Product Blending: A Simulation Case Study in Double-Time

This paper describes what a study team at Exxon did when faced with the need to determine rapidly the impact of new gasoline blend regulations on production. The study team designed and built from scratch in under a month a simulation model of the gasoline supply system at a refinery. The model basis used—letting the user control blending decisions through data input rather than building a model that can decide heuristically on blend decisions—yielded a very easy to use, accurate model applicable to a wide range of general blending/inventory control problems both inside and outside the oil business. The paper also demonstrates that a simple yet adequate model may sometimes yield better results than a more sophisticated model. It shows how factors like ease of understanding, low cost development, and timely response can be crucial in getting a study implemented. The use of this model was directly responsible for the determination that several tankage additions originally envisioned would not be required. As a result, Exxon was able to save at least 1.4 million dollars.