Sonic Logging Slowness Determination Using Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Sonic logging analysis is currently an expert-driven process, leading to variability in results and limitations in processing capacity and speed, which hinders accurate and timely data analysis, especially in real-time drilling operations.
Innovation Solution
The implementation of machine learning systems, specifically convolutional neural networks, to process sonic logging data for determining sonic slowness, enabling automated and accurate analysis in real-time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If expert-driven processing is used for sonic logging analysis, then analysis accuracy can be maintained through expert judgment, but processing capacity and speed are limited and results vary significantly based on expert skill
Solution Approach 1:
The system enables self-service by training a neural network to automatically perform sonic logging analysis without requiring expert intervention. The neural network learns from training data containing sonic waveforms and corresponding slowness values, then independently processes new sonic logs to determine slowness, eliminating dependency on expert operators while maintaining consistent results
Solution Approach 2:
The patent replaces the mechanical expert-driven analysis process with an automated neural network system. The neural network substitutes human experts by processing sonic waveforms through learned patterns and relationships, transforming the manual expert judgment process into an automated computational system that operates without human intervention
2Loss of time
If expert-driven processing is used for sonic logging analysis, then thorough analysis can be performed, but significant delay occurs between data recording and result availability
Solution Approach 1:
The neural network enables continuous processing of sonic logging data as it becomes available, eliminating the batch processing delays inherent in expert-driven methods. The system can continuously analyze sonic waveforms and generate slowness results in real-time, maintaining uninterrupted useful action from data acquisition to analysis output
Solution Approach 2:
The patent replaces the slow, sequential expert analysis process with rapid automated neural network computation. This substitution eliminates the time-consuming nature of manual expert review while maintaining analysis quality, enabling real-time or near-real-time results that significantly reduce the delay between data recording and availability
3Quantity of substance
If more sonic logging data is processed to improve availability, then data coverage increases, but the limited capacity of experts becomes a bottleneck
Solution Approach 1:
The neural network performs self-service by automatically processing large volumes of sonic logging data without requiring proportional increases in expert resources. Once trained, the system independently handles any quantity of data inputs, eliminating the bottleneck where expert availability limits data processing volume
Solution Approach 2:
The patent changes the operational parameters of the processing system by transitioning from manual expert analysis to automated neural network processing. This parameter change fundamentally alters the system's capacity to handle data volume, allowing scalable processing of large datasets without the linear increase in complexity associated with expanding expert teams
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
Sonic logging data including a sonic waveform associated with a plurality of shot gathers is accessed. A transformation operator is applied to the sonic logging data to provide a transformed sonic image, the transformation operator including at least one of a short time average long time average (STA/LTA) operator, a phase shift operator, and a deconvolution operator. A machine learning process is performed using the transformed sonic image to determine a sonic slowness associated with the sonic logging data. The sonic slowness is provided as an output.