Attendees will also receive an Excel spreadsheet with the d*2 factors that are required for the average and range method, and another with data for the worked examples. These appear in some references that are named in the handout, but they are not as widely available as the traditional control chart factors. The handout notes will include not only the slides, but also accompanying notes as well as instructions for performing a gage R&R analysis using Analysis of Variance (ANOVA) in Excel. This may be useful if you don’t have Statgraphics, and don’t wish to use the average and range method.
Seminar Agenda:
- Know the difference between accuracy, as ensured by calibration, and precision, as assessed from a MSA.
- Know the meaning of bias, stability, and linearity.
- Know how gage variation, a function of repeatability and reproducibility (R&R, alternatively equipment variation and appraiser variation respectively), affects outgoing quality, process performance estimation, and statistical process control.
- The chance of accepting nonconforming work, and rejecting good work, increases near the specification limit(s).
- If gage variation is built into the estimate of process variation, the estimated process performance index will be lower.
- Gage variation is included in the estimate of variation for statistical process control, which means the control limits will be wider and less able to detect genuine process shifts. The false alarm rate will not, however, increase.
- Awareness of:
- Gage resolution or discrimination
- Bias (example will be given, with data in the Excel handout)
- Stability
- Linearity (can be handled by StatGraphics or even Excel; an example will be provided in the handouts). StatGraphics includes data, used by permission, for the linearity example on page 99 of the 4th edition of AIAG’s MSA manual so StatGraphics users can therefore try this example on their own in addition to the one in the presentation.
- Know the recognized methods for gage studies, including number of parts, number of inspectors, measurements per part, and also the need to randomize the order in which the parts are measured.
- Know the components of gage variation (repeatability and reproducibility), and how to estimate them from data using the average and range method. Basic directions are provided for Excel and StatGraphics, and the handout adds instructions for Microsoft Excel. A worked example with data will be provided. (StatGraphics comes with an additional example from the AIAG MSA manual.)
- Instructions also are provided for the Analysis of Variance method.
- Know potential remedies for non-capable gages, including guard banding (tightening the acceptance limits to protect the customer from nonconforming work, at the expense of rejecting some borderline acceptable product).
- Reproducibility can be suppressed through multiple readings (if economically feasible).
- Optimize tightened acceptance limits (guard banding) to either minimize the total cost of inadequate gage and process capability, or ensure that the customer receives no more than a certain fraction of nonconforming work.
Who Should Attend:
- Quality Managers
- Quality Inspectors
- Quality Technicians
- Quality Engineers
- Manufacturing Departments
- Everyone with responsibility for dimensional measurements
FDB3374
Why You Should Attend:
Calibration (accuracy) has been traditionally required by ISO 9001, but measurement systems analysis (precision) has not—although it has been an element of the automotive standard IATF 16949 ever since the latter was QS 9000. Accuracy and precision are both, however, necessary to obtain reliable measurements of critical to quality (CTQ) characteristics. Attendees will learn the implications of measurement systems analysis (MSA) with regard to outgoing quality, estimation of process performance indices, and statistical process control, and how to perform and assess a MSA.
Procedures for MSA are well documented, but this webinar goes further to address the issue of guard banding. This means tightening acceptance limits to protect customers from nonconforming work when the process and gage capabilities are both relatively mediocre. The job can be performed in Excel with a Visual Basic for Applications (VBA) function that can quantify, given the process and gage standard deviations, the fraction of nonconforming work that will reach the customer along with the fraction of good work that will be rejected. This allows the user to either assure the customer that no more than a given fraction of nonconforming work will escape or, given the cost of external failure (bad items that reach the customer) and also of good parts that are rejected, optimize the acceptance limits to minimize the total cost.
