Multiphase Transient Flow in Pipes

Abstract

The development of oil and gas fields in offshore deep waters (more than 1000 m) will become more common in the future. Inevitably, production systems will operate under multiphase flow conditions. The two–phase flow of gas–liquid in pipes with different inclinations has been studied intensively for many years. The reliable prediction of flow pattern, pressure drop, and liquid holdup in a two–phase flow is thereby important.With the increase of computer power and development of modelling software, the investigation of two–phase flows of gas–liquid problems using computational fluid dynamics (CFD) approaches is gradually becoming attractive in the various engineering disciplines. The use of CFD as a modelling tool in multiphase flow simulation has enormously increased in the last decades and is the focus of this thesis. Two basic CFD techniques are utilized to simulate the gas–liquid flow, the Volume of Fluid (VOF) model, and the Eulerian–Eulerian (E–E) model.The purpose of this thesis is to investigate the risk of hydrate formation in a low–spot flowline by assessing the flow pattern and droplet hydrodynamics in gas– dominated restarts using the VOF method, and also to develop and validate a model for gas–liquid two–phase flow in horizontal pipelines using the Eulerian– Eulerian method; the purpose of this is to predict the pressure drop and liquid holdup encountered during two–phase (i.e. gas–oil, gas–water) production at different flow conditions, such as fluid properties, volume fractions of liquid, superficial velocities, and mass fluxes.In the first part of this thesis, the VOF approach was used to simulate the droplet formation and flow pattern at various levels of liquid patched and restart gas superficial velocities. The effect of restart gas superficial velocity on the liquid displacement from the low section of the pipe showed a decrease in the remaining liquid with an increase in gas superficial velocity, and the amount of liquid depends on the fluid properties, such as density and viscosity. Moreover, the flow pattern is also strongly dependent on the restart gas superficial velocity as well as the patched liquid in the low section. A low gas superficial velocity with different patched liquids illustrated no risk of hydrate formation due to the observed flow pattern that is often a stratified flow. However, as the restart gas superficial velocity is increased, regardless of initial liquid patching, hydrate formation is more likely to be observed due to the observed flow pattern, such as annular, churn or dispersed flow.In the second part, the E–E model was employed to establish a computational model to predict the pressure drop and liquid holdup in a horizontal pipeline. Due to the complicated process phenomena of two–phase flow, a new drag coefficient was implemented to model the pressure drop and liquid holdup in the 3D pipe. Different simulations were performed with various superficial velocities of two–phase and liquid volume fractions, and were carried out using RNG k-ε model to account for turbulence. Based on the results from the numerical model and previous experimental study, the currently used E–E model is improved to get more accurate prediction for the pressure drop and liquid holdup in horizontal pipes compared with the existing models of Hart et al. (1989) and Chen et al. (1997).The improved model is validated by previously reported experimental data (Badie et al., 2000). The deviation of pressure drop and liquid holdup obtained throughout the CFD simulation with regard to the experimental data was found to be relatively small at low superficial gas velocities. It was observed that the pressure gradient increased with the system parameters, such as the drop size, liquid and gas superficial velocity and the liquid volume fraction, where the liquid holdup decreased.The developed model provided a basis for studying the pressure drop and liquid holdup in a horizontal pipe. Different parameters have been examined, such as gas and liquid mass flux and liquid volume fraction. Two empirical correlations have been examined (Beggs and Brill (1973), and Mukherjee and Brill (1985)) against the CFD simulation results of pressure drop and liquid holdup, it was noted that they gave better agreement with the air–oil system rather than the air–water system, but shows reasonable agreement over the entire gas mass fluxIn the third part, the coupling of Eulerian–Eulerian multiphase model with the population balance equation (PBE), accounting for droplet coalescence and breakage, is considered. Strengths and weaknesses of each numerical approach for solving PBE have been given in details. The Quadrature Method of Moments (QMOM) is used and particular coalescence and breakup kernels were utilized to demonstrate the droplet size distribution behaviour. Numerical simulations on a two–phase flow in a horizontal pipe, including coalescence and breakage are performed. The QMOM is shown to give the solution of the PBE with reasonable agreement. The numerical data are compared with the experiment data of Simmons and Henratty (2001). The flow variables, such as liquid volume fractions, gas and liquid superficial velocities are employed to examine the droplet size distribution and the potential of the multiphase k–ε with population balance model for predicting the two–phase pressure drop and liquid holdup.The significance of this work is to assist in understanding the risk of hydrate formation in bend pipes at gas–dominated restarts with different patched liquid values. The knowledge gained from this work can be utilized to avoid the hydrate formation operating conditions. The developed of multiphase flow E–E model will provide an accurate prediction for two–phase pressure drop and liquid holdup in a horizontal pipe which will be of benefit to the design of tubing and surface facilities.