Flow Simulator With Proxy Models for Reservoir Management Forecasts
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional hydrocarbon reservoir workflows require significant computational resources and time, delaying decision-making and increasing costs due to the use of high-dimensional parameter space data models and full physics algorithmic models.
Innovation Solution
A system utilizing machine learning methods, model-order reduction techniques, and optimization techniques to generate less computationally expensive workflows that maintain accuracy, employing proxy flow modeling and cloud-based high-performance processing units.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If full physics algorithmic models and high-dimensional parameter space data models are used, then forecast accuracy is improved, but computational time and complexity increase significantly
Solution Approach 1:
The patent creates simplified proxy models that copy the essential behavior of complex full physics models. These proxy models are trained on data generated by the full physics models and then used to rapidly generate forecasts without requiring the computationally expensive full physics simulations, thus maintaining accuracy while reducing computational time
Solution Approach 2:
The patent extracts the most critical relationships and patterns from the high-dimensional parameter space by training machine learning models on subsets of data. This extraction process identifies key predictors and removes redundant parameters, creating streamlined workflows that maintain forecast accuracy while significantly reducing computational complexity
2Measurement precision
If complex workflows with full physics models are used, then forecast accuracy is improved, but device complexity and computational resources increase
Solution Approach 1:
The patent replaces complex full physics workflow steps with simpler proxy models that replicate their essential functionality. These proxy models use machine learning algorithms that are computationally lighter while maintaining the predictive capabilities needed for reservoir management decisions
Solution Approach 2:
The patent segments the complex workflow into distinct phases: data generation using full physics models, training of proxy models on this data, and deployment of proxy models for actual forecasting. This segmentation allows the complex full physics models to be used only when necessary for training, while the majority of forecasting operations use simpler proxy models
3Measurement precision
If high-dimensional parameter space data models are used, then forecast accuracy is improved, but computational resources and costs increase
Solution Approach 1:
The patent extracts the essential predictive information from high-dimensional parameter spaces by identifying and focusing on the most influential parameters and relationships. This extraction creates streamlined models that require fewer computational resources while maintaining the accuracy needed for effective reservoir management
Solution Approach 2:
The patent creates simplified proxy models that copy the predictive behavior of complex high-dimensional models. Once trained on comprehensive data, these proxy models can generate forecasts with minimal computational resource requirements, enabling frequent updates and scenario analysis that would be prohibitively expensive with full physics models
Data Source
AI summary
An apparatus used to generate forecasts from a high-dimensional parameter data space. The apparatus comprising a reservoir model and a flow simulator module. The reservoir model comprising a plurality input variables, output variables, and at least one algorithmic model. The input variables and output variables are generated by the flow simulator module and variables from a formation and reservoir properties database and a field production database. The flow simulator module generates the at least one algorithmic model and the output variables using at least one selected from a group comprising a full-physics flow simulator, proxy flow simulator for assisted history matching, and a proxy flow simulator for field development optimization. The full-physics flow simulator and the two proxy flow simulators generate the at least one algorithmic model using at least one selected from a group comprising the reservoir model, history matching input variables, and optimization input variables.