The scheduling of products pipelines was a very complex task that many decision variables and constraints needed to be considered. In this paper, a single source and multi-terminals products pipeline system was studied. A novel optimization model was developed, while the Simulated Annealing algorithm was applied. In contrast to existing literature, a new objective function was proposed, that the flow rate of each pipe segment changing with time during the scheduling horizon was kept to be minimum. The optimization process was divided into two stages. The first stage generated the initial batch plans by the Space Recursion method. In this stage, the flow rate and time constraints of delivering operations, the constraints of material balance, the constraints of batch tracking and the constraints of market demands were considered. But the initial solution was usually unfeasible in that the flow rate constraints of pipe segments along the pipeline weren't satisfied. For achieving a feasible and optimal solution, new solutions were continuously generated by the SA algorithm and the poorer new solutions were probabilistically accepted in the second stage. The approach successfully solved a real China products pipeline called LC involving a source and ten terminals, and transporting three different oil refined products.


During the past fifty years, the researchers have been studying on the optimization of the scheduling of products pipelines. The scheduling of products pipelines referred that the dispatchers made batch plans according to the supply capacity of oil source, the market demands of terminals and the transmission capacity of pipeline. Many optimization models were proposed, including the mathematical models, the heuristic models and the simulated models. The progress on this issue was reflected in that: the complexity of study objects, the strictness of constraints, the time expression methods, the scheduling horizon and the heuristic rules.

The study objects became more complex. The single-source and single-terminal pipeline system, the single-source and multi-terminals pipeline system, the multi-sources and multi-terminals pipeline system, the tree structure pipelines system and the mesh structure pipelines system were chronologically studied.

The constraints of optimization models became stricter and more applicable to the reality. The scheduling models of products pipelines were generally mixed integer non-linear programming (MINLP) or mixed integer linear programming (MILP). There were so many constraints and decision variables that the optimization models could not achieve a feasible solution in a reasonable time, let alone an optimal solution. For decreasing the computational time, the early literature simplified many constraints, causing that the optimization results could not be applied to the reality. For example, Diego C. Cafaro and Jaime Cerdá (2009) assumed that only a source could synchronously perform injecting operation. As the research continued, Diego C. Cafaro and Jaime Cerdá (2010) proposed an improved model, which allowed simultaneous batch injection at different sources.

This content is only available via PDF.
You can access this article if you purchase or spend a download.