International Society for Industrial Process Tomography

GRNN-based ultrasonic tomographic detection method for the gas holdup of pipeline cross-section

Jingyi Hu¹, Lina Wang¹, Nan Li^1*

 ¹Northwestern Polytechnical University,  University, Xi’an, China

^*Email: nan.li@hotmail.co.uk

ABSTRACT 

Ultrasonic transmission tomography (UTT) is an effective non-destructive testing method for detecting gas-liquid two-phase flow patterns and cross-section gas-holdup ratio (CGR) during the reaction process of the bubble column. For this study, 16 ultrasonic transducers were evenly leaned on the outer wall of the pipeline, and the signals were excited with a five-period Hanning window modulation signal of 300kHz. SLBP (sector-diffusion-matrix-based LBP) algorithm, Tikhonov regularization algorithm, and Landweber iterative algorithm combined with ATF (adaptive filtering) processing were used for image reconstruction. GRNN-based method to estimate the CGR was studied in this paper.  RE_CG (relative error of CGR) was selected as the evaluation index. Single bubble (Case1, CGR=2.56%), dual-bubble  (Case2, CGR=5.12%), circulation (Case3, CGR=50%) and laminar flows (Case4, CGR=50%) were designed. As a non-image-based method, GRNN had more stable estimation results than image-based methods. GRNN estimated the RE_CG range of 0.28%-12.14%, while the imaging method estimated the range of 0.05%-38.96%. Simulation experiments proved that the RE_GC estimated by GRNN will decrease as the number of training samples increases. When the number of training samples reaches 140, the decrease tends to be flat. In addition, the calculation of CGR based on the GRNN method only took 0.009s, which can meet the real-time requirements in industrial processes.

Keywords: Cross-section gas holdup ratio (CGR), UTT, GRNN, gas-liquid two-phase flow

Industrial Application: Chemical bubbling bed reactors