The control charting procedures discussed in Chapters 6 and 10 are based on predetermined levels of significance at which warning or action is taken. Alternative conditions may arise where the evidence for talking action is based on the result of several samples, which individually would not be significant. In the case presented here, a differing procedure must be found for accumulating this evidence. A charting method to achieve this was proposed by Ewan and Kemp — ‘Sampling inspection of continuous processes with no autocorrelation between successive results’ (Biometrika 1960,47, 363). These charts are known as ‘cumulative sum’ charts and the values entered represent the accumulated sum of all data to date (instead of, as previously considered, the sample value just obtained). The cumulative sum control chart is particularly well-suited to detecting abrupt changes in process level that remain with the process for about five sampling periods. Usually, each point on the chart equals the previous point, plus the value of a statistic computed from the last subgroup, hence the name ‘cumulative sum’.
KeywordsUnit Distance Cusum Chart Acceptance Sampling Decision Interval Acceptance Sampling Plan
Unable to display preview. Download preview PDF.