Skip to main content
Home
Membership
Log on
Cart
Toggle search
Search
Toggle navigation
Search
Society of Petrophysicists and Well Log Analysts
About
Toggle
Vision
SPWLA History
Presidents and Awards
Symposium Locations and Conferences
Board of Directors
Policies
Events
Toggle
Event List
Event Calendar
Annual Symposium
Event Roster
Publications
Toggle
SPWLA Today Newsletter
Petrophysics Journal
Publication Guidelines
Digital Papers
Publications Advertisement
Petrophysicists Covers
Resources
Toggle
Careers/Education
Job Opportunities
Frank S. Millard Training Center
Webinars/Videos
Petrophysics Skill Set Guidelines (PSSG)
SPWLA Nuggets of Wisdom
SPWLA The More You Know
YP Resources
Mnemonics
API Standards
Petrophysical Software Directory
Member Directory
Downloads
Glossary
Annual Symposium
Toggle
2024 Annual Symposium
Symposium Author Instructions
International Student Paper Contest
Chapters/SIGs
Toggle
Chapters
SIGs
Resources for Chapter Officers
Chapters at Large
Student Chapters
Distinguished Speaker Program
Foundation
Toggle
Make a Donation
Foundation Board Members
Foundation Overview
Scholarship and Grant Program
Store
Toggle
Digital Papers
Books, CDs and more
Additional Membership Dues
Make A Donation
All
Volunteer Opportunities
RESERVOIR SCALE CHEMOSTRATIGRAPHY AND FACIES MODELING USING
Unconventional reservoirs and laminated sand-shale reservoirs exhibit a high level of vertical heterogeneity in terms of geochemical, petrophysical, and geomechanical properties. These properties often change on the scale of centimeters across rock types or bedding, and as such, they cannot be accurately measured by low-resolution petrophysical logs. Nonetheless, the distribution of these properties within a flow unit can significantly impact targeting, stimulation, and production. This paper involves high-resolution laboratory measurements of geophysical properties over the whole core and analysis of such data using machine-learning (ML) techniques to build novel high-resolution facies models. Multiple core wells in two key US organic shale basins were evaluated. Thousands of feet of core were scanned at high sample rates (1 mm to 1 in.) using specialized equipment to acquire continuous high-resolution logs from CT, automated XRF, gamma bulk density, spectral GR (K, U, Th), 20+ element XRF, anisotropic compressional and shear velocities (H, V, and 45°), and magnetic susceptibility. The high-frequency sample rate data was processed through graphical analysis using dimensionality reduction (PCA, MPCA, and NLDR) and classification routines (K-means, random forests, etc.) to build high-resolution electrofacies, chemofacies, and lithofacies models. Basin models and facies associations were then established using pore-scale measurements (NMR, SEM, BET) in conjunction with core-scale chemostratigraphy and upscaling to more common openhole and borehole image logs. a. Multivariate plotting and dimensionality reduction were used to identify measurements that distinguish lithology, porosity, TOC, and maturation while discarding redundant measurements. b. PCA and Multilinear PCA worked best for data decomposition and identifying attributes for classification, while targeted projection pursuit proved useful in feature selection. c. Random forest proximity measures helped to identify the underlying structures within the high-dimensionality data. d. For facies modeling, K-means gave acceptable results for most of the data sets, while density-based clustering (DBSCAN) added clarity in some cases. e. Trace element abundances and major element ratios were used for mapping zones of highest organic preservation. f. Elemental proxies were effectively used to recognize the variability in organic sourcing and kerogen types. This is consequential to the transformation of organics to bitumen and the migration of bitumen byproducts at micro and nanoscales. g. High-resolution core data was upscaled to openhole log resolution and used for field-scale 3D geomodeling. The novel application of high-resolution data sets integrated with very high-resolution geophysical measurements over the entire length of the core allows for the correlation of chemofacies and electrofacies to nano-scale pore textural properties derived from NMR and BET. This procedure, along with upscaling from core scans to openhole log and borehole image data, lends itself well to ML algorithms that enable most representative facies models from high-dimensionality data to be identified. Finally, the ability to relate high-resolution vertical well core data sets to lateral well cuttings using XRF mapping has proven to be useful for evaluating lateral wellbore deviation from the intended target zone.
Standard price:
10.00
Discounted price:
1.00
Your price:
10.00
You could save:
90.0%
Quantity:
Author(s):
John J.Degenhardt, Jr., Safdar Ali, Mansoor Ali, Brian Chin, W. D. Von Gonten, Jr., Eric Peavey
Company(s):
W. D. Von Gonten Laboratories, Shell Fellow-UROC, Texas A&M University
Year:
{1}
##LOC[OK]##
{1}
##LOC[OK]##
##LOC[Cancel]##
{1}
##LOC[OK]##
##LOC[Cancel]##