Unlocking well productivity drivers in Eagle Ford and Utica unconventional resources through data analytics

Abstract

North America's unconventional resources are viewed as statistical plays because of the wide variability in the performance of hydraulically fractured wells. Attempts to unlock the primary drivers of productivity yield different interpretations because of the over-parameterized stimulation processes, diverse geoscience concepts and non-unique solutions of associated reservoir flow modeling. An opportunity to derive insights on the yet-to-be fully understood flow mechanisms lies in the interrogation of acquired stimulation and production data. In search of causation relationships, a data analytics approach was used to profile and associate well productivity with multiple hydraulic fracturing stimulation parameters for Eagle Ford and Utica horizontal wells in areas with similar geological settings. Data for the study was retrieved from public-accessible databases managed by the Rail Road Commission of Texas, Department of Natural Resources of Ohio and Chemical Disclosure Registry of FracFocus. Results from the approach show multiple apparent causal relationships between production and stimulation parameters with the latter evolving in tandem. As a result, a priori knowledge was incorporated in decoupling dependencies for sole contributions of parameters. Among the most critical trends, the amount of proppant and fracturing fluid volume correlated with productivity. The average proppant concentration, used to decouple this association, correlated poorly with cumulative hydrocarbon production. Better performing wells in which the amount of proppant was high coincidentally had been stimulated with large amounts of fracturing fluid; thus, motivating the latter to be interpreted as a primary driver for Eagle Ford and Utica production enhancement. It was speculated that the induced hydraulic fracture length is effective; thus, use of large amounts of high quality proppant to create conductive fractures might be yielding limited value. Additional results provided insights on interdependences between injected proppant, lateral length, fracturing stages, perforation clusters and fracturing fluid volume. The identification of stimulation parameters that correlated with well productivity provided ground for speculating that unconventional resources are not statistical as portrayed and repeatable well performance can be realized if stimulation operations are systematically constrained. Further reflection on the results backed postulations on attributes of reservoir flow physics such as matrix permeability, fracture conductivity and effective fracture length. In conclusion, data analytics shed light on dominant productivity drivers in the subject unconventional resources and provided ground for hypotheses that will improve knowledge-driven simulation models.