Subsurface Region Modeling with Decorrelated Property Transformations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing subsurface modeling methods, such as bivariate modeling, are time-consuming and produce inconsistent results when dealing with multiple subsurface properties with higher-order correlations and constraints like summation to unity in mineralogy proportions.
Innovation Solution
A multivariate modeling tool that uses reversible transformations to transform subsurface properties into a decorrelated and unconstrained space, allowing for independent modeling of subsurface properties while honoring constraints and correlations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If bivariate modeling is used to model multiple subsurface properties, then correlations between properties can be preserved, but the modeling process becomes time-consuming and produces inconsistent results
Solution Approach 1:
The patent transforms the original subsurface properties through a mathematical transformation to create new properties with different statistical characteristics. This transformation changes the parameter space from correlated original properties to uncorrelated transformed properties, enabling independent modeling while preserving the ability to recover original correlations through inverse transformation.
Solution Approach 2:
The patent introduces an intermediary transformation step between the original subsurface properties and the final modeled properties. This intermediary space with uncorrelated properties serves as a mediator that simplifies the modeling process while maintaining the ability to preserve correlations in the final output through inverse transformation.
2Reliability
If bivariate modeling is used for multiple subsurface properties, then some correlations can be maintained, but higher-order correlations and constraints are lost
Solution Approach 1:
The patent applies a parameter transformation that converts correlated subsurface properties into uncorrelated transformed properties. This transformation preserves the information content while changing the statistical relationships, allowing all correlations (including higher-order ones) to be captured in the transformed space and then recovered through inverse transformation.
Solution Approach 2:
The patent moves the modeling problem into a different dimensional space through mathematical transformation. By transforming to a space with uncorrelated properties, the method accesses a different representation of the same underlying information, enabling preservation of all correlation structures including higher-order correlations that are difficult to maintain in the original space.
3Productivity
If subsurface properties are modeled independently, then modeling speed increases, but correlations and constraints between properties are violated
Solution Approach 1:
The patent introduces a transformation intermediary that decouples the properties for independent modeling while preserving their relationships. The transformed properties serve as intermediaries that can be modeled independently without violating correlations, and the inverse transformation recovers the original correlation structures in the final output.
Solution Approach 2:
The patent changes the parameters from correlated original properties to uncorrelated transformed properties, enabling independent modeling. The parameter transformation preserves the information needed to maintain correlations, and the inverse transformation restores the original correlation structures after independent modeling of the transformed properties.
Data Source
AI summary
A multivariate modeling tool utilizes reversible transformations to transform subsurface properties of wells into a modeling space in which the subsurface properties are decorrelated and unconstrained. The subsurface properties of a region are independently propagated from the well(s) in the modeling space. Reverse transformations are applied to the subsurface properties in the modeling space to generate a subsurface representation for the region. The subsurface representation honors constraints on the subsurface properties and correlations between the subsurface properties.