--> Innovative Data Science Approach at Unconventional Pay Characterization and Production Prediction: Identifying Key Production Drivers in the Permian Basin Unconventional Plays

AAPG ACE 2018

Datapages, Inc.Print this page

Innovative Data Science Approach at Unconventional Pay Characterization and Production Prediction: Identifying Key Production Drivers in the Permian Basin Unconventional Plays

Abstract

We introduce a proprietary analytic workflow that ties unconventional reservoir (UCR) characteristics to production and that makes robust predictions of well performance from key reservoir properties. Our data analytic process is built on a core workflow that utilizes random forest machine learning technology. Random forest methods are preferred for UCRs because they capture complex interaction in noisy databases. We present details of the first of three inventions developed from this core technology, the Enhanced Pay Integrated Curve (EPIC) (U.S. Provisional Patent No. 62/564343). EPIC is used by Chevron’s UCR Factory to help identify geologic targets for horizontal development. EPIC is designed to recognize zones that enable high-end well recovery, not simply resource storage. Previous approaches to pay characterization were developed from either disparate unconventional analogs or relied on very limited empirical relationships to well productivity. EPIC identifies leading and statistically significant geologic variables that account for differences in long-term (e.g. 18 month) well production. Pay cutoffs are interpreted from partial dependency plots of key reservoir property predictor variables. These plots are graphical visualizations of the marginal effect of a given reservoir property on the production outcome. The closely coupled seal-source systems of the Wolfcamp have primary reservoir drivers that are very different than those assocaited with decoupled seal (semiconventional) systems of the Bone Spring UCR zones. An a posteriori assessment of wells previously landed in EPIC pay zones have upwards of 65% more cumulative production at 6 months than those landed outside of the pay zone. Data science and machine learning approaches have revolutionized Chevron’s understanding of its unconventional interests in the Permian Basin. The Permian’s unconventional plays were once thought to be relatively unpredictable, highly variable, and having little connection to reservoir properties. EPIC and other proprietary data analytic workflows enabled the identification of key reservoir performance drivers and to use them to predict well performance through time, routinely within 70%- 85% confidence. This new production predictability is a key competitive advantage for Chevron.