An MDP Model for Integrated Operational Decision Making Under Uncertainty in a Gas and Oil Separation Plant

Abstract

Oil and gas supply chain (SC) plays key role in the global economy. Gas and oil separation plants (GOSP) belong to the upper stream of the oil and gas SC. In this article we study a GOSP that pumps oil to a stabilizing plant through a 600 KM pipeline. The purpose of this research is to develop an integrated mathematical model to support operational decision making regarding the optimal pumps scheduling and the dosage level of a drag reduction chemical, which is injected in the oil pipeline to stabilize the pressure and achieve higher oil flow rate. To the best of our knowledge, there is no MDP model in the literature that jointly considers pumps scheduling and oil flow control. A novel Markov decision process (MDP) model and two intuitive heuristic polices are proposed and simulated based on historical data. The heuristic policies are to operate the oil pumps in a cyclic weekly or biweekly patterns and to use a maximum likelihood rule to select the dosage level. Compared to the heuristic policies, our results demonstrated that MDP can lead to a substantial amount of savings in terms of the total system operating and maintenance costs. It also provides some practical insights on the interaction ways between the frequency of operating the pumps, the maintenance costs and the costs of a chemical that is used to control the pressure and flow rate of oil in a 600 KM oil pipeline.

Keywords

• Decision making under uncertainty
• gas and oil separation plant (GOSP)
• Markov decision Process (MDP)
• scheduling