The focus of this paper is on the measurement and calculation model of void fraction for the vertical upward co-current air–water slug flow in a circular tube of 15 mm inner diameter. High-speed photography and optical probes were utilized

with water superficial velocity ranging from 0.089 to 0.65 m·s

-1

and gas superficial velocity ranging from 0.049 to 0.65 m·s

-1

. A new void fraction model based on the local parameters was proposed

disposing the slug flow as a combination of Taylor bubbles and liquid slugs. In the Taylor bubble region

correction factors of liquid film thickness

C

δ

and nose shape

C

Z

*

were proposed to calculate

α

TB

. In the liquid slug region

the radial void fraction distribution profiles were obtained to calculate

α

LS

by employing the image processing technique based on supervised machine learning. Results showed that the void fraction proportion in Taylor bubbles occupied crucial contribution to the overall void fraction. Multiple types of void fraction predictive correlations were assessed using the present data. The performance of the Schmidt model was optimal

while some models for slug flow p

erformed not outstanding. Additionally

a predictive correlation was correlated between the central local void fraction and the cross-sectional averaged void fraction

as a straightforward form of the void fraction calculation model. The predictive correlation showed a good agreement with the present experimental data

as well as the data of Olerni

et al.

indicating that the new model was effective and applicable under the slug flow conditions.