Optimization of Detailed Schedule for a Multiproduct Pipeline Using a Simulated Annealing Algorithm and Heuristic Rules

This paper addresses the optimization of delivery/injection plans of intermediate stations along a unidirectional multiproduct pipeline. A mixed-integer linear programming (MILP) model is developed to minimize pump rate variations in every pipeline segment along the pipeline during a specific scheduling horizon. Given the problem complexity, a computational framework is proposed consisting of three main stages: initial solution finding, solution improvement based on a simulated annealing (SA) meta heuristic, and solution refinement based on tailor-made heuristics. The algorithm for making an initial solution is based on a so-called space recursive method that the delivery/injection schedule of every intermediate station is gradually generated from upstream to downstream segments. The algorithm for generating a new solution by SA consists of three steps: selecting an intermediate station, choosing a batch passing by the selected station, and adjusting the selected operation. For improving the quality of solutions given by SA, the solutions are further fine-tuned based on a heuristic rule on the proper relay of delivery/injection operations carried out by different intermediate stations. Finally, the model and its algorithms are tested and analyzed using historical monthly plans' of two multiproduct pipelines.