Neural Network Wireline Log Prediction for Reservoir Characterization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for reservoir characterization in hydrocarbon production are limited by the need for direct measurement of mechanical earth properties, which often requires lab testing of core samples and provides incomplete borehole coverage, leading to inefficient hydrocarbon management.
Innovation Solution
A system using a neural network to predict wireline logs, specifically shear-slowness and bulk-density logs, from gamma ray and compressional slowness logs, enabling more accurate characterization of subsurface reservoirs and optimizing well drilling operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If direct measurement of mechanical earth properties is used through core sample testing, then measurement precision is improved, but productivity deteriorates due to time-consuming lab testing and incomplete borehole coverage
Solution Approach 1:
The patent replaces physical core sample testing with a computational model (neural network) that predicts mechanical earth properties from wireline log data. This substitution eliminates the need for time-consuming laboratory testing while providing continuous borehole coverage, thereby resolving the contradiction between measurement precision and productivity.
Solution Approach 2:
The patent creates a virtual copy of the mechanical earth properties through predictive modeling. Instead of physically measuring properties from core samples, the system generates predicted wireline logs that replicate the information obtained from direct measurement, enabling full borehole coverage without physical sampling constraints.
2Productivity
If complete borehole coverage is achieved through direct measurement, then productivity is improved, but device complexity increases due to extensive sampling and testing requirements
Solution Approach 1:
The patent replaces complex physical sampling and testing systems with a computational prediction model. The neural network processes existing wireline log data to generate mechanical earth properties, eliminating the need for complex core sampling equipment, laboratory testing apparatus, and associated handling systems, thereby achieving complete borehole coverage with reduced complexity.
3Measurement precision
If manual core sample testing is performed, then measurement accuracy is improved, but extent of automation deteriorates as the process requires manual laboratory analysis
Solution Approach 1:
The patent replaces manual laboratory analysis with an automated neural network model. The system automatically processes wireline log data through trained predictive models to generate mechanical earth properties, eliminating manual sampling, testing, and analysis steps while maintaining measurement accuracy through validated computational algorithms.
Solution Approach 2:
The patent enables the system to automatically generate mechanical earth properties from existing wireline log data without requiring manual intervention. The neural network model self-processes the input data through learned relationships, producing predicted wireline logs that provide complete borehole coverage with full automation.
Data Source
AI summary
Methods and systems, including computer programs encoded on a computer storage medium are described for implementing a system that predicts wireline logs used in well drilling operations at a subsurface region. The system derives inputs from a first wireline log and includes a predictive model based on a neural network trained to generate data predictions. The predictive model processes the inputs derived from the first wireline log through layers of the neural network to generate a prediction that identifies multiple second wireline logs for a reservoir in the subsurface region. Based on the multiple second wireline logs, the system controls well drilling operations that simulate hydrocarbon production at the reservoir.