Development of ultrasonic tomography system for liquid/gas flow measurement in a vertical column

Abstract

Liquid/gas two-phase flow widely exists in many applications including chemical and petroleum industries. Measurement of this two-phase flow is an essential parameter for these applications where accurate flow measurement is required. Two-phase flow, a phenomenon of critical importance to oil and gas sector where compared with singlephase flow, its flow characteristic is much more complex. Thus, the improvement of the instrumentation and measurement technology for the two-phase flow such as the development of two-phase flowmeters will have a growing demand since it will bring significant benefits to many industries. Based on the advantage of tomographic imaging technique, non-invasive ultrasonic sensing technique is realized by using electronic measurement circuits for transmitting and receiving the analog signals, data acquisition system for transferring the data to the PC and most importantly the suitable image reconstruction algorithm for providing visual access to the two-phase liquid/gas flow and estimating the component distribution for real-time measurement. The important characteristic for sensor selection is noted and balanced between high operating frequency for increased spatial sensitivity and the cross-sectional area of the vertical column to avoid undetected measurement signal caused by complete attenuation. The enhancement of liquid and gas component distribution imaging over the common Linear Back Projection (LBP) algorithm has been successful by implementing the new Eminent Pixel Reconstruction (EPR) algorithm. Simulative study on liquid/gas two-phase velocity measurement for dual-plane ultrasonic tomography system also has been made. Two different methods based on the use of cross-correlation technique have shown the usability of the technique for estimating the flow velocity. The image reconstructions quality of the two-phase flow is seen to have a significant increase by combining the EPR algorithm with Median Filtering technique that eliminated pixel values which are unrepresentative of their surroundings. Another advantage of such combination is the smoothen effect on the reconstructed images, resulting in better visualization of the twophase liquid and components because the outcome have approximating shape and size if compared with the actual flow condition. Linear regression method are also introduced for configuring the appropriate threshold values for imaging different size of gas component inside the investigated column especially on the successful detection of small gas bubbles. The results obtained can be useful for the online monitoring of liquid/gas flow in many industrial processes such as chemical mixing process, fluid transportation or process monitoring.