Flow-Based Coarsening Mask for Reservoir Grid Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Coarsened models in reservoir simulations often lose accuracy due to the averaging process, particularly when applied to highly influential grid-cells, leading to a trade-off between computational time and preservation of physical relationships.
Innovation Solution
A method involving a computer processor to generate a multilevel coarsening mask based on flow property data, allowing for the creation of a coarsened grid model that reduces computational time while preserving relevant physical relationships by selectively coarsening or refining grid cells based on flow values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If upscaling is applied to coarsen the grid model, then computational time is reduced, but accuracy is lost due to averaging details in highly influential grid-cells
Solution Approach 1:
The patent applies different coarsening levels to different regions of the reservoir model based on flow property data. Highly influential grid-cells are identified and assigned lower coarsening levels to preserve their detailed characteristics, while less critical regions are assigned higher coarsening levels. This local differentiation resolves the contradiction by maintaining accuracy where needed while achieving computational efficiency elsewhere.
Solution Approach 2:
The patent segments the reservoir grid model into multiple coarsening levels, creating a hierarchical structure where different regions are coarsened to different extents. This segmentation allows the model to retain fine details in critical areas while coarsening non-critical areas, thus reducing overall computational time without sacrificing accuracy in influential regions.
2Device complexity
If uniform coarsening is applied to all grid cells, then computational complexity is reduced, but relevant physical relationships are lost
Solution Approach 1:
Instead of uniform coarsening, the patent implements local quality by assigning different coarsening levels to different grid cells based on their flow properties. This ensures that physical relationships are preserved in regions where they are most relevant while allowing greater simplification in less critical regions, thus reducing computational complexity without compromising reliability.
Solution Approach 2:
The patent changes the coarsening parameter dynamically based on flow property data. By using flow-based metrics to determine coarsening levels, the model adapts its complexity to the physical characteristics of the reservoir, preserving important physical relationships while reducing overall computational complexity.
Data Source
AI summary
A method may include obtaining model data for a reservoir region of interest. The model data may include flow property data based on streamlines. The method may further include generating a multilevel coarsening mask describing various coarsening levels that correspond to different flow values among the flow property data. The method may further include generating a coarsened grid model using the model data and the multilevel coarsening mask. The method may further include performing a reservoir simulation of the reservoir region of interest using the coarsened grid model.