Bir kalıp atelyesinin SIMAN benzetim dili ile benzetimi

Abstract

Günümüzde benzetim artık imalat sistemleri ile ilgili tüm sorunların çözümlenmesinde karar vericiye yardımcı olan tekniklerinin başında gelmektedir. Bu tezin konusu ülkemizin önde gelen kuruluşlarından biri olan SIMKO A.Ş. kalıphane atelyesinde yapılan bir benzetim çalışmasıyla ilgilidir. Bu çalışmada ilk önce sistemle sistemle ilgili ve sistem performansını en iyi temsil eden verilerin toplanmasına çalışılmıştır. Verilerin elde etme güçlüğü ve güvenirliği önemli bir problem yaratmıştır. Elde edilen veriler düzenlenmiş ve kurulacak model üzerinde kullanılacak şekilde bir formata sokulmuştur. Daha güvenilir veri elde etmek amacıyla geçici bir süre sistem incelenmiştir. İncelenen bu süre zarfında sistemdeki 15 kalıp incelenmiş ve elde edilen veriler analiz edilmiş ve gerekli istatistiksel analizler yapılmıştır. Kullanıcıya sağladığı avantaşlarından dolayı SIMAN benzetim dili seçilmiş ve model bu dille kodlanmıştır. Modelin kurulmasından sonra model çalıştırılmış ve elde edilen sonuçlar ile gerçek sistemden elde edilen sonuçlar karşılaştırılmış modelin geçerliliği sağlanmıştır. Daha sonra model üzerinde yatırım stratejileri denenmiş ve elde edilen sonuçlar değerlendirilip en uygun strateji bulunmuştur. Alternatifler mevcut sistemde ki tezgahların sayılarının değişik kombinasyonlarının model üzerinde denenmesiyle ilgilidir. Kullanılan benzetim dili sayesinde kullanıcı, alternatifleri sadece deneysel dosyadaki verileri değiştirerek inceleyebilmektedir. Bu da kullanıcıya büyük kolaylık sağlamaktadır. Model kurma aşamasında karşılaşılan en büyük güçlüklerden biri SIMAN benzetim dilinin bize verdiği sınırlı hafızadır. Çünkü benzetim çalışması boyunca elde edilen tüm veriler dosya lara kaydedilir. Bu dosyalara kaydettiğimiz verilerin bü yüklüğünü bu hafıza sınırlamaktadır. Modelde kullandığımız ve sistemde dolaşan varlıkların özelliklerini temsil eden özalanlarda hafızada önemli bir yer tutmaktadır.

Simulation of a mold production facility in SIMKO A simulation model seeks to "duplicate" the behavior of the system under investigation by studying the inter actions among its components. Simulation must be treated as a statistical experiment. Unlike the mathematical models the output of the model represents a long-run steady-state behavior, the results obtained from running a simulation model are observations that are subject to experimental error. A simulation experiment differs from the regular laboratory experiment in that it can be conducted totally on the computer systems. By expressing the interactions among the components of the system as mathematical relations ships. &Je are able to gather the necessary information in very much the same way as if we were observing the real-life system ( subject, of course to the simlif ications built into model ). The nature of simulation thus allows greater flexibility in representing complex systems that are normally difficult to analyze by standard mathematical models. Although simulation is a flexible technique, both time consuming and costly, particularly when one is trying to optimize the simulated system. Simulation has been used to analyze problems of two distinct types ; xiv - 1. Theoretical problems in basic science areas such as mathematics, physics, an chemistry. 2. Practical problems in all aspects of real life. The technique used to solve the theoretical problems cited may be considered a forerunner of simulation in its present-day usage. Its called the Monte Carlo method and is based on the general idea of using sampling to estimate a desired result. Simulation like the Monte Carlo method and is based on estimating the output of a system through sampling. In this respect, many ideas, that were developed in conjunction with Monte Carlo are being using directly in the application of simulation. The presents success of simulation in modeling very complex systems rests squarely on the impressive advances in the capabilities and power of the digital computer. It's unimaginable to think that simulation could have reached any degree of success without the computer. System models are developed to analyze the behavior of systems as a function of time. From that stand point, there are two types of simulation. 1. Discrete simulation. 2. Continuous simulation. In discrete simulation, the simulated system is looked at only at selected points in time, whereas in continuous simulations the system is monitored at every point in time. XV In a typical simulation model, we view the system as comprising entities or transactions that, at any point in time during the simulation, may either be in service or waiting for service. To process waiting entities in desired order, a simulation language must provide automatic means for storing and retrieving these entities whenever needed by utilizing files or ordered lists. Available discrete simulation languages can be categorized into four groups ; 1. Event scheduling. 2. Activity scanning. 3. Process. 4. Combination of both process and event scheduling. The most flexible languages are those based on event scheduling and activity scanning. Flexibility, here implies that the language can readily model any complex situation. This flexibility, however, implies that the user must expand additional effort in development of the model. The subject of this master thesis is related to the simulation of a mold production facility in SIMKO producing telecominiquation devices and being one of the biggest company in TURKEY. In this simulation study. As the simulation language, SIMAN has been used to build the model representing real mold production facility. A SIMAN simulation is divided into three distinct activities ; system model development, experimental frame development and data a analysis. Within these three activities, the xvi - SIMAN software consists of five individual processors which interact through four data files ; 1. The model processor is used to construct a block diagram model. The data file that is generated is called the model file. 2. The experiment processor is used to define the experimental frame for the system model. The data file that is generated is called the experiment file. 3. The link processor combines the model file and the experiment file to produce the program file. 4. The program file is input to the run processor which executes the simulation runs and writes the results on the output file. If the system model includes an event or continuous model, the user-written FORTRAN subroutines are linked to the run processor before the simulations runs are executed. 5. The output processor is used to analyze, format and display the data contained in the output file. By using SIMAN simulation language, alternative designs which are related the real system can be made by changing parameters in the experiment file without changing the model file. this important feature of SIMAN provides the user to build model easily. After selecting simulation language to be used, datas which are related to the real system were collected so that the model of the system could be built as required. - xvii - After building of model, alternative designs were performed on model I built, the results of model were analyzed to find what alternative design satisfied our system and the best design was determined by using SIMAN output processors. The best investment strategy which made optimum the numbers of machines in the real system were found as the result of this simulation study and presented the managers who manage the facility. The memory limit of SIMAN was so important during the study. Because SIMAN gives 64 KB memory to record datum obtaining from model during model executing. If it is exceeded this value, SIMAN processor will give error. So we must use this memory effectively. In the same time, this memory limit determined the numbers of workpieces existing in the model at any specified time point while the model was being runned. In this simulation study, the memory problem which was expressed above, was a important problem for me. This problem was eliminated by simplifying model of the system.