Reservoir Productivity Estimation Using Machine Learning Parameter Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for characterizing reservoir productivity in subsurface volumes are inadequate, particularly for unconventional and tight rock plays, as they often rely on inferential relationships and simple regression techniques that fail to capture complex interactions in noisy data, leading to overfitting and poor characterization of rock properties affecting well performance.
Innovation Solution
A method and system that utilize subsurface and well data to generate production parameter maps, applying a trained parameter model to refine these maps, and allowing user input to limit parameters, ultimately estimating reservoir productivity as a function of depth using visual representations, thereby identifying recoverable pay zones and fracture barriers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If simple regression techniques are used to characterize reservoir productivity, then the method is easy to implement, but it fails to capture complex interactions in noisy data leading to overfitting and poor characterization
Solution Approach 1:
The patent transforms the approach by changing from simple regression parameters to machine learning model parameters that can capture complex non-linear interactions. The system uses trained models with multiple parameters to represent complex relationships between rock properties and well performance, avoiding overfitting while improving characterization accuracy.
2Device complexity
If inferential relationships are used to estimate reservoir productivity, then the method requires fewer direct measurements, but it leads to poor characterization of rock properties affecting well performance
Solution Approach 1:
The patent introduces machine learning models as intermediaries between available well data and reservoir productivity estimation. These models learn complex relationships from training data and serve as mediators to predict productivity without requiring direct measurement of all rock properties, thus maintaining simplicity while improving accuracy.
3Measurement precision
If complex machine learning models are applied to capture interactions in noisy data, then the characterization accuracy improves, but the model complexity and training requirements increase
Solution Approach 1:
The patent applies preliminary action by training machine learning models in advance using historical well data and rock properties. The models are pre-trained to capture complex interactions before being deployed for reservoir productivity estimation, which reduces real-time computational complexity while maintaining high accuracy.
Data Source
AI summary
Systems and methods for estimating reservoir productivity as a function of depth in a subsurface volume of interest are disclosed. Exemplary implementations may: obtain subsurface data and well data corresponding to a subsurface volume of interest; obtain a parameter model; use the subsurface data and the well data to generate multiple production parameter maps; apply the parameter model to the multiple production parameter maps to generate refined production parameter values; generate multiple refined production parameter graphs; display the multiple refined production parameter graphs; generate one or more user input options; receive the one or more user input options selected by a user to generate limited production parameter values; generate a representation of estimated reservoir productivity as a function of depth in the subsurface volume of interest using visual effects; and display the representation.