Well Network Steady-State Flow Partitioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Full-physics numerical simulations of large oil and gas reservoirs are computationally intensive and suffer from slow convergence due to non-linearities, leading to significant computational dependencies and messaging delays, especially in networks with over 1,000 wells.
Innovation Solution
The method involves decomposing the network into well sub-networks and pipe sub-networks, solving each sub-network in parallel with boundary conditions, and iteratively adjusting until convergence is reached, minimizing dependencies and messaging delays.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If full-physics numerical simulations are performed on large reservoir networks, then comprehensive simulation accuracy is improved, but computational time increases significantly
Solution Approach 1:
The network is partitioned into pipe sub-networks and well sub-networks, allowing independent parallel solving of each segment. This segmentation enables the simulation to process large networks efficiently by dividing the computational domain into manageable parts that can be solved simultaneously, reducing overall computational time while maintaining accuracy.
Solution Approach 2:
The solution approach transitions from solving the entire network as a single system to solving multiple independent sub-networks in parallel computational dimensions. This dimensional change in the solution space allows simultaneous processing of multiple sub-problems, dramatically reducing computational time while preserving the accuracy of the full-physics simulation.
2Productivity
If parallel processing is used to solve network equations, then computational speed is improved, but messaging dependencies increase significantly
Solution Approach 1:
By segmenting the network into pipe and well sub-networks with clearly defined boundaries, the patent minimizes the number of inter-dependencies between parallel processes. Each sub-network can be solved independently with local boundary conditions, reducing the messaging overhead and synchronization complexity associated with parallel processing.
Solution Approach 2:
The patent extracts and isolates the boundary conditions at sub-network interfaces, allowing each parallel process to solve its sub-network independently using local boundary data. This extraction of boundary conditions eliminates the need for continuous messaging and synchronization between parallel processes, reducing computational complexity while maintaining speed.
3Stability of the object's composition
If non-linear equations are solved with damping for convergence, then solution stability is improved, but convergence speed decreases
Solution Approach 1:
Segmenting the network into smaller sub-networks reduces the size of the non-linear systems that must be solved in each parallel process. This segmentation allows each sub-problem to converge faster while maintaining stability through damping, as the smaller system size reduces the propagation of non-linear effects throughout the entire network.
Data Source
AI summary
A method of managing a network of wells and surface facilities includes partitioning the network to obtain a pipe sub-network and two or more well sub-networks, and constructing an associated set of equations that represent a steady-state fluid flow in the sub-network. The method further includes setting boundary conditions for each well sub-network, and determining a steady-state flow solution for each well sub-network. The method further includes establishing boundary conditions for the pipe sub-network, and finding a steady-state flow solution for the pipe sub-network. If the solution does not match the estimated pressure, the method further includes adjusting the estimated pressure, repeating said setting, determining, establishing, finding, and adjusting operations until the calculated and estimated pressures converge, and analyzing flow rates of the steady-state flow solutions to evaluate suitability of a modification to the network.