Taguchi and TQM: quality issues for surface engineered applications

Industrial organisations, including manufacturers and users of surface coatings, are increasingly seeking to ensure conformance to target specifications and low variance; using measured variables in a product or process that can be used to assess quality, for example, using Taguchi's loss function. In a TQM environment, the drive is to constantly improve quality. However, an engineer may regard a system not exhibiting a normal distribution function as one lacking quality. We discuss how a "non-normal" physical phenomenon can be accepted as a "quality" system. One such system example is a coated cold forming tool. Using design of experiment techniques in an industrial environment, it has previously been shown that a TiN coating was only effective at reducing tool wear in a forming operation if it was combined with pre-polishing of the tool nose. Not only was the improvement quantifiable using the Weibull characteristic life, it also showed that the population density function changed dramatically, indicating that the failure mode had altered between the best and the rest of the tools - apparently increasing the variance as well as the mean response. This paper compares the previous analysis to one based on Taguchi's robust design ideas. The importance of choosing an applicable quality loss function is highlighted. The use of the Weibull characteristic life is proposed and demonstrated for the functional limit (Δ₀) in the quality loss function. In this application it is shown that the relative loss of quality can be reduced by over 40% even considering a 25% increase in tool cost by using polished and coated forming tools compared to uncoated tools.