Guided Bayesian Experimental Design for Subterranean Formation Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing experimental design techniques in subterranean formation analysis often neglect prior information, leading to increased uncertainty in data interpretation and inefficient experiment design.
Innovation Solution
A guided Bayesian experimental design method that utilizes prior information to calculate a sensitivity matrix, selecting optimal physical observations to reduce posterior uncertainties and improve data resolution, incorporating both noise and prior model uncertainties into the design process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If prior information is neglected in experimental design, then the design process is simpler, but uncertainty in data interpretation increases
Solution Approach 1:
The patent applies preliminary action by incorporating prior information (such as preliminary survey data, geological models, or previous measurement results) into the experimental design phase. This prior information is used to predict optimal measurement locations and parameters before actual data collection, thereby reducing uncertainty in the final interpretation without complicating the design process.
Solution Approach 2:
The patent implements feedback by using predicted posterior uncertainties to iteratively refine the experimental design. The system calculates expected uncertainties based on prior information, identifies areas with high uncertainty, and adjusts measurement locations or parameters to target those areas, creating a feedback loop that continuously reduces interpretation uncertainty.
2Measurement precision
If prior information is utilized in experimental design, then posterior uncertainties are reduced, but the design and calculation process becomes more complex
Solution Approach 1:
The patent replaces complex manual or iterative trial-and-error design methods with an automated computational system. The system uses algorithms to automatically incorporate prior information, calculate sensitivity matrices, predict posterior uncertainties, and optimize measurement locations, thereby reducing design complexity despite the sophisticated calculations involved.
Solution Approach 2:
The patent applies parameter changes by dynamically adjusting experimental parameters (such as measurement locations, source-receiver configurations, or acquisition parameters) based on calculated sensitivity analyses and uncertainty predictions. This allows the system to optimize measurements for maximum information gain while keeping the overall process manageable through systematic parameter adjustment.
3Productivity
If traditional experimental design methods are used, then the experiment can be completed quickly, but data quality and resolution are insufficient
Solution Approach 1:
The patent performs preliminary calculations of sensitivity matrices and predicted posterior uncertainties before field deployment. This allows optimal measurement locations and parameters to be predetermined, enabling rapid field execution without compromising data quality, as the experiment is already optimized on paper before actual data collection begins.
Data Source
AI summary
A Bayesian methodology is described for designing experiments or surveys that are improved by utilizing available prior information to guide the design toward maximally reducing posterior uncertainties in the interpretation of the future experiment. Synthetic geophysical tomography examples are used to illustrate benefits of this approach.