Sorry, you need to enable JavaScript to visit this website.

Skip to content Skip to navigation

ERE PhD Defense: Chuan Tian — Machine Learning Approaches for Permanent Downhole Gauge Data Interpretation

Date and Time: 
May 18, 2018 -
9:30am to 10:30am
Location: 
Green Earth Sciences Building, Room 014
Contact Email: 
jpcastro@stanford.edu
Contact Phone: 
650-725.9835
Event Sponsor: 
Energy Resources Engineering

PhD Defense: 
Chuan Tian, Dept of Energy Resources Engineering

Machine Learning Approaches for Permanent Downhole Gauge Data Interpretation

Permanent downhole gauges (PDGs) provide a continuous record of pressure, temperature, and sometimes flow rate during well production. One approach to analyze the rich information contained in PDG data is deconvolution, i.e. extracting a constant rate drawdown pressure from the multirate pressure-rate history. In this work, a machine learning based deconvolution approach was developed. The developed approach was
shown to identify the reservoir models successfully from the multirate data, and it outperformed conventional industry methods developed by von Schroeter et al. and Levitan et al. when noise or outliers were contained in the data for deconvolution. Other applications of machine learning for PDG data analysis including flow rate reconstruction and multiwell testing, were also investigated in this work.

More Information