روش بهبود فرایند چند بعدی یکپارچه برای سیستم های تولیدی
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|16837||2003||21 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Computers & Industrial Engineering, Volume 44, Issue 4, April 2003, Pages 673–693
In this paper, an integrated multidimensional process improvement methodology (IMPIM) is formulated to address the yield management, process control and cost management problems of a manufacturing system. Simulation is used as a platform to implement the integrated multidimensional process methodology by incorporating the productivity, quality and cost dimension in a unified, systematic and holistic manner. Total Quality Management (TQM) addresses the quality parameters and Activity-Based Costing is used to manage the cost dimension of the system. Discrete event simulation is then used as a platform to perform process reengineering (Business Process Reengineering) and process improvement (TQM). The general implementation framework of the IMPIM is given with a step-by-step explanation. A conceptual discussion is also provided for the integrated methodology. The generic IMPIM is then formulated and the detailed implementation procedures for two case studies are compared with the generic methodology.
2.1. General conceptual framework for the methodology In the IMPIM, simulation is used as a platform to implement process reengineering (BPR) and process improvement (TQM) (Leach, 1996 and Tomasek, 1992) (see Fig. 2). ABC was proposed by Cooper and Kaplan in their papers to measure and control the non-value-added costs better, such as quality costs (Cooper, 1988a, Cooper, 1988b, Cooper, 1989a, Cooper, 1989b, Cooper and Kaplan, 1988 and Kaplan and Cooper, 1998). Back-propagation neural networks was suggested as a new way to solve engineering problems (Rumelhart & McClelland, 1981). Montgomery gives a very good introduction of various Statistical Process Control (SPC) techniques (Montgomery, 1997). ABC, neural network and SPC are used to carry out BPR and TQM in the simulation environment. ABC and control charts can be easily programmed into the simulation software because they are process-oriented and time-based techniques. Artificial neural networks (Azouzi and Guillot, 1996, Gen et al., 1996, Hurrion, 1997, Wang, 1993 and Welstead, 1994) (artificial intelligent techniques) can be used to build the metamodel (a simple mathematical representation of the system) of the simulated system. Discrete event simulation can be considered a foundation to ‘glue’ together the productivity (traditional simulation), quality (TQM) and cost (ABC) dimension into a unified multidimensional systems model. Coding is relatively manageable with the modular components provided by the specialized simulation packages, such as ARENA, Witness and Promodel. These Simulation packages provide user-friendly syntax and coding standards for solving optimisation problems (Banks, Carson, & Nelson 1996). Time required to generate of neural network codes is also reduced significantly by having NeuralWork software that can generate the codes automatically based on the parameters inputted.The general implementation framework of the integrated multidimensional process methodology is illustrated in Fig. 3. At first, a conventional simulation model is built to address the productivity problem of the manufacturing system. SPC techniques or quality control charts can be programmed into the simulation software to address the quality dimension in the system. Next, an ABC system (Bharara and Lee, 1996, Jorgenson and Enkerlin, 1992 and Remer et al., 1992) is incorporated into the system which includes a process quality cost model (British Standard, 1992). An artificial intelligence metamodel (Friedman, 1996; Hurrion, 1997; Yu & Popplewell, 1994), DoE or scenario studies (Bezold, 1996) are/is constructed to support the decision making process of the simulation study and to provide an on-line learning mechanism.2.2. Formulation of IMPIM A novel IMPIM is proposed in this paper to solve the productivity, quality and cost problems of a manufacturing system in a methodical and unified manner (see Fig. 4). This methodology is a computer-based or information technology-based technique. Artificial intelligence and mathematical modelling can be easily incorporated into the computer-based system to find the optimal configuration for the system. Management concepts, such as process re-engineering, continuous process improvement and Activity-Based Management (ABM), are used as the backbone of the methodology. This methodology will help decision-makers to decide whether to perform process re-engineering or continuous process improvement efforts.(1) Define the objective of the process improvement program. In order to carry out a process improvement program, the objective must be defined by top management. The managers must be committed to carrying out process re-engineering or a continuous process improvement program. They must also give a realistic goal for the program by benchmarking with the best practices, for example the six-sigma quality level advocated by some (TQM) programs or 100% improvement by BPR proponents. In the case studies, the general major objective is to improve yield, reduce cost and achieve better quality control. (2) Collect data and analyse relationship between variables. (2.1) Collect data from the actual system. Data, such as processing time, downtime, maintenance time and the defects generation rate, are collected by the operator manually or by the machines automatically. The cost data are obtained from the accounting department. Data are verified by the engineers involved in the manufacturing processes. (2.2) Analyse the relationship between variables. The relationship between variables can be identified or analysed using mathematical analysis (such as Statistical techniques, Autoregressive Integrated Moving Average (ARIMA) methodology (Box and Jenkins, 1976 and Gaynor and Kirkpatrick, 1994) and Response Surface Methodology (RSM)) or artificial intelligence techniques (such as neural networks and rule-based systems). It is very difficult to identify the relationship between variables. Normally, the relationship between the variables is hidden in the data. Hence mathematical and artificial intelligence techniques can be used to determine the relationship between the variables. (3) Construct simulation model. A simulation model is then constructed to incorporate productivity, quality and costing analysis. (3.1) Incorporate the productivity dimension. Standard productivity measures (such as dispatching policies (for example, FIFO), number of buffers, number of pallets circulating in the system, travelling time, machine downtime, machine utilisation and material processing time) are included in the simulation for the productivity analysis. (3.2) Include the quality dimension. SPC is incorporated into the system for controlling the quality. A rule-based system can also be incorporated to identify the out-of-control condition for the control chart (Hoyer & Ellis, 1996). (3.3) Incorporate the costing dimension. ABC can be incorporated into the simulation by defining the cost drivers for various activities (Spedding & Sun 1999). As a result, the ABM concept can be used to improve the system (Kaplan & Cooper, 1998). (4) Real-time data acquisition system. Step 2 and 3 can be totally or partially replaced by a data acquisition system that can obtain real-time data from the system. (5) Perform optimisation and analysis of the system. Artificial Neural Networks, DoE and Scenario Studies are the techniques used to optimise the system. A Grid Search method or RSM is used to find the optimal conditions for the system. A decision support model is built based on the optimal results to assist the managerial decision making process. (6) Draw conclusion from the results. Finally, conclusions are drawn based on the results obtained from the optimisation analysis. Process re-engineering or continuous process improvement efforts are then carried out according to the recommendations given by the optimisation analysis. The enabling technology makes IMPIM unique compared to other methodologies. The enabling technology for the IMPIM can be divided into 4 layers: (1) Management concepts, such as Process Re-engineering (BPR), Continuous Process Improvement (TQM) and ABM, are the most important layer because management concepts are the core values that penetrates through the methodology and the organisation. The techniques applied in the methodology must be consistent with the ideas proposed by the management concepts adopted to solve the problems. (2) Information technology is essential for this methodology because it is computer-based. Discrete Event Simulation Software, Networking and Real-time Data Acquisition systems are examples of IT related technology. (3) Artificial intelligence techniques, such as Artificial Neural Networks and Rule-Based systems, can be used as the tools to learn the system behaviour and optimise the system parameters. (4) Mathematical models, such as SPC, ARIMA methodology, ABC, DoE (Montgomery, 1991) and RSM, are also used to characterise, optimise, control and solve the problems inherent in the system. In the following sections, case studies will be conducted in this paper to show the practical application of this methodology. The enabling technology of the methodology will be highlighted in the detailed step-by-step approach. The step-by-step approach will be compared with the generic methodology.
نتیجه گیری انگلیسی
Global competition, changing customers’ needs, increasing product complexity, volatile economic conditions and higher customer expectation are the changing set of business requirements that causesthe need for optimising productivity, improving quality and reducing cost. However, there is no scientific and methodical approach to address the productivity, quality and cost problems in an integrated and a unified manner. In this paper, an IMPIM is developed to solve the yield management, process control and cost management problem. The enabling technologies for IMPIM are management concepts, information technology, artificial intelligence techniques and mathematical models. This paper also formulates a step by step implementation of IMPIM. At first, the objective of the process improvement program is defined. Data collection is then carried out on the actual system. The relationship between the variables is identified and estimated using ARIMA methodology. Discrete event simulation is then used as a platform to perform process reengineering (Business Process Reengineering) and process improvement (TQM). Simulation is also a foundation to ‘glue’ together productivity (traditional simulation), quality (TQM) and cost (Activity-Based Costing). Next, optimisation and analysis of the system are performed using DoE, RSM and neural networks metamodel. Finally, conclusions are drawn based on the results obtained from the optimisation analysis. The practical application of IMPIM is presented in two case studies. The first case study is concerned with control charts system design and the second case study focuses on an on-line optimisation of quality in a manufacturing system to assist the six-sigma quality improvement effort with reference to the practical situation of having a process shift. The results of the first case study shows that the appropriate combination of different control charts in the system can achieve the conditional optimal process control. In the second case study, the grid search method with the NNM is an on-line quality improvement tool that can achieve optimal performance without costly intervention and complex analytical tools. IMPIM provides interesting insight and satisfactory results in both case studies.