تجزیه و تحلیل کل از سیستم های تولید با استفاده از پویایی های سیستم و فرایند تحلیل شبکه ای (ANP)
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|6013||2005||20 صفحه PDF||سفارش دهید||6480 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Computers & Industrial Engineering, Volume 49, Issue 1, August 2005, Pages 98–117
Aggregate analysis in manufacturing system design is a useful approach to relegate the non-feasible alternatives at earlier stages. A reusable System Dynamics model and the Analytic Network Process is proposed for a rapid and strategically consistent decision-making. The SD model captures the causal relationships and interdependence of the factors that can be simulated while the ANP provides the preferences towards the performance objectives consistent to the strategic objectives. The basis for the SD and ANP is the Causal Loop Diagram (CLD) that shows the relevant relationships and feedbacks among the model parameters. The approach is exemplified via a case to select the best among competing system configurations.
Three design stages can be recognised in manufacturing system development: Conceptual Design, System or Configuration Design and Detail Design. The conceptual design stage starts with a top-down analysis of the design requirement where transformation to a functional and then physical description are made (Paul & Beitz, 1988). The physical description includes technology selection and process planning from which a Master Flow Diagram or Technology Diagram is developed. Alternative possible routes of production can be realised by possibly multiple resource types and arrangements. The total design space of the system is thus formed as a set of all these alternatives. Given the requirements and the design space, selecting the best alternative out of the feasible alternatives is a typical design decision task. The scope and complexity of the alternatives vary depending on the requirement they are meant to address. When the complexity of the system makes analytical methods impractical for performance evaluation, simulation based tools with a rigorous experimental design and statistical output analysis are used. Evidently the time and budgetary resources required to assess multiple alternatives may be high that can be beyond limit for highly complex and/or several alternatives. On this backdrop rapid and aggregate analysis are necessary to relegate the unfeasible alternatives at earlier stage before proceeding to a more elaborate analysis. Though the methods and tools used for such an aggregate analysis vary depending on the underlying system complexity, it is generally desirable for the method to have the following properties: • Less time of analysis. • Reusability of the model for multiple alternative systems. • Capability of the model to evolve into a more detailed model that can be useful throughout the system life cycle. The use and research work on aggregate such as queuing networks is extensive and well established. Notable to these literatures include (Kleinrock, 1976 and Djik, 1993). However, the use of queuing networks is limited to steady state analysis and to certain types of production systems that can be modelled as a product form—the solution for the state of the system is dependent on the product of the utilization factor and the number of jobs. This is based on such assumptions as exponential service times, FIFO discipline, zero or negligible transport times, no scrap, etc. Even with some extensions or relaxations to these fundamental assumptions, for instance, processing time distribution be any type as long as it has a rational Laplace transform non-product form models can not satisfactorily be applied. In this paper, an aggregate performance analysis of MS based on a reusable System Dynamics model and the Analytic Network Process is proposed for a rapid and strategically consistent decision-making. The SD model is based on system thinking and captures the causal relationships of the factors and their interdependence that can be simulated to reveal the dynamic behaviour of the system (Sterman, 2000). Hence the SD simulation provides the system performance while the ANP provides the preferences/ weights of the lower level performance criteria consistent to the strategic objectives. The basis for the SD model and ANP is the Causal Loop Diagram (CLD) that captures the relevant relationships and feedbacks among the model parameters and performance variables. The SD model is developed by mathematically establishing the relations and transforming the factors in the CLD to rates, stocks (levels) and auxiliary variables that can be simulated with time. The reusability of the model is due to its generic structure from which the different competing alternative can be represented. The use of a generic model greatly decreases the model building time—a highly desirable advantage for aggregate analysis. Furthermore, the model can evolve into a more detailed model for a comprehensive performance analysis by integrating the different aspects of the manufacturing system to suit subsequent design or operational tasks. Due to their long term effects, design decision need to be consistent with the manufacturing strategy. Therefore mechanisms should be in place to ensure accurate interpretation of the strategic objectives at every level of the decision making process. One such mechanism is the assignment of preferences to the design criteria when there exist conflicting or competing objectives—which is often the case. Due to its ability to accommodate complex interdependencies and feed backs in the performance parameters, the Analytic Network Process, ANP, (Saaty, 1996), is applied to complement the SD based aggregate analysis.. The strengths of dependencies and feedback loops elucidated in the CLD is provides a basis to establish the network for the ANP model. The following diagram depicts the approach for aggregate analysis as proposed in the paper. The remainder of the paper is organized as follows. In Section 2 the concept of manufacturing strategy and performance parameters are described. In Section 3 SD and modelling in SD is discussed. Section 4 discusses ANP model for performance evaluation and ANP's synergy with SD. In Section 5 the illustrative case is presented. In 6 and 7 results and some concluding remarks are given.
نتیجه گیری انگلیسی
A System Dynamics and ANP based aggregate analysis of manufacturing system configurations design is proposed. Since SD and ANP models are based on the Causal Loop Diagram, the symbiosis help to make consistent considerations of the interdependence among the decision and model parameters. Besides a more rapid aggregate assessment of multiple alternatives is possible using the generic SD model since it can be reused with minor modifications to the alternatives with their respective parameter values. Compared to other methods used for aggregate analysis often applicable to a limited classes of manufacturing systems, such as queue network models, the SD based models are unlimited in application Fig. 4.Furthermore, not only is the analysis rapid and model reusable, the production flow model so developed can be integrated with other aspects of manufacturing such as logistics, marketing, human resources, etc at the same aggregation level and evolve into a comprehensive performance-measuring model of the system. This is particularly important in the analysis of supply chains, formulation of maintenance policy etc. Since SD modelling is based on system thinking the insights obtained greatly enhances the quality of the decision. The approach is exemplified in the extended theoretical case. Though the case is made as reasonable as possible to reflect the real case situation, It is not the purpose of the paper to make conclusive statements about different type of Configuration since the results merely refer to the hypothetical situation as per the assumptions taken.