Probabilistic/mechanistic modeling was carried out to develop a predictive model for initiated slug-length distribution at the lower elbow of a hilly-terrain pipeline. Statistical analysis suggested the appropriateness of a Log-Normal model over an Inverse Gaussian model. The Log-Normal model is correlated by two empirical relationships developed for mean slug length and slug-length standard deviation. Based on experimental observations, the approach of critical liquid level (instead of critical liquid volume) was adopted as the slug initiation criterion at the lower elbow. Consequently, the critical liquid level was mechanistically modeled and empirically correlated to the initiated mean slug length and standard deviation. A model validation study demonstrated the capability of the probabilistic/mechanistic models to reproduce experimental data with a satisfactory match. The match is improved when the developed correlations were tuned using the statistical confidence intervals of their coefficients.
展开▼
