Geological Model Updates via Discrete Cosine Transform
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current reservoir modeling techniques face challenges in efficiently generating large-scale three-dimensional geological models with multi-model facies, permeability, or porosity distribution, particularly due to high computational costs and difficulties in preserving geological connectivity and realism, especially in complex geometrical configurations.
Innovation Solution
The use of a Discrete Cosine Transform (DCT) method for updating geological models, which allows for rapid processing by transforming data into the wave-number domain, reducing computational times and maintaining geological realism through the retention of low-frequency DCT coefficients, thereby facilitating efficient generation of statistically unbiased model realizations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional covariance matrix inversion methods are used for parametrization, then geological model accuracy is improved, but computational expense increases dramatically
Solution Approach 1:
The patent extracts only the essential low-frequency DCT coefficients from the full permeability field representation. By retaining only a small subset of coefficients (e.g., the lowest frequency components), the method captures the dominant geological features while discarding high-frequency details that contribute minimally to overall accuracy. This extraction approach dramatically reduces the number of parameters from millions to a manageable subset, enabling fast computational updates without sacrificing critical geological realism.
Solution Approach 2:
The patent transforms the permeability field from physical space to frequency domain using Discrete Cosine Transform. This parameter transformation allows the system to work with frequency-domain coefficients instead of spatial permeability values. The low-frequency coefficients represent the dominant geological structures, and by operating in this transformed parameter space, the method achieves both computational efficiency and geological accuracy simultaneously.
2Productivity
If Fourier-filter based method with two-point geostatistics is used, then computational speed is improved, but facies distribution description becomes highly challenging
Solution Approach 1:
The patent uses training images to create a library of geologically realistic facies patterns. Instead of relying on simplistic two-point statistics, the method copies and adapts proven geological configurations from the training images. The DCT coefficients are constrained to reproduce these training image patterns, ensuring that the generated models accurately represent complex facies distributions like deltaic channels and fluvial deposits while maintaining computational efficiency.
Solution Approach 2:
The patent combines multiple approaches: DCT for computational efficiency, training images for geological realism, and low-frequency coefficient retention for capturing dominant structures. This composite methodology integrates the strengths of different techniques - the speed of Fourier-based methods with the geological accuracy of training image simulation - to achieve both fast computation and accurate facies distribution representation.
3Manufacturing precision
If MPS techniques with training images are used, then geological model realism is improved, but computational time increases very long
Solution Approach 1:
The patent extracts only the essential low-frequency DCT coefficients from the full permeability field representation. By retaining only a small subset of coefficients (e.g., the lowest frequency components), the method captures the dominant geological features while discarding high-frequency details that contribute minimally to overall accuracy. This extraction approach dramatically reduces the number of parameters from millions to a manageable subset, enabling fast computational updates without sacrificing critical geological realism.
Solution Approach 2:
The patent applies partial action by using only a fraction of the available DCT coefficients - specifically, only the low-frequency components are retained and updated, while high-frequency coefficients are either discarded or regenerated. This partial approach is sufficient to capture the essential geological structures and achieve realistic models, while avoiding the computational burden of processing all coefficients. The method recognizes that excessive detail (high-frequency components) is not necessary for achieving geological realism.
Data Source
Figure 1~2
Figure 3
Figure 4A
AI summary
Systems and methods for generating updates of large scale 3D geological models with multi-model facies, permeability or porosity distribution.