بهینه سازی شبیه سازی مبتنی بر نرم افزار سیستم پشتیبانی تصمیم گیری: ابزار الماس خط تولید در صنعت

A diamond tool manufacturing system simulation is developed to predict the number of machines and the number of workers necessary to maintain desired levels of production for a company in Ankara, Turkey. The current manufacturing system is analysed by a simulation model emphasizing the bottlenecks and the poorly utilized machines. Validated simulation outputs are collected and used to build a multiple regression meta-model as a simulation optimization based decision support system (DSS). The proposed DSS involves analysis and evaluation of the system’s behaviour through the use of a meta-model with an integrated optimization module. It enables the decision maker to perform sensitivity analysis by considering several combinations of decision variables. The aim of this study is two fold. The first is to represent a simulation optimization based DSS application for a real system by considering all the required steps. The second is to analyse the performance of the current production system and determine the optimum working conditions by simulation with greatly reduced cost, time, and effort.

In today’s highly competitive industry, a company must be able to adapt to its customers’ ever-changing needs and improve the quality of its products in order to survive. It is important that the company responds quickly to rapid changes in technology, demand fluctuations, and design changes. These needs have forced many companies to put emphasis on automated systems to improve productivity and quality, while reducing cost simultaneously. When the systems under investigation are complex—as is often the case in manufacturing environments—it becomes impossible to find analytical solutions. Because of the complex stochastic characteristics of such systems, simulation is used to predict their behaviour as a powerful management science and operations research (MS and OR) technique. In other words, simulation—an alternative method to analytic tools—overcomes the complexities of large-scale stochastic systems. However, the major drawback of simulation for practical applications is that it is computationally time consuming. This study demonstrates how the outputs from a complex DSS were transformed into a simple DSS and how a meta-model and an optimization module were integrated. Simulation models show the dependence between the controllable variables and the outcomes. Simulation models yield probabilistic (variable) outputs. A DSS is an interactive, flexible and adaptable computer-based information system that utilizes decision rules, a model, and a model base with a comprehensive database. Thus, a DSS supports a complex decision-making process and increases its effectiveness. Garry and Scott-Morton [9] and Keen and Scott-Morton [10] aid the decision maker in addressing unstructured or semi-structured decisions.

This paper discusses the design of a decision support system (DSS) based on simulation optimization integrated with a regression meta-model, which helps the decision makers evaluate the effect of manufacturing technologies on the performance of an organization and determine the inputs that affect the performance. The proposed DSS involves analysis and evaluation of system behaviour, as well as the optimization of the system configuration for a given range of parameter values. The proposed DSS model also enables the decision makers to perform sensitivity analysis quickly by allowing them to get immediate results for alternative configurations of decision variables, whenever changing the system is required. This research also demonstrates how simulation modelling and the regression meta-modelling approach can be used to design and optimise a multi-stage, multi-server real production system in detail. It also shows that analytical modelling can be used together with the modelling capability of simulation, providing (nearly) optimum solutions. Since the production requirements are customer oriented, the planning is dynamic. Based on the results of this study, it was decided by the company to balance and design a line capable of producing a production target of approximately 9042 sockets per day. The primary goal of the company was to meet customer demands at the lowest possible cost. The line was redesigned under some economic, technological and physical constraints (such as company budget restrictions and available space for the production line), to achieve the optimal production rate. The use of a regression meta-model allows defining any performance measure as a function of the factors that affect it. In this way, a feasible combination of input variables can be determined, to meet the production goals of management. The meta-modeling approach also reduces the computational cost of simulation. The following results are obtained for this case study: • The feasible combination of the main factors found to be x1 = 15, x2 = 3, x3 = 15, x4 = 1 and x5 = 6, gave the lowest total operating cost, satisfied all economical, physical and technological constraints, and improved daily productivity by 19.42%. • The average queue length of the hot press station identified as a bottleneck in the production system has been reduced to 0.136, being an improvement of 99.84%. • The mean cycle time came down from an average of 4114.6 min to an average of 1233.98 min, with an improvement of 70.01%. Finally, it can be stated that the state of the art of simulation adopted in this paper with the multiple regression meta-modelling approach as a DSS for modelling real multi-stage, multi-server production systems can prompt management to compare the existing system and the proposed new design, and to find near-optimum values of the decision variables. It is shown that the proposed DSS is used effectively to improve the production rate, and results in a substantial decrease of both bottleneck queues and the overall production time. If the company relaxed the restrictions on the operating parameters, they can use the DSS to investigate how the different scenarios can affect the throughput rate of the production system.