Batson, Robert G.; Phillips, J. Fred (1992, ASQC) University of Alabama, Tuscaloosa, AL 35487
American Cast Iron Pipe Company improved the quality of their ductile iron pipe by experimenting with two critical variables.
In the core-making experiment, the ACIPCO quality team chose four process parameters, then created an experimental layout studying parameters interaction based on Taguchi's L16 array. The sample pipes were randomized in order and for various chemical and machining factors. The team conducted analysis of variance and signal-to-noise analysis on the resulting pipe measurements. The team was able to justify the purchase of a computer-controlled system for mixing pipe resin.
In the silicon addition experiment, the team looked at three introduction and one timing method for the ferro-silicon additive. They chose Taguchi's L81 array to show the two-, three- and four-way interactions between introduction and timing methods. They varied the factors and levels for each pipe run. They found that one of the introduction methods offered a substantial shift in the mean, compensating for most over- or under-size conditions. After applying ANOVA to the results, the team was able to provide a means to select online control actions adjusting the additive. This experiment improved performance in the Melting Department and justified a new spectrograph for the Laboratory Department.
Taguchi method,Steel industry,Noise,Chemical and process industries,American Cast Iron Pipe Company (ACIPCO),Analysis of variance (ANOVA)