THE CRITICALITY OF MEASUREMENT SYSTEM ANALYSIS (MSA) By Doug Stohr "How reliable is your measurement system?" Introduction In the world of measurement and the criticality of good measurement information, a full company-wide measurement system analysis is a must. As we all know, there is no perfect measurement system – all systems have inherent as well as induced variation and error. A high-quality measurement system has solid statistical data supporting it. Many companies using measurement equipment have never done the proper measurement analysis to prove to themselves and their customers that they have adequate measurement systems. Not only must these systems be reliable and repeatable, they must also provide sound statistical data. In today's world of manufacturing where high quality is mandatory, a full measurement system analysis goes hand in hand with providing high quality data and product compliance proof. Failure to understand total capability of the measurement system can, and often does, lead to constant waste, error and unnecessary process adjustments. The unknowns of a measurement system will lead to questionable data, customer dissatisfaction and data challenges from your customers. Performing a full company measurement system analysis takes away the doubt and mystery and leads to improving the system. I recommend that a full system analysis be performed at least once a year. Just a good calibration system is not enough. I would like to present the following scenario regarding a full measurement system analysis. Mini Case Study The coating process under review produced a very soft material that must be held to a tight thickness tolerance, per a competitive industry standard of +/-.002 inches. The process engineer used this information in the real-time SPC program. The process inspection was performed with a contact micrometer, and the same micrometer was used by all operators in the study. The process data showed the process was unable to hold the stated tolerance of +/-.002 inches. This resulted in constant adjustment and uncontrolled process results. The operators then started using the measured values as a reference only, and no real process control existed. In the process variation analysis that followed, it was suspected that the measurement system was not capable of providing meaningful data. It was then decided to conduct a Gage R&R study. Gage Repeatability & Reproducibility and part variation were all studied, and the total variation was 87% of the stated tolerance. When the gage variation was extracted from the study (or EV%), the gage variation alone was discovered to be 69% of the total variation! Consequent research found that the soft plating material was deforming when even the slightest amount of pressure was being applied to the micrometer, thus inducing measurement error. It was determined that the instrumental cause was the decision to use the micrometer to measure this feature. The solution, of course, was to measure this feature using a non-contact measuring gage. It was further decided to install an off-line laser thickness measuring instrument that was calibrated using a known standard prior to each use. The gage variation dropped to less than 4%, and the total measurement variation dropped to below 7%! The laser thickness system was a robust and inexpensive measurement system upgrade – and only discovered and implemented when the MSA was performed. To summarize, the following are some considerations for a full measurement system analysis: 1. The Need to know measurement error 2. The Need to know elements of measurement variability 3. The Need to know measurement uncertainty The Basic Elements of an Effective MSA: Accuracy Linearity Stability Bias Precision Repeatability Reproducibility Resolution Environment Time The Basic Methods Used of an Effective MSA: Range – quick overall approximation of gage measurement variability Gage R&R / Average and Range Method (X bar & R) – determines repeatability & reproducibility X bar & R / X bar & S – for stability also for percentage bias ANOVA – determine interaction between measurement gage and gage operator Key Inclusion: All expected measurement sources of variation are understood and considered within the scope of the measurement study. Note: Current software available today will cover all of the measurement system analysis techniques well beyond the basic tools. Current software provides graphical analysis capability for better measurement system interpretation. About the Author Doug has been Quality Manager at the following companies: Bergquist Company, Taber Bushnell, Sheldahl and most recently Director of Total Quality with Clore Automotive. Some of Doug's background and qualifications include Master Black Belt in Six Sigma and Six Sigma Trainer, Implementation of TQM Programs, QS9000 Registration, Measurement Systems Analysis Trainer, Advanced Quality Planning, Continuous Improvement, Design Of Experiments, Calibration, SPC Trainer and Mechanical Inspector. Comment on this article