Machine Learning Geological Models for Subterranean Landing Zones
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods struggle to accurately determine landing zones in subterranean formations, particularly in closely stacked geologic formations, which affects hydrocarbon production predictions and leasing information.
Innovation Solution
A computer-implemented method using machine learning techniques, such as tree-based learning processes, to generate a geological model of a reservoir basin by correlating well data with landing zones, utilizing wellbore criteria like TVD and completion data, even with incomplete directional surveys, to predict accurate landing zones.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional methods are used to determine landing zones, then the process is simpler, but the accuracy of landing zone determination deteriorates, especially in closely stacked geologic formations
Solution Approach 1:
The patent segments the landing zone determination process into distinct computational stages: data collection from multiple wells, geological model generation, machine learning training, and prediction execution. This segmentation allows complex analysis to be broken down into manageable components that can be processed systematically, resolving the contradiction between accuracy and complexity.
Solution Approach 2:
The patent introduces a machine learning model as an intermediary between raw well data and landing zone predictions. This intermediary processes and interprets complex geological patterns, enabling high accuracy in closely stacked formations without requiring direct complex geometric calculations, thus resolving the accuracy-complexity trade-off.
2Measurement precision
If complete directional surveys are required for accurate landing zone determination, then the precision improves, but the loss of information increases when survey data is incomplete
Solution Approach 1:
The patent performs preliminary actions by collecting and storing well data from multiple wells during the model training phase, before actual landing zone predictions are needed. This preliminary data aggregation creates a robust geological model that can compensate for incomplete survey data in individual wells, resolving the contradiction between precision and information loss.
Solution Approach 2:
The patent implements feedback mechanisms where the machine learning model continuously refines its predictions by comparing predicted landing zones with actual known landing zones from wells with complete data. This feedback loop allows the model to learn from complete survey data and apply that knowledge to cases with incomplete data, maintaining precision while tolerating information loss.
3Productivity
If machine learning processes are used to correlate wells with landing zones, then the productivity of analysis improves, but the device complexity increases
Solution Approach 1:
The patent creates a computational copy of the geological formation through the machine learning model. Instead of performing complex physical or geometric analyses for each well, the system uses the trained model copy to rapidly predict landing zones, dramatically improving productivity while the computational complexity is encapsulated in the model creation phase rather than the prediction phase.
Data Source
AI summary
Techniques for generating a geological model include identifying a plurality of well data for each of a plurality of wells drilled into a reservoir basin from a terranean surface. The reservoir basin includes a plurality of landing zones formed under the terranean surface, each of the landing zone including a discrete geological layer. The techniques further include comparing the plurality of well data for each well with a reservoir basin database that associates the well data with one of the plurality of landing zones; correlating each of the plurality of wells with a particular landing zone of the plurality of landing zones based on the comparison; and generating a geological model of the reservoir basin based on the correlated wells.