This research continues the study of generic stochastic manufacturing systems. It includes the development of further results on simple almost optimal heuristics for parallel machine scheduling. It also examines a wide range of flow line problems, including probabilistic analysis of deterministic scheduling rules, new models for tandem lines with dependent service times, and some new performance measures aimed at reduced Work-In-Process and Just- In-Time manufacturing. New approaches to control of queueing networks and jobshops, such as Bandit problems and Brownian networks that may yield robust, simple, priority scheduling rules will be explored. Some recently started work in scheduling of a metal rolling mill, will provide a practical testbed for the combined application of stochastic and deterministic models. Here the stochastic models provide long term optimal throughput, while short term deterministic algorithms minimize setup and changeover costs. This is a realistic approach for the increasing uncertainty as the time horizon lengthens. Technology transfer will be achieved through the implementation of case studies and development of constructional manuals and tutorials for curriculum in manufacturing systems.
