Using CLP and MILP for Scheduling Commodities in a Pipeline

This paper addresses the problem of developing an optimization structure to aid the operational decision-making in a real-world pipeline scenario. The pipeline connects an inland refinery to a harbor, conveying different types of commodities (gasoline, diesel, kerosene, alcohol, liquefied petroleum gas, etc). The scheduling of activities has to be specified in advance by a specialist, who must provide low cost operational procedures. The specialist has to take into account issues concerning product availability, tankage constraints, pumping sequencing, flow rate determination, and a series of operational requirements. Thus, the decision-making process is hard and error-prone, and the developed optimization structure can aid the specialist to determine the pipeline scheduling with improved efficiency. Such optimization structure has its core in a novel mathematical approach, which uses Constraint Logic Programming (CLP) and Mixed Integer Linear Programming (MILP) in an integrated CLP-MILP model. Moreover, a set of high-level modeling structures was created to straightforward formulate the CLPMILP model. The scheme used for integrating techniques is double modeling (Hooker, 2000), and the CLP-MILP model is implemented and solved by using a commercial tool. Illustrative instances have demonstrated that the optimization structure is able to define new operational points to the pipeline system, providing significant cost saving.