Figures
Tables
Symbols
Acronyms
1. Introduction
1.1. Project background
1.2. Project aims and problem statement
1.3. Scope of the project
1.4. Hypotheses
2. Literature Review and Discussions
2.1. Review on work about process control variables
2.2. Review on previous work about process optimisation
2.3. Review on optimisation techniques used for injection moulding
2.3.1. Computer simulation
2.3.2. Integration of computer simulation and optimisation algorithms
2.4. Conclusion and discussions
3. Methodology
3.1. Sampling control factors
3.2. Data collection and data analysis
3.3. Framework of proposed approach
3.4. Virtual Simulation Model (VSM)
3.4.1. Virtual environment
3.4.2. Pre-processing
3.4.3. Post-processing
3.4.4. Application of VSM
3.5. Coupled Optimisation Model (COM)
3.5.1. DOE
3.5.2. SAPSO-based ANN
4. Requirements of the project
4.1. Timeframe
4.2. Supervision
4.3. Facilities
4.4. Estimated cost
5. Significance of the project
6. Conclusion and future work
7. References
Appendix A Table of quality response comparison of different approach
Appendix B Alias Relationship for Fractional 25 Designs
Appendix C 25-1 Design
Appendix D Timeframe of the project
Figure.1. Flow diagram of proposed CIOS
Figure.2. Structure of VSM
Figure.3. Flow diagram of COM
Figure.4. SAPSO logical sequence
Table.1. Process control factors and levels
illustration not visible in this excerpt
illustration not visible in this excerpt
PROCESS OPTIMISATION OF INJECTION MOULDING USING COMPUTER SIMULATION
Injection moulding is one of the most important processes for manufacturing commercial items due to its advantages such as short product cycles and easily moulded complicated shapes. It can be basically divided into four stages: plasticisation, filling, packing and cooling, and ejection (Rosato & DV Rosato, 1995). Of four stages, filling phase includes the process parameters that influence on the product quality as well as the productivity. The high demand of three dimensional parts for new application areas in different industries has promoted the development of its efficiency. Previously, trial-and-error approach based on the engineer’s experience and intuition was employed to identify the significant parameter setting. However, it is costly, time-consuming and inappropriate for complex manufacturing processes (Lam, Deng & Au 2004). Manzione (ed. 1987) suggested that a key to improve the capacity of handling important aspects of injection moulding process is the utilisation of computer simulation. However, there remains a gap between simulation results and optimisation objectives owing to the complexity and dynamic interaction of the injection moulding process. Moreover, the issues regarding materials, product design and quality, production time, and cost effectiveness make it more complicated. Accordingly, the requirement of more flexible optimisation method draws attention of many researchers. The ideal method is to check and measure the effects of process parameters in the real production line, but it has many disadvantages such as time-consuming, high-cost and inaccuracy (Bikas, Pantelelis & Kanarachos 2002, p.112). Therefore, in most cases, optimisation algorithms are integrated with computer simulation to bridge the gap between simulation and reality.
The purpose of this research is to develop a new sophisticated computer integration system to optimise the injection moulding process using computer simulation and optimisation algorithms. In dealing with this project, the study will focus on the three sub-problems:
i. The first sub-problem is to determine the process design variables that are dominant in the injection moulding process.
ii. The second sub-problem is to decide on the optimisation algorithms that are able to quickly search the best results for the moulding process performance.
iii. The third sub-problem is to discover the most efficient computer simulation package for injection moulding.
To eliminate irrelevancies to the identified objective and the research problem, the scope of the project is clearly stated as follows:
i. The study will be limited to plastic injection moulding process for most of manufacturing enterprises use plastic as a major material for both household and industrial products.
ii. The study will not attempt to optimise all stages of the injection moulding process but filling stage as it is critical.
iii. The study will be limited to the dominant control process design variables of the filling phase.
iv. The study will not evaluate the economic factors of the injection moulding process.
v. The study will not appraise simulation software used in other industries.
i. The appropriate combination of process control factors influence on the production efficiency and product quality.
ii. The preferred simulation package can facilitate the optimisation purpose, giving more flexibility in understanding and decision making of the injection moulding process.
iii. The optimisation algorithms designated are able to effectively search the global optimum solution for the entire process.
In the literature, a large number of papers regarding optimisation of injection moulding process can be found. The word “optimisation” in different papers means making an improvement of process factors affecting the product quality and productivity within the feasible limits that satisfy all specified constraints. These factors include design variables relating with process parameter settings and process design. Besides, the use of optimisation methods and computer simulations plays an important role for optimising the plastic injection moulding process.
A number of researches have been carried out with different aspects of moulding quality to optimise the process parameters that comprise not only physical and chemical parameters such as melt temperature, mould temperature, injection velocity, injection pressure, and packing pressure, but also time parameters such as injection time, packing time and cooling time (Chen et al. 2007; Kurtaran & Erzurumlu 2006). In the past studies, different combination of process parameters had been used. Wu and Liang (2005) used six control process parameters: melt temperature, mould temperature, packing pressure, injection velocity, injection acceleration and packing time to examine the effects of parameters on the width of weld line of an injection moulded plastic part. Chen, Wang, Fu, & Chen (2008) employed four process control parameters: Injection Time, Velocity Pressure Switch Position, Packing Pressure, and Injection Velocity to identify the optimum parameter setting for the moulded plastic push-button housing piece under consideration of single quality response: product weight. Chiang and Chang (2006) applied four process factors: melt temperature, mould temperature, injection pressure, and injection time to determine the optimal process parameter for a thin-shell plastic part with multiple quality characters. Zhou, & Turng (2007) exploited seven process parameters: melt temperature, mould temperature, packing pressure, injection time, packing time, cooling time, and velocity/pressure switch-over (V/P) by volume to optimise the volume shrinkage for Plexiglass optical lens with different thickness at the centre and outer rim.
In terms of process design optimization, a series of studies have been found, which focus on different perceptive of the moulding process design: gate location, runner system, cooling system, and part geometry. Subramanian, Tingyu, & Seng, (2005) attempted to optimise the gate location in order to minimize the warpage of the injection moulded part, analysing the distortion of the legs and reference pads in a plastic optical housing for a CD optical pickup. Zhai, Lam, & Au (2006) tried to specify the optimum solutions of gate locations and runner size of a multi gated moulding process, considering two critical affects: weld line and warpage. Lee, & Lin (2006) targeted to determine the optimal runner and gating system parameters for a multi-cavity injection mould in order to minimise the warp formation. Deng, Zheng, & Lu (2008) aimed to optimise multi-class variables including process parameter, gating system, runner system, cooling system and part geometry under consideration of multi-response moulding qualities: part warpage, weld lines, air traps and so on. Reviewing the above papers, one can conclude that appropriate combination of process parameters and process design variables is a very important issue to achieve the objective quality requirements.
Concerning the optimisation methods used for the injection moulding process, there are a significant number of papers in the literature, some of which focus on single criterion while another on multiple criteria. The techniques used for optimisation issue include traditional methods and artificial intelligence methods. Previously, trial-and-error method was used to determine the process parameter settings, depending on the experience and intuition of the engineers. However, it has many drawbacks and is unsuitable for complex processes to get the actual optimum results (Lam et al. 2004). Next, Taguchi’s parameter design method has been used for experimental design and process improvement as a central one in many papers. Liao et al. (2004) exploited L27 orthogonal array experimental tests based on Taguchi’s method to optimise the process conditions of a thin-wall injection moulding of a cellular phone cover made of amorphous PC/ABS resin plastics, taking into account the interaction effects between process parameters, and quality targets: shrinkage and warpage. Oktem, Erzurumlu & Uzman (2007) utilised Taguchi optimization tool to determine plastic injection moulding process parameters for thin shell parts. In this case, material property, part design, and injection-moulding conditions were considered as the factors or variables needed to be changed to minimize shrinkage. Nonetheless, Taguchi’s method is only able to find the best combination of parameter level that includes discrete values so it falls short when the parameter values are continuous (Chen et al. 2008). To deal with the continuous parameter variables, Artificial Neural Networks (ANNs) has been introduced as an alternative means (Chiu et al. 1997). Chen et al. (2009) claims that ANNs can map the relationship between input factors and output responses and it has different sub-categories such as Back Propagation Neural Network (BPNN), General Regression Neural Network (GRNN). However, it finds difficult to search the final optimal variables. Later on, the robust optimisation method Genetic Algorithm (GA) has been extensively used for random searching of the global optimum value in large dimensional space without being trapped in the local optimum (Tseng 2006; Shen, Wang & Li 2007). Lam, Den & Au (2006) argues that GA is opportunistic but not deterministic alone. Therefore, many researchers combine different nature of optimisation methods to approach the optimal design and process conditions. Chen et al. (2008) integrated Taguchi’s parameter design method, BPNN, and Davidon-Fletcher-Powell (DFP) method to optimize the process parameter settings of plastic injection moulding, considering single product quality (weight). Furthermore, Chen et al. (2009) combined Taguchi’s parameter design method, BPNN, GA and engineering optimization concepts to optimize the process parameters, for an experiment on a standard plastic piece, under multiple-input multiple-out (MIMO) consideration. They claim that their approach can effectively help engineers determine optimal process parameter settings and achieve competitive advantages of product quality and the costs.
On the other hand, some researchers used different optimisation methods for their particular objectives. Su & Chang (2000) proposed the combination of Neural Network (NN) and Simulated Annealing (SA) to optimise the parameter design of the injection moulding process. Loera et al. (2008) used Data Envelopment Analysis (DEA) optimisation strategy for a thermoplastic injection moulding, considering multiple criteria: design and process variables to meet several performances. Upon the application of their approach, they discussed the optimum parameter settings for the moulding of rear automotive lamps to control the part dimension and surface properties (aesthetic) using seven control process inputs. Moreover, Park & Ahn (2004) used DOE to reduce the cooling time and the injection pressure. This research proved that DOE is useful to discover the cause and effect relationship between the inputs and outputs of a process, and to determine the optimal process parameters with fewer testing trials. Furthermore, Deng (2008) coupled Multi-Response Signal to Noise Ratio (MRSN Ratio) with Analytic Network Process (ANP) to optimize the process parameters of multiple-response process in order to achieve higher production efficiency. MRSN is one of reformed Taguchi’s methods for multiple-quality response process parameter optimization and ANP is a systemic process that applies ratio scales to evaluate internal relationship of dimensions, criterions, and alternatives. They used five control parameters: mould temperature, pipe temperature, injection velocity, injection pressure and packing time to achieve the targeted four quality responses: weight, length (dimensional), warpage & shrinkage (surface property) for case study of a bottom cover of polypropylene Modem. In addition, Da & Xiurun (2004) presented SAPSO-based ANN method using particle swarm optimisation (PSO) with simulated annealing (SA) in order to achieve global optimum. Their approach modelled the relationship between confining pressure, peak stress and corresponding strain, and showed its flexibility in escaping local optimum. For that reason, it can be concluded that proper selection and/or combination of different techniques to meet quality criterion and production efficiency becomes one of the most important factors in optimisation of injection moulding process.
