International Society for Industrial Process Tomography

Spatio-temporal void fraction visualization on air-water two phase flow in vertical and horizontal pipe by combination of multiple current-voltage and machine learning (MCV-ML)

T.Kanamoto¹, Y.A.K. Prayitno^1,2, S.Miwa³, and M.Takei¹

¹Chiba University, Faculty of Engineering, Department of Mechanical Engineering, Chiba, Japan

²Universitas Gadjah Mada, Vocational College, Department of Mechanical Engineering, Yogyakarta ,Indonesia

³The University of Tokyo, Nuclear Professional School of Engineering, Ibaraki, Japan

ABSTRACT

One of the essential parameters for the safety and optimal control of industrial facilities is the void fraction α, which represents the percentage of gas phase volume in a unit volume of a pipeline. However, accurate α estimation is difficult due to the unsteady and inhomogeneous flows. In this study, Spatio-temporal void fraction α was visualized by combination of multiple current-voltage and machine learning (MCV-ML).MCV-ML has two phases which are 1) ML training and 2) ML evaluation. In the 1), several ML were developed for visualizing the spatio-temporal void fraction. In order to make training data for ML, in-situ void fraction is measured using WMS, and MCV voltage is simulated. In the 2), the actual MCV measured voltage and inputted into ML for visualizing the spatio-temporal void fraction. Experiments were conducted on air-water two phase flow in vertical pipe. As a result, the images of spatio-temporal void fraction is successfully generated. The spatial-averaged temporal void fraction estimated by trained ML shows qualitative agreement with void fraction measured by Wire-mesh sensor.

Keywords: void fraction, two-phase flow, multiple current-voltage, machine-learning

Industrial Application: General