Sequential Decision-Making Methods on Real-time Optimization of Pump Scheduling of Refined Oil Pipelines

The pump scheduling optimization plays an important role in the pipeline transportation of refined oil. The accurate and efficient pump scheduling can not only ensure safety of the pipelines but also help reduce the cost and carbon emission and accelerate the carbon neutrality process. The operation conditions of the refined oil pipelines are complex and there is often flowrate adjustment during the actual operation of the pipelines, not exactly according to the initial batch schedule. Therefore, it’s necessary to optimize the pump scheduling of refined oil pipelines based on the real-time operation conditions to guarantee safety of the operation. In this work, sequential decision-making methods, including myopic policy and model predictive control (MPC) algorithm, are utilized to refactor the conventional MILP model for realizing the real-time optimization. Finally, several cases are designed to verify the availability of proposed methods. The results validate the effectiveness and superiority of the proposed MPC model in real-time optimization of pump scheduling of refined oil pipelines.