AI Formation Model for Real-Time Drilling Precision
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current sub-surface geological interpretation methods in oil and gas exploration rely on simplistic, one-dimensional models that may inaccurately predict rock layer boundaries, and data transmission from downhole tools is limited, preventing real-time data usage during drilling operations.
Innovation Solution
A method utilizing machine-learning models trained with formation-measurement pairings to generate detailed formation models, allowing for real-time adjustments to well trajectories without surface intervention, using sensors and processors positioned in the well to preprocess and analyze data, and employing both model-based and pixel-based formulations for accurate representation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine-learning models with multiple dimensional formulations are used, then formation model accuracy is improved, but device complexity increases
Solution Approach 1:
The system dynamically switches between one-dimensional and pixel-based model formulations based on computational resource availability and real-time processing requirements. The machine-learning model adapts its complexity level during operation, using simpler models when resources are constrained and switching to more complex pixel-based models when computational capacity allows, thus resolving the contradiction between accuracy and complexity
Solution Approach 2:
The formation model is segmented into different representation levels: simplified one-dimensional models for rapid processing and pixel-based detailed models for high-accuracy analysis. The system processes formation data through multiple segmentation levels, allowing accurate representation of complex formation structures while managing computational complexity through hierarchical processing
2Productivity
If real-time data processing is implemented, then drilling efficiency is improved, but data transmission requirements increase
Solution Approach 1:
The system performs preliminary processing of formation data locally at the wellsite before transmission to the surface. By pre-processing and filtering data in advance, the system reduces the volume of data that needs to be transmitted while maintaining the ability to perform real-time analysis, thus resolving the contradiction between processing speed and transmission capacity
Solution Approach 2:
An intermediary processing layer is introduced between the downhole sensors and the surface processing systems. This intermediary layer performs data aggregation, filtering, and preliminary analysis, acting as a buffer that decouples the real-time processing requirements from the limited transmission capacity, allowing efficient drilling decisions without overwhelming the transmission system
3Ease of operation
If one-dimensional models are used, then ease of operation is improved, but manufacturing precision deteriorates
Solution Approach 1:
The system dynamically adjusts model complexity based on operational context, using simple one-dimensional models for routine operations where ease of operation is prioritized, and switching to complex pixel-based models when high precision boundary prediction is required, thus resolving the contradiction between simplicity and precision through adaptive model selection
Solution Approach 2:
The system changes the dimensional parameters of the model formulation based on the specific drilling conditions and accuracy requirements. By adjusting the model from one-dimensional to pixel-based formulations depending on the complexity of the formation and the criticality of the boundary prediction, the system achieves both ease of operation and manufacturing precision as needed
Data Source
AI summary
A method for drilling includes obtaining formation-measurement pairings, training a machine-learning model using the formation-measurement pairings, receiving measurements obtained by a tool positioned in a well formed in a formation, and generating a formation model of at least a portion of the formation using the machine-learning model and the measurements. The formation model represents one or more physical parameters of the formation, one or more structural parameters of the formation, or both.