טכניון מכון טכנולוגי לישראל
הטכניון מכון טכנולוגי לישראל - בית הספר ללימודי מוסמכים  
M.Sc Thesis
M.Sc StudentPoltielov Sonya
SubjectDevelopment of Improved Repeatability and Reproducibility
Procedures for Products Effected by the
Measurements
DepartmentDepartment of Quality Assurance and Reliability
Supervisor Dr. Pavel Grabov


Abstract

Statistical process control is a part of company-wide efforts towards continuous improvement. Every process is directed with a database that is essential for assessing the variability in the process. Measurement systems characterize this source of the database. Since measurement system itself is subject to variation, it is important to assess measurement system variation. Too much variation in the measurement system may mask variation in the process.

The study that addresses the issue of measurement systems variation is called Gage Repeatability and Reproducibility (GR&R).

The main objective of the study is to determine whether a measurement procedure is adequate for monitoring a process, also by providing criteria for judging acceptability of a measurement system.  

Analysis of existing methods showed that an existing tool is not able to evaluate the precision of measurement systems when products are affected by the measurement performance. The purpose of this work was to develop an improved GR&R procedure that will provide a correct estimate of the precision of the measurement systems for products that are affected by the measurement performance. In order to provide the requested result, an improved   approach has been employed.

    This includes:
  1. Application of statistical tests, in order to detect presence of a non-random component.
  2. If a non-random component is present, we have to evaluate and remove this component.
  3. Precision evaluation on the 'purified' data.

This work also provides application of the proposed procedure to numerous industrial situations where measurements are effectively accompanied by a non-random component due to the measurement performance.