Machine Learning Simulation for Flow-After-Flow Well Testing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current flow-after-flow tests in hydrocarbon wells require personnel to travel to the well location, leading to curtailed production and reduced testing frequency due to environmental conditions.
Innovation Solution
A simulation system that trains a machine learning model using historical well production data to simulate flow-after-flow tests, allowing for the prediction of flowing pressures and temperatures without taking the well offline or requiring on-site personnel.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If personnel travel to the well location to conduct flow-after-flow tests, then the test can be performed, but the well production is curtailed and testing frequency is reduced
Solution Approach 1:
The patent creates a virtual copy of the flow-after-flow test through machine learning simulation. The ML model replicates the physical testing process by processing historical well data and generating simulated flow rates, pressures, and temperatures, eliminating the need for actual well shutdowns while maintaining test validity.
Solution Approach 2:
The patent replaces the mechanical/physical testing system with a computational simulation system. Instead of physically conducting tests at the well site requiring personnel presence and well shutdown, the system uses software algorithms to simulate the entire flow-after-flow test process, substituting physical operations with digital computation.
2Reliability
If personnel travel to the well location to conduct flow-after-flow tests, then the test can be performed, but environmental conditions may prevent testing
Solution Approach 1:
The patent creates a virtual copy of the flow-after-flow test through machine learning simulation. The ML model replicates the physical testing process by processing historical well data and generating simulated flow rates, pressures, and temperatures, eliminating the need for actual well shutdowns while maintaining test validity.
Solution Approach 2:
The patent replaces the mechanical/physical testing system with a computational simulation system. Instead of physically conducting tests at the well site requiring personnel presence and well shutdown, the system uses software algorithms to simulate the entire flow-after-flow test process, substituting physical operations with digital computation.
3Measurement precision
If the well is taken offline to perform the test, then accurate flow potential data can be obtained, but production loss occurs
Solution Approach 1:
The patent performs preliminary actions by training the machine learning model on historical well data before conducting actual tests. The model is pre-trained to recognize flow patterns and reservoir characteristics, enabling it to accurately simulate flow potential without requiring real-time well shutdowns or offline conditions.
Solution Approach 2:
The patent creates a virtual copy of the flow-after-flow test through machine learning simulation. The ML model replicates the physical testing process by processing historical well data and generating simulated flow rates, pressures, and temperatures, eliminating the need for actual well shutdowns while maintaining test validity.
Data Source
AI summary
Disclosed are methods, systems, and computer-readable medium to perform operations including: receiving historical production data associated with a hydrocarbon well; preprocessing the historical production data to remove noise from the historical production data; using one or more machine-learning algorithms and the preprocessed historical production to train a simulation model to simulate a flow-after-flow test for the hydrocarbon well; and testing the simulation model to determine that the simulation model passes predetermined testing criteria.