Content area
Full Text
Acceptable results always require expert human intervention to manage the interactions among process variables and to balance correlated outputs. When such interventions are incorrect, however, they can cause quality problems and equipment damage. Even when they are correct, they are stressful because of the high degree of skill and attention required.
The objective in manufacturing, therefore, should be to minimize human intervention by using robustness concepts to detach and automate process interactions or to detach the correlation between output characteristics and eliminate the need for delicate balancing. This requires much more than merely adding mistake-proofing mechanisms that act only upon known human errors, without providing knowledge-based human intervention.
We create robust manufacturing processes that minimize failure by first understanding the interactions between uncontrollable variables (noise) and controllable variables (signal) and then designing process mechanisms to deal with them without human intervention. Classic definitions of robustness, however, do not include the case of correlated characteristics--that is, situations where any attempt to improve one characteristic will degrade the other. Yet robustness can be used to detach this correlation so that...