In order to simulate the transfer of mass, momentum and energy in the gas-liquid two-phase flow system, tremendous work focused on the phenomenon, mechanisms and models for two-phase flow in different channels, such as circular pipe, rectangular channel, rod bundle and annulus. Drift-flux model is one of the widely used models for its simplicity and good accuracy, especially for the reactor safety analysis codes (RELAP5 and TRAC et al.) and sub-channel analysis code (COBRA, SILFEED and NASCA et al.). Most of the adopted drift-flux models in these codes were developed based on the void fraction measured in pipe and annulus, which were different with the actual nuclear reactor. Although some drift-flux models were developed for rod bundles, they were based on the void fraction on the whole cross-section not in subchannel in rod bundles due to the lack of effective measuring methods. A novel sub-channel impedance void meter (SCIVM) has been developed to measure the void fraction in sub-channel of 5 × 5 rod bundles, which is adopted to evaluate these existing drift-flux models for rod bundles. By comparison, the values of drift-flux parameters have large differences among different correlations, which are suggested to be reconsidered. Based on the experimental data and physical laws, Lellouche-Zolotar and Chexal-Lellouche correlations show a better performance for drift velocity. If the predicting error of void fraction is the only concerned parameter, Chen-Liu, Ishizuka-Inoue and Chexal-Lellouche correlations are recommended for averaged relative error less than 30%. More experiments are suggested to focus on the distribution parameter and drift velocity through their definition.
