Subsurface Material Abundance Prediction from Well Logs and Seismic
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods fail to accurately determine material abundance in unsampled intervals of hydrocarbon wells, where existing technologies are inefficient in addressing the extraction of specific challenges in addressing the material abundance in unsampled intervals of hydrocarbon wells, where existing technologies have not effectively addressed the material needs of the subsurface, particularly in the determination of the material composition in unsampled regions of the subsurface, particularly in the determination of material abundance in unsampled intervals of hydrocarbon wells.
Innovation Solution
A multivariate approach using machine learning models to correlate well logs and seismic attributes with core sample data to predict material abundance without requiring core samples from every well section, utilizing wireline logs, logging-while-drilling techniques, and seismic surveys to generate a continuous material abundance map.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If core sampling is performed at every well section to determine material abundance, then measurement precision is improved, but device complexity and operational difficulty increase
Solution Approach 1:
The patent introduces well logs and seismic surveys as intermediary measurements that correlate with core sample data. These intermediaries allow prediction of material abundance in unsampled intervals without requiring physical core extraction at every section, thus reducing operational complexity while maintaining measurement precision through statistical correlation models.
Solution Approach 2:
The patent creates a computational model that copies the information content of core samples into well log and seismic survey data. By training a machine learning model on the correlation between core sample material abundance and well log/seismic attributes, the system generates predictive copies of core sample data from easier-to-obtain well log and seismic measurements.
2Measurement precision
If core sampling is performed throughout the entire well to obtain continuous material abundance data, then measurement precision is improved, but loss of time and productivity decrease
Solution Approach 1:
The patent performs preliminary correlation analysis between core sample data and well log/seismic survey data during the exploration phase. This preliminary training of the predictive model allows subsequent wells to be evaluated rapidly using only well logs and seismic data, eliminating the need for time-consuming core sampling at every interval while maintaining continuous material abundance data through the predictive model.
Solution Approach 2:
The patent replaces the mechanical core sampling process with a computational prediction system. Instead of physically extracting and analyzing core samples at every well section, the system uses machine learning models to substitute the mechanical sampling process with data-driven predictions from well logs and seismic surveys, significantly reducing time loss while maintaining measurement precision.
3Device complexity
If single-variable correlation methods are used to estimate material abundance, then device complexity is reduced, but measurement precision deteriorates
Solution Approach 1:
The patent merges multiple data sources (well logs and seismic surveys) and multiple measurement variables (gamma ray levels, neutron porosity, density, sonic shear, and seismic attributes) into a unified predictive model. This combination of multiple variables and data types provides more comprehensive information about subsurface properties, thereby improving measurement precision while the integrated approach is managed through a single machine learning model framework.
Solution Approach 2:
The patent creates a composite predictive model that integrates multiple data types and variables into a unified system. The machine learning model processes composite inputs from well logs (multiple physical properties) and seismic surveys (seismic attributes) to generate comprehensive material abundance predictions, achieving higher precision through the composite nature of the integrated model.
Data Source
AI summary
A method for recovering resources from a subsurface region that includes obtaining log data from multiple wells located in a first region, obtaining seismic data from a seismic survey of the first region, obtaining material abundance data from core samples from at least one of the wells, correlating the log-seismic data with the material abundance data, logging a second well to generate log data for the second well, obtaining seismic data from a seismic survey of a second region that includes the second well, processing the log-seismic data of the second region with a machine learning model trained on the log-seismic-abundance data of the wells in the first region to generate predicted material abundance data of the second well, and generating a pseudo-log of material abundance of the second well based at least in part on the predicted material abundance data.