Virtual Flare Flowmeter Validation for Faulty PFF Readings
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Physical Flare Flowmeters (PFF) frequently malfunction and provide inaccurate readings, leading to incorrect flaring data, which existing methods fail to accurately validate or correct in real-time.
Innovation Solution
Implementing a machine learning-based Virtual Flare Flowmeter (VFF) that validates PFF readings by comparing them with predicted data, using image analysis to identify faulty devices and correlating operating conditions to determine accurate flow measurements, thereby selecting the correct reading between PFF and VFF.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If physical flare flowmeter is used to measure gas flare flow, then real-time flow measurement is obtained, but frequent malfunction and inaccuracy occur
Solution Approach 1:
An image processing system acts as an intermediary between the physical flare flowmeter and the validation process. The system captures images of the flare stack, processes them to extract flame characteristics, and uses these as a mediator to validate or correct the physical flowmeter readings, thereby improving reliability while maintaining real-time measurement capability
Solution Approach 2:
The patent creates a virtual copy of the flare flow measurement by processing images of the flame to derive flow rate information. This virtual flare flowmeter reading serves as a backup and validation mechanism, allowing comparison with the physical flowmeter to identify and correct malfunctions, thus improving measurement reliability
2Measurement precision
If virtual flare flowmeter based on machine learning is implemented, then measurement accuracy is improved, but device complexity increases
Solution Approach 1:
The patent replaces complex mechanical validation methods with machine learning-based image processing. Instead of using additional physical sensors or complex mechanical validation systems, the solution uses software-based machine learning models to process flame images and predict flow rates, achieving high measurement precision while managing system complexity through software rather than hardware
Solution Approach 2:
The patent transforms the measurement approach by changing from direct physical sensing to image-based parameter extraction. By capturing flame characteristics (length, intensity, color) from images and using machine learning to correlate these parameters with flow rates, the system achieves accurate measurements while avoiding the complexity of additional physical measurement devices
3Measurement precision
If manual analysis is used to determine root cause of deviation, then accurate diagnosis is achieved, but time consumption increases
Solution Approach 1:
The patent implements a self-diagnosing system that automatically compares physical flowmeter readings with virtual flowmeter predictions, identifies deviations, and determines potential root causes without requiring manual intervention. The system self-validates measurements and triggers alerts only when necessary, significantly reducing time loss while maintaining diagnostic accuracy through automated machine learning-based analysis
4Reliability
If image processing is used to validate flowmeter readings, then measurement validation is improved, but use of energy increases
Solution Approach 1:
The patent implements periodic image capture and processing rather than continuous processing. The system captures flare stack images at intervals, processes them to update the virtual flowmeter reading, and compares with the physical flowmeter. This periodic approach maintains validation reliability while significantly reducing energy consumption compared to continuous image processing
Data Source
AI summary
A computer-implemented method and system for validating physical flare flowmeter readings using virtual flare flowmeter predictions can include receiving physical gas flare flow measurement data from a physical flare flowmeter coupled upstream from a flare stack; receiving predicted gas flare flow measurement data from a virtual flare flowmeter; determining a quantitative deviation value between the physical gas flare flow measurement data and the predicted gas flare flow measurement data; and for quantitative deviation values less than a threshold deviation value, determining that the physical gas flare flow measurement data is accurate. The method and system can automatically identify abnormalities in any of physical flow meter (PFF) and virtual flow meter (VFF) and automatically select the accurate reading between the PFF and VFF to be used based on online plant and equipment conditions as inputs.