The oSTN graph-set: an informed input that improves the efficiency of continuous-time scheduling formulations

DIEGO M. GIMÉNEZ; GABRIELA P. HENNING

Cambridge, United States

Conferencia; Foundations of Computer-Aided Process Operations 2008; 2008

CACHE organization

Continuous-time formulations for the short-term scheduling of multipurpose batch plants are based on a set of event points (unit specific or global ones)/time slots, which are non-uniformly distributed along the scheduling horizon. The main limitation of these approaches is that the number of event points/time slots required to represent the optimal schedule is unknown a priori, so a given problem needs to be solved many times in order to reach the optimal solution. Therefore, two major challenges must be addressed to improve the computational efficiency. First, it is essential to reduce the number of iterations. Second, the computational load required to reach the optimal solution at each iteration should also be diminished. While the first goal can be achieved by properly estimating the number of event points/time slots to be considered at the first iteration, the second one can be tackled by a model size reduction. This paper formalizes a novel tool, called oSTN (ordered State-Task Network), that addresses these goals and renders continuous-time formulations that are potentially more efficient.