00192 'aa328750125e1243c8afa938aa44d2

194

Messina, Montgomery, Keats & Runger

procedurę developed by MacGregor (1990) in the context of the funnel experiment is augmented with SPC rules in the form of control charts on the control actions taken on the manipulated variable. A simulation study demonstrates that the combined EPC/SPC scheme reduces overall variability in comparison to EPC alone when assignable causes are present. Montgomery etal. (1994) have previously demonstrated the effectiveness of integrating SPC and EPC for the deviation from target. This study will focus on SPC or statistical monitoring rules applied to the manipulated variable. The results of both approaches (monitoring the deviation from target and monitoring the manipulated variable) using simulation studies are employed to develop combined EPC/SPC strategies for control engineers to employ in practice.

Introduction

There continues to be substantial current interest in the subject of integrating Statistical Process Control (SPC) and Engineering Process Control (EPC) . See Montgomery, Keats, Runger, and Messina (1994) for a review of the relevant literaturę. A recent related paper on this topie is that of English and Case (1990). They used SPC procedures as controllers rather than monitoring devices. The bulk of the literaturę on the EPC/SPC relationship, including the present study, uses SPC only as a monitoring tool. The key to successful integration of these two different approaches for ąuality improvement is to develop strategies that control engineers can implement in practice. Strategies that are robust to a variety of assignable causes that can routinely occur in the continuous process industries such as sustained shifts and trends in the process.

Montgomery, Keats, Runger, and Messina (1994) have examined through simulation the effectiveness of integrating statistical monitoring and EPC for the deviation from target on the reduction of overall variability for the continuous process industries when assignable causes of sustained shifts or trends are present in the process. In that paper, a model of the continuous process industries proposed by MacGregor (1990) which is a modification of the funnel experiment discussed by Deming (1993) and described in morę detail by Neave (1990) was examined. Integration of MacGregor's EPC scheme with the following SPC schemes were examined: Shewhart Chart for Individuals, ExponentialIy Weighted Moving Average (EWMA) control charts with a weighting factor of X = 0.1 or 0.4, and Cumulative Sum control charts with H = 5a and K = 0.5a. The results of the study demonstrated that the addition of an SPC control chart to monitor output deviation from target in a system with EPC resulted in a reduction of overall variability in the process.