ABSTRACT

A good policy for avoiding deadlock should enforce the least restrictions on flexible manufacturing systems (FMS) to increase resource utilization and productivity, or it should be maximally permissive. Toward this, much outstanding work based on various discrete event models was reported (Kim and Kim, 1997; Reveliotis and Ferreira, 1996; Fanti et al., 1997; Xing et al., 1996; Lawley et al., 1998; Wu, 1999). A necessary and sufficient condition for deadlock-free operation and a maximally permissive control policy for a class of FMS are derived using the resource-oriented Petri net (ROPN) model (Wu, 1999). Its significance lies in the identification of a boundary for deadlock to occur in automated manufacturing systems (AMS). However, a new question arises regarding whether such a maximally permissive policy is optimal in terms of production rate in an environment where dispatching rules dominate and optimal scheduling is impossible due to either the unaffordable computation required or changing operational parameters and structures.