Restricted (Penn State Only)
Da, Lina
Graduate Program:
Petroleum and Mineral Engineering
Master of Science
Document Type:
Master Thesis
Date of Defense:
December 04, 2017
Committee Members:
  • Turgay Ertekin, Thesis Advisor
  • 123, Committee Member
  • Artificial Neural Network
Since the development of the oil industry, engineers have been working to improve the production of wells. In order to maximize the production of a wells, many parameters need to be considered, and the angle of the wellbore trajectory is one of the most important parameters. In recognition of this importance, this thesis focuses on methods of choosing different slanted well configurations’ angles for reservoirs with various characteristics using artificial neural network (ANN) expert system. To achieve this goal, a three-phase black-oil reservoir model was built at the beginning of the study. The reference data for the model come from Ertekin (2001). Two programs, MATLAB1 and software IMEX2 from Computer Modeling Group (CMG), were used to create the reservoir model; these tools were also used to generate the various slanted well configurations. A total of 26 different wellbore trajectories were built to validate the results of this study. The ANN expert system is a built-in MATLAB that is based on the Monte Carlo method. This expert system requires a large amount of random sampling data to obtain numerical results. Accordingly, hundreds of different reservoir models were randomly built with their parameters changed in certain ranges. In the end, three kinds of ANN expert systems were shown to be able to predict different expected values. The forward ANN model was found to predict oil, gas, and water production data. The first inverse ANN model was shown to predict reservoir thickness, permeability, porosity, area, and bottomhole pressure. Finally, the second ANN model was found to predict the optimum slanted well configuration design angle.