An approach in statistical control chart design

Abstract

Several questions arise when designing statistical process control charts Is the control chart sensitive enough to detect a particular process shift with a particular frequency? When should a process shift be considered critical and why? What should the optimum sample size be? What should the optimum sample inspection frequency be? These and some other questions are associ­ated with a control chart "sensitivity analysis." We will try to answer these questions through a general step by step procedure for designing statistical control charts. This procedure will not end up with a unique answer because we are dealing with design and not with analysis. Criteria for the different situations and the degree of impact of a particular defective product will govern the possibilities. Most of the time the original control chart design will lead to particular improvement projects due to the process characterization through sensitivity analysis.