טכניון מכון טכנולוגי לישראל
הטכניון מכון טכנולוגי לישראל - בית הספר ללימודי מוסמכים  
M.Sc Thesis
M.Sc StudentMantel Yaniv
SubjectSimple Processing System Simulation Using Arena Software
DepartmentDepartment of Quality Assurance and Reliability
Supervisor Dr. Leonid Burstein
Full Thesis text - in Hebrew Full thesis text - Hebrew Version


Abstract

This paper examines the capabilities of simulation in general , and specifically the Arena software package , as a tool for quality control. The examination was performed by building simulation models in three software packages: Arena , Mathcad and Matlab , comparing their working processes , identifying their relative advantages and disadvantages , and examining the capabilities of the simulation tool in answering questions that arise during the work of a quality engineer. On its basis , conclusions and recommendations were formulated. As first step , two versions of models were built describing a simple processing system and simulating the production of a batch of 1,000 items on a single machine and on two machines. The one-machine models were based on Arena vs. Mathcad , the two-machines models on Arena vs. Matlab. The Mathcad and Matlab models were validated by a deterministic input which was calculated manually as well as by the simulation model. That done , a stochastic input was used in the Mathcad and Matlab models and in their Arena counterparts. The output from the two one-machine models was compared by each parameter , and so was that from the two-machines models. The intention was to show that each parameter has the same distribution throughout , thus proving that the two model types are equivalent. The analysis was carried out using the Kolmogorov-Smirnov test )where applicable ( or the chi-square goodness-of-fit test. The equivalence of the models was proven for each parameter at a 95% confidence level. As the next step , two additional models were built in Arena which simulated the production of the same batch , incorporating additional quality factors such as machine failures , periodic maintenance , inspection and costs. Finally , the three software packages were compared by several parameters such as convenience in building the model , simplicity of the debugging process , the total time required for realizing the simulation , etc. The main conclusions from the comparison of the three software packages are as follows: Arena provides the easiest way for building a simulation model , as well as the fastest and most intuitive one. It is best suited for operators with no knowledge or experience in computer programming. Matlab has the widest array of functions , thus is the most flexible of all software packages. It can be used to simulate almost everything at the macro level )systems ( as well as at the micro level )processes ( . Mathcad does not offer any specific advantages in simulation.