Optimizing Corporate Decisions for Dominant Hydrocarbon Producers under Uncertainty

Open Access
Bukhari, Abdulwahab
Graduate Program:
Energy and Mineral Engineering
Doctor of Philosophy
Document Type:
Date of Defense:
October 23, 2014
Committee Members:
  • Turgay Ertekin, Dissertation Advisor
  • Anastasia Shcherbakova, Dissertation Advisor
  • Luis Ayala, Committee Member
  • Sarma V Pisupati, Committee Member
  • Paul Griffin, Committee Member
  • Yilin Wang, Committee Member
  • Anastasia Shcherbakova, Special Member
  • Optimization
  • Risk Analysis
  • Decision Analysis
  • VOI
  • petroleum economics
  • optimization
  • monte carlo simulation
  • stochastic
The production strategy of a dominant hydrocarbon producer can significantly affect the shape of the future of the hydrocarbon market (Nakov, A., and Nuno, G. 2011). This study focused on the role of a dominant producer in the hydrocarbon market; however, these findings could be applied to any industry extracting finite resources. The thesis was divided into four major chapters: 1) Capacity management, 2) Predicting decline parameters and development strategy for a capacity management model—an artificial expert system, 3) Optimal spare capacity, and 4) Integrated capacity management and spare capacity. Collectively, information offered in these four chapters can optimize the corporate decisions made by a dominant hydrocarbon producer that operates oil and/or gas fields. Each of these studies was discussed in each chapter. The first study had to do with Capacity Management (CM) that is a complex dynamic optimization problem whose goal is to manage production capacity across a portfolio of producing assets to maximize the total portfolio plateau length and/or NPV. Field-level constraints included facility limitations, costs, and technical properties (i.e., reservoir characteristics, and crude or gas types). Moreover, there also were company-level constraints (e.g., supply and spare capacity commitments). Although the problem could have been solved deterministically, the presence of uncertainty, in reservoir performance and the hydrocarbon market, made the probabilistic approach more realistic. This problem was tackled in three sections: (1) an integrated stochastic optimization model was built to solve for the optimal production rate allocation for a portfolio of oil or gas fields under the uncertainties of the markets and reservoir performances, (2) A sensitivity analysis was conducted for different market and reservoir models and parameters, and (3) A value of information (VOI) analysis was implemented to estimate the value of a more accurate expected demand. The various production allocation decisions, resulting from different economic and reservoir models, were compared with the goal of understanding the effect of these model parameters on decision-making. A Genetic Algorithm was implemented as the optimization algorithm. Results showed that the CM integrated model effectively maximized plateau length and mean NPV for the whole portfolio of fields. In addition, the optimal decisions in the CM problem were the function of the reservoir (i.e., reservoir quality, size, and maturity) and price (i.e., long-run equilibrium price) parameters. To maximize the total plateau length and/ or the NPV, production from youg fields with large decline exponent should be prioritized; then, mature fields with large decline expoenets. In other words, decline expoenet is the most impacting factor on the optimization problem. Capacity Management can add reserves with respect to time, delay development projects, increase profit, and reduce the uncertainty of the NPV. The relationship between the optimal production allocation decision and reservoir properties can be investigated for future studies; this may be achieved via regression analysis or machine learning techniques. This relationship can save significant amount of optimization time. The second section looks at developing an artificial expert system for the Capacity Management Model. The real-world applicability of a research study can be a key element in its success and enhance its value to oil and gas companies. Boosting a research study with this applicability can increase the reliability of the results since actual field data are being used rather than making an educated guess. The Capacity Management Model derived the proper production allocation for a portfolio of fields that maximized production plateau length while meeting the total target rate of output. This study was divided into two parts. In the first part, we predicted input parameters (decline curve parameters) for the CM model for given reservoir and development parameters. In the second part, we predicted a field development plan (drainage area) for a given set of reservoir properties, production rate and plateau length; production rate and plateau length were output from the CM model. Artificial Neural Network (ANN) and Genetic Programming (GP) methods were implemented to build the two prediction models. Predicting decline curve parameters and an optimal field development plan for any given set of reservoir properties constitutes a complex nonlinear optimization problem. Large variation in the magnitudes of various parameters (from 10e-04 for decline parameters to 10e07 for gas flow rates) further added to the computational complexity. These expert systems predicted decline parameters and a development plan for a specified system in a timely manner. In addition, no software can predict these parameters explicitly. In other words, the proposed expert system can predict endogenous parameters within seconds. Several existing studies addressed parameter estimation for the oil and gas industry (e.g., well testing results and multi-lateral well design) using artificial neural network methods. The prediction systems were tested against results from reservoir simulation; both systems showed excellent agreement with the simulation results. Moreover, the two implemented mechanisms (Artificial Neural Network and Genetic Programming) resulted in similarly accurate predictions that increased confidence in the constructed systems. The proposed models eliminated the need for a reservoir simulator in describing an appropriate production profile for a specified set of reservoir properties and development plans, and describing a development plan that achieved a required rate and plateau length for a given reservoir. Results showed that ANN and GP expert systems predicted decline parameters and dvelopoment plans with sufficient accuracy. For future studies, the expert systemts built in this study can be enhanced for more complicated reservoirs like unconventional and hetrogenous reservoirs. The third was spare capacity that can be a powerful tool for a major producer seeking to optimize its profit and increase its market power. Managing spare capacity under demand and price uncertainty is a complex optimization problem that requires sophisticated stochastic modeling and high computational capacity. This problem is best specified probabilistically, since it involves major investment in building and maintaining spare capacity volume, and may require a few years to develop the production facility. In other words, expected demand (at planning time) can be significantly different from actual realized demand. The problem was approached in three stages: 1) a stochastic optimization model was built to identify the optimal spare capacity under demand and crude price uncertainty, 2) sensitivity analysis was conducted for different demand and price forecasting models, and 3) a value of information analysis was conducted to quantify the impact of the accuracy of demand and price models. Acquiring spare capacity can increase profit and market power for a major crude oil producer. Spare capacity set the major producer as a price shock absorber (buffer mode), preventing prices from increasing (triggering alternative energy development) or from falling (reducing profit). Moreover, increasing crude prices can reduce the market share of a major producer by motivating the development of non-conventional and alternative energy sources. This study focused specifically on optimizing economic profits. We analyzed this issue by specifying an integrated stochastic optimization model that simulated demand for oil, and mapped demand forecasts on production decisions. The optimum spare capacity, estimated from different price and demand model assumptions, was evaluated with the purpose of understanding the impact of these assumptions on spare capacity decisions. Results showed a reasonably narrow range of spare capacity levels that maximized the total profits of a major oil producer. We recommend, for future studies, analyzing the relationship between the optimal spare capacity to acquire and market parameters include crude prices, world reserves and global demand. This will speed up the process and enhance the understanding of the problem. Moreover, our method implemented can be used to analyze the problem of optimal gas storage capacity given the seasonality in demand for natural gas. The last part of the study integrated the first two studies (capacity management and spare capacity). This was done by assuming that a major producer was producing a portfolio of existing oil fields at maximum production potential. The integrated spare capacity model recommended a level of spare capacity to build. After that, the capacity management model suggested the optimal production rate allocation. Results showed that the optimal production allocation was a function of reservoir properties and maturity.