Dynamic Reservoir Front Prediction Using Adaptive Grid Refinement
Solution Overview
Problem
Current reservoir simulation methods face challenges in accurately modeling and predicting the movement of fronts, such as water or surfactant fronts, within complex geological structures, leading to inefficiencies in enhanced oil recovery (EOR) processes due to limitations in spatial discretization and computational resources.
Innovation Solution
A method involving a trained machine model to predict the position of fronts within a reservoir, allowing for dynamic grid refinement using techniques like kD-tree discretization, which adapts the spatial reservoir model's resolution based on predicted front positions, enabling more accurate and efficient simulation results.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If uniform spatial discretization is used throughout the reservoir model, then measurement precision of front position is improved, but computational resources and device complexity increase significantly
Solution Approach 1:
The patent applies local quality by using non-uniform spatial discretization where grid cells are refined (smaller size) in regions near predicted front positions and coarsened (larger size) in distant regions. This allows high measurement precision for front position where needed while reducing overall device complexity and computational burden in regions where fine resolution is not required.
Solution Approach 2:
The patent implements dynamic spatial discretization where the grid refinement level changes over time based on predicted front positions. As the front moves through the reservoir, the refined grid regions are dynamically updated to follow the front position, maintaining measurement precision while adapting computational resources to current simulation needs rather than using static uniform refinement throughout.
2Measurement precision
If high-resolution spatial discretization is applied throughout the model, then simulation accuracy is improved, but computational resources and processing time increase
Solution Approach 1:
The patent applies local quality by concentrating computational resources (fine grid resolution) only in local regions where the front is expected to be, rather than applying high-resolution discretization uniformly across the entire reservoir model. This maintains simulation accuracy in critical regions while significantly improving computational efficiency by using coarser grids in regions that do not require fine resolution.
Solution Approach 2:
The patent applies partial action by using fine spatial discretization only partially - specifically in regions near the front position - rather than applying it excessively throughout the entire model. The trained machine model predicts where front positions will be, and refinement is applied only in those specific local regions, achieving necessary accuracy without the computational cost of universal fine discretization.
3Device complexity
If static grid configuration is used, then device complexity is reduced, but adaptability to dynamic front movement decreases
Solution Approach 1:
The patent implements dynamics by making the grid configuration adaptive rather than static. The spatial reservoir model dynamically adjusts its discretization levels based on predictions from the trained machine model about front positions. This allows the grid to automatically adapt to moving fronts without requiring complex manual reconfiguration, achieving high adaptability with manageable device complexity through automated dynamic adjustment.
Solution Approach 2:
The patent applies self-service by using the trained machine model to automatically predict front positions and trigger grid refinement in the appropriate regions without requiring external intervention or complex control systems. The system serves itself by using its own prediction capability to guide the discretization adaptation, improving adaptability to dynamic front movement while keeping the overall device complexity relatively low.
Data Source
AI summary
A method of operating a reservoir simulator can include performing a time step of a reservoir simulation using a spatial reservoir model that represents a subterranean environment that includes a reservoir to generate simulation results for a first time where the simulation results include a front defined by at least in part by a gradient at a position between portions of the spatial reservoir model; predicting a position of the front for a subsequent time step for a corresponding second time using a trained machine model; discretizing the spatial reservoir model locally at the predicted position of the front to generate a locally discretized version of the spatial reservoir model; and performing a time step of the reservoir simulation using the locally discretized version of the spatial reservoir model to generate simulation results for the second time.