Font Size: a A A

Economic statistical design and analysis for the Poisson CUSUM control char

Posted on:1993-03-15Degree:Ph.DType:Dissertation
University:Arizona State UniversityCandidate:Kim, Do-HyungFull Text:PDF
GTID:1479390014497915Subject:Industrial Engineering
Abstract/Summary:
Considerable research has been done for control chart design based on either pure statistical or economic criteria. However, very little attention has been given to the economic statistical design of control charts which consider both economic and statistical criteria. Although this scheme might cause an inevitable cost increase, it improves upon the pure economic-based schemes by achieving desirable statistical properties.;The research follows the general cost modelling structure presented by Lorenzen and Vance. The Average Run Lengths (ARL) are incorporated as constraints to bound certain statistical criteria. The ARLs are calculated by converting the movement of the test statistics into a Markov chain.;Of the four different classes of schemes (Pure Statistically-based, Pure Economically-based, and Economic statistical control schemes with simple constraints and constraints based on process shifts), the economic statistical control schemes yield a statistical improvement but produce a little cost penalties over the pure economically-based control scheme. With careful selection of the constrained values for ARL1s, based on the magnitude of the process shifts, an economic statistical control scheme with constraints based on process shifts provides tighter control but yields slightly more cost penalties over an economic statistical control scheme with simple constraints.;The study also investigates the sensitivities of the 12 cost and process parameters on the lost-cost and design parameter responses. The behavior of the cost function in the vicinity of the optimum for design parameters is explored. The cost surface is most sensitive to the reference value, k, and the study indicates reasonable flatness in the vicinity of the optimum. The impacts of process misspecification with alternate manufacturing process models are investigated. The study shows that the use of a continuous process model for a discontinuous production generates the cost benefits while using a discontinuous process model for a continuous production produces a high magnitude of cost penalties.
Keywords/Search Tags:Statistical, Economic, Cost, Process
Related items