Wellbore Bed Boundary Interpretation Using Depth, Dip, and Uncertainty
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Geosteering engineers face challenges in fusing interpretations from different logging-while-drilling sensors with varying depths of investigation, leading to inconsistent and error-prone subsurface formation boundary interpretations due to reliance on human expertise and empirical confidence.
Innovation Solution
A method that combines primary and secondary measurements using weighted averages and machine learning techniques to correct boundary interpretations, incorporating depth and dip data from multiple sensors, and provides uncertainty analysis for improved accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual fusion of interpretations is used, then engineer expertise can be applied, but human bias and empirical confidence introduce errors
Solution Approach 1:
The patent replaces manual human fusion processes with an automated machine learning-based fusion system. The system uses a neural network to automatically combine interpretations from multiple LWD sensors, eliminating human bias and empirical confidence limitations while maintaining the ability to process complex multi-sensor data relationships.
Solution Approach 2:
The system creates a virtual model of the fusion process through machine learning algorithms that replicate and optimize the expert engineer's decision-making logic. The neural network learns from training data to automatically reproduce accurate boundary interpretations that would otherwise require manual expert judgment.
2Quantity of substance
If multiple LWD sensors with different depths of investigation are used, then more comprehensive data is available, but fusion complexity increases
Solution Approach 1:
The patent replaces complex manual fusion processes with an automated machine learning-based fusion system. The system uses a neural network to automatically combine interpretations from multiple LWD sensors, eliminating human bias and empirical confidence limitations while maintaining the ability to process complex multi-sensor data relationships.
Solution Approach 2:
The system processes each sensor's data independently through separate interpretation modules, then combines the results through a unified neural network fusion layer. This segmentation allows complex multi-sensor data to be processed in manageable segments before integration, reducing overall system complexity.
3Adaptability or versatility
If manual fusion is used, then flexibility in interpretation is maintained, but consistency and reproducibility decrease
Solution Approach 1:
The patent replaces manual human fusion processes with an automated machine learning-based fusion system. The system uses a neural network to automatically combine interpretations from multiple LWD sensors, eliminating human bias and empirical confidence limitations while maintaining the ability to process complex multi-sensor data relationships.
Data Source
AI summary
Systems, methods, and apparatus, including computer programs encoded on computer-readable media, for determining subsurface formation boundary information for a wellbore using a well drilling system. Primary boundary information for a formation bed boundary of a subsurface formation may be determined based on primary measurement data. Secondary measurement data may be obtained from one or more sensors of the well drilling system. The primary boundary information may be corrected using the secondary measurement data to determine corrected boundary information for the formation bed boundary of the subsurface formation. An uncertainty may be determined for the corrected boundary information. A drilling operation or a drilling attribute in the wellbore may be modified based on the corrected boundary information and the uncertainty.