Multiple response robust design and yield maximization

Abstract

A design is robust if the design values for selected performance characteristics (i.e. responses) are chosen to be invariant to the variations the product will experience. For a design to be acceptable, it must conform to the design specifications. However, due to the existence of variation, this conformance is satisfied probabilistically, i.e. yield. Optimal manufacturing yield design is defined as a design that maximises the probability of satisfying the design specifications. Methods to achieve robust design for a single response and to achieve yield maximization are well established. A new method of achieving high yield and robust design for multiple responses is presented using the Cp and Cpk capability indices used in on‐line quality control techniques. The proposed method is applied to a single response problem and two multiple response problems. The results showed that the proposed method is capable of producing good manufacturing yield and robust design simultaneously.

Development of a sequential optimization procedure for robust design and tolerance design within a bi-objective paradigm

“…While allowing some process bias, the resulting process variance is less than or at most equal to the variance obtained from the Vining and Myers model (1990): hence, the mean-squared error model may provide better (or at least equal) RD solutions unless the zero bias assumption must be met. Jayaram and Ibrahim (1999), Cho et al (2000), Cho (2000, 2002), Yue (2002), Miro-Quesada and Del Castillo (2004) and Shin and Cho (2005 discussed further modifications to the mean-squared error model.…”

Optimal Mean and Tolerance Allocation Using Conformance‐based Design

“…In the quality literature to date [1][2][3][4][5][6][7][8][13][14][15] a single objective function comprises the production cost and the socalled loss of quality cost. The production cost follows from (2).…”

Applying surrogate variables to deal with destructive characteristics for robustification using Six Sigma methodology

“…Myers, Brenneman and Myers [36] developed modified dual-response models using generalized linear model. Further modifications to the mean-squared error model have been discussed by Jayaram and Ibrahim [37], Cho, Kim, Kimbler and Phillips [38], Kim and Cho [39], Yue [40], Xu and Albin [41], Park and Cho [42], Miro-Quesada and Del Castillo [43], Shin, Gavindaluri and Cho [44], Shin and Cho [45], Lee, Park and Cho [46], and Shaibu, Cho and http://dx.doi.org/10.12776/mie.v14i3-4.454 3…”