SPC for the Process Industries

Process industries such as paper and chemical present rather unique problems for SPC implementation.  Namely that several assumptions of X-Bar charts are violated:

Problems

  • Samples or subgroups are usually limited to a size of one
  • A control chart with a sample size of one is not sensitive to small process shifts.   It would take on average over 16 samples to detect just a one and a half sigma shift in the process
  • Successive data points in continuous processes are often autocorrelated, i.e. related in value.  For example in the diagram below successive samples will be correlated do to mixing in the tank

Solutions

  • Exponentially Weighted Moving Average (EWMA) and Cumulative Sum (CUSUM) Charts provide an increased sensitivity to small shifts
  • Autoregressive Integrated Moving Average (ARIMA) models provide a means to remove the noise in the data caused by autocorrelation.  These tools are mentioned in the course outline on the SPC page