Parallel Reservoir Simulation Using Sub-domain Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current reservoir simulation models face challenges in accurately simulating complex oil and gas reservoirs with multi-lateral wells and varying geological structures using structured grids, leading to inefficient processing times and inaccurate results due to the need for upscaling and the limitations of parallel computing with global hash tables.
Innovation Solution
A method for scalable parallel reservoir simulation that partitions the reservoir into sub-domains using geological characteristics, allowing for unstructured grids and structured grids to be simulated without upscaling, using a system of application servers that reorder local cell identification references and transmit data between sub-domains for efficient simulation of production characteristics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If structured grids are used for reservoir simulation, then parallel processing efficiency is improved, but geological accuracy deteriorates due to inability to represent complex structures
Solution Approach 1:
The patent divides the reservoir into multiple sub-domains, each handled by a separate application server. Within each sub-domain, unstructured grids are used to accurately represent complex geological structures while maintaining parallel processing capability across multiple servers. This segmentation allows each server to work independently on its sub-domain with high-resolution unstructured grids without requiring a global hash table.
2Manufacturing precision
If unstructured grids are used to accurately represent geological structures, then geological accuracy is improved, but computational complexity and memory requirements increase
Solution Approach 1:
The reservoir is divided into multiple sub-domains, each assigned to a separate application server. This segmentation distributes the computational complexity and memory requirements across multiple servers, allowing each server to handle a manageable portion of the unstructured grid data independently without requiring the entire grid to fit in single-machine memory.
Solution Approach 2:
Each sub-domain is assigned to a specific application server that handles it with locally optimized data structures and algorithms. The local cell identification references are reordered within each sub-domain to optimize memory access patterns and reduce communication overhead, allowing each server to efficiently process its portion of the unstructured grid without being burdened by the global complexity.
3Manufacturing precision
If global hash tables are used for unstructured grid simulation, then geological accuracy is improved, but memory requirements and processing overhead increase
Solution Approach 1:
The global hash table approach is replaced by local cell identification references within each sub-domain. Each application server maintains its own local references for the grid cells in its assigned sub-domain, eliminating the need for a large global hash table that would require excessive memory. This segmentation of the reference system allows unstructured grid simulation without the memory overhead of global hash tables.
4Manufacturing precision
If reservoir models with over one billion grid cells are simulated, then simulation accuracy is improved, but runtime exceeds acceptable limits
Solution Approach 1:
The reservoir model with over one billion grid cells is divided into multiple sub-domains, each assigned to a separate application server. This segmentation enables parallel processing where multiple servers simultaneously compute flow equations for their respective sub-domains, reducing the overall runtime from days to hours or minutes while maintaining high simulation accuracy through the use of unstructured grids.
Solution Approach 2:
The patent performs preliminary reordering of local cell identification references within each sub-domain before simulation begins. This preliminary action optimizes the data structure for the upcoming parallel computation, ensuring efficient memory access patterns and minimizing communication overhead during the simulation, thereby reducing runtime for models with over one billion grid cells.
Data Source
AI summary
A machine, computer program product, and method to enable scalable parallel reservoir simulations for a variety of simulation model sizes are described herein. Some embodiments of the disclosed invention include a machine, methods, and implemented software for performing parallel processing of a grid defining a reservoir or oil/gas field using a plurality of sub-domains for the reservoir simulation, a parallel process of re-ordering a local cell index for each of the plurality of cells using characteristics of the cell and location within the at least one sub-domain and a parallel process of simulating at least one production characteristic of the reservoir.