Subsurface Multi-Domain Modeling for Uncertainty-Based Field Planning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing modeling processes in oilfield exploration and production often introduce bias and inefficiencies due to the manual selection of representative realizations, which fail to maintain self-consistency and accurately propagate uncertainty across different modeling domains.
Innovation Solution
A method involving the generation of a multi-domain model with an ensemble of realizations, statistically sampling these realizations based on uncertainty, and using a field development planning engine to simulate and generate a field development plan, incorporating machine learning and k-means clustering to select suitable realizations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If manual selection of representative realizations is used to capture uncertainty range, then the process can be simplified, but bias and unintended side-effects are introduced
Solution Approach 1:
The system automatically selects representative realizations using statistical sampling methods (e.g., Latin Hypercube Sampling, Monte Carlo methods) without requiring manual intervention. The algorithm independently identifies and selects realizations that best represent the uncertainty space, eliminating human bias while maintaining operational simplicity
Solution Approach 2:
The system transforms the uncertainty representation from manual discrete selection to automated continuous statistical sampling. By changing the parameter selection methodology from subjective expert judgment to objective statistical algorithms, the system maintains ease of operation while significantly improving reliability and reducing bias
2Reliability
If statistical sampling of realizations is performed, then uncertainty propagation is improved, but computational complexity increases
Solution Approach 1:
Instead of processing all possible realizations, the system performs partial action by selecting a statistically representative subset of realizations for detailed analysis. This subset is sufficient to capture the uncertainty characteristics without requiring exhaustive computation of every possible scenario, thus reducing computational complexity while maintaining reliability
Solution Approach 2:
The system segments the large space of all possible realizations into manageable statistical groups using sampling methods. By dividing the computational task into smaller, statistically representative segments rather than processing the entire realization space, the system achieves accurate uncertainty propagation with reduced computational complexity
3Ease of operation
If representative realizations are selected from different modeling domains, then domain-specific analysis is simplified, but self-consistency of information is lost
Solution Approach 1:
The system creates a universal statistical sampling framework that works across multiple modeling domains (geological, reservoir, economic, etc.). This single multi-functional approach selects realizations consistently across all domains simultaneously, maintaining self-consistency while enabling domain-specific analysis without the need for separate selection processes in each domain
Data Source
AI summary
A method for field development includes receiving input data representing a subterranean volume, generating a multi-domain model of the subterranean volume, statistically sampling one or more of the realizations of the multi-domain model based at least in part on an uncertainty associated therewith, simulating the sampled one or more of the realizations using a field development planning engine, and generating a field development plan based at least in part on the simulated one or more of the realizations.