Systematic trial and error

The application of a highly fractional factorial design (HFFD) to a manufacturing process that makes electromagnetic clutch coils is described. An HFFD or 'screening design' is used to separate the few vital factors from the many peripheral ones, so that the root cause of a problem can be isolated and dealt with. The results collected made it possible to compute the effects due to each process parameter considered, and to identify the critical parameters that lead to significantly more problems. An effect caused by a two-level parameter refers to the change in average response caused by a change in average response caused by a change in the level of the parameter.