Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Automated Flow Pattern Classification in Multi-phase Systems Using AI and Capacitance Sensing Techniques

Nian Ran, Fayez M. Al-Alweet, Richard Allmendinger, Ahmad Almakhlafi

TL;DR

This work tackles the problem of robust, real-time flow-pattern classification in multiphase systems where traditional visualization is subjective. It introduces an AI-driven platform that relies on simple capacitance-sensor data from two sensors, using 5-second, 100 Hz time-series to classify seven patterns, with the 1D SENet achieving over 0.85 accuracy on experiment-based data and about 0.71 on pattern-based data. The authors demonstrate improved robustness and practical viability for real-time monitoring, supported by a lightweight ~12 MB model and a comprehensive testing platform that includes high-speed imaging as reference. The study also analyzes generalization under distribution shifts, suggesting data augmentation and more diverse experiments to enhance out-of-distribution performance, with significant implications for industrial predictive modeling and process control.

Abstract

In multiphase flow systems, classifying flow patterns is crucial to optimize fluid dynamics and enhance system efficiency. Current industrial methods and scientific laboratories mainly depend on techniques such as flow visualization using regular cameras or the naked eye, as well as high-speed imaging at elevated flow rates. These methods are limited by their reliance on subjective interpretations and are particularly applicable in transparent pipes. Consequently, conventional techniques usually achieve context-dependent accuracy rates and often lack generalizability. This study introduces a novel platform that integrates a capacitance sensor and AI-driven classification methods, benchmarked against traditional techniques. Experimental results demonstrate that the proposed approach, utilizing a 1D SENet deep learning model, achieves over 85\% accuracy on experiment-based datasets and 71\% accuracy on pattern-based datasets. These results highlight significant improvements in robustness and reliability compared to existing methodologies. This work offers a transformative pathway for real-time flow monitoring and predictive modeling, addressing key challenges in industrial applications.

Automated Flow Pattern Classification in Multi-phase Systems Using AI and Capacitance Sensing Techniques

TL;DR

This work tackles the problem of robust, real-time flow-pattern classification in multiphase systems where traditional visualization is subjective. It introduces an AI-driven platform that relies on simple capacitance-sensor data from two sensors, using 5-second, 100 Hz time-series to classify seven patterns, with the 1D SENet achieving over 0.85 accuracy on experiment-based data and about 0.71 on pattern-based data. The authors demonstrate improved robustness and practical viability for real-time monitoring, supported by a lightweight ~12 MB model and a comprehensive testing platform that includes high-speed imaging as reference. The study also analyzes generalization under distribution shifts, suggesting data augmentation and more diverse experiments to enhance out-of-distribution performance, with significant implications for industrial predictive modeling and process control.

Abstract

In multiphase flow systems, classifying flow patterns is crucial to optimize fluid dynamics and enhance system efficiency. Current industrial methods and scientific laboratories mainly depend on techniques such as flow visualization using regular cameras or the naked eye, as well as high-speed imaging at elevated flow rates. These methods are limited by their reliance on subjective interpretations and are particularly applicable in transparent pipes. Consequently, conventional techniques usually achieve context-dependent accuracy rates and often lack generalizability. This study introduces a novel platform that integrates a capacitance sensor and AI-driven classification methods, benchmarked against traditional techniques. Experimental results demonstrate that the proposed approach, utilizing a 1D SENet deep learning model, achieves over 85\% accuracy on experiment-based datasets and 71\% accuracy on pattern-based datasets. These results highlight significant improvements in robustness and reliability compared to existing methodologies. This work offers a transformative pathway for real-time flow monitoring and predictive modeling, addressing key challenges in industrial applications.

Paper Structure

This paper contains 19 sections, 5 equations, 13 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Literature Review
  3. Methodology
  4. Experimental Setup
  5. Experimental Apparatus
  6. Methodology and Experimental Approach
  7. Data Acquisition and Pre-processing
  8. Artificial Intelligence Models
  9. Data preparation
  10. Model Choices
  11. SENet
  12. Experimental Results and Analysis
  13. Flow Patterns Observed on the Experimental Rig
  14. Experiments on AI models
  15. Hyperparameter Tuning
  16. ...and 4 more sections

Figures (13)

  • Figure 1: Illustration of common two-phase flow patterns observed in pipelines. From top to bottom: (a) Dispersed Bubble Flow, characterized by small gas bubbles distributed within the liquid phase; (b) Plug Flow, featuring larger gas bubbles separated by liquid plugs; (c) Elongated Bubble Flow, where bullet-shaped gas bubbles dominate; (d) Slug Flow, alternating liquid slugs and gas pockets; (e) Churn Flow, a chaotic mixture of frothy liquid and gas; (f) Annular Flow, with a liquid film along the pipe walls and gas flowing centrally; and (g) Stratified Flow, where gas and liquid phases are clearly separated.
  • Figure 2: (a) Photographic view of flow rig, (b) Gas injection panel and flow speed control devices, (c) Gas-liquid mixer.
  • Figure 3: A Schematic layout of the flow rig.
  • Figure 4: Photographic viewing box.
  • Figure 5: Capacitance sensor clamped on the test-section pipe and its electronic measurement device.
  • ...and 8 more figures