Subsurface Property Modeling With ML Feedback Validation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing methods for calculating subsurface properties in oilfields are time-intensive, prone to errors, and expensive due to the reliance on human intervention and disjointed processing techniques.
Innovation Solution
A method utilizing multiple machine learning models and physics-based models to automate the subsurface property modeling process, including training and validation through a feedback loop system, to enhance accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If human-based efforts and disjointed processing techniques are used for inversion, then measurement interpretation and subsurface property calculation can be performed, but the process becomes time-intensive and expensive
Solution Approach 1:
The patent replaces manual human-based processing and traditional physics-based modeling with machine learning models. Specifically, trained ML models automatically perform measurement interpretation and subsurface property calculation, eliminating the need for human experts to manually process data through disjointed techniques. This substitution of mechanical/manual processes with automated intelligent systems directly reduces inversion process time while maintaining calculation accuracy.
Solution Approach 2:
The patent implements self-service through automated machine learning models that independently perform the complete inversion workflow. The ML models automatically interpret measurements, calculate subsurface properties, and generate results without requiring human intervention at each processing stage. This self-service automation eliminates the time-consuming nature of human-based efforts while preserving the analytical depth previously provided by experts.
2Measurement precision
If human-based efforts are used for data interpretation, then accurate subsurface property calculation can be achieved, but the process becomes expensive
Solution Approach 1:
The patent substitutes expensive human expert labor with cost-effective machine learning models. The ML models, once trained, automatically perform data interpretation and subsurface property calculation that previously required human experts. This replacement eliminates the high costs associated with human-based efforts including expert salaries, training, and operational expenses, while maintaining the accuracy of subsurface property calculations through the intelligent capabilities of the trained models.
3Measurement precision
If traditional processing and modeling techniques are used, then subsurface property inversion can be performed, but the process becomes prone to errors
Solution Approach 1:
The patent replaces traditional processing and modeling techniques with machine learning models that are less prone to human errors. The ML models provide consistent, repeatable processing without the variability and error-proneness inherent in manual human-based efforts. The automated nature of ML model execution ensures reliable and reproducible inversion results, eliminating errors that can occur during manual data interpretation and processing steps.
Solution Approach 2:
The patent implements feedback mechanisms where the machine learning models are trained using ground truth data and their predictions are continuously refined. The models learn from training data and can be retrained to improve accuracy, providing a feedback loop that enhances reliability. This feedback-driven training process ensures the models adapt to minimize errors and maintain high inversion accuracy across different datasets and conditions.
4Measurement precision
If multiple processing and modeling techniques are combined, then comprehensive subsurface analysis can be achieved, but the process becomes complex and disjointed
Solution Approach 1:
The patent merges multiple discrete processing and modeling techniques into a unified machine learning framework. Instead of using separate human-based techniques for different processing stages, the ML models integrate these functions into a cohesive automated workflow. This merging eliminates the disjointed nature of traditional multi-technique approaches while maintaining comprehensive subsurface analysis capabilities through the integrated intelligent processing of the unified ML system.
Data Source
AI summary
A method for modeling a subsurface property for a subterranean volume of interest includes receiving input measurement data representing a subterranean volume of interest, predicting a subsurface property based at least in part on the input measurement data using a first machine learning model, predicting a subsurface property model based at least in part on the subsurface property, the input measurement data, or both, using a second machine learning model, predicting synthetic measurement data based at least in part on the subsurface property model using a third machine learning model, a physics-based model, or both, comparing the synthetic measurement data and the input measurement data, and training the first machine learning model, the second machine learning model, or both based at least in part on the comparing.