Proxy Flow Model for Reservoir Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current petroleum reservoir simulation methods, such as full physics flow simulators, are computationally intensive and time-consuming, making it impractical to perform numerous iterations required for accurate history matching and uncertainty estimation, which affects the accuracy of production forecasts.
Innovation Solution
Implementing a proxy model based on a deep neural network (DNN) to reduce computational requirements, enabling quicker history matching and uncertainty estimation by calibrating dynamic reservoir properties with static parameters, and using an ensemble Kalman filter for data assimilation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If full physics flow simulators are used for reservoir simulation, then prediction accuracy is improved, but computational time and resource consumption increase significantly
Solution Approach 1:
The patent creates a proxy model that copies the essential behavior of the full physics flow simulator but with reduced computational complexity. The proxy model is trained on simulation data to replicate the simulator's predictions, enabling rapid iteration while maintaining acceptable accuracy for history matching and uncertainty estimation
Solution Approach 2:
The patent employs simplified proxy models that are computationally inexpensive and can be rapidly executed multiple times. These proxy models sacrifice some physical fidelity but provide sufficient accuracy for iterative processes like history matching, where numerous simulations are required
2Measurement precision
If numerous iterations are performed for history matching and uncertainty estimation, then forecast accuracy is improved, but computational resource consumption becomes impractical
Solution Approach 1:
The proxy model serves as a computational copy that requires minimal resources, allowing hundreds or thousands of iterations for history matching and uncertainty quantification without the prohibitive cost of running full physics simulators for each iteration
Solution Approach 2:
The proxy model is pre-trained on a comprehensive dataset generated from full physics simulations. This preliminary action captures the complex physics relationships in advance, enabling the model to provide accurate predictions during iterative processes without requiring the full computational power during each iteration
Data Source
AI summary
Using production data and a production flow record based on the production data, a deep neural network (DNN) is trained to model a proxy flow simulation of a reservoir. The proxy flow simulation of the reservoir is performed, using an ensemble Kalman filter (EnKF), based on the trained DNN. The EnKF assimilates new data through updating a current ensemble to obtain history matching by minimizing a difference between a predicted production output from the proxy flow simulation and measured production data from a field. Using the updated current ensemble, a second proxy flow simulation of the reservoir is performed based on the trained DNN. The assimilating and the performing are repeated while new data is available for assimilating. Predicted behavior of the reservoir is determined based on the proxy flow simulation of the reservoir. An indication of the predicted behavior is provided to facilitate production of fluids from the reservoir.