IP Library Granted Patent US 12699197
Granted Patent B2
US 12699197 · App. 18/259,969 · Granted Aug 4, 2026

Rock facies identification method based on seismic attribute classification using a machine learning network

Inventors: Si-Hai Zhang (Dhahran, SA); Xuekai Sun (Beijing, CN); Ammar Y. Balilah (Dhahran, SA); Yijun Hu (Dhahran, SA)
Assignee: SAUDI ARABIAN OIL COMPANY
G01V1/302G01V1/307G01V1/364G01V1/50G01V2210/51G01V2210/643
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12699197
App. No.
18/259,969
Granted
Aug 4, 2026
Kind
B2
Abstract

Methods and systems for determining a rock facies map are disclosed. The method includes obtaining a three-dimensional (3D) seismic image and a plurality of well logs, and identifying a horizon and determining a set of bandlimited 3D seismic images. The method further includes determining a set of mono-frequency maps by applying spectral decomposition to the 3D seismic image and determining a seismic attribute map based on the set of mono-frequency maps and a machine learning network. The method still further includes identifying a set of rock facies based, at least in part, on the plurality of well logs, determining a transformation function that maps a subset of rock facies to values of the seismic attribute map, and determining the rock facies map based on the seismic attribute map and the transformation function.

Claims (38)

1 . A method of determining a rock facies map comprising:

obtaining a three-dimensional (3D) seismic image of a subterranean region;

obtaining a plurality of well logs recorded in each of a plurality of wellbores within the subterranean region;

identifying a horizon within the subterranean region on the 3D seismic image;

determining a set of bandlimited 3D seismic images of the subterranean region by applying a filter to the 3D seismic image;

determining a set of mono-frequency maps of the horizon by applying spectral decomposition to the 3D seismic image, wherein a frequency of each mono-frequency map is selected based, at least in part, on a spectrum of the 3D seismic image and the set of bandlimited 3D seismic images;

determining a seismic attribute map of the horizon based, at least in part, on the set of mono-frequency maps and a machine learning network;

identifying a set of rock facies based, at least in part, on the plurality of well logs recorded in each of the plurality of wellbores and the set of bandlimited 3D seismic images;

determining a transformation function that maps a subset of rock facies to values of the seismic attribute map;

determining the rock facies map of the horizon based, at least in part, on the seismic attribute map and the transformation function;

identifying a hydrocarbon reservoir within the subterranean region based, at least in part, on the rock facies map; and

planning and drilling a wellbore within the subterranean region to recover the hydrocarbon reservoir.

2 . The method of claim 1 , wherein the 3D seismic image comprises a time migrated image.

3 . The method of claim 1 , wherein the plurality of well logs comprises at least one of a gamma ray log, a resistivity log, and a density log.

4 . The method of claim 1 , wherein a transform for spectral decomposition comprises a constrained least-squares spectral analysis.

5 . The method of claim 1 , wherein the machine learning network comprises an unsupervised variational Bayesian Gaussian mixture model.

6 . The method of claim 1 , wherein the set of rock facies comprises a high porosity sandstone.

7 . The method of claim 1 , wherein the subset of rock facies comprises an intersection of the set of rock facies and the horizon.

8 . A system comprising:

a seismic acquisition system;

a well logging system;

a computer system configured to:

receive a three-dimensional (3D) seismic image of a subterranean region using the seismic acquisition system,

receive a plurality of well logs recorded in each of a plurality of wellbores within the subterranean region using the well logging system,

identify a horizon within the subterranean region on the 3D seismic image,

determine a set of bandlimited 3D seismic images of the subterranean region by applying a filter to the 3D seismic image,

determine a set of mono-frequency maps of the horizon by applying spectral decomposition to the 3D seismic image, wherein a frequency of each mono-frequency map is selected based, at least in part, on a spectrum of the 3D seismic image and the set of bandlimited 3D seismic images,

determine a seismic attribute map of the horizon based, at least in part, on the set of mono-frequency maps and a machine learning network,

identify a set of rock facies based, at least in part, on the plurality of well logs recorded in each of the plurality of wellbores and the set of bandlimited 3D seismic images,

determine a transformation function that maps a subset of rock facies to values of the seismic attribute map,

determine a rock facies map of the horizon based, at least in part, on the seismic attribute map and the transformation function,

identify a hydrocarbon reservoir within the subterranean region based, at least in part, on the rock facies map, and

plan a wellbore within the subterranean region to recover the hydrocarbon reservoir; and

a drill system configured to drill the wellbore within the subterranean region to recover the hydrocarbon reservoir.

9 . The system of claim 8 , wherein the 3D seismic image comprises a time migrated image.

10 . The system of claim 8 , wherein the plurality of well logs comprises at least one of a gamma ray log, a resistivity log, and a density log.

11 . The system of claim 8 , wherein a transform for spectral decomposition comprises a constrained least squares spectra analysis.

12 . The system of claim 8 , wherein the machine learning network comprises an unsupervised variational Bayesian Gaussian mixture model.