Reservoir Model Simulation Using GANs for Geological Connectivity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional geostatistical methods struggle to generate realistic geology models in subsurface reservoirs due to limitations in capturing complex geological features, leading to erroneous predictions and misplaced well locations, while object-based modeling faces challenges in data conditioning and computational efficiency.
Innovation Solution
Employing neural networks, specifically Generative Adversarial Networks (GANs) and Conditional Generative Adversarial Networks (CGANs), to simulate reservoir models that honor subsurface constraints and generate geologically realistic models efficiently, incorporating training images and well observations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Shape
If Multipoint Statistics (MPS) is used to generate geology models, then geological realism is improved, but connectivity of geological objects is lost
Solution Approach 1:
The patent introduces a connectivity constraint mechanism as an intermediary between the MPS simulation process and the final model output. This mediator ensures that while MPS generates geologically realistic patterns, the connectivity of geological objects is preserved through additional constraint enforcement during the simulation process.
2Measurement precision
If multiple training images are used in MPS modeling to improve uncertainty evaluation, then accuracy is improved, but computational cost increases
Solution Approach 1:
The patent performs preliminary analysis to identify and focus computational resources on the most influential training images and parameters. By pre-screening training images based on their potential contribution to uncertainty evaluation, the system achieves accurate uncertainty quantification with reduced computational expense.
3Shape
If object-based modeling is used to generate realistic geobodies, then shape reproduction is improved, but data conditioning becomes difficult
Solution Approach 1:
The patent uses training images as templates or copies of realistic geological patterns. These training images serve as reference models that guide the object-based modeling process, allowing the system to replicate complex geobody shapes while maintaining consistency with observed data through the training image constraints.
4Productivity
If conventional geostatistical methods are used for model building, then computational efficiency is maintained, but geological realism is lost
Solution Approach 1:
The patent transitions from conventional two-point statistics to multipoint statistics, fundamentally changing the statistical parameters used in the modeling process. This parameter change enables the system to capture complex geological patterns and curvilinear features while maintaining computational efficiency through optimized MPS algorithms.
Data Source
AI summary
A method, computer program product, and computing system are provided for defining one or more injector completions and one or more producer completions in one or more reservoir models. One or more edges between the one or more injector completions and the one or more producer completions in the one or more reservoir models may be defined. The one or more edges between the one or more injector completions and the one or more producer completions may define a graph network representative of the one or more reservoir models. The one or more reservoir models may be simulated along the one or more edges between the one or more injector completions and the one or more producer completions.