(64a) An Alternative Control Chart to Monitor Low Level Count Data

Control charts use statistics to help translate patterns in data into decisions. Data-to-Decisions is critical in CPI since early detection of deviations helps ensure safety, high quality, and productivity. Control charting is mainly applied to normally distributed data, but control charting techniques are available for tracking count results on failures, spills, equipment breakdowns, defects in polymers, punctured films, thin spots in yarn, etc. For these kinds of discrete data, the count control chart (c-chart) is the de facto industry standard.

The c-chart will perform well at high count rates, but can be slow to respond to process changes at low count rates. In failure-process monitoring, the time-between-events chart (t-chart) was developed as an alternative to the c-chart for low count rates. At The Dow Chemical Company we have extended the use of the t-chart to monitoring of defects in bulk chemical product. The t-chart allows for the detection of changes by the basic SPC rule, where the c-chart requires supplementary rules. More importantly, the time efficiency of the t-chart was found to be considerably better than the efficiency of the c-chart for detecting process changes in test data sets, especially for detecting process deterioration. Here the efficiency of the t-chart is several orders of magnitude better than the efficiency of the c-chart, greatly improving Data-to-Decision speed.

To paraphrase a statement from Leo Chiang in the Big Data feature in March 2016 edition of CEP magazine, ‘Big Data Analytics expresses that the potential value is from the thoughtful collection of data combined with knowledge to answer complex questions.’ If you have a large volume and/or velocity of data with low counts, the t-chart has potential to accelerate the detection of change relative to the c-chart, putting this Big Data Analytics concept into practice.