Human Behavior, Computation, and the Design of Manufacturing Systems Human Behavior, Computation, and the Design of Manufacturing Systems

This paper suggests that a major deeciency in computational approaches to manufacturing design is the lack of applicable models of human behavior. Starting with a discrepancy between the practices of lean manufacturing|practices that recommend small interprocess buuer sizes|and the practices of manufacturing simulation|practices that treat human behavior as an uncorrelated, random process, which when optimized recommends large interprocess buuer sizes|two plausible models of diierent facets of human behavior are created, simulated, and reconciled with the smaller buuers of lean manufacturing systems. The rst model, the local demotivation-motivation model, suggests that lean thinking can be partially explained by the general tendencies of human beings to (1) get behind as their work piles up, and (2) work much harder as a deadline or constraint approaches. Representative simulations demonstrate this eeect and show how keeping buuers small can oftentimes be the wisest course of action. The second model, the line-wise helpfulness model, suggests that in systems where co-workers can be encouraged to help one another that a form of adaptive or dynamic buuering can take place between many processes; this kind of buuering can be more eeective than the static kind when used alone, and the buuering eeect can be maximized by using small buuers, which tend to permit more eecient propagation of information between stations. The paper concludes by suggesting that more modeling combining insight into human behavior, manufacturing savvy, and an appropriate computational-analytical basis is necessary if eeective computational design optimization of manufacturing systems is to become more streetwise and thus more widely believed and used.