Solving a Supply-Delivery Scheduling Problem with Constraint Programming

We describe a constraint programming approach for a supply-delivery problem in the petrochemical industry, in which barges transport liquid material from supplier locations to downstream processing plants. The problem is to design a pickup-and-delivery route for each barge such that given minimum and maximum inventory levels at each location are met for a given fleet size. This optimization problem is part of a larger planning system to determine the fleet size, negotiate pickup windows and quantities, and design operational schedules. We evaluate our model on representative supply networks provided by BP North America, and contrast our results with those obtained by a mixed-integer programming approach.