A scheduling approach with uncertainties in generation and consumption for converter gas system in steel industry

Abstract Linz–Donawitz converter gas (LDG) is a class of significant secondary energy in steel enterprises, and its adequate recycling can help save energy, reduce emission and improve profit. Aiming at the scheduling problem with uncertainties in both gas generation and consumption, a three-layer-causal-network-based approach for LDG system is proposed in this study. Two causality-based methods by employing causal exception estimation and working condition clustering are designed to construct the prediction intervals (PIs) of the gas generation and consumption flows respectively. Given that the surplus or shortage range of the LDG that can keep the tank levels balance can be calculated, a three-layer causal network of “generation-storage-consumption” is established to describe and evaluate the scheduling rules. Then, the solution is finally optimized based on the objective function and its corresponding constraints constructed by the safety, environmental and economic indicators. To verify the performance of the proposed method, the experiments by using real-world data coming from a steel plant are carried out, where the manual approach and the existing mixed-integer-linear-programming-based one in literature are also employed as comparative ones. The results indicate that the proposed method is capable of providing superior performance for such an industrial application.