FORECASTING THE PRODUCTION PERFORMANCE OF WELLS LOCATED IN TIGHT OIL PLAYS USING ARTIFICIAL EXPERT SYSTEMS

Open Access
Author:
Bansal, Yogesh
Graduate Program:
Petroleum and Mineral Engineering
Degree:
Doctor of Philosophy
Document Type:
Dissertation
Date of Defense:
November 28, 2011
Committee Members:
  • Turgay Ertekin, Committee Chair
  • Zuleima Karpyn, Committee Chair
  • Luis F. Ayala, Committee Member
  • Li Li, Committee Member
  • Kultegin Aydin, Committee Member
Keywords:
  • forecasting production
  • Tight oil
  • infill drilling
  • ANN
  • optimized completion design
Abstract:
The potential of unconventional oil and gas reservoirs is promising to account for the declining conventional supplies in the future. However, because of their complex nature, it is uneconomical to produce from of these resources with the current state of technology. In addition, these resources are relatively new (from the development point of view), thus, it is difficult to completely characterize these resources in the absence of their respective analogs. This study focuses on tight oil reservoirs. Characterization of these resources is a complex problem as tight oil systems are discontinuous hydrocarbon sources. Developing these resources by identifying the location to drill, estimating well performance and suggesting a completion strategy will be a challenge in the absence of a representative reservoir model. An inexpensive and field-deployable expert systems-based tool has been proposed in this study to characterize such unconventional reservoirs. A group of inter-assisting expert systems are developed, where the individual capabilities lie in suggesting completion parameters and predicting quarterly cumulative production of oil, water and gas for a two-year period. These expert systems are grouped together to suggest the best infill drilling location in the field with a forecast of their respective cumulative productions by the end of two years. The predictions from the expert systems-based tool are found to be in good agreement with field performance. Production surfaces generated by these expert systems are found to reflect the actual productions obtained in the field. In addition, a hybrid optimization method is also developed in this work. The method is used to optimize the well completion parameters in a tight oil reservoir. The tools developed in this work will help in a quick evaluation of tight oil reservoirs. The results discussed in the dissertation show the accuracy of predictions made by the expert systems. The production characteristics and optimized completion design parameters of a well predicted by the expert systems will help in developing a tight oil reservoir more efficiently and economically.