Advances in Dynamic Scheduling Problems

One of the core assumptions in classical scheduling theory is that job processing times are constant over the planning horizon; however, in practice, this assumption is rarely satisfied. Various dynamic scheduling models have been proposed to capture these scenarios. In this thesis, we consider three of these modelling methods: rate-modifying activities, which is an optional task that, when completed, allows subsequent jobs to be completed more efficiently; step-learning, which allows jobs that are started at or after a given date to be completed more efficiently; and agent-dependent scheduling, where skill sets of different agents are taken into consideration. The first two modelling mechanisms are cases of the established position-dependent and time-dependent time-varying processes, respectively, whereas the latter scenario extends on the classical multi-agent scheduling literature. This thesis consists of three parts. Chapters 2 and 3 examine how rate-modifying activities can be applied in competing two-agent scheduling problems. In Chapters 4 and 5, we study step-learning with due-date related scheduling criteria. In Chapter 6, we propose a new agent-dependent approach to scheduling.