The short-term scheduling of refined products pipelines is a very complex problem with many operational constraints to be considered. Available nondiscrete planning approaches just provide the input schedule and the set of aggregate product deliveries to be performed from in-transit lots to depots at every batch injection. To determine the sequence of individual cuts on each batch to be accomplished by the pipeline operator, it is still necessary to refine such an aggregate plan. To do so, new computational tools for efficiently generating the detailed schedule of single-source pipelines with multiple distribution terminals are presented. Two types of methodologies are proposed. On one hand, it is developed a continuous-time mixed-integer linear programming (MILP) formulation that seeks to minimize both the total flow restart volume and the number of single-delivery pumping runs over the planning horizon. In this way, substantial savings in energy consumption and pump maintenance costs are achieved. Effective solution strategies for the MILP model are also designed to deal with large pipeline scheduling problems. On the other hand, three different heuristic rules for selecting the receiving terminal are introduced. By applying those rules in combination with a discrete-event simulation model, not only alternative detailed schedules can be generated in a very short CPU time but also some of them are near-optimal solutions. Three instances of a case study aimed at finding the detailed schedule of a real-world single-source pipeline system are solved through the proposed optimization and discrete-event simulation methods. Results are analyzed to assess the quality of the generated solutions and the required computational costs.

Continuous formulations for the scheduling of single-source refined-product pipelines with multiple distribution terminals have already been published. They help the scheduler make key operating decisions such as the sequence of batch injections at the input station, the batch sizes, the start and end times of pumping runs, and the set of aggregate product deliveries to perform during each run. In this way, the aggregate pipeline schedule is defined. However, the scheduler has yet to decide not only on the order of execution of such product extractions but also the number of batch cuts to make to accomplish each aggregate delivery. Therefore, it becomes necessary to refine the aggregate schedule. Most previous contributions providing a refined delivery plan have assumed that product withdrawals at receiving locations over a single-source pipeline should be made one at a time. However, such pipelines do not actually work in this way because their operators usually perform simultaneous product deliveries to cut down on pump operating and maintenance costs. By doing so, they seek to reduce the number of pipeline segments where the liquid flow should be restarted. This article presents a new mixed-integer linear programming (MILP) continuous formulation for developing the detailed schedules of single-source pipelines assuming that simultaneous deliveries to multiple receiving terminals can be performed. The problem goal is to minimize the overall flow restart and stoppage volumes through the smallest number of pipeline operations. Solutions to two examples obtained with the proposed approach present significant reductions in both the operating cost and the CPU time with respect to previous contributions.

No abstract