Development and Utilization of Integrated Artificial Expert Systems for Designing Multi-lateral Well Configurations, Estimating Reservoir Properties and Forecasting Reservoir Performance

Open Access
Author:
Almousa, Talal Saeed
Graduate Program:
Petroleum and Natural Gas Engineering
Degree:
Doctor of Philosophy
Document Type:
Dissertation
Date of Defense:
June 24, 2013
Committee Members:
  • Turgay Ertekin, Dissertation Advisor
  • Turgay Ertekin, Committee Chair
  • Michael Adebola Adewumi, Committee Member
  • Li Li, Committee Member
  • Mirna Urquidi Macdonald, Committee Member
Keywords:
  • ANN
  • Artificial Neural Network
  • Expert Systems
  • Well Design
  • Characterization
  • Forecasting
  • Reservoir Simulation
Abstract:
Reservoir simulation is one of the main tools if not the most important one reservoir engineers use to forecast a reservoir performance. Nevertheless, developing and operating a reservoir simulator in the first place can be an arduous task that requires a set of highly skilled individuals in science, advanced mathematics, programing, and reservoir engineering and powerful computational models. The reliability of a reservoir simulator depends on the availability and the quality of the reservoir properties. These properties are obtained from open-hole logs, core studies and well testing analysis which can sometimes be prohibitively cost intensive. Another important component of the overall process affecting the reservoir performance is the multilateral well configuration. Achieving the right design of a multilateral well configuration is a complex problem due to the vast possibilities of well forms that need to be evaluated. In light of the above, this dissertation demonstrates the development and the application of a set of integrated artificial expert systems in the area of forecasting, reservoir evaluation and multilateral well design. The applied method has gradually progressed in degrees of complexity from addressing a preliminary case of volumetric single phase gas reservoirs completed with only dual-laterals towards an expanded form of the same system with varying multi-laterals and reservoir properties to eventually and successfully implementing it to multiphase reservoirs with bottom water drive systems completed with multi-laterals (choice of 2-5 laterals). The developed method and tools cover a wide spectrum of rock and fluid properties spanning tight to conventional sands. The developed approach successfully delivers a total of five distinct artificial expert systems, three of them serve as proxies to the conventional numerical simulator for predicting reservoir performance in terms of cumulative oil recovery, cumulative oil and gas productions and estimating the end of plateau and abandonment times and a third one for predicting cumulative fluid production. These aforementioned systems are categorized as forward-looking solutions. Whereas the other two artificial expert systems are categorized as inverse-looking solutions, one that addresses the multi-lateral well design problem and the other that estimates critical reservoir properties that can be used at the very least as first estimators in assist history matching problems and for improving the assessment of nearby prospects in field development or in-fill drilling exercises. Furthermore, graphical user interfaces (GUIs) in conjunction with the expert systems structured are developed and assembled together for standalone installation. These GUIs allow the engineer to edit and input data, produce results numerically and graphically, compare results with a commercial numerical simulator, and generate an interactive 3-D visualization of the multilateral well. It is expected that the developed integrated artificial expert systems will immensely reduce expenses and time requirements and effectively enhance the overall decision-making process. However, it is worth noting that the proposed expert systems are not to replace the conventional and well established procedures and protocols but rather as auxiliary or complementary applications, where applicable, to relief some of the computational overhead, provide educated estimation of key reservoir properties or the very least help fine tune them and present the inverse-looking solution to the multi-lateral well design problem or the least a starting point.