טכניון מכון טכנולוגי לישראל
הטכניון מכון טכנולוגי לישראל - בית הספר ללימודי מוסמכים  
M.Sc Thesis
M.Sc StudentPaykin Victoria
SubjectDevelopment of a Model for Key Characteristics
Identification
DepartmentDepartment of Quality Assurance and Reliability
Supervisor Dr. Pavel Grabov
Full Thesis text - in Hebrew Full thesis text - Hebrew Version


Abstract

The objective of this study was development of the Key Characteristic (KC) identification model for an electronic devices manufactured for the military industry. The KC represents some characteristic of a product or a part, whose variation has a significant influence on its suitability, performance, service life, or manufacturability as per AS9100/EN-9100/JISQ 9100. Identification of KC permits determination of the order of priorities in the design, production and inspection processes. Identification and control of the KC’s are mandatory in certain areas of the automotive and aerospace industries, but the standards mostly merely specify the requirements and recommend control procedures, without setting clear guidelines for  KC identification. The study started with a literature review it included a comparative analysis which was performed using some methods covered by the review. These findings motivated the development of a KC identification methodology based on a quantitative estimation of the influence of KC variance on that of the product output. The software used was Crystal Ball, an Excel add-in with a high potential for the sensitivity analyses involved. The sensitivity chart is an important tool in this software, permitting quantitative evaluation of the influence of individual KC’s. Crystal Ball uses a mathematical model of the product behavior. Around this deterministic model, a stochastic environment is generated via Monte Carlo simulation for the KC uncertainties evaluation. The final paper presents the developed methodology for KC identification and specifies the tools for its different stages. The bottom up approach based on a product tree should be used for KC identification. The approach implies application of the sensitivity analysis supported by Crystal Ball for KC identification. Obviously, application of this technique is conditioned by existence of the mathematical models describing the part / subassembly / assembly behavior at these levels. As a rule such models could be set at the lower levels of a product tree. At the higher levels where such models setting are rather problematic, the conventional methods could be applied for KC identification (such as Pareto analyses, FMEA, etc.).