A Distributionally Robust Chance Constrained Model to Hedge Against Uncertainty in Steelmaking-continuous Casting Production Process

This paper proposes a distributionally robust chance constrained (DRCC) model to handle the daily small disruptions in steelmaking and continuous casting process. The processing time of each charge is assumed to be a random variable belonging to an ambiguous distribution set which is described by support set, mean and variance information. The proposed DRCC model aims to minimize the objective function and at the same time ensure each constraint is established with a certain probability, i.e., distributionally robust individual chance constrained (DRICC) model, or all constraints together are established with a certain probability, i.e., distributionally robust joint chance constrained (DRJCC) model even when the uncertain parameters are in their worst cases. We transform DRICC model into a linear programming model and propose an iterative improvement method to tackle the DRJCC model. To test the robustness of the models, we evaluate the obtained robust schedules under different distributions of the processing time. Experimental results show that both DRCC models are able to provide more robust schedules than compared to the deterministic model.