Frackable Interval Prediction Using Rock Fabric and Well Log Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional methods for selecting hydraulic fracturing intervals in wells are inefficient, leading to 40% of perforated intervals being unfrackable, resulting in unnecessary costs and inefficient stimulation operations.
Innovation Solution
A machine learning model is trained using rock fabric data from well logs and historical performance data to predict frackable intervals, decoupling geomechanics and leveraging advanced data analytics to identify suitable fracture locations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional manual methods are used to select hydraulic fracturing intervals, then the process is simple and requires less technology, but the accuracy is poor leading to 40% of perforated intervals being unfrackable
Solution Approach 1:
The patent replaces traditional manual mechanical selection methods with a machine learning-based automated system. The machine learning model processes rock fabric data from well logs to predict frackable intervals, substituting human expertise with an automated intelligent system that achieves 95% accuracy compared to the 60% accuracy of manual methods.
Solution Approach 2:
The patent introduces rock fabric data as an intermediary parameter that bridges the gap between available well log data and the target of predicting frackable intervals. This intermediary data type enables more accurate predictions by providing specific information about rock properties that influence fracturability, without requiring direct measurement of all possible parameters.
2Productivity
If manual processes are used to identify frackable intervals, then operational simplicity is maintained, but time and resources are wasted on unfrackable zones
Solution Approach 1:
The patent applies preliminary action by using the machine learning model to predict and identify frackable intervals before actually performing the hydraulic fracturing operation. This advance prediction allows operators to plan stimulation treatments more effectively, focusing resources only on intervals with high probability of success and avoiding time-wasting attempts on unfrackable zones.
Solution Approach 2:
The patent incorporates feedback mechanisms where historical performance data from previous fracturing operations is used to train and continuously improve the machine learning model. This feedback loop enables the system to learn from past successes and failures, progressively improving prediction accuracy and reducing wasted time on unfrackable intervals over time.
3Reliability
If rock fabric data is integrated with well drilling data and historical performance data, then prediction accuracy reaches 95%, but data processing complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the complex data integration task into distinct manageable components: rock fabric data extraction from well logs, well drilling data processing, and historical performance data analysis. The machine learning model then integrates these segmented data types systematically, making the overall complex process more manageable and scalable.
Solution Approach 2:
The patent utilizes parameter changes by transforming and normalizing different data types (rock fabric parameters, drilling parameters, historical performance metrics) into a unified format that the machine learning model can process effectively. This involves adjusting parameters to appropriate scales and formats, enabling reliable integration of diverse data sources without overwhelming complexity.
Data Source
AI summary
A computer implemented method that enables predicting frackable intervals. The method includes extracting rock fabric data from well logs and integrating rock fabric data with well drilling data and corresponding historical performance data to create labeled rock fabric data. The method also includes training a machine learning model to predict a probability of being a successful fracture for at least one interval using the labeled rock fabric data.