Int J Plast Technol DOI 10.1007/s12588-015-9115-2 R E V I E W A RT I C L E
Process parameter optimization of plastic injection molding: a review Satadru Kashyap 1 & Dilip Datta 1
Received: 7 August 2014 / Accepted: 22 September 2015 # Central Institute of Plastics Engineering & Technology 2015
Abstract Over the years, injection molding has been a premier manufacturing technique in the production of intricate polymer components. Its molding efficiency rests on the shoulders of multiple process and machine parameters, which dictate the final product quality in terms of multiple output responses. It is imperative to state that a precise optimization of various input parameters is paramount for achieving the desired quality indices. In this article, a review of different techniques employed till date for optimizing various injection molding parameters is presented along with their advantages and limitations. It is found in the review that a complete intelligent technique operable without human interference is yet to be developed. Keywords Injection molding . Process parameter . Optimization
Introduction With the advent of high grade production technology; apart from miniaturization of components; intricacy, lightness and precision to design have become important factors today for commercial viability of plastic products. Due to its versatility in terms of mold and process design, injection molding has been paramount for decades in massproduction of intricate plastic products, which produces components in their near net shape with excellent dimensional tolerance. Consequently, in order to obtain a product with optimum quality, harnessing machine and process parameters of injection molding has been a premier area of research over the years [39].
* Dilip Datta
[email protected];
[email protected] Satadru Kashyap
[email protected] 1
Department of Mechanical Engineering, Tezpur University, Napaam, Tezpur 784 028, India
Int J Plast Technol
The injection molding process incurs considerable changes in rheological and thermo-mechanical properties of polymeric materials and their composites due to highly varying stresses at various points of the process, materials processing at melt temperatures, and high cooling rates of the final product. It is imperative to state that properties of injection molded materials play a crucial role in obtaining the net shape of a final product. A material endures large amounts of shear stresses during injection molding, and rheological properties can be quantified to understand the stability of feedstock during molding. Rheological properties consisting of variation in viscosity with shear rates were studied by many researchers [34, 41, 53, 86, 87, 105]. Feng et al. [34] showed viscosity as a function of shear rate at different temperatures for high density polyethylene (HDPE) and ultra-high molecular weight polyethylene (UHMWPE)-HDPE blends (Fig. 1), where change in the melt viscosity was attributed to entanglement/disentanglement of molecular chains under high external forces of injection molding. In nonlinearly behaved (non-Newtonian behaviour) injected materials a small variation in shear rate results in large changes in melt viscosity, which is undesirable in setting up process parameters. Moreover, the thermal properties of extruded feedstocks, which can be obtained from DSC (Differential Scanning Calorimetry) plots, provide basic guidelines for the subsequent molding steps by showing a clear picture of various phases and their corresponding melting points [34, 41, 86, 87, 105]. As obtained by Feng et al. [34], Fig. 2 shows DSC curves for UHMWPE-HDPE polymer blends at different melt temperatures of injection molding, denoting the change in the crystal structure of the blends. Additionally, the PVT (Pressure-Volume-Temperature) behaviour of injection molded products should also be consulted beforehand as it provides specific volume of melt in the mold cavity as a function of cavity pressure and temperature, which helps in understanding the compressibility and the effects of temperature of the melt during molding operation. In order to avoid part relaxation problem, the holding pressure is to be selected from the PVT diagram in such a way that the cavity pressure comes near to the atmospheric pressure on mold opening. PVT diagrams for
Fig. 1 Plots of steady viscosity versus shear rate at different temperatures (reprinted from Feng et al. [34] with permission from Springer Science and Business Media)
Int J Plast Technol
Fig. 2 DSC melting curves for UHMWPE/HDPE blend samples at three melt temperatures of core melts (190, 210 and 230 °C) (reprinted from Feng et al. [34] with permission from Springer Science and Business Media)
different materials are available in literature [79, 86–88, 93], as an example Fig. 3 shows the PVT diagram of polystyrene for specific volume versus temperature at different pressures. In view of above, various factors affecting the injection molding have to be analyzed thoroughly before deciding the feasibility of producing a product with the desired quality and intricacy. Such factors can be classified into three categories as follows [12]: 1. Machine parameters (independent variables): Barrel temperature, nozzle temperature, coolant temperature, packing pressure, holding pressure, back pressure, injection pressure, sequence and motion, switch-over point, injection speed, screw speed, shot volume, cushion, etc.
Fig. 3 PVT diagram of polystyrene: specific volume versus temperature at different pressures (reprinted from Quach and Simha [93] with permission from AIP Publishing LLC, ©1971)
Int J Plast Technol
2. Process parameters (dependent variables): Mold temperature, melt temperature, cooling temperature, melt pressure, melt-front advancement, shear stress, injection time, filling time, packing time, holding time, cooling time, mold open time, injection rate, material flow rate, rate of heat dissipation and cooling, pressure switch-over, etc. 3. Quality indices (final responses): Part dimension, shrinkage, warpage, sink marks, appearance and strength at weld lines, and other aesthetic defects such as burn marks, gate blushes, surface texture, etc. The machine parameters can be controlled either by using effective sensors in the machine or by upgrading various machine components. The process parameters are dependent on process conditions, material properties, and mold design. On the other hand, the quality indices are the target variables, which are to be precision controlled for obtaining the desired net shape and intricacy. However, the extent of effect of each of the machine and process parameters on the final responses may not be the same or inter-related. Further, the machine and process parameters can form multiple sets of input, resulting in different output responses. However, a huge hindrance occurs in identifying the significant input parameters along with their inter-dependency on each other and their relationships with output responses. In earlier days, a machine was set up on trial and error basis by testing different sets of process and machine parameters based on the intuition and experience of machine operators. Such a manual process often led to wastage of resources as the effects of various underlying parameters could not be predicted beforehand. Moreover, drift in the machine settings over a period of time also resulted in upsetting of product quality. Nowadays, advanced technologies have given machines with better varieties and better control over parameters. However, it is still a daunting and expensive task to experimentally examine the output responses corresponding to various combinations of input parameters, or to quantitatively determine the inter-dependency between input parameters and output responses. Hence, there has been requirement of techniques which can design experiments with minimum number of trials, identify significant variables affecting the process, precisely predict final responses, and include material and machine constraints in addition to process constraints. With the progression of computation, many sophisticated studies are being conducted in these directions. This paper reviews the past and present research conducted till date towards using various optimization techniques for precise prediction of machine and process parameters so as to achieve the net shape of an injection molded product with desired quality at the minimum cost (resource and time). The article is organized as follows: the formulation of the injection molding optimization, followed by research conducted so far using different optimization techniques and subsequently, the article is concluded with some future research directions.
Int J Plast Technol
Formulation of the injection molding optimization The injection molding optimization problem can be formulated as in Eq. (1). T Determine x ≡ x j ; j ¼ 1; 2; …; m and T to optimize Qðx; yÞ ≡ fq ðx; yÞ; q ¼ 1; 2; …; n subject to hi ðx; yÞ ≤ 0 ; ðlÞ ðuÞ x j ≤ x j≤ x j ; ðlÞ
ðuÞ
yk ≤ yk ≤ yk ;
9 y ≡ ðyk ; k ¼ 1; 2; …; pÞT > > > > > = i ¼ 1; 2; …; c ð1Þ > > > j ¼ 1; 2; …; m > > ; k ¼ 1; 2; …; p
where x and y are design vectors representing machine and process conditions with (u) (l) (u) (x(l) j ,xj ) and (yk ,yk ) as the ranges of parameters xj and yk, respectively; Q is the desired overall quality to be optimized (maximized or minimized), subject to constraints hi representing limitations on cost, loss, efficiency, etc.; and fq are quality indices, such as shrinkage, warpage, defects, sink marks, etc. The number and type of parameters, quality indices and constraints in Eq. (1) may vary in different investigations. But the essence of most of the works conducted till date was to optimize the overall quality of injection molded components by taking into consideration all the machine and process parameters and constraints affecting final responses. Dang [28] broadly classified the optimization techniques, implemented to plastic injection molding, into direct discrete methods (no mathematical function defining the relation between input and output) and meta-model based methods (approximate models built before applying an optimization process). Many works have been conducted using mathematical, computational and statistical techniques towards development of the injection molding process in terms of better predictability of processing parameters and establishing the inter-dependency of multiple input parameters on multiple output responses. These techniques can broadly be classified as past approaches, Taguchi method, modern computational methods, and real time techniques. Different works conducted so far under these four categories are discussed under their respective headings.
Past approaches During the incipient stage of injection molding, trial and error methods, based on experience and instincts of molding operators, were used to obtain products with lesser defects. Many miscellaneous approaches were proposed in the 1980s and 1990s towards predicting either the ranges of input parameters or the process on case to case basis [80]. These approaches can be classified as process window approach, case based reasoning approach and expert system based approach. In the process window approach, the relationship between injection pressure and melt temperature is used to create a process window, within which the molding process is carried out. Boundaries of the window are fixed based on two facts: (a) a too high
Int J Plast Technol
injection pressure creates flash, while a too low one creates short shot, and (b) a too high melt temperature creates material degradation, while a too low one requires high injection forces. Using this approach, various process parameters were optimized [7, 32, 80, 82, 83, 90, 91]. However, the approach suffers from the inability for tackling multiple inputs and multiple outputs (MIMO). The case based reasoning approach matches a new problem with one or more previous cases from a library and the most comparable ones are used to provide a solution to the case at hand. The approach was implemented for cost estimation of injection molds and mold parts [32, 114], for determination of molding parameters [56, 57], and for setting of initial parameters [99]. Although it was a quick solution approach, but its effectiveness is dependent upon the size and relevance of the case library. Moreover, the use of this approach restricts the development of any new concept or method. The expert system based approach is a subset of artificial intelligence technique, and it has two components: (a) a knowledge based component containing facts and heuristics about the topic and their inter-relationships, and (b) an inference engine for simulating the end results. Expert systems based approaches were developed to provide either qualitative decisions on parameter changes [71]; or quantitative decisions on parameter settings, viz. significant parameters that might develop defects in products [43], likely solutions based on certainty factors [8, 45] and fuzzy logic [112]. However, in order to provide accurate solutions to different cases, a huge pool of knowledge from various domains (molding engineer, machine manufacturer, heat transfer, rheology, material selection and others) is required for developing expert systems and their regular upgradation based on technology advancement.
Taguchi method for design of experiments The Taguchi method is used for design of experiments (DOE) with the motto to obtain desired results by establishing the best combinations of design parameters through reduced time, cost and number of experiments [106, 107]. It employs some orthogonal arrays taking into account the number of parameters and their levels as well as the number of trials to be conducted, and subsequently determines the best combinations of parameters based on the signal to noise ratio. As an example, Taguchi L27313 implies a system comprising 13 parameters each with 3 levels, which requires only 27 trial experiments, instead of 313, for obtaining the desired result. In the injection molding of polyethylene and styrene, Chang and Faison [10] used the Taguchi method for finding the best combination of input parameters so as to minimize the shrinkage in the components, and subsequently applied Analysis of Variance (ANOVA) for finding the relative contribution of each input parameter and the most significant parameters affecting the shrinkage. The application of similar procedures in injection molding are found in some other works, e.g., minimization of friction coefficient and strain [118], minimization of warpage and shrinkage [84], and minimization of shrinkage in polypropylene and polystyrene components [4]. Various forms of the Taguchi method were applied in many other works concerning with injection molding optimization, e.g., determination of the optimum parameters based on visual quality and shrinkage [111], evaluation of the process parameters by avoiding silver streak defects in injection molded automobile plastic bumpers [13], evaluation of
Int J Plast Technol
the influence of process parameters on the dimensional variation of injection molded ceramic [109], evaluation of process parameters by minimizing warpage in thin-shell plastic products [39], identification of parameters affecting shrinkage and warpage in thin-wall cellular phone covers [66], study of the effects of design parameters on shrinkage [15], determination of the optimal values of process parameters by minimizing warpage of a rectangular plate [108], study of the effect of input parameters on optimizing mechanical properties of recycled plastic parts [76–78], etc. In the recent application of the Taguchi method, Kang [46] optimized input parameters by minimizing shrinkage and warpage, Wang et al. [116] optimized input parameters in order to improve the compressive property of a brake booster valve body, and Karasu et al. [47] reduced the number of trial runs needed before mass production. Although the Taguchi method has been widely accepted in manufacturing including injection molding, certain hindrances are still associated with it. Firstly, the optimum level of a process parameter is to be determined from a prefixed set of levels. However, since the range of a parameter is continuous, the estimated optimum level of the parameter may not match with any of the chosen levels. Secondly, the selection of process parameters and their ranges and levels requires thorough understanding of the whole process, and as such meticulous experimental planning and interpretation are needed. Lastly, although the method reduces the number of experiments, still it requires certain experiments to be performed beforehand as per the chosen orthogonal array, thus causing wastage of resources and human errors.
Modern computational methods With the ever increasing intricacy of product design and hindrances developed in injection molding, a gradual paradigm shift towards modern sophisticated computational techniques has occurred in the last few years. It was essential to confront with the challenges posed by MIMO parameters, requirement of huge experimental analysis, and limited knowledge about the intra- and inter-relationships between input and output parameters. The modern computational techniques consist of hard computing and soft computing. The hard computing usually involves finite element or finite difference based simulation for obtaining acceptable input parameters, while the soft computing predicts the performance of injection molding by modeling the process environment through artificial intelligence techniques. Hard computing techniques Hard computing in injection molding is performed using either commercial software packages or optimization modules. Some commercial software packages were developed years ago by applying finite difference or finite element methods to the filling and post filling stages of injection molding [27, 38, 113]. Nowadays, many sophisticated software packages are available, which can simulate, analyze, and generate injection molding data based on selected input and output parameters. Numerous works using such software packages have been reported, e.g., Moldflow® in the fabrication of window frames using rice husk filled polyethylene composites [94], Moldflow® combined with Autodesk Inventorr for
Int J Plast Technol
modeling and simulating shallow and flat thin walled molds of a polymer composite [5, 6], Moldflow® to determine significant parameters affecting the shrinkage of molded components [69], Moldflow® to study material distribution and melt flow behaviour in sandwich molding process [92], Moldflow® combined with the Taguchi method to determine the optimal design parameters by minimizing the warpage of gas assisted molding components [14], and Moldflow® integrated with the Taguchi method to study the effects of processing parameters on the molding of ultra-thin wall polymer components [102]. Apart from that, Moldflow® was applied to some other similar works also, e.g., simulation of feedstock properties for powder injection molding of thermal management devices [86], feedstock properties and injection molding simulations of bimodal mixtures [49], measurements of powder-polymer mixture properties and their use in powder injection molding simulations [48], powder injection molding of ceramic engine components for transportation [61], effects of nanoparticle addition on processing of alloys [85], powder injection molding of parts for UAV engine components [74], etc. In the case of optimization modules, significant works include Kriging models to minimize the warpage in plastic cellular phones with reduced computational resources than that required by Moldflow® [35–37], translational propagation algorithm for minimizing the warpage in terms of process parameters in the molding of an automotive car handle [68], goal programming to optimize the performance of a plastic injection molding by formulating the product defect in terms of imprecise fuzzy values [3], sequential simplex method combined with Moldflow® to minimize the warpage and shrinkage of an automotive ventiduct grid in terms of process parameters [33], probability-based concept integrated with a Gaussian process surrogate model for improving the accuracy and robustness of injection molding with reduced computation [117], heat expansion method integrated with a 3D registration and the NewtonRaphson method for optimizing the size of shrinkmolds used in show manufacturing [44], etc. The hard computing of injection molding process involves finite element analysis requiring a huge computational time, which may not be available in a real-time shop floor. Moreover, the commercial software packages may fail to consider many machinespecific constraints, such as specifications and capabilities. In addition, the adjustment of multiple input values in the simulation and interpretation processes of these techniques requires expert knowledge of the injection molding process. Soft computing techniques The widely applied artificial intelligence in soft computing of injection molding includes artificial neural network, evolutionary algorithms and hybridized techniques. Artificial neural network (ANN) The primary advantage of applying ANN for optimization of injection molding process is its capability for mapping out nonlinear relationships between a devious set of input and output data through a learning procedure, as well as its ability to interpret MIMO parameters. A huge number of works applying ANN to injection molding has been reported in the specialized literature, which include the prediction of injection pressure
Int J Plast Technol
and injection time by training and testing ANN using process dataset generated by the C-Mold® software [120], injection pressure and injection time of metal molding process by a feed forward ANN integrated with the Gauss training method [119], nonlinear tensile modulus of injection molded polycarbonate samples as a function of process parameters [50], changes required in process parameters in order to achieve the desired final response in terms of dimensional accuracy [60], optimum control of the injection molding ram velocity using an ANN-based predictive learning controller [40], quality of molding process and machine parameters using an ANN-based flash monitoring system integrated with vibration monitoring and threshold prediction based on process parameter settings [16], process parameter based shrinkage using a forward mapping ANN model and shrinkage based optimal set of process parameters using a reverse mapping ANN model [73], optimum molding parameters for minimum defects in molded parts [89], better influence of process parameters on shrinkage in comparison to Moldflow® [115], etc. Due to their fast response and higher accuracy, the back propagation neural networks (BPNN) is also widely preferred in injection molding [11, 72], e.g., computation of mold complexity index based on difficulty in manufacturing injection molds [95], reduction in time requirement for planning and optimizing injection molding process parameters by supporting BPNN with experimental data [97], BPNN trained and tested with data generated by the Taguchi method for more accurate and effective prediction of warpage and shrinkage behaviour of injection molded thin walled parts than those of C-Mold® software and Taguchi method [67], BPNN trained and tested with data generated by Moldflow® for evaluating the optimal set of process parameters and warpage of injection molded automobile glove compartment by deriving relationships among them [123], etc. Although applied, ANN to injection molding has certain drawbacks. ANN works like a black box as a user has no control over output after feeding input and defining a general architecture of the molding process. It becomes difficult also to assess the accuracy and inter-relationships of input and output as peeking inside the hidden layers of ANN is not possible. Additionally, identifying the appropriate numbers of nodes and hidden layers is also a tough job, requiring the process to be retested if any data goes beyond its predictable range. Further, due to the requirement of increasing number of epochs with increasing nonlinearity of test data, BPNN models may consume extra time making them slower to train the process. Evolutionary algorithm (EA) In modern times, EAs have become influential in optimizing injection molding operations due to their capability of finding global optima without or little problem information and also to accommodate multiple objectives. Major works on injection molding process optimization using EAs include genetic algorithm (GA) for optimizing processing parameters [2, 29, 51] and for minimizing shrinkage in runner diameters [1], multi-objective GA for estimating runner size in a multiple cavity injection mold [124], optimization of cooling system layout in plastic injection molding [58, 62, 96], multiobjective GA for investigating performance characteristics of injection molding machine [31], GA to optimize input parameters so as to get the desired response in terms of dimensional shrinkage of injection molded plastic parts [23], GA-based intelligent conceptual mold layout design system to avoid human errors and design flaws in
Int J Plast Technol
producing optimum conditions [9], and particle swarm optimization (PSO) for estimating process parameters with the assumption that most of the injection molded specimens have sheet like morphology [125]. Although applied widely, EAs also have certain drawbacks. The effectiveness of an EA is dependent upon the selection of fitness functions, process parameter ranges and elite preservation mechanism. Additionally, sometime the non-deterministic features of EAs make the search process tedious and difficult to reach the optima. Hybrid approaches Due to various drawbacks of an individual algorithm like ANN or EA, sometimes two or more techniques are combined as a hybrid approach to exploit their specific advantages which would not have been possible with their sole usage. Many such hybrid techniques have been investigated to injection molding optimization also, like GA with the Taguchi method for minimizing warpage of molded components [98], GA with ANN for optimizing the initial process settings [81], genetic neural fuzzy system with 2-stage hybrid learning algorithm to predict product weight [63, 64], GA with BPNN to achieve the optimal quality in terms of shear stress [101] and to minimize volumetric shrinkage [100], the Taguchi method combined with ANN and GA to achieve the minimal single response output in terms of warpage in a bus ceiling lamp base [55] and to save energy by multi-objective optimization of process parameters [70], the Taguchi method combined with BPNN and GA to determine the set of data in multiple-input single-output (MISO) by optimizing product weight [17] and to achieve multi response outputs [20], the Taguchi method and response surface method combined with BPNN and GA for predicting mechanical properties by estimating an optimal set of process parameters [110], the Taguchi method with Moldflow® for finding the efficient frontier for a thin digital camera cover in a MIMO environment [24], Moldflow® and orthogonal experiment method integrated with BPNN and GA to determine the optimal set of process parameters for optimizing warpage and clamp force [122], the variable complexity method combined with BPNN and GA to mice manufacturing for optimizing multiple objectives [26], GA with response surface methodology to achieve the optimal single response in terms of warpage in thin shell plastic parts [54] and to minimize sink depth in thermoplastic components [75], simulated annealing with ANN to predict part warpage in runner system by optimizing the runner dimensions [121], GA with a gradient-based method to find the optimum process parameters [59], PSO with ANN to optimize process parameters [103], BPNN with the Taguchi method and Davidson-Fletcher-Powell method to determine multiple input process parameters in order to achieve the desired product weight as the single output [18, 19], the Latin hypercube sampling method combined with the Kriging method and multi-objective PSO to achieve a better Pareto frontier by reducing simulation cost [21], the response surface method integrated with Moldflow® and Lingo software to optimize process parameters with corresponding output of warpage and shrinkage [22], GA with the mode-pursuing sampling method for achieving minimum warpage [30], ANN and artificial bee colony algorithm to determine the set of process parameters by minimizing warpage of molded components [42], the Taguchi method and ANOVA combined with GA and PSO to obtain optimal input and output process parameters by decreasing the process variance [25], Moldex3D software
Int J Plast Technol
along with the radial basis function network based sequential optimization method in order to reduce warpage in the final plastic molded products with process parameters as design variables and short shot as design constraint [52], etc. Although various hybrid models are being put forward, a wide confusion has arisen nowadays among end users due to the huge variety and sheer quantity of those models as to which one is superior and commercially viable. Moreover, ability of those models, to be easily integrated with real time machines in shop floors and customized as per end users’ requirements, still leaves a big question mark on their practical utility. Additionally, barring a very few works, almost all others consider approximate techniques and optimize injection molding parameters as a single-objective function, either by adding weighted average of different criteria or by considering only a single criterion at a time, which is not an appropriate approach in the case of MIMO parameters of injection molding.
Real time techniques The need of the hour is to formulate a process with additional capabilities to customize it as per the requirements of users and to incorporate changes in the process due to real time optimization during injection molding operation. However, very few works have been conducted to formulate a novel real time process, which would not only precisely set the initial parameters but also optimize the process during the molding operation. Three such works are addressed here. Li et al. [65] proposed a GA-based real-time process optimization system, which not only caters to the initial injection molding parameter settings but also provides for online correction of any defect. In this system, trial runs are performed, using a fuzzy inference model to replace defective parameters from the GA generated initial process parameters, until a satisfactory product quality is obtained and the final fuzzy inference engine is fully developed for integration with the molding machine for real time process optimization and correction of defects. In the optimization process, however, some process parameters are predicted based on the geometric approximation of a rectangular edge gate plate. In addition, relations among various process parameters are formulated with some simplified assumptions. Although the system provides time savings compared to other simulating techniques, it has the drawbacks of using a search engine whose library data may not be capable for handling intricate molding components. Moreover, the optimization model considers only three quality indices, and their values are predicted by approximate methods which might not be accurate in certain cases. Stanek et al. [104] employed Moldflow® which offers automated process set up through a series of velocity and pressure phase configuration sequences, online process optimization through a robust process window formulated by an automated DOE, and production control through maintenance of the obtained optimized conditions. These facilities provide for better control of the injection molding process and minimization of the defects in components. This software seems to be a good solution for real time usage in the shop floor. However, no information is provided regarding the construction of the automated DOE in the optimization process. In a recent work, Zhao et al. [126] have developed an algorithm which intakes signals from the sensors fitted to the injection molding machine and continuously
Int J Plast Technol
monitors the process in real time identifying different stages of the process. Based on the data obtained from the technique, the filling-to-packing switchover point and packing time are optimized accordingly.
Future scope Based on the literature survey carried out, it is observed that an ample scope is still there for conducting computational research in the area of injection molding process optimization. The following are few suggestions in which directions research can be conducted in future: 1. To develop a procedure having enough capability for precise prediction of process parameters, not only during the initial machine setup but also during operation, with additional features to be easily integrated into the machine setup. Such procedures also should be less time consuming and ergonomically sound, so as to be feasible in any shop floor. 2. Instead of developing a host of different models, more work may be conducted in developing a few standard models which would have the capabilities to avoid any confusion at the level of end users. 3. Although different input parameters may influence different output responses, many researchers simplify the problem by considering only a few assorted parameters and desired quality indices. Since injection molding is a complex process with MIMO features, effective study may be performed by considering it as a constrained multi-objective optimization problem with MIMO parameters. 4. Since injection molding parameters and machine constraints may be continuous as well as discrete based on the type and capability of machines and processes, models may be framed with the provision to customize for both discrete and continuous variables as per the requirements of end users.
Conclusions Intricate design, stringent quality tolerances and time constraints in today’s industrial scenario definitely calls for optimization techniques with a capability for precise prediction of process and machine parameters used in injection molding. A review of different techniques, used over the years in optimization of injection molding parameters for obtaining the desired quality indices, is conducted in this article. It is discussed how the earlier trial and error methods, based on the experiences and instincts of molding operators, have been gradually improved for better and precise prediction of different input and output parameters by optimizing injection molding through formal techniques developed by exploiting the advancement of technology and soaring ideas. Further, various drawbacks, faced by the proposed approaches in their practical implementations, are also discussed. Finally, some future research scopes in the area of injection molding process optimization are suggested.
Int J Plast Technol Acknowledgments The first author would like to thank the Department of Science and Technology – Scientific and Engineering Research Board (DST-SERB), Govt. of India, for providing the funding for this research vide SERB Project No: SB/FTP/ETA-88/2013.
References 1. Alam K, Kamal MR (2004) Runner balancing by a direct genetic optimization of shrinkage. Polym Eng Sci 44(10):1949–1959 2. Alam MR, Lee KS, Rahman M, Zhang YF (2003) Process planning optimization for the manufacture of injection moulds using a genetic algorithm. Int J Comput Integr Manuf 16(3):181–191 3. Al-Refaie A, Li MH (2011) Optimizing the performance of plastic injection molding using weighted additive model in goal programming. Int J Fuzzy Syst Appl 1(2):43–54 4. Altan M (2010) Reducing shrinkage in injection moldings via the Taguchi, ANOVA and neural network methods. Mater Des 31(1):599–604 5. Azaman MD, Sapuan SM, Sulaiman S, Zainudin ES, Abdan K, Mansor MR (2013) An investigation of the processability of natural fibre reinforced 13 polymer composites on shallow and flat thin-walled parts by injection moulding process. Mater Des 50:451–456 6. Azaman MD, Sapuan SM, Sulaiman S, Zainudin ES, Khalina A (2013) Shrinkages and warpage in the processability of wood-filled polypropylene composite thin-walled parts formed by injection molding. Mater Des 52:1018–1026 7. Bourdon K (1989) The selection and setting of injection molding machines by means of process simulation. In Annual Technical Conference (ANTEC-1989), pages 308–311, New York. Society of Plastic Engineers (SPE) 8. Catic I, Slavica M, Sercer M, and Baric G (1996) Expert system aided troubleshooting in polymer engineering. In Annual Technical Conference (ANTEC-1996), pages 837–841, Indiana Polis. Society of Plastic Engineers (SPE) 9. Chan IW, Pinfold M, Kwong CK, Szeto WH (2014) Automation and optimisation of family mould cavity and runner layout design (FMCRLD) using genetic algorithms and mould layout design grammars. Comput Aided Des 47:118–133 10. Chang TC, Faison E (2001) Shrinkage behaviour and optimization of injection molded parts studied by the Taguchi method. Polym Eng Sci 41(5):703–710 11. Chen WC, Hsu SW (2007) A neural-network approach for an automatic LED inspection system. Expert Syst Appl 33(2):531–537 12. Chen Z, Turng LS (2005) A review of current developments in process and quality control for injection molding. Adv Polym Technol 24(3):165–182 13. Chen RS, Lee HH, Yu CY (1997) Application of Taguchi’s method on the optimal process design of an injection molded PC/PBT automobile bumper. Compos Struct 39(3):209–214 14. Chen CS, Cheng WS, Wang TS, Chien RD (2005) Optimum design of gas-assisted injection molding. J Reinf Plast Compos 24(15):1577–1586 15. Chen J, Li Q, Shen C, and Wang L (2005) Effects of process conditions on shrinkage of the injectionmolded part. In Annual Technical Conference (ANTEC-2005), vol 2 Boston, Society of Plastic Engineers (SPE), page 275 16. Chen J, Savage M, Zhu JJ (2008) Development of artificial neural network- based in-process mixedmaterial-caused flash monitoring (ANN-IPMFM) system in injection molding. Int J Adv Manuf Technol 36(1–2):43–52 17. Chen WC, Lai TT, Fu GL, and Chen CT (2008) A systematic optimization approach in the MISO plastic injection molding process. In International Conference on Service Operations and Logistics, and Informatics (SOLI-2008), volume 2, pages 2741–2746. IEEE 18. Chen WC, Wang MW, Fu GL, and Chen CT (2008) Optimization of plastic injection molding process via Taguchis parameter design method, BPNN, and DFP. In International Conference on Machine Learning and Cybernetics, volume 6, pages 3315–3321. IEEE 19. Chen CP, Chuang MT, Hsiao YH, Yang YK, Tsai CH (2009) Simulation and experimental study in determining injection molding process parameters for thin-shell plastic parts via design of experiments analysis. Expert Syst Appl 36(7):10752–10759 20. Chen WC, Fu GL, Tai PH, Deng WJ (2009) Process parameter optimization for MIMO plastic injection molding via soft computing. Expert Syst Appl 36(2):1114–1122
Int J Plast Technol 21. Chen W, Zhou XH, Wang HF, Wang W (2010) Multi-objective optimal approach for injection molding based on surrogate model and particle swarm optimization algorithm. J Shanghai Jiaotong Univ 15:88– 93 22. Chen WL, Huang CY, Hung CW (2010) Optimization of plastic injection molding process by dual response surface method with non-linear programming. Eng Comput 27(8):951–966 23. Chen CC, Su PL, Chiou CB, Chiang KT (2011) Experimental investigation of designed parameters on dimension shrinkage of injection molded thinwall part by integrated response surface methodology and genetic algorithm: A case study. Mater Manuf Process 26(3):534–540 24. Chen WL, Huang CY, Huang CY (2013) Finding efficient frontier of process parameters for plastic injection molding. J Ind Eng Int 9(1):25 25. Chen WC, Liou PH, and Chou SC (2014) An integrated parameter optimization system for MIMO plastic injection molding using soft computing. International Journal of Advanced Manufacturing Technology, pages 1–10 26. Cheng J, Liu Z, Tan J (2013) Multi-objective optimization of injection molding parameters based on soft computing and variable complexity method. Int J Adv Manuf Technol 66(5–8):907–916 27. Chiang HH, Hieber CA, Wang KK (1991) A unified simulation of the filling and post-filling stages in injection molding. Part II: experimental verification. Polym Eng Sci 31(2):125–139 28. Dang XP (2014) General frameworks for optimization of plastic injection molding process parameters. Simul Model Pract Theory 41:15–27 29. Deng YM, Lam YC, Britton GA (2004) Optimization of injection moulding conditions with userdefinable objective functions based on a genetic algorithm. Int J Prod Res 42(7):1365–1390 30. Deng YM, Zhang Y, Lam YC (2010) A hybrid of mode-pursuing sampling method and genetic algorithm for minimization of injection molding warpage. Mater Des 31(4):2118–2123 31. Ding LP, Tan JR, Wei Z, Chen WL, Gao Z (2011) Multi-objective performance design of injection molding machine via a new multi-objective optimization algorithm. Int J Innov Comput, Inf Control 7(7):3939–3949 32. Duverlie P, Castelain JM (1999) Cost estimation during design step: parametric method versus case based reasoning method. Int J Adv Manuf Technol 15(12):895–906 33. Farshi B, Gheshmi S, Miandoabchi E (2011) Optimization of injection molding process parameters using sequential simplex algorithm. Mater Des 32(1):414–423 34. Feng J, Wang L, Zhang RY, Wu JJ, Wang CY, Yang MB, Fu XR (2014) Formation of double skin-core orientated structure in injection-molded polyethylene parts: effects of ultra-high molecular weight polyethylene and temperature field. J Polym Res 21(5):1–14 35. Gao Y, Wang X (2008) An effective warpage optimization method in injection molding based on the Kriging model. Int J Adv Manuf Technol 37(9–10):953–960 36. Gao Y, Wang X (2009) Surrogate-based process optimization for reducing warpage in injection molding. J Mater Process Technol 209(3):1302–1309 37. Gao Y, Turng LS, Wang X (2008) Adaptive geometry and process optimization for injection molding using the Kriging surrogate model trained by numerical simulation. Adv Polym Technol 27(1):1–16 38. Hieber CA, Shen SF (1980) A finite-element/finite-difference simulation of the injection-molding filling process. J Non-Newtonian Fluid Mech 7(1):1–32 39. Huang MC, Tai CC (2001) The effective factors in the warpage problem of an injection-molded part with a thin shell feature. J Mater Process Technol 110(1):1–9 40. Huang SN, Tan KK, Lee TH (2004) Neural-network-based predictive learning control of ram velocity in injection molding. IEEE Trans Syst Man Cybern Part C Appl Rev 34(3):363–368 41. Huang J, Lu X, Zhang G, Qu J (2014) Study on the rheological, thermal and mechanical properties of thermoplastic polyurethane/poly (butylene terephthalate) blends. Polym Test 36:69–74 42. Iniesta AA, Alcaraz JLG, Borbón MR (2013) Optimization of injection molding process parameters by a hybrid of artificial neural network and artificial bee colony algorithm. Rev Fac Ing Univ Antioquia 67: 43–51 43. Jan TC, OBrien KT (1993) A user-friendly, interactive expert system for the injection moulding of engineering thermoplastics. Int J Adv Manuf Technol 8(1):42–51 44. Jeng YR, Liu DS, Yau HT (2013) Fast numerical algorithm for optimization mold shape of direct injection molding process. Mater Manuf Process 28(6):689–694 45. Jong WR and Hsu SS (1997) An integrated expert system for injection molding process. In Annual Technical Conference (ANTEC-1997), Tornto, Society of Plastic Engineers (SPE) pages 541–544 46. Kang GJ (2014) Process factor optimization for reducing warpage and shrinkage in injection molding using design of experiments. Appl Mech Mater 541:359–362
Int J Plast Technol 47. Karasu MK, Cakmakci M, Cakiroglu MB, Ayva E, Demirel-Ortabas N (2014) Improvement of changeover times via Taguchi empowered SMED/case study on injection molding production. Measurement 47:741–748 48. Kate KH, Onbattuvelli VP, Enneti RK, Lee SW, Park SJ, Atre SV (2012) Measurements of powderpolymer mixture properties and their use in powder injection molding simulations for aluminum nitride. JOM 64(9):1048–1058 49. Kate KH, Enneti RK, Onbattuvelli VP, Atre SV (2013) Feedstock properties and injection molding simulations of bimodal mixtures of nanoscale and microscale aluminum nitride. Ceram Int 39(6):6887– 6897 50. Kenig S, Ben-David A, Omer M, Sadeh A (2001) Control of properties in injection molding by neural networks. Eng Appl Artif Intell 14(6):819–823 51. Kim SJ, Lee KI, and Kim YI (1996) Optimization of injection-molding conditions using genetic algorithm. In SPIE Fourth International Conference on Computer-Aided Design and Computer Graphics, volume 2644, pages 173–180 52. Kitayama S, Onuki R, Yamazaki K (2014) Warpage reduction with variable pressure profile in plastic injection molding via sequential approximate optimization. Int J Adv Manuf Technol 72(5–8):827–838 53. Krauss VA, Pires EN, Klein AN, Fredel MC (2005) Rheological properties of alumina injection feedstocks. Mater Res 8(2):187–189 54. Kurtaran H, Erzurumlu T (2006) Efficient warpage optimization of thin shell plastic parts using response surface methodology and genetic algorithm. Int J Adv Manuf Technol 27(5–6):468–472 55. Kurtaran H, Ozcelik B, Erzurumlu T (2005) Warpage optimization of a bus ceiling lamp base using neural network model and genetic algorithm. J Mater Process Technol 169(2):314–319 56. Kwong CK, Smith GF (1998) A computational system for process design of injection moulding: combining a blackboard-based expert system and a case-based reasoning approach. Int J Adv Manuf Technol 14(5):350–357 57. Kwong CK, Smith GF, Lau WS (1997) Application of case based reasoning injection moulding. J Mater Process Technol 63(1):463–467 58. Lam YC, Zhai LY, Tai K, Fok SC (2004) An evolutionary approach for cooling system optimization in plastic injection moulding. Int J Prod Res 42(10):2047–2061 59. Lam YC, Deng YM, Au CK (2006) A GA/gradient hybrid approach for injection moulding conditions optimisation. Eng Comput 21(3):193–202 60. Lau HCW, Ning A, Pun KF, Chin KS (2001) Neural networks for the dimensional control of molded parts based on a reverse process model. J Mater Process Technol 117(1):89–96 61. Lenz J, Enneti RK, Onbattuvelli VP, Kate KH, Martin R, Atre SV (2012) Powder injection molding of ceramic engine components for transportation. JOM 64(3):388–392 62. Li CG and Wu Y (2010) Evolutionary optimization of plastic injection mould cooling system layout design. In International Conference onIntelligent System Design and Engineering Application (ISDEA2010), vol 1 I.E. Press, pages 693–696 63. Li E, Jia L, Yu J (2002) A genetic neural fuzzy system-based quality prediction model for injection process. Comput Chem Eng 26(9):1253–1263 64. Li E, Jia L, Yu J (2004) A genetic neural fuzzy system and its application in quality prediction in the injection process. Chem Eng Commun 191(3):335–355 65. Li D, Zhou H, Zhao P, Li Y (2009) A realtime process optimization system for injection molding. Polym Eng Sci 49(10):2031–2040 66. Liao SJ, Chang DY, Chen HJ, Tsou LS, Ho JR, Yau HT, Su YC (2004) Optimal process conditions of shrinkage and warpage of thinwall parts. Polym Eng Sci 44(5):917–928 67. Liao SJ, Hsieh WH, Wang JT, Su YC (2004) Shrinkage and warpage prediction of injectionmolded thinwall parts using artificial neural networks. Polym Eng Sci 44(11):2029–2040 68. Liao X, Yan X, Xia W, and Luo B (2010) A fast optimal Latin hypercube design for Gaussian process regression modelling. In Third International Workshop onAdvanced Computational Intelligence (IWACI-2010), pages 474–479. IEEE 69. Lotti C, Ueki MM, Bretas RES (2002) Prediction of the shrinkage of injection molded iPP plaques using artificial neural networks. J Injection Molding Technol 6(3):157–176 70. Lu NY, Gong GX, Yang Y, Lu JH (2012) Multi-objective process parameter optimization for energy saving in injection molding process. J Zhejiang Univ Sci A 13(5):382–394 71. Luong LH, Chan FT, and Ip RW (1997) A knowledge-based system for fault diagnosis in plastic injection moulding. In International Conference of Manufacturing Automation (ICMA’97), Hong Kong, volume 2, pages 947–952
Int J Plast Technol 72. Maier HR, Dandy GC (1998) Understanding the behaviour and optimising the performance of backpropagation neural networks: an empirical study. Environ Model Softw 13(2):179–191 73. Manjunath PG, Krishna P (2012) Prediction and optimization of dimensional shrinkage variations in injection molded parts using forward and reverse mapping of artificial neural networks. Adv Mater Res 463:674–678 74. Martin R, Vick M, Enneti RK, Atre SV (2013) Powder injection molding of ceria-stabilized, zirconiatoughened mullite parts for UAV engine components. JOM 65(11):1388–1394 75. Mathivanan D, Parthasarathy NS (2009) Sink-mark minimization in injection molding through response surface regression modeling and genetic algorithm. Int J Adv Manuf Technol 45(9–10):867–874 76. Mehat NM, Kamaruddin S (2011) Investigating the effects of injection molding parameters on the mechanical properties of recycled plastic parts using the Taguchi method. Mater Manuf Process 26(2): 202–209 77. Mehat NM, Kamaruddin S (2011) Multi-response optimization of injection moulding processing parameters using the Taguchi method. Polym-Plast Technol Eng 50(15):1519–1526 78. Mehat NM, Kamaruddin S (2011) Optimization of mechanical properties of recycled plastic products via optimal processing parameters using the Taguchi method. J Mater Process Technol 211(12):1989–1994 79. Menges G, Thienel P (1977) Pressurespecific volumetemperature behavior of thermoplastics under normal processing conditions. Polym Eng Sci 17(10):758–763 80. Mok SL, Kwong CK, Lau WS (1999) Review of research in the determination of process parameters for plastic injection molding. Adv Polym Technol 18(3):225–236 81. Mok SL, Kwong CK, Lau WS (2001) A hybrid neural network and genetic algorithm approach to the determination of initial process parameters for injection moulding. Int J Adv Manuf Technol 18(6):404– 409 82. Murakami T, Okamura T and Tanizaki N (1993) An operation assisted system for injection molding machines based on constraints processing. In Twelveth Triennial World Congress of the International Federation of Automatic Control, volume 4, pages 455–460 83. Nagarsheth PS (1989) An integrated system approach for injection molded plastic parts from design to fabrication. In Annual Technical Conference (ANTEC-1989), pages 220–223. Society of Plastic Engineers (SPE) 84. Oktem H, Erzurumlu T, Uzman I (2007) Application of Taguchi optimization technique in determining plastic injection molding process parameters for a thin-shell part. Mater Des 28(4):1271–1278 85. Onbattuvelli VP (2010) The effects of nanoparticle addition on the processing, structure and properties of SiC and AlN. PhD thesis, Oregon State University, USA. 86. Onbattuvelli VP, Vallury S, McCabe T, Park SJ, Atre SV (2010) Properties of SiC and AlN feedstocks for the powder injection moulding of thermal management devices. Powder Injection Moulding Int 4(3): 64–70 87. Onbattuvelli VP, Enneti RK, Sohn SB, McCabe T, Park SJ, Atre S (2011) Micro-powder injection moulding of barium titanate. Powder Injection Moulding Int 5(4):59–65 88. Osswald T, Hernández-Ortiz JP (2006) Polymer processing. Hanser Publishers, Munich 89. Özek C, Çelık YH (2012) Calculating molding parameters in plastic injection molds with ANN and developing software. Mater Manuf Process 27(2):160–168 90. Pandelidis I, Zou Q (1990) Optimization of injection molding design. Part II: molding conditions optimization. Polym Eng Sci 30(15):883–892 91. Pandelidis I, Zou Q and Lingard TJ (1988) Optimization of gate location and operational molding conditions for injection molding. In Annual Technical Conference (ANTEC-1988), volume 46, pages 18–20. Society of Plastic Engineers (SPE) 92. Patcharaphun S, Mennig G (2006) Simulation and experimental investigations of material distribution in the sandwich injection molding process. Polym-Plast Technol Eng 45(6):759–768 93. Quach A, Simha R (1971) Pressurevolumetemperature properties and transitions of amorphous polymers; polystyrene and poly (orthomethylstyrene). J Appl Phys 42(12):4592–4606 94. Rahman WA, Sin LT, Rahmat AR (2008) Injection moulding simulation analysis of natural fiber composite window frame. J Mater Process Technol 197(1):22–30 95. Raviwongse R, Allada V (1997) Artificial neural network based model for computation of injection mould complexity. Int J Adv Manuf Technol 13(8):577–586 96. Ribeiro CJ and Viana JC (2011) Optimization of injection moulded polymer automotive components. In Chiaberge, M., editor, New Trends and Developments in Automotive System Engineering, ISBN: 978953-307-517-4 97. Sadeghi BHM (2000) A BP-neural network predictor model for plastic injection molding process. J Mater Process Technol 103(3):411–416
Int J Plast Technol 98. Sahu R, Yao D and Kim B (1997) Optimal mold design methodology to minimize warpage in injection molded parts. In Annual Technical Conference (ANTEC-1997), volume 3, pages 3308–3313, Toronto. Society of Plastics Engineers (SPE) 99. Shelesh-Nezhad K, Siores E (1997) An intelligent system for plastic injection molding process design. J Mater Process Technol 63(1):458–462 100. Shen C, Wang L, Li Q (2007) Optimization of injection molding process parameters using combination of artificial neural network and genetic algorithm method. J Mater Process Technol 183(2):412–418 101. Shi F, Lou ZL, Zhang YQ, Lu JG (2003) Optimisation of plastic injection moulding process with soft computing. Int J Adv Manuf Technol 21(9):656–661 102. Song MC, Liu Z, Wang MJ, Yu TM, Zhao DY (2007) Research on effects of injection process parameters on the molding process for ultra-thin wall plastic parts. J Mater Process Technol 187-188: 668–671 103. Spina R (2006) Optimisation of injection moulded parts by using ann-pso approach. J Achiev Mater Manuf Eng 15(1–2):146–152 104. Stanek M, Manas D, Manas M, Suba O (2011) Optimization of injection molding process. Int J Math Comput Simul 5(5):413–421 105. Szücs A, Belina K (2012) Rheological and thermal analysis of the filling stage of injection moulding. Express Polym Lett 6(8):672–679 106. Taguchi G, Konishi S (1987) Orthogonal arrays and linear graphs: tools for quality engineering. American Supplier Institute, Allen Park 107. Taguchi G, Elsayed EA and Hsiang TC (1989) Quality Engineering in Production Systems. McGrawHill College 108. Tang SH, Tan YJ, Sapuan SM, Sulaiman S, Ismail N, Samin R (2007) The use of Taguchi method in the design of plastic injection mould for reducing warpage. J Mater Process Technol 182(1):418–426 109. Tseng WJ (1998) Statistical analysis of process parameters influencing dimensional control in ceramic injection molding. J Mater Process Technol 79(1):242–250 110. Tzeng CJ, Yang YK, Lin YH, Tsai CH (2012) A study of optimization of injection molding process parameters for SGF and PTFE reinforced PC composites using neural network and response surface methodology. Int J Adv Manuf Technol 63(5–8):691–704 111. Väätäinen O, Järvelä P, Valta K (1994) The effect of processing parameters on the quality of injection moulded parts by using the Taguchi parameter design method. Plast Rubber Compos Process Appl 21(4): 211–217 112. Vagelatos GA, Rigatos GG, Tzafestas SG (2001) Incremental fuzzy supervisory controller design for optimizing the injection molding process. Expert Syst Appl 20(2):207–216 113. Wang VW, Hieber CA, Wang KK (1986) Dynamic simulation and graphics for the injection molding of three-dimensional thin parts. J Polym Eng 7(1):21–45 114. Wang H, Ruan XY, Zhou XH (2003) Research on injection mould intelligent cost estimation system and key technologies. Int J Adv Manuf Technol 21(3):215–222 115. Wang R, Zeng J, Feng X, Xia Y (2013) Evaluation of effect of plastic injection molding process parameters on shrinkage based on neural network simulation. J Macromol Sci B 52(1):206–221 116. Wang YQ, Kim JG, Song JI (2014) Optimization of plastic injection molding process parameters for manufacturing a brake booster valve body. Mater Des 56:313–317 117. Xia W, Luo B, Liao XP (2011) An enhanced optimization approach based on Gaussian process surrogate model for process control in injection molding. Int J Adv Manuf Technol 56(9–12):929–942 118. Yang YK (2006) Optimization of injection-molding process of short glass fiber and polytetrafluoroethylene reinforced polycarbonate composites via design of experiments method: a case study. Mater Manuf Process 21(8):915–921 119. Yarlagadda PK (2002) Development of an integrated neural network system for prediction of process parameters in metal injection moulding. J Mater Process Technol 130:315–320 120. Yarlagadda PK, Khong CAT (2001) Development of a hybrid neural network system for prediction of process parameters in injection moulding. J Mater Process Technol 118(1):109–115 121. Yen C, Lin JC, Li W, Huang MF (2006) An abductive neural network approach to the design of runner dimensions for the minimization of warpage in injection mouldings. J Mater Process Technol 174(1):22– 28 122. Yin F, Mao H, Hua L (2011) A hybrid of back propagation neural network and genetic algorithm for optimization of injection molding process parameters. Mater Des 32(6):3457–3464 123. Yin F, Mao H, Hua L, Guo W, Shu M (2011) Back Propagation neural network modeling for warpage prediction and optimization of plastic products during injection molding. Mater Des 32(4):1844–1850
Int J Plast Technol 124. Zhai M, Lam YC, Au CK (2009) Runner sizing in multiple cavity injection mould by non-dominated sorting genetic algorithm. Eng Comput 25(3):237–245 125. Zhao P, Zhou H, Li Y, Li D (2010) Process parameters optimization of injection molding using a fast strip analysis as a surrogate model. Int J Adv Manuf Technol 49(9–12):949–959 126. Zhao P, Zhou H, He Y, Cai K, Fu J (2014) A nondestructive online method for monitoring the injection molding process by collecting and analyzing machine running data. Int J Adv Manuf Technol 72(5–8): 765–777