Process setup adjustment with quadratic loss

A classic adjustment method, the harmonic adjustment rule, minimizes expected quadratic loss where adjustments are easy and cheap, and the items produced during the adjustment process should be as close to target as possible. This paper generalizes the harmonic adjustment rule by explicitly taking into account measurement and adjustment costs. The generalization is static, in that adjustments are performed at predetermined points. An optimal model and a near-optimal approximation, in which adjustments are spaced roughly according to a geometric sequence, are presented. Skipping small adjustments dynamically may offer further savings, but dynamic models require further research.