Interfacial wave velocity of vertical gas-liquid annular flow at different system pressures

In two-phase annular flow, the velocity of interfacial waves play a crucial role in mass transport of liquid film, radial velocity distribution of liquid film, friction pressure drop, momentum transfer of two phases and heat-transfer characteristic of liquid film. Majority of existing interfacial wave velocity predictive correlations is established by the experimental data available in literatures which are limited to near atmospheric conditions. So it is necessary to further study the accuracy and applicability of predictive correlations at intermediate and high pressure conditions. The near infrared (NIR) sensor is designed for the interfacial wave velocity measurement in this paper. Air-water annular flow experiments have been conducted at five pressure conditions (from 0.2 MPa to 0.9 MPa) in a stainless steel tubular test section having an inside diameter 50.0 mm. A comparison of the performance of four existing predictive correlations (i.e. Kumar et al. [3] correlation, Omebere-Iyari et al. [5] correlation, Schubring et al. [8] correlation and Al-Sarkhi et al. [9] correlation) based on experimental dataset mentioned above (154 data points) is made. Comparison analysis between the correlations are made and appropriate recommendations drawn (Kumar et al. [3] correlation and Omebere-Iyari et al. [5] correlation). On the basis of the Kumar et al. [3] correlation, both the density increasing of gas core caused by entrained droplet and relative velocities of gas core and liquid film interface are considered, the structure of model is optimized reasonably, a modified wave velocity correlation which could acceptably handle the data of different system pressures is proposed. Laboratory results indicate that the mean absolute error (MAE) of the modified correlation is 5.37% and the relative deviations are within ±10% error band. By extending the data range for empirical contants fitting, the 84.98% relative error of the improved model is within ±20% error band for the all available data (authors' and literature). The improved correlation has a certain extrapolation for different system pressure and pipe diameter conditions.