(485e) Optimization of Crude Oil Purchasing and Blending Under Uncertainties | AIChE

(485e) Optimization of Crude Oil Purchasing and Blending Under Uncertainties


Wen, Y. - Presenter, Lamer University
Zhang, J. - Presenter, Lamar University
Xu, Q. - Presenter, Lamar University

Crude oil blending is a very common practice in petroleum refineries, where the main purpose is to minimize the total purchasing cost of crude oils and obtain qualified blending oils for processing. Obviously, the crude oil blending is subject to the crude oil inventories, which in turn will be dynamically affected by the plant purchase plan including the crude oil types, amounts, and their shipment dates over a planned period of time. Thus, the crude oil purchase and blending schedule should be simultaneously optimized to maximize the potential profitability of a refinery plant prior to the actual manufacturing. This becomes more necessary when the uncertainties of the crude oil price and availability arise.

In this paper, a bi-level MINLP-based methodology is presented to integrate the optimization of crude oil purchasing and blending under uncertainties. In the first level, the detailed schedule to guide the short-term blending operations will be optimally identified based on current inventories and purchase orders that have already been placed. The optimization results including the timing and quantity of both inflow and outflow for all the tanks and blenders. In the second level, the actual crude-oil purchase orders are unknowns. They will be optimally determined to obtain the best trade-off solution between the long-term plant profitability and its operational flexibility. An inventory-related index is developed to characterize such flexibility. Purchase orders generated in the level contain the crude oil type, price, and shipping arrival date. Finally, the crude oil purchasing and blending strategy with short-term and long-term profitability and flexibility will be implemented through the time horizon movement. The efficacy of the developed methodology is demonstrated by several case studies.