This paper is focused on the modelling of liquid droplet concentration profiles in horizontal stratified-annular flows. Two approaches are studied. First a review of the current state of 1D prediction models for liquid droplet concentration profiles is made. The limitations and assumptions are also discussed. Second, a new methodology is proposed as an alternative for the droplet concentration profiles prediction assuming an exponential droplet distribution in the vertical diameter. The methodology is built by obtaining empirical correlations using genetic algorithms. The algorithm implementation is made by using Binary trees and Prüfer encoding. As a result two empirical correlations are presented for the droplet concentration at the gas-liquid interface and the profiles decay coefficient. The correlations are developed for two-phase gas-liquid flows and are expressed in terms of three non-dimensional parameters including the effects of the physical fluid properties and operational conditions. The obtained two-phase flow correlations are extended to the three-phase oil-water-gas flows. The model and correlations are tested against recent experiments and available data from the literature.
Separated gas-liquid flows are very common in the oil and gas industry. One important mechanism to be considered at high gas velocities is the liquid droplet entrainment. In some cases a large fraction of the liquid transport in the pipe occurs in the droplet field. Predicting the liquid holdup in transport pipes is important so an accurate droplet model should be considered.