A TREE PARTITIONING ALGORITHM FOR ACTIVITY FORMATION IN AGGREGATE SCHEDULING

In project-oriented production systems where hundreds of thousands of operations are carried out each month subject to precedence and resource constraints, long-term scheduling is a challenging problem for any branch of operation research or artificial intelligence. In a dynamically changing environment, fixing a schedule for several months in advance would be pointless, and, furthermore, the solution of the explicit scheduling problem on atomic operations could easily turn computationally prohibitive. The customary technique at this point is aggregation. In this paper we address aggregate scheduling in components manufacturing, where the operation precedence graphs of projects form trees, and there are no alternative routings.