Surrogate Reservoir Model for Well Completion Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The hydrocarbon production industry faces high computational costs and time requirements for generating well completion plans and designs, which involve optimizing the placement and settings of flow control devices and packers in reservoir simulations.
Innovation Solution
A method and system utilizing a processor to evaluate multiple well completion plans, develop a surrogate reservoir model, and apply intelligent sequential sampling to iteratively update the model until it meets validation criteria, significantly reducing computational time by using machine learning techniques and Bayesian optimization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If reservoir simulation is used to model and optimize well completion plans, then the quality and accuracy of the completion plan is improved, but the computational time and cost increase significantly
Solution Approach 1:
The patent creates a surrogate model that copies the essential behavior of the complex reservoir simulator. This surrogate model is trained on a subset of simulation data and can predict completion plan performance with much lower computational cost, enabling rapid optimization iterations without running full reservoir simulations for every scenario evaluation
Solution Approach 2:
The patent performs preliminary reservoir simulations to generate training data before the optimization process begins. This pre-computed data is used to build the surrogate model, which then serves as the basis for rapid evaluation of multiple completion plan scenarios during optimization, avoiding the need to run full simulations for each scenario
2Manufacturing precision
If the number of reservoir simulations is increased to optimize flow control device placement and settings, then the quality of the completion plan is improved, but the computational cost increases
Solution Approach 1:
The surrogate model serves as a computationally inexpensive copy of the reservoir simulator, allowing numerous optimization iterations to be performed with minimal energy expenditure. The surrogate model captures the essential physics and relationships, enabling high-quality optimization without the energy cost of running the full reservoir simulator repeatedly
3Reliability
If traditional optimization methods are used for well completion plans, then comprehensive evaluation is achieved, but the time required to generate optimized plans is excessive
Solution Approach 1:
The surrogate model enables comprehensive evaluation of multiple completion plan scenarios with dramatically improved speed. By using the trained surrogate model instead of the full reservoir simulator for evaluation, the system maintains evaluation thoroughness while reducing computation time from hours or days to minutes or seconds
Solution Approach 2:
The patent implements an adaptive optimization approach that dynamically adjusts the evaluation process based on the surrogate model predictions. The system can rapidly explore the solution space, identify promising regions, and focus computational resources on those areas, achieving comprehensive evaluation with improved productivity
Data Source
AI summary
A method for generating a well completion plan includes: evaluating a plurality of different well completion plans using a reservoir simulator to calculate dynamic flows of fluid through a subsurface formation, each well completion plan having a flow control device with location and associated flow setting or rating, and optionally a packer and location to provide output data for each well completion plan evaluation; developing a surrogate reservoir model using the output data and input data for each well completion plan evaluation; using intelligent sequential sampling of the output and input data for each well completion plan evaluation to provide intelligent sequential sampling data in response to the surrogate reservoir model not meeting a validation criterion; updating the surrogate reservoir model using the intelligent sequential sampling data; and iterating the using and the updating using a latest surrogate reservoir model until the latest surrogate reservoir model meets the validation criterion.