Control policies conciliating deadlock avoidance and flexibility in FMS resource allocation

Deadlock arises in FMS when a job set is in "circular wait", i.e. each job in the set waits for a resource held by another job in the same set. This is an unfavourable situation that can occur in production systems with high routing flexibility. Facing this problem requires control policies that avoid deadlock conditions by ruling resource allocation and deallocation. Often, by reducing the allowed options in resource usage, these policies lead to poor production system performance. On the other hand, more flexible approaches to deadlock avoidance usually need more complex control algorithms. This paper considers two deadlock avoidance policies that use feedback of the production system state to allow or inhibit resource allocation or deallocation. Both the policies are stated using a digraph-theoretic approach describing possible and current job-resource interactions. The former policy has been already introduced by the authors in a previous work. The latter is original and, by extending the basic ideas that motivate the previous one, allows wider flexibility in resource use, even if it involves higher computational complexity. The paper compares the two policies and a third one proposed by other authors, by analysing two examples by discrete event simulation, This confirms that the new policy results in better performance measures.