Despite significant developments in the past few years in the application of machine learning algorithms for the lithologic classification of rock samples, publicly available labeled data sets are very scarce. We open source a fully labeled data set containing more than 16,000 scanning electron microscopy (SEM) images of drill cutting samples—mounted on thin sections—from a low-permeability reservoir in western Canada. We develop a simplified image processing workflow to segment and isolate the rock chips into individual SEM images, which in turn are used to identify, classify, and quantify rock types based on textural characteristics. In addition, using this data set, we explore the use of convolutional neural networks (CNNs) as a baseline tool for acceleration and automatization of rock-type classification. Without significant modifications to popular CNN models, we obtain an accuracy of approximately 90% for the test set. Results demonstrate the potential of CNN as a fast approach for lithologic classification in low-permeability siltstone reservoirs. In addition to making the data set publicly available, we believe our workflow to segment and isolate drill cutting samples in individual images of rock chips will facilitate future research of drill cuttings properties (e.g., lithology, porosity, and particle size) using machine learning algorithms.

The current practice for multi-fractured horizontal well development in low-permeability reservoirs is to complete the full length of the well with evenly spaced fracture stages. Given methods to evaluate along-well variability in reservoir quality and to predict stage-by-stage performance, it may be possible to reduce the number of stages completed in a well without a significant sacrifice in well performance. Provision and demonstration of these methods is the goal of the current two-part study. In Part 1 of this study, reservoir and completion quality were evaluated along the length of a horizontal well in the Montney Formation in western Canada. In the current (Part 2) study, the along-well reservoir property estimates are first used to forecast per-stage production variability, and then used to evaluate production performance of the well when fewer stages are completed in higher quality reservoir. A rigorous and fast semi-analytical model was used for forecasting, with constraints on fracture geometry obtained from numerical model history matching of the studied Montney well flowback data. It is concluded that a significant reduction in the number of stages from 50 (what was implemented) to less than 40 could have yielded most of the oil production obtained over the forecast period.

Completion design for horizontal wells is typically performed using a geometric approach where the fracturing stages are evenly distributed along the lateral length of the well. However, this approach ignores the intrinsic vertical and horizontal heterogeneity of unconventional reservoirs, resulting in uneven production from hydraulic fracturing stages. An alternative approach is to selectively complete intervals with similar and superior reservoir quality (RQ) and completion quality (CQ), potentially leading to improved development efficiency. In the current study, along-well reservoir characterization is performed using data from a horizontal well completed in the Montney Formation in western Canada. Log-derived petrophysical and geomechanical properties, and laboratory analyses performed on drill cuttings, are integrated for the purpose of evaluating RQ and CQ variability along the well. For RQ, cutoffs were applied to the porosity (>4%), permeability (>0.0018 mD), and water saturation (<20%), whereas, for CQ, cutoffs were applied to rock strength (<160 Mpa), Young’s Modulus (60–65 GPa), and Poisson’s ratio (<0.26). Based on the observed heterogeneity in reservoir properties, the lateral length of the well can be subdivided into nine segments. Superior RQ and CQ intervals were found to be associated with predominantly (massive) porous siltstone facies; these intervals are regarded as the primary targets for stimulation. In contrast, relatively inferior RQ and CQ intervals were found to be associated with either dolomite-cemented facies or laminated siltstones. The methods developed and used in this study could be beneficial to Montney operators who aim to better predict and target sweet spots along horizontal wells; the approach could also be used in other unconventional plays.

The Triassic Montney Formation hosts major oil and gas resources in Western Canada. Despite significant historical development of these resources, the origin of its hydrocarbons remains unclear, partly due to limited evidence of primary organic matter within the formation. Most of the hydrocarbons in the Triassic Montney Formation are trapped in low-permeability siltstone facies. For the purposes of this study, however, we investigated the lesser-known and understudied organic- and clay-rich mudstone beds, typically ranging from 2 to 8 cm in thickness, which are interbedded with the siltstone reservoir facies. Petrographic and organic geochemical analyses were conducted on core samples from a well located in west-central Alberta with thermal maturity of 0.95-1.16%Ro. Petrographic analysis, including kerogen maceral composition and fluorescence of petroleum fluid inclusions, reveals that the organic matter in the Montney Formation is closely linked to its host lithology. Organic-rich mudstone facies, with TOC values up to 3.5 wt %, contain primary kerogen (32-74 vol %) and solid bitumen (29-68 vol %). In contrast, the more organic-lean siltstone facies (TOC < 0.8 wt %) contain primarily pore-filling solid bitumen and lack structured kerogen. Kerogen within the organic-rich mudstone facies is mainly composed of liptinite macerals, including amorphinite, liptodetrinite, and alginite, subordinate amounts of acritarch and sporinite, and traces of inertinite and vitrinite macerals. The primary organic matter in the mudstone facies is presumed to be derived from marine sources (i.e., phytoplanktonic algae and bacterial biomass). Its preservation may result from the balance between periodic high productivity (potentially driven by river floods), rapid sedimentation to bury and preserve organic matter, and low dilution by siliciclastic material. Organic petrology analysis suggests an in situ generation of hydrocarbons in the mudstone beds and a potential subsequent short-distance migration to adjacent siltstone beds. Biomarker analysis supports the genetic linkage between rock extracts from these beds. Additionally, three families of petroleum fluid inclusions were systematically observed (using fluorescence microscopy) in several sets of contiguous siltstone and mudstone beds. These findings suggest that the organic-rich mudstone facies partially contributed to the present-day hydrocarbons of the Middle Montney Member in the study area.