TY - GEN

T1 - Selective assembly for maximizing profit in the presence and absence of measurement error

AU - Matsuura, Shun

AU - Shinozaki, Nobuo

PY - 2010/1/1

Y1 - 2010/1/1

N2 - Selective assembly is an effective approach for improving the quality of a product assembled from two components when the quality characteristic is the clearance between the mating components. A component is rejected if its dimension is outside specified limits of the dimensional distribution. Acceptable components are sorted into several classes by their dimensions, and the product is assembled from randomly selected mating components from the corresponding classes. We assume that the two component dimensions are normally distributed with equal variance, and that measurement error, if any, is also normally distributed. Taking into account the quality loss of a sold product, the selling price of an assembled product, the component manufacturing cost, and the income from a rejected component, we discuss the optimal partitioning of the dimensional distribution to maximize expected profit, including the optimal choice of the distribution limits or truncation points. Equations for a set of optimal partition limits are given and its uniqueness is established in the presence and absence of measurement error. It is shown that the expected profit based on the optimal partition decreases with increasing variance of the measurement error. In addition, some numerical results are presented to compare the optimal partitions for the cases when the truncation points are and are not fixed.

AB - Selective assembly is an effective approach for improving the quality of a product assembled from two components when the quality characteristic is the clearance between the mating components. A component is rejected if its dimension is outside specified limits of the dimensional distribution. Acceptable components are sorted into several classes by their dimensions, and the product is assembled from randomly selected mating components from the corresponding classes. We assume that the two component dimensions are normally distributed with equal variance, and that measurement error, if any, is also normally distributed. Taking into account the quality loss of a sold product, the selling price of an assembled product, the component manufacturing cost, and the income from a rejected component, we discuss the optimal partitioning of the dimensional distribution to maximize expected profit, including the optimal choice of the distribution limits or truncation points. Equations for a set of optimal partition limits are given and its uniqueness is established in the presence and absence of measurement error. It is shown that the expected profit based on the optimal partition decreases with increasing variance of the measurement error. In addition, some numerical results are presented to compare the optimal partitions for the cases when the truncation points are and are not fixed.

UR - http://www.scopus.com/inward/record.url?scp=78649889691&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=78649889691&partnerID=8YFLogxK

U2 - 10.1007/978-3-7908-2380-6_12

DO - 10.1007/978-3-7908-2380-6_12

M3 - Conference contribution

AN - SCOPUS:78649889691

SN - 9783790823790

T3 - Frontiers in Statistical Quality Control 9

SP - 173

EP - 190

BT - Frontiers in Statistical Quality Control 9

PB - Physica-Verlag

T2 - 9th International Workshop on Intelligent Statistical Quality Control

Y2 - 1 September 2007 through 1 September 2007

ER -