On the gas expansion and gas hold-up in vertical slugging columns—A simulation study

https://doi.org/10.1016/j.cep.2007.01.031Get rights and content

Abstract

A study on the gas phase expansion and gas hold-up occurring in free bubbling vertical slug flow is reported. A slug flow simulator (SFS) supported by air–water experimental data was used for this purpose. The study was accomplished by implementing in the simulator the gas phase expansion along the column. Effects over bubble lengths and bubble velocities were considered. The flow in 6.5 and 20 m long columns with internal diameter of 0.032 m was simulated. Expansion of gas phase along the column is shown to slightly decrease the occurrence of bubble coalescence. Liquid free surface oscillations (due to bubble burst events and continuous inlet of liquid and gas in the column) were found to affect the expansion of the gas phase, especially for high gas flow rates. The gas phase expansion for different outlet column configurations was studied. The use of a high and large cross sectional tank (to dampen free surface oscillations) is shown not to assure a permanent expansion rate of the gas phase. Simulations with and without gas expansion along the column were compared for the computation of average flow parameters. Approximate approaches (with constant UG, corrected for the mid-column pressure) were found suitable for the prediction of the average slug length and gas hold-up. Those approaches are, however, inadequate for the computation of the average bubble length and velocity along the vertical coordinate of the column.

Introduction

Gas–liquid mixtures flow in pipes in different flow patterns, which depend on flow rates, fluid properties, pipe diameter, inclination and configuration. Slug flow is one of these flow patterns and can be characterized by the flowing of long bubbles (known as Taylor bubbles) occupying most of column cross sectional area, separated by more or less aerated liquid plugs (termed slugs). It is a complex, irregular and intermittent phenomenon that can be found in several engineering applications (various types of reactors, membrane processes or extraction and transportation of hydrocarbons, just to mention a few) and even in natural phenomena (e.g. volcanic events).

Much of the primary modelling of slug flow was based on the early works of Dumitrescu [1], Davies and Taylor [2] and Nicklin et al. [3]. They set the bases for the first understanding of two-phase slug flow pattern. Several works followed focussing different aspects of such flow pattern (e.g. [4], [5], [6]).

The numerical simulation of two-phase vertical slug flow pattern has been attempted by several researchers (e.g. [7], [8], [9]). It serves as a tool for the study of the influence of several phenomena over the development of the flow, as well as an outcome predictor for any process/application in which this flow occurs. The usual approach requires the input of bubble-to-bubble interaction correlations relating the trailing bubble velocity to the length of the liquid slug ahead of the bubble. Different interaction correlations have been proposed (e.g. [8], [9], [10]) depending, for instance, on experimental conditions, fluid properties, flow regimes, etc. The simulation of slug flow pattern is often achieved, however, without accurate implementation of the gas phase expansion along the column (in terms of effect over bubble length and over bubble velocity). Two workarounds to address this problem are often implemented. The simplest one involves performing the flow simulation based on gas related parameters given at ambient pressure (e.g. [7], [8]). A more elaborate approach involves correcting those parameters for the pressure at the middle of the column (e.g. [11]). These are, nevertheless, approximate approaches which comprise limitations that should be considered while elaborating on data obtained in that way. In addition, there can be operating conditions whose simulation may not produce reasonable results when using such approximate approaches (for instance regarding flow simulation in long columns). There is thus a need for input in this area.

Two-phase flows are known to play a relevant role in volcanic events. Slug flow is believed, for instance, to be responsible for Strombolian eruptions at basaltic volcanoes [12]. There are also reports associating bubble coalescence and rise to both tremor and eruption seismic signals (e.g. [13]). In addition, bubble formation, ascent and their bursting at the surface are often related to strong pressure oscillations during volcanic events [14]. But these issues are also relevant for Industry in terms of the structural integrity of facilities (for instance in hydrocarbon and natural gas extraction plants). Thus, the implementation of the gas phase expansion in a slug flow simulator can be an asset to promote a deeper understanding of the flow dynamics at the source of those phenomena.

The main goal of this work is to provide information on the influence of the gas phase expansion over the evolution of the slug flow pattern in vertical columns. An algorithm for implementation of gas phase expansion along the column is proposed and issues like column outlet configuration and its influence on gas expansion rate, the pressure and bubble velocity oscillations and the gas hold-up inside the column are addressed.

Section snippets

Experimental work

A series of air–water co-current slug flow experiments [10] were performed in a 6.5 m long acrylic vertical column (0.032 m internal diameter). An image analysis technique [15] was used to collect data on the flow pattern characteristics, at two vertical coordinates (3.25 and 5.40 m from the base of the column) and for several superficial gas and liquid velocities (UG and UL up to 0.26 and 0.20 m/s, respectively). The operating conditions were designed to have turbulent regime in the main liquid

Onset of the simulation

A given number of randomly distributed liquid slugs (and Taylor bubbles) is assumed to enter the column at its base. These distributed variables “introduce” in the simulation the effect of the gas injection system (in terms of the length of the gas bubbles and liquid slugs formed). The slug length (normal) distribution is prepared using Box Muller algorithm [16] and the bubble length distribution is prepared as a dependent distribution (i.e. a function of the slug length distribution). Assuming

Simulation results

Three major topics are addressed in this section: the validation/benchmarking of the simulator, the gas expansion along the column and the gas hold-up in the column.

Conclusions

A simulation study on the gas phase expansion and gas hold-up in co-current slug flow is reported. The simulations including gas expansion and approximate approaches are shown to produce very similar estimates of the average liquid slug length. Similar matching can be observed regarding the average bubble velocity and length, provided that the UG estimates, used in the approximate approaches, are corrected for the vertical coordinate in question. In agreement with this, the gas phase expansion

Acknowledgments

The authors gratefully acknowledge the financial support of Fundação para Ciência e a Tecnologia through project POCTI/EQU/33761/1999 and scholarship SFRH/BD/11105/2002. POCTI (FEDER) also supported this work via CEFT.

References (18)

There are more references available in the full text version of this article.

Cited by (8)

  • 3D numerical study of a single Taylor bubble rising along an inclined tube through Newtonian and non-Newtonian liquids

    2023, Chemical Engineering and Processing - Process Intensification
    Citation Excerpt :

    So, our purpose was to assess the impact on slug flow main hydrodynamic features of wider liquid viscosity ranges and rheology, since it is more representative of systems found in the targeted applications. Different options are available to perform a numerical study on this flow pattern [30–32], but CFD tools (3D approach) are attractive to provide more precise and descriptive data. A detailed analysis of the flow regions identified around the bubble (nose, film and wake) covers the changes in the hydrodynamic features between systems with Newtonian and non-Newtonian liquids in tubes with different inclinations.

  • The dynamics of slug trains in volcanic conduits: Evidence for expansion driven slug coalescence

    2017, Journal of Volcanology and Geothermal Research
    Citation Excerpt :

    This raises a question regarding the extent to which coalescence could modulate the magnitude and timing of strombolian explosions during such rapid multiple-slug activity, by altering the distribution of slug gas masses within the conduit. In order to further consider the role of coalescence, we report here on the first experimental study into the behaviour of rising trains of interacting slugs in a vertical conduit, where expansion rates have been scaled to the volcanic scenario, which previous coalescence studies (Mayor et al., 2008b; Santos et al., 2008; Santos and Coelho Pinheiro, 2014) have not considered. We refer to the continuous slug flow investigated here as ‘slug train’ flow, preferring this nomenclature to ‘multiple slugs’.

  • Review on vertical gas–liquid slug flow

    2016, International Journal of Multiphase Flow
    Citation Excerpt :

    Several of the experimental and numerical works found in the literature, that provide important understanding on the characteristics of continuous co-current vertical slug flow, have neglected gas expansion along the vertical pipe (Barnea and Taitel, 1993; Hasanein et al., 1996; Pinto et al., 2001; Van Hout et al., 2001, 2002a, 2003). Implementing a simulation approach, Mayor et al. (2008b) reported this to be an acceptable assumptions since gas expansion only accounts for a slight decrease in the coalescence phenomenon. The data from Fig. 12 suggests the mixed scenario to be an intermediate between the fully turbulent and the fully laminar regimes, in line with the experimental findings of Mayor et al. (2008a, 2007a, 2008c), who observed strong interaction between consecutive bubbles for distances inferior to 3–4D in the mixed scenario, whereas for the turbulent flow pattern in the liquid and in the wake lmin = 8–10D, and for the fully laminar scenario lmin = 2D.

  • Stable and self-adaptive performance of mechanically pumped CO<inf>2</inf> two-phase loops for AMS-02 tracker thermal control in vacuum

    2011, Applied Thermal Engineering
    Citation Excerpt :

    It might be related to the vertical orientation of the evaporators, inside which two-phase CO2 flows up and down. Vapor bubbles may prefer staying at the top of the evaporator bends, leading to temporal vapor locking [12–14] until the bubbles were flushed away by a stronger flow. To conclude, the mechanically pumped CO2 two-phase loop cooling system for the AMS-02 tracker performs quite well in the thermal vacuum test.

View all citing articles on Scopus
View full text