A computer-aided approach for optimal statistical design

An approach for design centering and tolerancing to achieve 100% yield is introduced. The procedure consists of steps solving three optimization problems: first, centering the design point to the vicinity of the optimum point in the design parameter space, second, solving a fixed tolerance problem, and third, solving a variable tolerance problem, in which parameter tolerances are reassigned to achieve 10% yield and minimum cost. Two numerical examples testing the approach are presented.<
>