Selecting the Optimal Deadlock Resolution Strategy in Buffer Space Allocation of Flexibly Automated Production Systems: An Analytical Perspective

Abstract

Motivated by the emerging need to reconsider the problem of selecting the optimal deadlock resolution strategy for buffer space allocation in some contemporary production environments — e.g., semiconductor manufacturing cluster tools — this work extends past results on deadlock resolution in sequential resource allocation systems in a number of different ways.First, it extends the formal framework modeling the resource allocation taking place in these environments, by introducing a probabilistic structure on the considered job routing schemes, and bringing, thus, the whole analysis in the context of kprobabilistic automata [17, 14]. In this new modeling paradigm, it introduces the concept of randomized deadlock avoidance policies (R-DAP’s), which establishes a continuum between the two extreme options of deadlock prevention avoidance and detection & recovery, traditionally recognized as the possible deadlock resolution strategies.Finally, it outlines an analytical framework for selecting optimally the deadlock resolution strategy that maximizes the system production throughput. The defining logic and the effectiveness of this framework are demonstrated by applying it on a prototype case study, while the obtained results admit an interesting intuitive interpretation.