C;D3 is recom-Department of Mechanical Engineering. National Central Universitv Taiwan.mended for the multi-quality characteristic problem.China. His research interests incude injection molding optimization, Taguchi(3) From the contribution of each property, the flling time ismethod, grey systen, virual reality, etc.the most influential factor to fracture toughness. The melt tem-Tel: +886 3-4228550; E-mail: zctetic@mail2000.com.twperature and mold temperature are the most influential factor to Fumg Chinping is an asciatce professor in Departnent of System Engineering,ultimate strength and elongation respectively.Chung Cheng Institute of Technology, Taiwan, China. He received a MS degree(4) Using the PCA, the three original quality responses werein mechanical engineering fom National Central University, Taiwan, China, intransformed to obtain the frst principal component, The melt1988, and a PhD degree in engineering in 1995, from the University of Man-chester, UK. He has publisbed some refereed papers in the area of injectiontemperature is found to be the most influential factor.fnotot Tistord pastcs His man rsearch ares ne!lude plate teriesReferencescomputer-tided engineering and virtual realtechnology, design of experiments,Tel: +886-3-3809257; E-mail: cpfung@ccit.edu.twwith poly (acrylonitrile butadiene styrene). Polymer, 1994， 35(12):Chang Shihhsing is a professor in Department of Industrial Management De2 631-2 635partment, Van Nung Institute of Technology, Taiwan, China. His researchGreco R, Astarita M F, Dong L D, et al. Polyarbonate/ABS blends: procinterests include manufacuring process optimization, poto-electronic meas-esbilit, thermal properties, and mechanical and impect behavior. Ad- urenent, nano & semiconductor proces technology and equipment, dreetor ofvance in Polymer Technology, 1994, 13(4): 259-274PICO lab. (process identification, clustering & control, optimization).Balakrishman s, Nelaknten N R Mechanical propertics of blends of Tel: +886-939924099; E-mail: prof.chang@sma.hinet.etpolycarbonate with unmodifed aund malefic anhydride grafted ABS. Polym-e International, 1998, 45(4): 347-352Chang S H The stuodies on manufucturing optimal design using the taguchitional Central University, Taiwan, China His Trescarch interests include fatigueeregmoncadeignoagdererelaetoeaeel analysis (PhD Diseretion. Chun-and fracture of engineering naterials, durability analysis and life prediction ofmechanical components, laboratory road simulation of vehicles, etc.Chang S H, Hwang J R, Doong J L Manufacturing process optimization ofTel: +886-3-4267314; E-mail: jrhwang@cc.ncu.edu.twshort glass fiber reinforced polycarbonate composites in injection molding.Journal of Reinforced Plastics and Composites, 2000, 19(4): 301~321Doong Jiliang is 8 professor in Department of Mechanical Engineering, Na-Ho K C. The studies of mechanical and wear behaviors in short glass fibertional Central University, Taiwan, China. His research interests include virtualreinforced polyarbonate composies: (PhD Dserarion], Chungli: National relity fatigue and fracture of enineering materials, design of mechanical,Central Univensity, 1997accideat reconstuction, etcHo M s. The study on mechanical poperties in poylrateteroririri Tel: +86-3-4267359; E-mil: j@l@nil000om.w中国煤化工MYHCNMHG