Page 129 - Six Sigma Demystified
P. 129

110        Six SigMa  DemystifieD


                          Considering that the measurement system is used to determine acceptability
                        of the process relative to customer requirements, it should be apparent that as
                        the error increases, so does the chance that

                          •  Products or service truly outside requirements could be measured as being
                             within requirements and subsequently considered acceptable.
                          •  Products or service truly within requirements could be measured as being
                             outside requirements and subsequently rejected as unacceptable.
                          Measurement error also influences control charting and process capability
                        analysis. The calculation of statistical control limits and the estimation of pro-
                        cess capability require a sound estimate of the process variation. As measure-
                        ment error increases, so does the chance that
                          •  Subgroups from a controlled process will be determined to be out of control.

                          •  Subgroups from an  out- of- control process will be determined to be from
                             an  in- control process.

                          In some projects, the use of R&R studies has detected this measurement
                        system variability as the most significant source of  common- cause variation. In
                        such cases, the remainder of the project addresses improvement and control of
                        the measurement system. Of course, the loss to the organization (before this
                        was determined) was quite real, so significant savings were achieved by the
                        project team in eliminating these measurement errors.
                          R&R analyses can be performed on variable or attribute (count) data. In the
                        case of attribute data, measurement error is often the result of the subjectivity
                        of the inspector, which ultimately results from requirements that have been
                        poorly defined. For example, if five inspectors are asked to classify a painted

                        sample as having a color imperfection, then, unless proper standards are estab-
                        lished, error often will result from the subjectivity of that assessment. How
                        much variation in color is acceptable? Over what distance of painted part? As
                        viewed under what lighting and optical conditions? An excellent discussion of
                        attribute gauge analysis is provided in The Six Sigma Handbook.
                          Examples of the use of R&R analysis may be found in Part 3.



                          Project examPle: Measurement Systems analysis

                          The process metrics are cycle time, measured in minutes, and errors, measured as
                          a count by error type. The precise start and end points for the cycle time measure-
   124   125   126   127   128   129   130   131   132   133   134