Probabilistic/mechanistic modeling of slug-length distribution in a horizontal pipeline

A probabilistic/mechanistic modeling was carried out to develop a predictive model for fully developed slug-length distribution in a horizontal pipeline. Statistical analysis suggested the appropriateness of a log-normal model over an inverse-Gaussian model for predicting slug-length distribution. A total of 64 data sets were used to empirically correlate the log-normal model. Two empirical relationships for mean slug length and slug-length standard deviation were developed. A statistical analysis revealed that, in addition to pipe diameter and mixture velocity, volumetric flow rate of the liquid film in the bubble region and the momentum exchange between the slug body and the liquid film are significantly correlated to the mean slug length. The slug-length standard deviation was found to have a significant correlation with film liquid holdup and momentum exchange. A model-validation study demonstrated the capability of the probabilistic/mechanistic model to reproduce the experimental data with a satisfactory match. The match was improved when the developed correlations were tuned using the statistical confidence intervals (CIs) of their coefficients.