Heeran, Joseph P. (1987, ASQC) Armstrong Tire Company, New Haven, CT
Process control must begin with the identification of those factor which drive the process and regulate its output. Statistical analysis applied to the task of process control is well suited for the responsibility of identifying these factors. The primary statistical tool used to exercise 'control' over a process in the implementation of statistical process control is control charting.The ability to exercise 'control' however, is directly proportional to the degree of statistical control which the process demonstrates. All too often, trial control charts describe a process which is not a state of statistical control. Plotted points, which provide evidence of lack of control, abound. Attempts to identify those factors which promote the 'out-of-control' situation seem hopeless. The exercise of 'control' over the process is, therefore, impossible, because the factors which define the process are unknown.The inability to identify why plotted points demonstrate an unstable process are most often due to the influence of not one but rather many factors operating simultaneously on that sample. Reliance on the operator's experience for solutions often fails because of the inability of the operator to isolate the influence of each factor and thereby identify the source of process variability. The tool of control charting is not enough. Control charts in such instances prove ineffective. Although they signal when special causes are operating in a process, they do little to diagnose why statistical instability exists. What is needed for process control is a tool to isolate and identify the factors which promote unexpected process variability.In addition to pinpointing special causes of process variation, a statistical tool for isolating and identifying common cause variation is needed. In other words, how can we reduce the variation inherent in a statistically stable process?The answer to such a dilemma and the successful implementation of statistical process control is a systems approach, that is, a functional approach which incorporates a multitude of statistical tools that analyze why process variation is excessive. This paper proposes such an approach. Specifically, the proposal is a strategy which utilizes experimental design and analysis and integrates its utilization with SPC in a manufacturing environment.