Advances in model based dual chart process control

Traditional statistical process control (SPC) is based on a single control chart of the process variable, assumed to be an independent identically distributed random variable. In practice there are significant correlations over time, resulting in false readings. The modem approach is to fit a model to the variable, thereby decomposing it into a systematic (the fitted values) and a random component (the residuals). A dual chart system is then implemented. One chart is a special cause chart of one of a kind random effects. The other is a common cause chart, symptomatic of a systematic problem that is developing. Also, the potential to predict systematic out of control conditions forms the basis for avoiding them altogether. Significance: The separation of statistical process control (SPC) charts into special cause and common cause charts will reduce the rate of false alarms and focus the search for the true cause of a breach of the SPC chart. implementation of this technique may result in a reduction in down time and an increase in productivity and profitability.