Bayesian Inversion for Subterranean Facies Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for predicting facies in subterranean regions are suboptimal as they do not consider spatial relationships, leading to inaccurate predictions, such as high probability for brine above gas due to lack of involvement of spatial relationships in mapping.
Innovation Solution
The method considers all facies combinations in a neighborhood of a location, approximating the rock physics distribution and marginalizing elastic parameters to compute posterior probabilities for facies configurations, using a Gaussian approximation to reduce computational complexity and account for non-Gaussian distributions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If local mapping methods are used to predict facies probabilities, then computational complexity is reduced, but prediction accuracy deteriorates due to lack of spatial relationship consideration
Solution Approach 1:
The patent divides the continuous spatial domain into discrete neighborhood regions around each location. Instead of considering the entire subsurface volume, the method segments the problem into local neighborhoods, evaluating facies probabilities within these segmented regions while maintaining computational efficiency. This segmentation allows spatial relationship consideration without excessive computational complexity.
Solution Approach 2:
The patent applies local quality by considering facies probabilities specifically within neighborhoods of each location rather than using uniform global models. The method evaluates local geological characteristics, seismic attributes, and facies configurations within each neighborhood, allowing predictions to reflect local geological reality while maintaining manageable computational complexity through localized analysis.
2Measurement precision
If all facies combinations in a neighborhood are considered, then prediction accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent applies partial action by considering only facies combinations within a defined neighborhood radius rather than all possible facies combinations in the entire subsurface. This partial consideration of spatial relationships provides sufficient accuracy for practical purposes while dramatically reducing computational complexity compared to exhaustive global analysis.
Solution Approach 2:
The patent transforms the problem from a high-dimensional global facies combination space to a lower-dimensional neighborhood-based approach. By defining neighborhoods with specific radius parameters and evaluating facies probabilities within these constrained spatial regions, the method reduces the effective dimensionality of the search space while maintaining predictive accuracy.
Data Source
AI summary
A method of modelling a subterranean region of the Earth at a first location comprises the steps of: providing geological data; selecting a plurality of facies; providing a distribution of rock physics probability with spatial dependencies; approximating the rock physics probability at the first location with at least one distribution per facies utilizing the spatial dependencies in the rock physics probability distribution; and performing a Bayesian inversion at the first location using the approximation of the rock physics probability distribution. The method may also employ a window comprising the first location and at least one further location.