Sweet Spot Prediction via Segmented Data Normalization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current optimization and design methods in petroleum engineering, relying on geology and geophysics data, are not comprehensive or efficient for predicting parameters like sweet spots for oil well placement, as they can be dominated by completion data, leading to incomplete or inefficient optimization.
Innovation Solution
A device and processing method that normalize production data using both geology and geophysics data, and completion data separately, to mitigate dominance issues and facilitate sweet spot-based machine learning (SSML) and completion-based machine learning (COMML) for improved prediction and optimization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If geology and geophysics data are used for optimization and design, then the method is simple and easy to implement, but the prediction accuracy and comprehensiveness are insufficient due to dominance by completion data
Solution Approach 1:
The patent segments the data normalization process into two distinct parts: normalization based on geology and geophysics data, and normalization based on completion data. This segmentation allows each data type to be processed independently and then integrated, preventing completion data from dominating the overall prediction while maintaining implementation feasibility through modular processing steps.
2Device complexity
If only geology and geophysics data are used, then the processing is straightforward, but the optimization is not comprehensive due to missing completion data insights
Solution Approach 1:
The patent merges two separate normalization processes: one based on geology and geophysics data and another based on completion data. By combining these normalized datasets, the system achieves comprehensive optimization that incorporates both geological insights and completion performance data, while maintaining manageable processing complexity through the use of separate but parallel normalization streams.
3Measurement precision
If completion data is used for normalization, then the prediction is more accurate, but the method becomes complex and completion data dominates other data types
Solution Approach 1:
The patent applies local quality by performing completion data normalization specifically for completion-related parameters and performance metrics, while using geology and geophysics data normalization for spatial and geological parameters. This localized approach ensures completion data contributes accurately to its relevant domains without overwhelming the entire prediction system, maintaining balanced processing complexity.
Data Source
AI summary
Provided herein is a device suitable for sweet spot-based machine learning (SSML). There is further provided a processing method in association with completion-based machine learning (COMML). The device can include an input portion configured to receive at least one output signal, the output signal based on at least one production data normalized based on geology and geophysics (G&G)-based data. The device can further include a processing portion coupled to the input portion, the processing portion configured to process the output signal by manner of machine learning-based processing to produce at least one prediction signal, wherein the prediction signal corresponds to at least one visually perceivable graphics-based signal displayable as a three-dimensional (3D) productivity volume for identifying at least one sweet spot location for placement of a structure.