DEVELOPMENT OF ARTIFICIAL NEURAL NETWORKS FOR HYDRAULICALLY FRACTURED HORIZONTAL WELLS IN FAULTED SHALE GAS RESERVOIRS

Open Access
Author:
Oz, Sinan
Graduate Program:
Petroleum and Natural Gas Engineering
Degree:
Master of Science
Document Type:
Master Thesis
Date of Defense:
April 23, 2014
Committee Members:
  • Turgay Ertekin, Thesis Advisor
Keywords:
  • artificial neural networks
  • hydraulic fracture
  • horizontal well
  • shale gas reservoir
Abstract:
There is no alternative energy to replace fossil fuels yet, demand for hydrocarbon is still increasing all over the world. In addition to that, productions from conventional reservoirs are not sufficient enough to fulfill the demands. This has led to increase interest of production from unconventional reservoirs especially from shale gas reservoirs. Numerical simulations of reservoirs are complex problem, and can be time consuming. Accordingly, Artificial Neural Network (ANN) technology can be used for the non-linear relationships to avoid excessive time requirements of reservoir modeling. The purpose of this study is to develop network models that can generate accurate results from large amount of historical data in shale gas reservoirs. Variables used in this study include reservoir and well design properties such as well length, porosity, permeability, etc. Three network models are developed and a total of 1040 sample runs were generated to train these networks. The first network predicts production profile for a given reservoir and well design parameters. Second neural network estimates well design parameters by using reservoir parameters and production profiles. Finally, the third calculates the reservoir parameters as it predicts unknown reservoir parameters from given gas rate and cumulative production data, and well design parameters. Results of this study show that ANNs are able to estimate the unknowns of the problem within error margin of 5%.