Stratigraphic variability of Mississippian Meramec chemofacies and petrophysical properties using machine learning and geostatistical modeling, STACK trend, Anadarko Basin, Oklahoma

Hardisty, Laynie; Pranter, Matthew J.; Devegowda, Deepak; Marfurt, Kurt J.; Sondergeld, Carl H.; Rai, Chandra; Gupta, Ishank; Han, Heyleem; Dang, Son; McLain, Chris T.; Larese, Richard E.

doi:10.1190/int-2020-0169.1

Cited by 3 publications

(1 citation statement)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Al and K represent the clay content, while Ti and Zr signify the continental-derived sediments (Sageman and Lyons, 2004;Tribovillard et al, 2006). Ca and Sr are indicators of carbonates, and the Sr/Ca ratio shows the degree of diagenesis (Banner, 1995;Hardisty et al, 2021). The Si/Al ratio represents sand/clay concentration in mixed beds.…”

Section: Chemostratigraphymentioning

confidence: 99%

Petrophysical Rock Typing in Uinta Basin Using Models Powered by Machine Learning Algorithms

Arengas

Curtis

Dang

et al. 2023

Day 3 Wed, April 19, 2023

View full text Add to dashboard Cite

The development of unconventional resources in the Uinta basin requires more investigation. Petrophysical characterization is the key to identifying different rock types to optimize hydrocarbon production. Rock-typing can be performed using wireline measurements, such as triple combo and special logs; however, this identification needs to be verified using laboratory characterization to enhance the accuracy of rock-typing prediction models. In this work, we implement an integrated characterization workflow for 600 ft of the core interval, including total organic carbon, source rock analysis, elemental (X-ray Fluorescence) and mineral (Fourier- transform Infrared Spectroscopy) composition, total porosity (High-pressure pycnometer, Nuclear Magnetic Resonance), pore throat size distribution (Mercury Injection Capillary Pressure), elastic moduli (Ultrasonic velocity and nanoindentation) and microstructure (Scanning Electron Microscopy). Wireline measurements include the triple combo and the sonic logs. Principal Component Analysis and K-means (as unsupervised machine learning algorithms) were applied to both datasets to cluster and classify different rock types. In parallel, the petrophysical systematic for each rock type was evaluated. The Uinta group is vastly diverse, having a wide range of porosity (2-18%) and TOC (0.5-10%). Three main rock types were identified type 1-siliceous rich, type 2-calcite rich, and type 3-dolomite rich. The relative contribution of types 1, 2, and 3 is 47, 31, and 22 %, respectively. The top section of the analyzed core is dominated by rock type 1, which generally has the highest porosity and relatively higher TOC. Most of the bottom section is carbonate-rich rock types, in which calcite-rich and dolomite-rich layers are interbedded; SEM analyses suggest that a fraction of the porosity is associated with organic matter. Between rock types 3 and 2, further studies indicate that the high dolomite rock type tends to have higher porosity, larger pore size, and better-sorted grains, while the high calcite rock type has lower porosity and small pore size. There is a fair agreement in rock type identification between using core-derived and log- derived models. The Uinta basin leads the hydrocarbon production in Utah. The study provides a comprehensive core analysis dataset highlighting the vertical complexity of the Uinta group. The agreement in rock-typing using core and wireline inputs suggests that log-derived rock-typing can be utilized to identify sweet zones.

show abstract