SINDy-Based Fluid Flow Forecasting for Scalable Well Surveillance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for assessing well performance in multi-fractured horizontal wells are not scalable, reliable, or efficient, particularly in unconventional reservoirs, due to complex fluid flow dynamics and lack of robust, automated techniques for estimating fluid flow characteristics.
Innovation Solution
A hybrid sparse nonlinear regression (SNR) method using sparse identification of nonlinear dynamics (SINDy) to estimate fluid flow dynamics in multi-fractured horizontal wells, leveraging time-dependent basis functions and routinely available data like rates and pressures, without relying on fracture and reservoir flow mechanism assumptions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional manual interpretation methods are used for well performance evaluation, then interpretive accuracy may be maintained for simple cases, but the method does not scale for large fields or wells with large data volumes
Solution Approach 1:
The patent replaces manual mechanical interpretation methods with an automated computational system using sparse nonlinear regression and machine learning algorithms. The system automatically processes well test data, pressure data, and production data to estimate fluid flow dynamics, eliminating the need for manual surveillance while handling large data volumes from multiple wells simultaneously.
Solution Approach 2:
The patent introduces an intermediary computational model that acts as a bridge between raw well test data and performance predictions. The sparse nonlinear regression framework serves as an intermediary that processes complex multi-fractured horizontal well data, incorporating fracture and reservoir flow mechanisms to generate accurate performance estimates without requiring direct manual analysis.
2Extent of automation
If robust automated methods are developed for estimating fluid flow dynamics, then surveillance scalability is improved, but the complexity of the estimation method increases
Solution Approach 1:
The patent segments the complex fluid flow estimation problem into distinct components corresponding to different flow mechanisms (fracture flow, reservoir flow, boundary effects). The sparse nonlinear regression model separately estimates parameters for each segment, allowing the complex overall system to be managed through modular, independent parameter estimation that can be automated.
Solution Approach 2:
The patent transforms the complex qualitative interpretation task into quantitative parameter estimation by changing the problem parameters from interpretive categories to measurable physical parameters (pressure, flow rate, time). This parameter transformation enables automated computational processing while maintaining physical meaningfulness of the results.
3Quantity of substance
If manual well performance evaluation is performed, then data volume requirements are reduced, but the method cannot handle large data volumes from multiple wells
Solution Approach 1:
The patent replaces manual evaluation mechanics with automated computational mechanics that can process large data volumes efficiently. The sparse nonlinear regression algorithm automatically handles pressure data, flow rate data, and well test data from multiple wells simultaneously, transforming the data volume limitation into an advantage where more data improves estimation accuracy.
Data Source
AI summary
A method for determining fluid flow dynamics in a well comprises determining a first set of basis functions that is dependent on time. The method further comprises conducting a sparse identification of nonlinear dynamics (SINDy) evaluation based on the first set of basis functions. The method further comprises determining a rate normalized pressure for a primary phase rate. The method further comprises forecasting the primary phase rate for a plurality of bottomhole pressure measurements. The method further comprises determining a gas oil ratio based on a second set of basis functions that are dependent on both time and bottomhole pressure. The method further comprises calculating a secondary phase rate based on the determined gas oil ratio and the forecasted primary phase rate.