(368c) Material Pipeline Scheduling and Inventory Planning Under Uncertainty

Inventories of raw materials for chemical manufacture are managed at various distributed sites within Dow based on supply and demand. Various sources of supply and demand exist at each of these sites, some dedicated and some shared. Multiproduct pipelines for material transport connect these different sites facilitating material movement to manage inventory. Various planning, scheduling and forecasting activities take place in this complex framework involving uncertainties around demand, supply and pipeline flows. In this work, we will present case studies where multiple optimization methods were applied to scheduling a multiproduct pipeline while simultaneously optimizing material inventories. Uncertainties around pipeline flows were factored in based on historical data.  Results from stochastic optimization and deterministic optimization as well as a hybrid method will be compared. Additionally, results from these optimization methods will be compared to manually planned schedules. Certain practical caveats to applying scheduling optimization methods to this problem will also be discussed.