Resilient scheduling under uncertain processing times: a hybrid CP/TOC approach

Abstract

Uncertainty and variability are inherent characteristics of industrial environments that affect production schedules and could turn them unfeasible or economically unattractive. Process-inherent uncertainty, one of the multiple stochasticity sources, influences task processing times, transforming them into one of the main uncertain parameters. This work proposes a scheduling approach to consider this situation in a proactive fashion at the decision stage, without resorting to the generation of scenarios. It relies on a Constraint Programming (CP) model that focusses on the Capacity Constrained Stage (CCS) to reduce its complexity and size. The proposal is tested with various instances of three case studies, showing that the approach is computationally efficient. The attained agendas can effectively absorb the processing times variabilities in an efficient way, exhibiting a resilient behaviour. To assess the schedules, they are compared with those reached by a deterministic CP formulation and with the agendas obtained by means of another proactive methodology proposed by the same authors.