Rosenzweig, George (1989, ASQC) G.R. Technologies Ltd., Toronto, Ontario, Canada
We are realize that without measurements we have limited chances for economical improvement of our manufacturing processes. Without facts we may not even know if an improvement took place or whether it was significant or transient.
Many times our measurements are taken at face value without our knowing if they are accurate and reliable. We must first determine that they are and then statistically evaluate.
Traditional Quality Control and Inspection mostly focused on product acceptance and used the measurements (attribute and variable) on a "GO-NOGO" basis. The concept of product consistency within specification limits was ignored at large.
Concerns for product verification instruments were always present in the form of calibrations and administrative controls. The after the fact product control approach was extended to "good measurements" through calibration. The calibration procedures were considered the answer to this problem. Frequent calls for re-inspection and re-testing did not trigger major revisions to the measurement systems, and controls continued on the "believe the most favourable value."
The rapid transformation of the Quality Management Philosophy saw after the fact inspection replaced by real time process control. We need reliable data before defective products are created. Our dependence on decisions based on measurements reaches a new level of importance.
This paper is a result of our observations based on the industry-wide need to know, what the measurement system capabilities are, recognizing its limitations and their impacts on the quality of decision making and on product quality.
As the statistical way of thinking by management and engineers gained momentum, the search for sources of product variation expanded to the study of the process variability itself. In this quantitative cause-effect analysis between process parameters and the resulting productivity and quality fluctuations, the question is "How good are these results?" or "Are these measurements realistic?"
This paper illustrates different statistical Measurement System Analysis techniques that have proven successful. Ease of application and practical interpretation are important characteristics and basic conditions for everyday use.
The time when statistical analysis was the sole domain of a few engineers and scientists has passed. Successful companies achieved their improvements by bringing analytical tools to the supervisory and operator levels. This of course required cultural changes within the organization and well planned training courses for the different need levels.
The common aspects of the different techniques are summarized as follows:
Statistics,Metrology,Continuous quality improvement (CQI)