Reservoir Model Calibration via Self-Organizing Topological Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for modeling reservoirs of compressible fluids, such as hydrocarbons, face challenges in predicting future operation and performance due to limited knowledge of reservoir structure and behavior, leading to complex and delicate calibration processes with risks of non-convergence or unrealistic solutions.
Innovation Solution
A method that optimizes in the space of classes with lower dimensionality, allowing for integration of graphical representations and attribute extraction, using self-organizing topological classification techniques to classify and optimize reservoir models, thereby improving convergence and reducing the complexity of calibration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a sophisticated reservoir model with multiple maps and parameters is used to comprehensively represent reservoir properties, then the model accuracy and completeness improve, but the calibration process becomes more complex and time-consuming with higher risks of non-convergence
Solution Approach 1:
The patent segments the reservoir model into multiple independent maps (porosity map, permeability map, thickness map, etc.), where each map represents a specific reservoir property. This segmentation allows the calibration process to be broken down into separate optimization steps for each map, reducing the overall complexity while maintaining comprehensive model accuracy.
Solution Approach 2:
The patent introduces a vertical dimension to traditional 2D property maps by creating stacked maps arranged in a calibration sequence. This dimensional transformation allows the optimization process to proceed through multiple levels (iterations), where each level refines specific maps based on production data, systematically reducing calibration complexity.
2Reliability
If traditional calibration methods are used to adjust numerical parameters and maps, then the model can be calibrated to match production measurements, but the process is time-consuming and may converge to unrealistic solutions
Solution Approach 1:
The patent applies preliminary action by pre-defining a calibration sequence for multiple maps before the actual calibration process begins. Each map is assigned a specific position in the calibration sequence based on its importance and interrelationships. This preliminary arrangement guides the optimization process to adjust maps in a logical order, preventing unrealistic solutions and reducing calibration time.
Solution Approach 2:
The patent implements feedback mechanisms where each map's calibration results are evaluated against production measurements, and the performance information is fed back to adjust subsequent map calibrations. This iterative feedback process ensures that each map adjustment contributes to overall model reliability while avoiding unrealistic solutions.
3Adaptability or versatility
If the number of elements to be optimized in the reservoir model is increased to capture more reservoir aspects, then the model comprehensiveness improves, but the calculation resources and convergence risks increase
Solution Approach 1:
The patent divides the comprehensive reservoir model into multiple specialized maps, each representing a specific reservoir property (porosity, permeability, thickness, etc.). This segmentation allows the optimization to focus on one property at a time, reducing computational resources required for each optimization step while maintaining overall model comprehensiveness.
Solution Approach 2:
The patent transforms the single-step optimization problem into a multi-level optimization process by arranging maps in a calibration sequence dimension. This dimensional change allows the system to process multiple map elements systematically, reducing memory requirements and computational burden at each step while maintaining comprehensive model adaptability.
Data Source
Figure 1~2
Figure 3
Figure 4~6
AI summary
A method for defining a calibrated model for an underground fluid reservoir comprising, - a step (S100) of generating a library (100) of graphic representations describing the reservoir for different embodiments thereof, - a step of self-organising classification (S200) of the graphic representations of the library in a space (200) of classes organised in at least one dimension, - and a step (S300) of optimising an objective function at least in said space of classes, in order to define an embodiment (300) of the reservoir calibrated on operating measurements (ME).