Title: A modern two-stage stochastic programming portfolio model for an oil refinery with financial risk management

Authors: Patrick Johnson O'Driscoll

Addresses: Department of Economics, Mathematics and Statistics, Birkbeck, University of London, Malet Street, London WC1E 7HX, UK

Abstract: The proposal which we wish to make is a two-stage stochastic programming model for a competitive oil refinery with stochastic crude and fuel prices. Most models for refineries are deterministic, and those considering the stochastic problem do so by utilising a Gaussian assumption on profits - implementing variance as the risk measure. Our model falls into the category of optimisation with coherent risk measures where robustness, rather than ambiguity, is the focus. The objective is to maximise the refiner's profit under raw material, product inventory constraints and a financial risk constraint. The two-stage model leverages off a unique discrete scenario generation technique alongside an admissible and computational tractable drawdown risk measure. The expected value of perfect information calculation of each model gives a value for the additional benefit, which the decision-maker receives in considering the uncertainty inherent in the problem.

Keywords: stochastic programming; refinery planning; optimisation; uncertainty; probability fitting; mean variance; conditional drawdown-at-risk; CDaR; oil refineries; financial risk management; profitability; raw materials; product inventory.

DOI: 10.1504/IJOR.2017.080602

International Journal of Operational Research, 2017 Vol.28 No.1, pp.121 - 140

Received: 04 Jun 2014
Accepted: 09 Apr 2015

Published online: 01 Dec 2016 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article