Brittleness is one of the most important reservoir properties for unconventional reservoir exploration and production. Better knowledge about the brittleness distribution can help to optimize the hydraulic fracturing operation and lower the costs. However, there are very few reliable and effective physical models to predict the spatial distribution of brittleness. We propose a machine learning based method to predict the subsurface brittleness by using a multi-discipline dataset, which includes seismic attributes, rock physics, and petrophysics information, allowing the implementation of brittleness prediction without using a physical model. The method is applied on a dataset from Tuscaloosa Marine Shale (TMS) and the predicted rock physics template (RPT) labels are close to the calculated labels from conventional inverted elastic parameters. Therefore, the proposed method helps to determine reservoir areas that have optimal geomechanical properties for successful hydraulic fracturing.
Presentation Date: Monday, October 12, 2020
Session Start Time: 1:50 PM
Presentation Time: 2:40 PM
Location: Poster Station 1
Presentation Type: Poster
