Minimax Model Selection for Reservoir Uncertainty

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current methods for selecting representative models from a large ensemble of reservoir models are suboptimal, often relying on manual selection or clustering, which can lead to suboptimal decisions due to the lack of efficient approaches for matching target percentiles of multiple output responses and maximizing model diversity in the input uncertainty space.

Innovation Solution

The minimax model selection approach, which involves solving a complex multi-objective optimization problem to simultaneously select models that match target percentiles of multiple output responses while maximizing their diversity in the input uncertainty space, using a combination of global exhaustive search and greedy methods for efficient model selection.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If manual selection or clustering methods are used to select representative models, then the selection process is simple to implement, but the selection quality is suboptimal and decision-making is compromised

Engineering Contradiction:

Improvesimplicity of model selection processVSAvoidaccuracy of representative model selection

Core Design Contradiction:

Ease of operationVSMeasurement precision

Solution Approach 1:

The patent transforms the model selection problem into a multi-objective optimization problem by changing the selection criteria from simple clustering or manual choice to a systematic optimization approach that simultaneously matches target percentiles and maximizes diversity. This parameter change enables automated, high-accuracy selection while maintaining operational simplicity through standardized algorithms.

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent replaces manual mechanical selection processes with automated computational optimization methods. The minimax optimization algorithm automatically identifies representative models based on multiple criteria (percentile matching and diversity maximization), eliminating the need for manual intervention while achieving superior selection accuracy.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Productivity

If traditional clustering methods are used for model selection, then the approach is computationally efficient, but it fails to simultaneously match target percentiles of multiple output responses and maximize model diversity

Engineering Contradiction:

Improvecomputational efficiency of model selectionVSAvoidstatistical representativeness of selected models

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The patent segments the model selection task into two distinct optimization objectives: matching target percentiles for multiple output responses and maximizing model diversity in the input uncertainty space. This segmentation allows each objective to be optimized independently through the minimax framework, ensuring both computational efficiency and statistical representativeness are achieved simultaneously.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The minimax optimization framework serves multiple functions simultaneously: it matches target percentiles for multiple output variables, maximizes model diversity, and maintains computational efficiency. This multi-functional approach replaces traditional single-objective clustering methods, providing comprehensive model selection capability that addresses multiple requirements at once.

Inventive Principle:

Principle #6Universality (Multi-functionality)

3Adaptability or versatility

If multiple output responses and percentiles are required, then the decision-making process becomes more comprehensive, but the selection difficulty increases significantly

Engineering Contradiction:

Improvecompleteness of decision-making analysisVSAvoidcomplexity of model selection process

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent merges multiple selection criteria (percentile matching for multiple output responses and diversity maximization) into a single unified minimax optimization framework. This consolidation handles complex multi-objective problems systematically, maintaining comprehensive decision-making analysis while reducing operational complexity through standardized optimization procedures.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The optimization framework incorporates feedback mechanisms that continuously evaluate selected models against target percentiles and diversity criteria. This feedback loop ensures that the selection process adapts to multiple output responses and maintains statistical representativeness, making the complex selection process more manageable through systematic iteration and validation.

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS9069916B2Model selection from a large ensemble of models

Publication Date: 2015.06.30 CHEVRON USA INC

AI summary

A method, system and processor readable medium containing computer readable software instructions for selecting representative models from a large ensemble of models is disclosed. An ensemble of reservoir models that define an input uncertainty space is provided. A target number of representative models and one or more target percentiles of output variables that the representative models are to approximate are input. The representative models from the ensemble of reservoir models that match the one or more target percentiles of output variables are selected while maximizing the spread between the selected representative models in the input uncertainty space.