Data mining the shale plays

Ryder Scott reports on Baker Hughes’ "boosted tree" multivariate big data analytics.

Interviewed in the Q1 2015 issue of Ryder Scott’s Reservoir Solutions newsletter Baker Hughes’s Randy LaFollette argues that data mining is to play a critical role for shale plays in today’s low price environment. Public domain data from over 65,000 onshore US wells from IHS’ data library have been studied to reveal that several simplistic prior analyses fail to identify the best completion strategies. Cross-plotting peak gas rates vs. stimulated fluid volumes in the Barnett shale play for instance failed to account for reservoir-quality differences.

LaFollette recommends using ‘boosted-tree, multivariate analysis’ of injection rates, fluid volumes, well architecture and other parameters. Combining such analyses with GIS data, geology and geochemistry produced ‘strong predictors of well production.’ Notably with the top 10% of wells drilled at some distance from faulting. The study also found that shorter laterals were more efficient in the Bakken while stage count was far less influential.

Baker Hughes has now rolled-up its shale analysis into a Google Earth-based interactive map of the Eagle Ford and other shale plays that makes ‘mining for data as easy as looking at a computer screen.’

Click here to comment on this article

Click here to view this article in context on a desktop

© Oil IT Journal - all rights reserved.