Permian Basin SPWLA April 23, 2024 Technical Lunch Talk

Abstract:

Regional mapping and 3D geocellular modeling of geomechanical properties greatly benefit unconventional resource development. However, this comes with many challenges, including the need for appropriate and good-quality data with adequate vertical and areal coverage, comparable data vintages, etc. We present an integrated methodology to estimate and map different geomechanical properties of the Wolfcamp-Spraberry interval across the entire Midland Basin. Our method integrates well logs, core (triaxial tests), petrophysical inversion-based multi-mineral models, and class-based machine learning. We utilize a long short-term memory (LSTM) network for machine learning.

We estimated and mapped Young’s modulus, Poisson’s ratio, bulk modulus, and shear modulus of the Wolfcamp-Spraberry interval. The machine learning results showed approximately 0.97 accuracy and a root-mean-squared error of 7.0 in predicting shear sonic velocity used in log-based estimates from dipole sonic logs. There are uncertainties in log-based estimates when dealing with only compressional sonic logs and limited triaxial tests. We use triaxial data to convert log-based dynamic properties to static geomechanical properties. Lithologies control geomechanical properties; however, the differences in geomechanical responses are smaller in the case of mixed lithologies. Extensive data QA/QC, wide coverage of wells, and a good understanding of the basin geology are needed for meaningful model training; otherwise, the prediction can be poor. Additionally, model performance is affected if data from the southern Midland Basin is used to predict rock properties in the central and northern parts of the basin, and the reverse is true as well; this indicates varying geologic controls.

Speaker Biography:

Shuvajit Bhattacharya is an applied geophysicist/petrophysicist with demonstrated research expertise in petrophysical analysis, log-based multi-mineral modeling, 3D seismic interpretation, seismic attributes, machine learning, and integrated subsurface characterization. He integrates diverse quantitative tools for energy resources exploration (including geothermal), and carbon and hydrogen storage in the subsurface.
Prior to joining BEG, he worked with the University of Alaska Anchorage, Battelle, and other organizations in different roles, such as an assistant professor, petroleum geoscientist, project assistant, and summer intern. He has taught, mentored, and graduated students, many of whom are employed in the state/federal agencies/industry.

He has published several peer-reviewed research articles and scientific conference abstracts related to subsurface geology, petrophysics, 3D seismic attributes, inversion, and machine learning.

Event Venue & Nearby Stays

Bush Convention Center, 105 North Main Street, Midland, United States