Spatial Clustering Model for Type Curve Region Delineation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for identifying type curve regions in hydrocarbon-producing regions are subjective and fail to account for natural clustering of well performance, reservoir properties, and continuous reservoir changes, leading to inaccurate geographic boundary connections and lack of statistical significance testing.
Innovation Solution
The use of a spatial clustering model based on machine learning, incorporating adjustable hyperparameters such as reservoir original oil in place, porosity, and geomechanics, to delineate type curve regions objectively, recognizing patterns in well and production parameter data to identify clusters with distinct mean production values and minimal internal variation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If subjective methods are used to identify type curve regions, then the process is simpler and faster, but the accuracy and statistical significance are reduced
Solution Approach 1:
The patent replaces subjective manual methods with an automated machine learning-based spatial clustering system. The system uses algorithms (such as k-means clustering, hierarchical clustering, or DBSCAN) to objectively analyze well data and production parameters, generating type curve regions through computational processes rather than human judgment. This substitution maintains speed while dramatically improving accuracy and statistical validity.
Solution Approach 2:
The patent transforms the identification process by changing from qualitative subjective assessment to quantitative parameter-based analysis. The system incorporates multiple measurable parameters including productivity values, reservoir properties (porosity, permeability, thickness), and spatial coordinates to objectively define region boundaries. This parameter-driven approach enables precise, reproducible, and statistically significant type curve region identification.
2Ease of manufacture
If existing methods are used to delineate geographic boundaries, then the process is simpler, but the boundaries lack statistical significance and fail to account for natural clustering
Solution Approach 1:
The patent replaces simple geometric boundary drawing with a machine learning-driven spatial clustering system. The system automatically identifies natural clusters of wells and reservoir features by analyzing multiple parameters simultaneously, generating boundaries that reflect actual subsurface conditions rather than arbitrary geometric divisions. This maintains ease of operation while dramatically improving statistical reliability.
Solution Approach 2:
The system performs self-service by automatically detecting natural clustering patterns and determining optimal boundary locations without requiring manual intervention. The machine learning algorithms autonomously analyze the data, identify statistically significant groupings, and generate boundaries that naturally separate distinct reservoir characteristics, eliminating the need for subjective boundary drawing while ensuring statistical validity.
3Productivity
If type curve regions are identified without accounting for natural clustering, then the process is faster, but the distinction between regions and prediction accuracy are reduced
Solution Approach 1:
The patent substitutes rapid but imprecise manual identification with a computational system that simultaneously achieves speed and precision. The machine learning algorithms process large datasets of well data and production parameters efficiently, automatically identifying natural clusters that maximize the distinction between regions while maintaining fast processing times through optimized computational algorithms and data structures.
Solution Approach 2:
The patent enhances region distinction by incorporating multiple quantitative parameters including mean productivity values, standard deviations, reservoir properties (porosity, permeability, thickness), and spatial coordinates. The system analyzes the distribution and variation of these parameters to identify clusters with maximum statistical distinction, thereby improving prediction accuracy while maintaining computational efficiency through optimized parameter selection and processing.
Data Source
AI summary
Methods, systems, and non-transitory computer readable media for identifying type curve regions as a function of position in a region of interest are disclosed. Exemplary implementations may include: obtaining a spatial clustering model from the non-transitory storage medium; obtaining well data from the non-transitory storage medium; obtaining production parameter data from the non-transitory storage medium; and delineating each of the type curve regions in the region of interest by applying the spatial clustering model to the well data and the production parameter data.