Comparison of Continuous-Time Models for Adjustable Robust Optimization in Process Scheduling under Uncertainty

Employing Robust Optimization to address uncertainty in process scheduling has been traditionally done in a static manner, where the entirety of the decisions are takenhere-and-now, without any flexibility to later adapt-and capitalize upon-the actual uncertainty realizations. We recently introduced an Adjustable Robust Optimization framework, which has the ability to incorporate wait-and-see decisions to be taken at multiple stages during the scheduling horizon, and we showed how it can provide solutions that are less conservative, i.e., solutions that are more profitable while insuring against the same level of uncertainty, than the solutions obtained with the traditional approach. In this paper, we extend our previous results by utilizing a number of alternative deterministic scheduling models as the basis of our framework, and we investigate the computational tractability of each model via a comprehensive computational study based on a series of literature benchmark instances.