Font Size: a A A

Strategies for the integration of statistical and engineering process control

Posted on:1993-05-02Degree:Ph.DType:Dissertation
University:Arizona State UniversityCandidate:Messina, William SamuelFull Text:PDF
GTID:1479390014996868Subject:Engineering
Abstract/Summary:PDF Full Text Request
Statistical process control and engineering process control are two strategies for quality improvement that have developed in relative isolation from each other. In general, both techniques have reduction of variability as their objective, although they seek to accomplish this objective in different ways.; The purpose of this research is to develop for feedback control specific strategies for integrating statistical and engineering process control when assignable causes in the form of sustained shifts or trends are introduced into the process. Two different situations are examined: (1) Integration of engineering and statistical process control for the continuous process industries, and (2) The integration of statistical process control with proportional-integral-differential controllers.; The general findings of this research indicate: (1) The effect of assignable causes on engineering feedback control schemes increases the overall variability in the process. (2) The integration of Shewhart Charts for Individuals, Cumulative Sum, or Exponentially Weighted Moving Average statistical control schemes with engineering feedback control schemes significantly decreases this overall variability in the process. However, none of these three schemes consistently outperforms the others. Therefore, the following strategies are proposed for usage by the control engineers in practice.; Three alternate integration strategies are proposed for use in the continuous process industries. A combined prediction error Exponentially Weighted Moving Average control chart with weighting constant {dollar}lambda{dollar} that determines the amount of information secured from historical data and appropriate tracking signal on the control action with one of the following three strategies proposed for use on the deviation from target: (1) An Exponentially Weighted Moving Average scheme with {dollar}lambda{dollar} = 0.4, (2) A Shewhart Chart for individuals, and (3) A Shewhart Chart for individuals with associated Western Electric rules. The choice is based on cost considerations.; Three alternate integration strategies are proposed for use with proportional-integral-differential controllers. A Shewhart chart for individuals combined with one of the following three schemes: (1) An Exponentially Weighted Moving Average chart with {dollar}lambda{dollar} = 0.4, (2) A Cumulative Sum chart with H = 5{dollar}sigma{dollar} and K = 0.5{dollar}sigma{dollar}, and (3) The applications of Western Electric rules to the chart. The choice is based on cost considerations.
Keywords/Search Tags:Process control, Strategies, Statistical, Exponentially weighted moving average, Integration, Shewhart chart for individuals
PDF Full Text Request
Related items