Dynamics of WIP Regulation in Large Production Networks of Autonomous Work Systems

In this paper, dynamic behavior is compared for two methods of local work in progress (WIP) regulation in autonomous work systems in production networks. In one method, work systems do not share information regarding the expected physical flow of orders between them; in the other, order-flow information is shared to compensate for the variable dynamic effects of physical order-flow coupling. In both methods, the work systems adjust production rate with the objective of maintaining a desired amount of local WIP. A linear discrete-time dynamic model of the flow of orders between work systems is used, which promotes identification of fundamental properties such as characteristic times and damping. The results demonstrate the need for order-flow information sharing in establishing desired network dynamic behavior. Examples are used to illustrate behavior in the general case of omnidirectional order flows and the special case of unidirectional order flows.