Multi-Channel Electrical Conductivity Sensor for Flow-Regime Classification
Supervisior: Univ.-Prof. Dr. Marco Da Silva
Co-Supervisor: DI Dr. Stefan Puttinger
For the observation of liquid-gas flow mixtures in vertical pipes a multichannel electrical conductivity sensor (MECS) was developed and evaluated. The primary goal is to create a sensor system with a high repetition rate, which can be used to assist in the validation of flow simulations and classify flow patterns.
The developed sensor system features an electrical excitation circuit with adjustable amplitude and frequency, as well as 16 transimpedance amplifiers for precise conductivity measurements of the multiphase flow. The experimental setup includes tests of the sensor in a controlled environment with a transparent pipe.
Sensor measurements were visualized and compared with recordings from a high-speed camera.
Machine learning algorithms were used to classify flow regimes based on extracted statistical features. The sensor demonstrates high effectiveness in distinguishing various flow patterns, confirming its potential for real-time monitoring in industry and the validation of CFD simulations.
This work provides a reliable, cost-effective, and robust sensor system for the classification of flow regimes and flow monitoring. It has potential applications for CFD validation, process monitoring and process optimization, thereby contributing to potential improvements in safety, efficiency and performance across various industries.
Keywords: flow patterns, conductivity sensor, flow visualization, classification
June 12, 2024