PROEFSCHRIFT
ter verkrijging van de graad van doctor aan de Technische Universiteit Delft,
op gezag van de Rector Magnificus prof. ir. K.C.A.M. Luyben; voorzitter van het College voor Promoties,
in het openbaar te verdedigen op maandag 15 juni 2015 om 15:00 uur
door
Andreas Teunis van Nimwegen natuurkundig ingenieur
en copromotor: Dr. Eng. L.M. Portela
Samenstelling promotiecommissie:
Rector Magnificus voorzitter
Prof. dr. ir. R.A.W.M. Henkes Technische Universiteit Delft, promotor
Dr. Eng. L.M. Portela Technische Universiteit Delft, copromotor
Prof. dr. C. Sarica University of Tulsa
Prof. dr. P. Angeli University College London
Prof. dr. W.R. Rossen Technische Universiteit Delft
Prof. dr. ir. M.T. Kreutzer Technische Universiteit Delft
Prof. dr. R.F. Mudde Technische Universiteit Delft
This work was supported by NAM (Nederlandse Aardolie Maatschappij), a Dutch subsidiary of Shell and ExxonMobil.
Printed by: Proefschriftmaken.nl || Uitgeverij BOXPress Published by: Uitgeverij BOXPress, ’s-Hertogenbosch
ISBN: 978-94-6295-205-8
Copyright © by A.T. van Nimwegen
All rights reserved. No part of the material protected by this copyright notice may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying, recording, or by any information storage and retrieval system without written permission from the author.
Liquid loading is a major problem in the natural gas industry, in which gas production is limited by the accumulation of liquids in the well tubing. Liquid loading can be prevented by the injection of surfactants at the bottom of the well. The surfactants cause the liquid in the well to foam, thereby changing the gas-liquid flow in the well. The flow is characterized by the TPC (Tubing Performance Curve), which relates the average pressure gradient in the tubing to the gas flow rate. This work has two main goals: (i) To improve the understanding of the effect of surfactants on gas-liquid flow in pipes, which we characterize by a change in the generalised TPC. The generalised TPC relates the average pressure gradient to the gas and liquid flow rates in the pipe. (ii) To provide subsidies for the development of simple physically-based models for the effect of surfactants on gas-liquid flow.
We performed experiments in intermediate-scale pipes (lengths of 12 m to 18 m and diameters of 34 mm, 50 mm, and 80 mm) with air and water at atmospheric conditions, without and with surfactants. Multiple parameters, that also vary between different gas wells in the field, were varied: the gas and liquid flow rates, the pipe diameter, the pipe inclination, the surfactant type and the surfactant concentration. We performed a visualisation of the flow without and with surfactants to obtain qualitative results on the effect of surfactants on the flow morphology, and we related these results to quantitative measurements of the generalised TPC and the liquid holdup.
The behaviour of the generalised TPC is to a large extent determined by the transition between annular flow and churn flow. In annular flow without surfactants, at large gas flow rates, the water is present in a film along the pipe wall and in entrained droplets in the gas core; the water always moves upwards, which leads to a relatively regular flow morphology. In the churn flow regime, which occurs at low gas flow rates, the liquid film reverses, as the interfacial friction between the gas and the liquid, which drags the liquid upwards, no longer exceeds the gravitational force on the film. This leads to a complex flow morphology, a large liquid holdup and a large pressure gradient.
Surfactants cause the formation of foam through the hydrodynamics of the flow. The foam decreases the density and increases the volume of the film at the wall. This changes the balance between the interfacial friction and the gravitational force, which shifts the transition between churn flow and annular flow to lower gas flow rates. As a result, the generalised TPC is changed by the surfactants, leading to a decrease in the pressure gradient at low gas flow rates. An optimum surfactant
decreasing gas flow rate, increasing liquid flow rate, increasing pipe diameter, and decreasing inclination from horizontal. Qualitatively, these results are unaffected by the type of surfactant that is used.
From the results obtained in this work, we qualitatively understand the effect of surfactants on the gas-liquid flow, and we understand why surfactants are able to deliquify gas wells. However, a physically-based model is required to translate the results obtained in this work in a quantitative way to the large-scale gas wells. Such a model requires a characterization of the foaming behaviour of the surfactant-liquid mixture using a small-scale setup. We determined that a small-scale sparging setup, often used in the gas industry, is not suitable, because the hydrodynamics in the sparging setup differ too much from the hydrodynamics of annular flow and churn flow. A small-scale shaking test, in which the hydrodynamics more closely resemble churn flow, shows more potential to characterize the foaming behaviour of the surfactants in the context of gas-liquid flows.
Een belangrijk probleem in de aardgasindustrie is liquid loading, waarbij de gaspro-ductie wordt beperkt door een opeenhoping van vloeistoffen in de progaspro-ductiebuis van de gasput. Liquid loading kan worden voorkomen door oppervlakte-actieve stoffen onderin de gasput te injecteren. Door toevoeging van oppervlakte-actieve stoffen zal de vloeistof in de put gaan schuimen, waardoor een verandering van de gas-vloeistofstroming in de productiebuis optreedt. De stroming wordt gekarakteri-seerd door de TPC (Tubing Performance Curve), die het verband tussen de gemiddelde drukgradiënt en het gasdebiet weergeeft. Deze studie heeft twee doelstellingen: (i) het vergroten van het begrip van het effect van de oppervlakte-actieve stoffen op gas-vloeistofstroming door het beschouwen van de verandering van de gegenerali-seerde TPC. Deze gegeneraligegenerali-seerde TPC relateert de gemiddelde drukgradiënt aan zowel het gas- als het vloeistofdebiet. (ii) Een basis leggen voor de ontwikkeling van een simpel fysisch model voor het effect van oppervlakte-actieve stoffen op gas-vloeistofstroming.
We hebben stromingsexperimenten uitgevoerd in buizen van middelgrote schaal (lengtes van 12 m tot 18 m en diameters van 34 mm, 50 mm en 80 mm) met lucht en water onder atmosferische omstandigheden, zowel zonder als met toegevoegde oppervlakte-actieve stoffen. Verscheidene parameters, die ook tussen gasputten on-derling verschillen, werden gevarieerd: de gas- en vloeistofdebieten, de buisdiameter, de buishelling en de soort en concentratie van de oppervlakte-actieve stoffen. We visualiseerden de stroming, zonder en met oppervlakte-actieve stoffen, om kwalita-tieve resultaten te verkrijgen over het effect van oppervlakte-ackwalita-tieve stoffen op de stromingsmorfologie en we relateerden deze resultaten aan kwantitatieve metingen van de gegeneraliseerde TPC en de vloeistof-holdup.
Het gedrag van de gegeneraliseerde TPC wordt voornamelijk bepaald door de overgang tussen annulaire stroming en churn-stroming. Zonder het gebruik van oppervlakte-actieve stoffen doet zich bij hoge gasdebieten een annulaire stroming voor, waarin het water te vinden is in een film aan de buiswand en in druppels in de gaskern. Het water stroomt dan altijd naar boven, wat resulteert in een regelmatige stromingsmorfologie. In het churn-stromingsregime keert de stromingsrichting van de vloeistoffilm om, omdat de wrijving aan het oppervlak tussen het gas en de vloei-stof, die de vloeistof omhoog sleept, niet langer groter is dan de zwaartekracht op de film. Dit leidt tot een complexe stromingsmorfologie, een grotere vloeistof-holdup en een grote drukgradiënt.
de dichtheid en verhoogt het volume van de film aan de buiswand. Dit verandert de balans tussen de wrijving aan het oppervlak tussen gas en film enerzijds en de zwaartekracht anderzijds en verschuift de overgang tussen annulaire stroming en churn-stroming naar lagere gasdebieten. Daardoor veranderen de oppervlakte-actieve stoffen de gegeneraliseerde TPC, wat leidt tot een lagere drukgradiënt bij lage gasde-bieten. Er bestaat een optimale concentratie van oppervlakte-actieve stoffen waarbij de drukgradiënt het minimaal is. Deze optimale concentratie is hoger naarmate de film dikker is; daardoor is deze hoger bij lagere gasdebieten, hogere vloeistofdebieten, grotere buisdiameters en kleinere buishellingen ten opzicht van het horizontale vlak. Kwalitatief worden deze resultaten niet beïnvloed door het soort oppervlakte-actieve stof die wordt gebruikt.
De resultaten die in deze studie zijn verkregen geven een kwalitatief begrip van het effect van oppervlakte-actieve stoffen op de gas-vloeistofstroming en maken duidelijk waarom oppervlakte-actieve stoffen liquid loading in gasputten kunnen voorkomen. Een model op fysische grondslag is echter noodzakelijk om een kwantitatieve vertaal-slag te maken van de resultaten die in deze studie verkregen zijn naar grootschalige gasputten. Voor een dergelijk model is een karakterisering van het schuimgedrag van de mengsels van vloeistof en oppervlakte-actieve stoffen door middel van een kleinschalige opstelling nodig. We hebben vastgesteld dat een kleinschalige bruisop-stelling voor schuimopwekking, zoals die vaak wordt toegepast in de gasindustrie, niet geschikt is, omdat de hydrodynamica in deze opstelling te veel verschilt van die van annulaire stroming en churn-stroming. Een kleinschalige schudopstelling lijkt meer geschikt voor de karakterisering van het schuimgedrag van oppervlakte-actieve stoffen, omdat de hydrodynamica in een dergelijke opstelling meer lijkt op die van een churn-stroming.
Abstract vii
Samenvatting ix
1 Introduction 1
1.1 Liquid loading . . . 2
1.2 Gas-liquid flow . . . 4
1.3 Surfactants and foam . . . 9
1.4 Objectives and strategy . . . 17
1.5 Outline . . . 19
2 Flow Visualisation 23 2.1 Introduction . . . 23
2.2 Flow patterns in upward vertical air-water flow . . . 24
2.3 Surfactant solutions and foaming . . . 26
2.4 Experimental setup . . . 30
2.5 Results . . . 33
2.6 Conclusion . . . 44
3 Pressure gradient and holdup 47 3.1 Introduction . . . 47
3.2 Vertical air-water flow, with and without surfactants . . . 49
3.3 Experimental setup . . . 53
3.4 Results . . . 56
3.5 Conclusion . . . 70
4 Flow pattern map 73 4.1 Introduction . . . 73
4.2 Flow patterns in air-water flow . . . 74
4.3 Surfactants and foaming . . . 75
4.4 Flow-loop . . . 76
4.5 Experimental results . . . 76
5 Surfactants 87
5.1 Introduction . . . 87
5.2 Air-water flow in vertical pipes . . . 89
5.3 Surfactants and foaming . . . 92
5.4 Experimental setup . . . 94
5.5 Results of the surfactant characterization . . . 98
5.6 Pressure gradient measurements . . . 100
5.7 Flow visualisation . . . 105
5.8 Future work . . . 109
5.9 Conclusion . . . 110
5.10 Comparison of three surfactants . . . 111
6 Inclinations 117 6.1 Introduction . . . 117
6.2 Vertical and inclined air-water flow . . . 120
6.3 Surfactants and foam . . . 121
6.4 Experimental setup . . . 123
6.5 Results . . . 126
6.6 Conclusions . . . 144
7 Diameters 147 7.1 Introduction . . . 148
7.2 Vertical air-water pipe flow . . . 151
7.3 Previous work . . . 156
7.4 Experimental setup . . . 158
7.5 Results . . . 160
7.6 Film thickness δ and interfacial friction factor fi . . . 179
7.7 Conclusions . . . 184 8 Small-scale experiment 189 8.1 Introduction . . . 189 8.2 Theoretical background . . . 191 8.3 Previous work . . . 193 8.4 Experimental setup . . . 193 8.5 Results . . . 197 8.6 Conclusion . . . 203
8.7 Additional measurement data . . . 205
9 Conclusions and outlook 209 9.1 Conclusions for the flow loop experiments . . . 210
9.2 Outlook towards a flow model . . . 216
9.3 Small-scale experiments . . . 219
9.4 Concluding thoughts . . . 220
List of publications 233
Acknowledgements 235
Introduction
The production of hydrocarbon gas and oil requires to transport the fluids from the subsurface reservoir to the surface processing facilities. The transport system includes a subsurface well, and surface pipelines and risers. The processing facility can be a an offshore platform, or an onshore gas plant or refinery. Often the wells, pipelines, and risers transport multiphase flow. In the case of a gas reservoir, the liquid can be the liquid water present in the reservoir, which is produced along with the gas, or the liquid can be formed due to condensation of water vapour and hydrocarbons from the gas, as the pressure and temperature are decreasing along the trajectory of the well tubing. In the case of oil wells, often some associated gas is produced along with the oil. In this work, we focus on the production of natural gas.
When the pressure in the gas reservoir is large, the gas velocity in the well tubing is sufficient to drag the liquid upwards to the surface. However, towards the end of field life (which typically is a number of years), the pressure in the reservoir has become so low that the liquid can no longer be dragged upwards by the gas. This leads to an accumulation of liquid down-hole in the well, which severely limits the gas production. This phenomenon is known as liquid loading. Eventually, the production will cease entirely and the well has to be shut in, even through there is still natural gas remaining in the reservoir (see e.g. Lea et al. (2008)).
In the gas industry, a variety of techniques has been developed to postpone the onset of liquid loading to an as low as (economically) possible gas velocity. These so-called deliquification techniques can be divided into two categories: (i) methods that use only the energy of the well fluids to lift the liquids and (ii) methods that use an external energy source to lift the liquids (e.g. down-hole pumps or the injection of compressed lift gas). In this work, we focus on the deliquification of gas wells by injecting surfactants at the bottom of the well, which is an example of the first category. The surfactants cause the liquid in the well to foam, which decreases the density of the liquid phase. This has a large effect on the multiphase flow in the well, and it decreases the gas velocity of the onset of liquid loading. However, not much is known about the precise effect of surfactants on gas-liquid flow and no model is available to predict the pressure gradient, liquid holdup, and flow regime of such flows. In this work, we systematically investigate the effect of the surfactants on air-water pipe flow. These results will provide subsidy for the development of models on the effect of surfactants on gas-liquid flow.
In the remainder of this introduction chapter, first liquid loading is considered in more detail. Subsequently, the theory of gas-liquid pipe flow is briefly summarised, and a literature overview on the subject of foam is presented, where we focus on foam in gas-liquid pipe flow. Finally, we present the objectives and strategy of this work.
1.1
Liquid loading
To understand the occurrence of liquid loading, the so-called nodal analysis can be used. In this analysis, the gas well is considered to be a system of two components in series.
The first component is the flow of the gas through the reservoir, a porous medium, towards the bottom of the production tubing of the well, known as the bottom-hole. This flow leads to a pressure drop in the reservoir. Because this pressure drop increases with increasing gas flow rate, the bottom-hole pressure decreases as the gas flow rate increases. This behaviour is shown in the reservoir curve, which is also referred to as the Inflow Performance Relation (IPR), as indicated in figure 1.1. As gas is produced from the reservoir, the reservoir pressure slowly decreases over the years and the reservoir curve moves towards lower gas flow rates and lower bottom hole pressures.
The second component of the gas well is the flow from the bottom-hole, through the production tubing, to the surface. At the surface, the pressure is usually fixed by the compressor that is installed or by the downstream surface pipeline and separator. The pressure at the bottom-hole location is, therefore, determined by the multiphase flow in the tubing. If there is no liquid present, the pressure gradient in the tubing, and therefore the bottom-hole pressure, increases approximately quadratically with the gas flow rate. However, there is also liquid present, which changes the behaviour of the pressure gradient. At sufficiently large gas flow rates, the bottom-hole pressure increases with increasing gas velocity. At low gas velocities, liquid loading causes the bottom-hole pressure to increase with decreasing gas velocity. Therefore, there exists a certain gas flow rate for which the bottom hole pressure is minimum. This is shown in the Tubing Performance Curve (TPC) in figure 1.1. Unlike the IPR, the TPC is only dependent on the size and inclination of the tubing and on the properties and flow rates of the gas and liquid in the well, and in general it does not change towards the end of the field life (Lea et al., 2008).
Since in a gas well, the pressure at the surface, i.e. at the top of the tubing, is fixed, and the TPC relates the pressure at the bottom of the tubing to the gas flow rate, the TPC can also be seen as a curve that relates the gas flow rate to the average pressure gradient in the tubing. Therefore, the TPC characterises the dynamics of the flow in the tubing.
Of course, there can only be a single value of the bottom-hole pressure, which is found at the gas flow rate where the TPC and the IPR cross. Because stable production occurs only at gas flow rates above the minimum in the TPC, where the well is not liquid loaded, at most one stable operating point is obtained. As shown in figure 1.1,
Reservoir
~ 3km Tubing
Gas flow rate
Bottom-hole pressure
Tubing Performance Curve (2)
Reservoir curve (1) (declined pressure) Stable operating point Unstable operating points (1) (2)
Figure 1.1:Schematic of a gas well showing the two components in series: (1) The flow from the reservoir to the bottom-hole, represented by the reservoir curve and (2) the flow from the bottom-hole to the surface, represented by the Tubing Performance Curve. Stable operation is only possible at the conditions where the two curves cross, which is to the right of the minimum in the TPC (Lea et al., 2008). In the left image the two components are drawn in a schematic of a gas well, in the right image the curves corresponding to the two components are drawn.
at low reservoir pressure there is no stable operating point and production of gas is no longer possible. For a certain IPR (i.e. a certain reservoir condition) the occurrence of liquid loading is determined by the TPC: it occurs when the TPC does not cross the IPR or when the crossing occurs below the minimum in the TPC.
The goal of the deliquification of a gas well is to change the TPC such that at lower reservoir pressure a stable operating point can still be found. When surfactants are introduced, the TPC changes due to a change in fluid properties, because foam has a much lower density than water. We perform experiments in a flow loop to understand the effect of the surfactants on the TPC.
There are differences in boundary conditions between gas wells in the field and our flow loop in the laboratory. In gas wells, (i) there is a constant gas-to-liquid ratio, and (ii) the pressure gradient in the flow in the well is fixed by the pressure at the surface and the conditions in the reservoir. Therefore, the pressure gradient determines the gas and liquid flow rates and operation at gas flow rates below the minimum in the
TPC is not possible. In the flow loop experiments, the superficial gas velocity (usg)
and the superficial liquid velocity (usl) are fixed (where a superficial velocity is the
volumetric flow rate divided by the cross sectional area of the pipe), and usgand usl
loop: uslis set, and the liquids are always produced from the flow loop. However, like in a gas well, the relation between the gas and liquid flow rates and the pressure gradient still characterises the dynamics of the flow in the tubing. In the experiments, the gas-to-liquid ratio is not fixed, and therefore we construct a generalised TPC, i.e.
−∇P = f (usg, usl), which is the pressure gradient as a function of the superficial gas
and liquid velocities.
The goal of this work is to characterise the dynamics of the flow with and without surfactants in the tubing, by measuring the generalised TPC. We expect that changes in the generalised TPC in our experiments due to the surfactants are qualitatively the same as the effect of surfactants on the TPC in gas wells, and therefore that our experiments allow us to understand the deliquification of gas wells using surfactants. Note that in our experiments, we consider constant fluid properties (air and water at atmospheric conditions), unlike in gas wells, where changes in temperature and pressure along the length of the tubing lead to changes in the properties of the fluids. In our experiments, we vary parameters that also vary between actual gas wells, i.e. the tubing diameter, the tubing inclination, the surfactant type and the surfactant concentration.
1.2
Gas-liquid flow
Although in this thesis we focus on the flow of air and surfactant-water solutions in pipes, the study of air-water flow without surfactants (or, more generally, gas-liquid flows) in pipes is also very relevant. This is because we study the effect of the surfactants on air-water flow, and, therefore, we use air-water flow as a reference case. Furthermore, the surfactant solution does not foam yet when it is introduced into the setup and the foam is only formed through the hydrodynamics of the air-water flow. Throughout this thesis, we will observe how the nature of the air-water flow affects the properties of the resulting foam.
There exists a long history of research on gas-liquid pipe flow, because it is encoun-tered in many applications. The research started with visual observations of the flow, which led to the categorisation of the flow into different flow patterns, which are described in the work by e.g. Hewitt and Hall-Taylor (1970). The transitions between the flow patterns were quantified through experiments and models to obtain flow pattern maps. From these maps, the flow pattern can be predicted based on the gas and liquid flow rates, the fluid properties and the pipe diameter. These flow pattern maps have been developed for horizontal (Mandhane et al., 1974; Taitel and Dukler, 1976), inclined (Spedding and Nguyen, 1980; Barnea et al., 1985) and vertical pipe flow (Taitel et al., 1980; McQuillan and Whalley, 1985). An overview of these different flow pattern maps is given in a review paper by Cheng et al. (2008).
The different flow patterns occurring in vertical gas-liquid flow are presented in figure 1.2. At large gas velocities, there exists annular flow, in which gas is the continuous phase, and the liquid is present in a film at the wall of the pipe and in dispersed droplets in the gas core. There are waves at the interface of the liquid
Figure 1.2:Schematic of the flow patterns occurring in vertical gas-liquid flow (van ’t Westende, 2007).
film and two types of waves can be distinguished. First, there are ripple waves, which are small capillary waves that continuously form and disappear. Second, there are roll waves, which are much larger inertial waves, that are continuous along the pipe circumference and coherent along a large axial distance. Roll waves have a complex morphology, consisting of smaller waves and ligaments. The roll waves give a dominant contribution to the interfacial friction between the gas and the liquid phases and are the largest cause of entrainment of droplets into the gas core (Belt, 2007).
In annular flow, the liquid film continuously moves upwards. When the gas velocity is decreased, at a certain point the liquid film will start to move downwards inter-mittently. This marks the transition to churn flow. As the gas velocity is decreased further, flooding waves appear. These are large, aerated waves that transport the liquid upwards. Behind the flooding waves, first the liquid film moves upwards, next it reverses and flows down into the next flooding wave. The morphology of the liquid film for churn flow with flooding waves is much more irregular than for annular flow, and the liquid film in churn flow contains many bubbles, droplets and ligaments (Hewitt and Jayanti, 1993).
At even lower gas flow rates, there exists a transition to slug flow, in which large Taylor bubbles are alternated with liquid slugs. As the gas flow rates decreases even further, there exists a bubbly flow, where dispersed bubbles are present in a continuous liquid phase. Liquid loading is related to the transition between annular flow and churn flow. Therefore, we focus our research on these flow patterns. There are three closely related phenomena in gas-liquid flows: (i) liquid loading, which was explained in section 1.1, (ii) the transition between churn flow and annular flow and (iii) flooding. To understand flooding, we consider a vertical pipe, where liquid is introduced in the middle, and the liquid is removed when it reaches either the bottom or the top of the pipe. Gas flows upwards through the pipe. There are now
two options: either the gas velocity is sufficient to drag the liquid upwards (co-current flow), such that the average velocity of the liquid in the pipe is upwards, or the gas velocity is insufficient and the liquid moves downwards on average (counter-current flow). For sufficiently long pipes, either all the liquid will reach the top of the pipe, or all liquid will reach the bottom of the pipe. The transition between these two regimes is the flooding point (Richter, 1981). Because criteria for the prediction of flooding consider the reversal of the liquid film as the flooding point (neglecting the entrained droplets), and liquid loading is so strongly related to the transition between annular flow and churn flow, in practice correlations for flooding and liquid loading are often applied to predict the transition between churn flow and annular flow.
In the gas industry, the Turner criterion (Turner et al., 1969) is most often used to predict liquid loading. It assumes that liquid loading occurs when the largest droplets in the flow can no longer be moved upwards by the gas. This requires an estimate of the size of these large droplets. This is done by introducing a critical Weber number, which is the ratio of the inertia of the gas acting on the droplet and the surface tension:
Wecr=
ρgu2gd
γ = 30, (1.1)
where ρgis the gas density, ugis the gas velocity, d is the droplet diameter and γ
is the surface tension. Once the critical droplet diameter is known, the gas velocity below which liquid loading occurs is obtained from a balance of the drag force and the gravitational force acting on the droplet, which yields:
ug= 40 (ρl− ρg) γg ρ2 gCD 14 , (1.2)
where ρlis the liquid density, g is the gravitational acceleration, and CDis the drag
coefficient (about 0.44 for large droplets). The Turner criterion works reasonably well to predict liquid loading in gas wells, although some improvements have been proposed (Veeken et al., 2010). Liquid loading for air-water flow under atmospheric
conditions is predicted at usg ≈ 14.5 m/s using this criterion, which is close to the
gas velocity at the transition between churn flow and annular flow (usg = 15m/s)
obtained in the lab experiments by van ’t Westende (2007). The minimum of the TPC
under these conditions is found at usg= 20m/s.
A relation often used to predict flooding is the Froude number criterion, proposed by Wallis (1969). The Froude number is a dimensionless number indicating the ratio of inertial forces due to the gas flow and the gravitational force acting on the liquid film. The onset of flow reversal occurs when these forces balance (Wallis, 1969), i.e. at
Fr = ρ 1/2 g usg (gD(ρl− ρg)) 1 2 ≈ 1. (1.3)
Note that this leads to a flooding point (and therefore a transition between annular flow and churn flow) that is dependent on the pipe diameter D.
Another way to estimate the flooding point is using the Kutateladze number, which is a balance of inertia of the gas phase with surface tension forces and buoyancy (Pushkina and Sorokin, 1969). The transition occurs at:
Ku = ρ
1/2 g usg
(gγ(ρl− ρg))
1/4 ≈ 3.2 (1.4)
Note that the Kutateladze criterion is almost equivalent to the Turner criterion
(equa-tion 1.2) because ug≈ usg, although the Turner criterion was derived from a balance
of forces on the droplets to predict liquid loading, whereas the Kutateladze criterion is derived to predict flooding from the film reversal.
Richter (1981) states that the Froude number criterion is valid to predict flooding for smaller pipe diameters (D < 0.05 m), while the Kutateladze criterion is accurate for large pipe diameters (D > 0.15 m). Richter (1981) performed a balance of forces on a liquid film with a semi-spherical wave, from which he derived an expression valid for all pipe diameters. In the limits of small and large pipe diameters the expression by Richter (1981) reduces to equations 1.3 and 1.4, respectively.
The above correlations do not yet give us a good understanding of the physics of the transition between churn flow and annular flow. The Turner criterion is able to predict the transition in a 50 mm vertical pipe at atmospheric conditions, but predicts droplets with a diameter of about 7 mm to be present in the flow, while measurements indicate that such large droplets do not appear in the gas core of annular flow (van ’t Westende, 2007). Richter (1981) made several bold assumptions in deriving his expression: he neglected entrainment and suggested that for the largest waves on the liquid film the pressure force and the surface tension force balance, while for such large roll waves inertia is much more important than surface tension.
In the Kramers laboratory, a number of studies have been carried out related to annular flow and the transition between annular flow and churn flow. Belt (2007) and van ’t Westende (2007) studied the liquid film and the droplets in annular flow in vertical and inclined pipes. Subsequently, Kalter (2010) developed an inside visualisation technique to study the air-water interface in annular flow in a horizontal pipe, which was improved by Khosla (2012) to also study annular flow in a vertical pipe. Near the transition between churn flow and annular flow, the morphology of the air-water interface changes and the number of ligaments and droplets that are formed starts to increase as the gas flow rate decreases. Khosla (2012) found a few of the large droplets predicted by Turner et al. (1969) near the interface in the churn flow regime. However, in later measurements in the same experimental flow loop, Tolboom (2014) found an increase in the size of the droplets near the interface around the annular-churn transition, but not as large as predicted by Turner et al. (1969). These studies indicate there are possibly some large droplets near the interface that are close to the droplet size predicted by Turner et al., but there are so few that they would not have a large effect on the flow.
From the previous work, it is clear that the large changes in the morphology of the interface and the droplet entrainment are key to explain the transition between annular flow and churn flow and liquid loading. For this reason, the experiments presented in this thesis include visualisation results, from which the morphology of the air-water interface is studied, both for flow with and for flow without surfactants. We discuss the flow patterns in detail, because they have a large effect on the pres-sure gradient, and, therefore, on the tubing performance curve. This is most easily explained from a balance of forces on the gas core in annular flow in a vertical pipe:
−αg
∂P
∂x − αgρgg − Fi= 0 (1.5)
where αgis the gas holdup, ρgis the gas density, g is the gravitational acceleration,
∂P
∂x is the pressure gradient, and Fiis the interfacial friction force between the gas and
the liquid phase per unit volume, and indicates the transfer of momentum between the two phases. Most of this transfer of momentum occurs at the interface between the gas and the liquid film at the wall, and only a small part is due to the entrainment of droplets (see Belt et al. (2009)). Note that there is no friction between the gas phase and the wall, as the wall is completely covered by a liquid film. In annular flow at
low liquid-to-gas ratios, the liquid holdup is small and αgis thus only slightly below
1. Furthermore, the hydrostatic pressure of the gas is negligible. Therefore, we can make the following approximation:
∂P ∂x ≈ 1 αg Fi= fi 4 D − 2δ ρg(ug− ui)2 2 ≈ fi 4 D ρg(ug)2 2 . (1.6)
This is similar to the pressure gradient for single-phase pipe flow with a rough wall.
In equation 1.6, ui is the velocity at the gas-liquid interface, and ug is the actual
gas velocity in the pipe. Most importantly, fi is the interfacial friction factor and
it is mostly determined by the morphology of the gas-liquid interface, i.e. by the structure and height of the liquid film (similarly, for a rough pipe, the friction factor
f is determined by the height and the structure of the roughness topography (van
Nimwegen, 2010)). The morphology of annular flow and churn flow is very different, leading to a very different behaviour of the interfacial friction factor. In annular flow, the morphology is relatively little affected by the gas flow rate, and the pressure gradient increases with increasing gas velocity. In the churn flow regime, the flow morphology quickly becomes more irregular as the gas flow rate is decreased, leading
to a large increase of fiand of the pressure gradient with decreasing gas flow rate. In
between, there is a minimum in the TPC, which plays a key role in the phenomenon of liquid loading.
In this thesis, we use a high-speed camera to study the flow morphology of the flow with and without surfactants. These results are related to the pressure gradient
through fi. The changes in the pressure gradient at different gas and liquid flow
rates due to surfactants and how they relate with the flow morphology will help us understand how and why surfactants are effective in gas well deliquification. They
will also provide subsidies for the development of physically based models, and for
this, fiplays an important role.
1.3
Surfactants and foam
In the experiments described in this thesis, surfactants are added to the water-phase in air-water pipe flow. The word surfactants is short for surface active agents. Surfactants are molecules with an apolar, or hydrophobic, part and a polar, or hydrophilic, part. Due to this structure, they will preferentially adsorb at the air-water interface. As the surfactant concentration is increased, more surfactants are adsorbed at the interface, decreasing the surface tension of the air-water interface by up to a factor
2 − 3(de Gennes et al., 2004). At the critical micelle concentration (cmc), it becomes
favourable for the surfactants to group in micelles in the bulk solution. A micelle is an aggregate of surfactants in which the hydrophobic tails of the surfactant molecules are sheltered from the water phase by the hydrophobic heads. Above the cmc the surface tension no longer decreases. The equilibrium spatial distribution of the surfactants in solution, as a function of the concentration, is illustrated in figure 1.3. For a more extended discussion of the adsorption of surfactants, see e.g. Myers (2005).
Since it takes time for the surfactants to diffuse towards, and be adsorbed at, the interface, thereby decreasing the surface tension of the water, the surface tension of newly created interfaces initially is still equal to the surface tension of water with air. Over time, the surface tension decreases. The rate of this decrease is determined by the diffusion coefficient of the surfactant in water and the details of the adsorption process. The time-dependency of the surface tension is illustrated in figure 1.4 (see also Eastoe and Dalton (2000)). The dynamics of the surface tension allow the formation of a stable foam. Pure liquids can never foam, because perturbations of the liquid films are never restored, causing the films to rupture quickly. Therefore, bubbles at a pure water interface disappear within a second. The dynamic effects of the surfactants at the interface stabilise the foam films. When a foam film is locally stretched and, therefore, becomes thinner, the local surface area of the film is also increased, which leads to a decrease of the surfactant concentration at the interface. Therefore, the local surface tension will increase, causing a surface tension gradient. The area of larger surface tension will “pull“ the liquid along the surface causing a flow along the surface proportional to the surface tension gradient. This Marangoni flow leads to a flow of liquid towards the stretched and thinned part of the liquid film, thereby stabilising the foam (Pugh, 1996).
Foam is a complex substance, because processes occurring at a wide variety of length-scales determine its bulk properties. This is illustrated in figure 1.5. At a very small scale, there exist diffusion and adsorption of surfactant molecules at the interface, as was described above. On the scale of the individual bubbles, there are Plateau’s rules, which determine the local structure of the foam. The bubbles in the foam are polyhedral in shape, and are surrounded by liquid films. The edges of the bubbles
cmc Gas Gas Gas Liquid Liquid Liquid Micelle γeq (mN/m) log (C)
Figure 1.3:Illustration of the decrease in surface tension with increasing surfactant concen-tration, due to adsorption of surfactants at the air-water interface. The cmc is also indicated, above which micelles form and usually the surface tension no longer decreases. The image is taken from Kawale (2012b).
of these Plateau borders meet at vertices of the bubbles, at angles of 109.5◦, to form
nodes. Note that these Plateau’s rules only apply to dry foams; at water contents above 3 percent, Plateau borders where four foam films meet start to appear (Weaire and Hutzler, 1999).
On the scale of several bubbles, rearrangements of the foam structure occur. These are caused by two types of events, denoted as T1 and T2, which are illustrated in figure 1.6. These processes are closely linked to the rheological behaviour of the foam. The flow of foam requires internal rearrangements of the bubbles through T1 processes, in which two bubbles that were previously separated become neighbours and vice-versa. The structure of the foam changes irreversibly when the foam is flowing. Over time, the foam undergoes coarsening as gas diffuses from smaller to larger foam bubbles. This diffusion is caused by the larger Laplace pressure in the small bubbles, due to their larger interface curvature. The disappearance of a small bubble through this gas diffusion is a T2 event. The structure of the foam, therefore, also changes irreversibly when the foam is not flowing (Höhler and Cohen-Addad, 2005).
γt (mN/m) t (s) Gas Gas Gas Gas Liquid Liquid Liquid Liquid
Figure 1.4:Illustration of the dynamic surface tension, i.e. the decrease of the surface tension of a freshly created air-water interface in time, due to diffusion and subsequent adsorption of the surfactants at the interface. The image is taken from Kawale (2012b).
The rheological behaviour of the foam has been summarised in several review papers (Kraynik, 1988; Höhler and Cohen-Addad, 2005; Cohen-Addad et al., 2013). In general, foam can be described as a shear-thinning fluid with a stress. The yield-stress is caused by the fact that a finite yield-stress is required for bubble rearrangements. Mathematically, this is expressed as a Hershel-Buckley relation between the shear stress τ and the shear rate du/dy:
τ = τy+ µhb
du dy
n
(1.7)
In this equation, µhb is called the plastic or Bingham viscosity. Foam is a
shear-thinning fluid, i.e. the effective viscosity decreases with increasing shear (n < 1). The viscosity of the foam depends on the water content. When the water content is large, the bubbles are not packed together, and in a liquid with bubbles there is no yield-stress: no bubble reorientations are required to make the water flow. At a gas volume content of 64%, corresponding to the packing fraction in a random packing of spheres, the bubbles attach to each other and the liquid with bubbles becomes a foam.
Chemistry
Physics
Engineering
molecular effects equilibrium of films continuum model of foam
~ 1 nm ~ 1 mm ~ 1 m
Figure 1.5:Illustration of the different length scales encountered when studying foam. The smallest scale is that of the surfactant molecules adsorbing at the surface. The internal structure of the foam is determined at the scale of several bubbles. In some engineering applications, such as gas wells, the relevant scales are much larger, as such wells are several kilometres long. Image taken from Weaire and Hutzler (1999).
At this point, reorientation of the bubbles is required for the foam to flow. In general, a larger gas fraction leads to a larger yield-stress (Höhler and Cohen-Addad, 2005). The rheological properties of foam vary by orders of magnitude, depending on the water content and the internal structure of the foam. Initially, the hydrodynamics
T1
T2
Figure 1.6:Illustration of bubble reorganisation processes in foams. Two bubbles that were originally separate become neighbours in a T1 event and a bubble disappears in a T2 event. Image taken from Höhler and Cohen-Addad (2005).
of the foam formation and the physical-chemical properties of the foam determine the bubble size distribution and the water content of the foam. Subsequently, the liquid content of the foam decreases over time due to drainage and the bubble size distribution changes through gas diffusion and T2 events, such that the average bubble size increases over time. Note that drainage is faster, for the same water content, when the bubble size of the foam is larger, leading to a complex interaction between drainage and coarsening. Furthermore, the drainage leads to a larger water content at the bottom of a foam layer than at the top. There is not yet a complete model of foam rheology taking all these factors into account (Hutzler and Weaire, 2011).
The complexity of the foam and the many factors affecting its internal structure make it difficult to perform reproducible experiments with foam. As indicated by Calvert and Nezhati (1987), experimental work on foam tends to be unrepeatable unless the conditions are very carefully controlled. The authors do indicate that they obtained reproducibility within 20%. The large variation of the properties of foam is shown in a review paper by Herzhaft (1999), who summarises results from the literature of the effective viscosity of foam with a gas fraction (or quality) of 95%, and finds that the results can differ by an order of magnitude, as illustrated in figure 1.7.
Figure 1.7:Review of different rheological measurements for foam with a gas fraction of 95 %. Image taken from Herzhaft (1999).
In this thesis, the focus is not on the "single-phase" flow of foam, but, instead, on the effect of foam on air-water flow in pipes. Therefore, in the first subsection below, publications from the gas industry that show the deliquification performance of surfactants in actual gas wells will be summarized. In the second subsection, we
refer to different laboratory experiments that considered the effect of surfactants on air-water pipe flow.
1.3.1
Surfactants for gas well deliquification
The main challenge in the deliquification of gas wells using surfactants is finding a surfactant that is effective at the actual operational conditions in the well, i.e. that leads to the creation of a stable foam at the bottom-hole temperature and pressure.
As the temperature at the bottom of gas wells can be above 100◦C, the surfactants
need to be chemically stable up to such temperatures. The pressure at liquid loading conditions typically is between 30 and 80 bara. Furthermore, the surfactants should be compatible with the liquid phase in gas wells, which can be a brine that contains
high valence ions (such as Ca2+ and Mg2+). These ions can precipitate together
with anionic surfactants (surfactant molecules that are negatively charged ions). In Chapter 5, we show that even the ions in tap water can lead to precipitation of a surfactant. The foam should be stable in the presence of hydrocarbon condensate from the well, which can act as a defoamer. Note that in gas wells, usually the water is foamed, although condensate foamers exist as well (Orta et al., 2007). The surfactant should not affect the reservoir rock, should not corrode the tubing, and should not freeze when stored outside in cold weather. Furthermore, there should be a chemical defoamer available, such that the foam can be collapsed when it reaches the surface. The surfactant should not create oil-water emulsions, or there must be a demulsifier available. For a more exhaustive list of the requirements of a surfactant for gas well deliquification, see e.g. Jelinek and Schramm (2005) and Bremner et al. (2010). Small-scale lab tests are used to evaluate surfactants on the criteria mentioned above. One of the tests often used in the gas industry is a standard sparger test, in which foam is created by sparging nitrogen through a porous plate, or through small needles, into the bottom of a column partially filled with the surfactant solution. In gas well deliquification, it is most relevant that the foam that is formed carries the liquid all the way to the top of the well. Therefore, in the tests with a sparger column, the amount of liquid carried out of the column by the foam is measured. In these tests brine and condensate can be used instead of pure water and the temperature can be raised to mimic the conditions at the bottom of the well. Surfactants that can carry more liquid out of the column are deemed more suitable for gas well deliquification (Nguyen, 2009; Bremner et al., 2010). A sparging test is considered in Chapter 8 of this thesis. The literature on the deliquification of gas wells contains many success stories describ-ing increased gas production after injection of surfactants. Some examples are given in the work of Xu and Yang (1995), Campbell et al. (2001), Bowman and Collins (2006) and Bremner et al. (2010). There exists, however, little knowledge about the influence of the surfactants on the flow in a gas well. A movie (Miller, 2009) of the actual flow with surfactants shows that most of the foam is transported along the wall of the pipe, while some droplets are entrained in the gas core. The foam holdup in the movie is low, indicating a small liquid-to-gas ratio. All foam is moving upwards, albeit slowly,
indicating an annular flow regime. A snapshot of this movie is presented in figure 1.8.
Figure 1.8:Snapshot of a movie made by Miller (2009) of the inside of a gas well that is being deliquified using surfactants.
Some basic modelling of the pressure gradient in the well in the presence of surfactants is described by Soni et al. (2009). Several models were considered, and a model based on plug flow of foam in the tubing yields the best results. However, such flow does not occur in gas wells, as shown in the movie (Miller, 2009). Furthermore, Soni et al. (2009) assume values of the surface tension that are much lower than in reality. A model of the flow with surfactants based on the actual physics of the flow in the well has not yet been developed.
In summary, there are many examples from the gas industry that show that surfactants are able to deliquify gas wells, and that the compatibility of the surfactants with the bottom-hole conditions is the most important criterion for surfactant selection. However, there exists very little knowledge on the actual flow with surfactants in the well.
1.3.2
Effect of surfactants on air-water pipe flow
Some early experiments on the effect of surfactants on air-water flow were performed by Saleh and Al-Jamae’y (1997). They determined the gas velocity required to lift liquid upwards in a 25.4 mm diameter vertical pipe in flow with and without surfac-tants. The surfactants reduced the superficial gas velocity corresponding to the onset of liquid fall-back by a factor 4, from 3.1 m/s to 0.8 m/s. However, they do not report liquid flow rates in their paper and they describe that their pipe is completely filled with foam, which is a different flow behaviour compared to that found in actual gas wells.
Christiansen (2006) performed experiments in a 50.8 mm vertical pipe at atmospheric pressure, also for air-water flow without and with surfactants. Unlike most authors, Christiansen did not fix the superficial gas velocity, but instead introduced air through a blower operated at a fixed electrical power. By decreasing the power to the blower, Christiansen determined the corresponding gas flow rate at which the liquid was no longer dragged upwards. The surfactants did not significantly change the gas flow rate required to lift the liquid, which is contrary to experiments by Saleh and Al-Jamae’y (1997). Unfortunately, no results of flow visualisation are presented, and we thus do not know how the flow in the experiments compares to that in an actual gas well. The pressure measurements by Christiansen also did not show a significant effect of the surfactant.
Duangprasert et al. (2008) performed experiments in a 19 mm diameter vertical pipe at atmospheric pressure, in all two-phase flow regimes indicated in figure 1.2. The superficial liquid velocity was varied between 0 and 0.136 m/s. The authors found no significant shift in the annular-churn flow transition due to the presence of surfactants. This is contrary to expectation, as liquid loading is related to the annular-churn flow transition. However, they did find a significant reduction of the pressure gradient due to surfactant in the churn-slug regime, at superficial gas velocities around 1 m/s. Sawai et al. (2004) investigated the effect of surfactants on air-water flow in a vertical pipe with a 25.8 mm diameter at atmospheric pressure. They found that the surfac-tants shift the transition from annular flow to churn flow to smaller gas velocities. This is consistent with the deliquification behaviour of the surfactants. Sawai et al. also observed that the pressure gradient in the churn flow regime is decreased by the surfactants. Using conductivity measurements, they determined the amount of water in the film, and found that it decreases due to the surfactants; i.e. the surfactants reduce the liquid holdup. The results are confirmed by the work of Rozenblit et al. (2006) who also studied the effect of surfactants on vertical pipe flow (25.4 mm di-ameter). Rozenblit et al. (2006) obtained a smaller effect of the surfactants compared to Sawai et al. (2004), which is most likely because they have used a low surfactant concentration.
Recently, Liu et al. (2014) and Liu (2014) studied the effect of surfactants on air-water flow in a vertical pipe with a diameter of 40 mm at atmospheric pressure. They made only limited observations of the flow pattern. They report a very large decrease of the pressure gradient due to the presence of the surfactant (up to 96% reduction) at gas flow rates corresponding to the churn and slug flow regimes for air-water flow. For surfactants to be effective in the deliquification of gas wells, the TPC has to be adjusted such that stable operating points are available up to lower reservoir pressure. This requires a shift of the minimum in the TPC to lower gas flow rates, which is related to a lower pressure gradient in the churn flow regime, and is consistent with the results by Liu (2014).
Further research on the effect of surfactants on air-water flow was performed by Liu and Gao (2007) for a capillary (1.6 mm), who found that the transition from
annular flow to churn flow shifted to slightly lower gas velocities when surfactants are introduced.
Xia and Chai (2012) performed measurements in pipes under small inclinations from the horizontal using electronic tomography. They found that the surfactant suppresses capillary waves and thereby increases the range of gas velocities of the stratified smooth regime.
Overall, most research on deliquification with surfactants in the gas industry is focused on the efficiency of the surfactants for the conditions at the bottom of a gas well. Not much research has been performed on the actual flow with surfactants in gas wells. The only model that is available assumes a plug flow of foam, which is not consistent with the flow observed in an actual gas well. There are several investigations on the effect of surfactants in air-water pipe flow. Most of them are consistent with the behaviour of surfactants in gas wells. In two studies (Christiansen (2006) and Duangprasert et al. (2008)), a different behaviour is obtained, but both these studies considered only few measurements at the relevant gas flow rates. No systematic studies exist that not only show the effect of the surfactant for different gas and liquid flow rates and concentrations, but also consider the surfactant type, the pipe diameter and the pipe inclination. Particularly the effect of the surfactants on the morphology of the flow, which can be observed using flow visualisation, has not been studied in detail yet. Furthermore, there is no model to predict the effect of surfactants on the pressure gradient for air-water pipe flow.
1.4
Objectives and strategy
The main motivation of the work described in this thesis is to improve the under-standing of the deliquification of gas wells using surfactants. There are two important facets of this deliquification: (i) the effect of surfactants, and of the generated foam, on the multiphase flow in the well and (ii) the compatibility of the surfactant with the fluids in the well. In this work, the focus is on the first of these two facets.
The effect of surfactants on the flow in the well has several aspects: 1. The hydrodynamics of the flow.
2. The formation of foam, caused by the hydrodynamics of the flow. 3. The interplay between (i) and (ii).
4. The consequences for the flow patterns, and for liquid loading.
These aspects are presented schematically in figure 1.9, and the investigation of these aspects forms the bulk of this thesis. The effect of surfactants on the flow is dependent on different parameters, which can be divided into three categories. First, there are design parameters, which are chosen fixed when the well completion is selected, or when deliquification measures are pursued:
1. Tubing diameter. 2. Tubing inclination.
3. Physical-chemical properties of the surfactant.
Second, there are reservoir parameters, which are determined by the conditions in the reservoir, and which vary over the years as gas is produced:
1. Pressure and temperature.
2. Rheological and physical properties of the gas and the liquid.
Third, the design parameters and the reservoir parameters together determine the following generated parameters:
1. Flow rates of gas and liquid. 2. Rheological properties of the foam.
These parameters are also illustrated in figure 1.9. In this work, we do not consider the reservoir parameters and limit ourselves to air and water at atmospheric conditions. Furthermore, the gas and liquid flow rates, which are now not determined by the reservoir conditions, are varied systematically.
The major objective of this research is to improve the fundamental understanding of the effect of surfactants on gas-liquid flow, and thus of the deliquification of gas wells for the different parameters mentioned, and to provide subsidies to create a physically-based model of gas-liquid flow with surfactants.
We build on a significant body of knowledge on gas-liquid annular and churn flow without surfactants and on previous experiments in the existing 50 mm flow loop in the Kramers laboratory by van ’t Westende (2007) and Belt (2007).
Our strategy is as follows:
1. We perform a visualisation of the flow in the 50 mm flow loop at a vertical orienta-tion, for air-water flow both without and with surfactants, in which we consider a single surfactant that is used in the gas industry (Trifoam 820 Block, Oilchem GmbH, Dessau-Roßlau, Germany). In the experiments, we vary the gas and liquid flow rates and the surfactant concentration. From the images of the flow morphol-ogy, we qualitatively consider the hydrodynamics of the flow without surfactants, the resulting foam formation when surfactants are added, and the effect of the foam on the flow morphology and the flow patterns.
2. We measure the pressure gradient and the liquid holdup in the 50 mm flow loop at a vertical orientation, at different gas and liquid flow rates both without and with surfactants. We consider only one surfactant (Trifoam 820 Block) and vary its concentration. From these measurements, we construct the generalised TPC,
i.e. −∇P = f (usg, usl), and determine the hydrostatic and frictional pressure
gradients and the interfacial friction. We use our qualitative visualisation results to better understand the quantitative holdup and pressure measurements. 3. We investigate the effect of the type of surfactant on the flow morphology, the
pressure gradient and the liquid holdup, in the 50 mm vertical flow loop..
4. We investigate the effect of the pipe inclination on the flow in the 50 mm flow loop. 5. We investigate the effect of the diameter on the flow, by additionally considering a
34 mm and a 80 mm flow loop, both at a vertical orientation.
These experiments are all performed at an intermediate scale (with a length of 12 m for the 34 mm and 50 mm diameter pipes and a length of 18 m for the 80 mm diameter pipe), in which the annular and churn flow patterns that exist in actual gas wells are obtained. When developing a model, the rheological and physical properties of the foam should be included, and these are related to the properties of the surfactant, to the reservoir parameters and to the second facet of the deliquification: the compatibility of the surfactant with the fluids in the well. It is difficult to test for these properties at our intermediate scale. For our model, we therefore want to determine the foaming characteristics of the surfactant solution in a small-scale test, and use the results of this test as an input for the model. In the gas industry, a sparging test is often used, hence we perform experiments to obtain more fundamental knowledge of this test. We are also considering a small-scale shaking test, and try to relate the results from this test to the experimental results at the intermediate scale.
1.5
Outline
The outline of the thesis is related to the objectives and strategy in figure 1.9. As a base case in this work, we consider a perspex pipe with an internal diameter of 50 mm in a vertical orientation with a Trifoam 820 Block surfactant. This base case is considered in chapters 2, 3, and 4, in which we perform measurements where we systematically vary the gas and liquid flow rates and the surfactant concentration. In chapter 2, we visualise the flow and study in a mostly qualitative way the effect of the surfactants on the morphology of the air-water interface. The surfactants reduce the superficial gas velocity that marks the transition from annular flow to churn flow. Therefore, the morphology is regular up to lower gas flow rates and consists of an almost stagnant foam film at the wall, with upwards moving foam waves that are superposed on this film. We also discuss the bubble size of the foam at different gas and liquid flow rates.
Flow hydrodynamics Foam formation
Change in flow pattern (Chapter 4) Tubing diameter (Chapter 7)
Tubing inclination (Chapter 6) Surfactant properties (Chapter 5)
Pressure and temperature (not considered) Gas and liquid properties (not considered) Gas and liquid flow rates (Chapters 2 - 4) Foam properties
Design parameters Intermediate scale-tests (Chapters 2 - 7)
Small -scale tests (Chapters 8 - 9)
Hydrodynamics Surfactant properties
Foam properties
Generated parameters
Reservoir parameters
Future flow model (Chapter 9)
Gas Wells Experiments
Figure 1.9:Schematic illustrating the scope of the current work. We consider different parame-ters in gas wells that affect the gas-liquid flow with surfactants in the production tubing (left column). We perform experiments at an intermediate scale, in which we study the effect of surfactants on air-water flow, while varying different parameters (top-right box). Furthermore, we characterise the foaming properties of surfactant solutions in small scale tests (bottom-right box). In the future, we want to use small and intermediate scale results to create a model for gas-liquid flow with surfactants. The chapters in which certain parameters and experiments are considered are indicated.
In chapter 3, we present the results for the pressure gradient and the liquid holdup. At large gas velocities, surfactants increase the total pressure gradient by increasing the frictional pressure gradient. At lower gas velocities, the surfactants decrease the total pressure gradient through a decrease of the liquid holdup, which in turn decreases the hydrostatic head. This also corresponds to a lower interfacial friction due to a more regular flow morphology.
In the literature, flow pattern maps for air-water flow are available, from which the flow pattern can be predicted as function of the gas and liquid flow rates. In chapter 4, we determine the effect of surfactants on the flow pattern map from visual observations of the flow. In this way, we classify the results that we obtained in chapter 2. The effect of the liquid flow rate and the surfactant concentration on the
gas flow rate corresponding to the transition between churn flow and annular flow is considered. Furthermore, we consider changes in the flow morphology due to changes in gas and liquid flow rates within a single flow pattern. We, therefore, identify the features of the morphology of the different flow patterns (e.g. ripples and roll waves for annular flow) and specify for which gas flow rates they occur. This analysis is performed for flow without and with surfactants, at two liquid flow rates and at two surfactant concentrations.
In chapter 5, we consider a different surfactant from the one considered in chapters 2-4, with very different physical-chemical properties, and show that qualitatively the two surfactants have a very similar effect on the flow. In section 5.10, we consider one more surfactant and show that it also has a similar effect on the flow.
Chapter 6 considers the effect of the inclination on flow without and with surfactants. For inclined flow, the transition between churn flow and annular flow occurs at larger gas flow rates, due to the thicker film at the bottom part of the wall. This effect is even stronger for flow with surfactants than for flow without surfactants: surfactants are thus more effective for larger inclinations from the horizontal (therefore, they are more effective in vertical pipes).
Chapter 7 considers the effect of the pipe diameter on air-water flow with and without surfactants in vertical pipes. Diameters of 34 mm, 50 mm and 80 mm are considered. Flow visualisation, pressure gradient measurements and holdup measurements are performed. The results show that surfactants are less able to decrease the superficial gas velocity at the transition between annular flow and churn flow at larger diameters. Chapter 7 is the last chapter that presents measurements from the intermediate-scale flow loop, as all design parameters of gas wells have been considered. Obviously measuring the effect of the reservoir parameters would be interesting as well, but is outside the scope of this thesis. In chapter 8, we consider the small-scale sparging experiment that is commonly used in the gas industry to evaluate the foaming perfor-mance at the conditions encountered at the bottom-hole location of a well. To gain a more fundamental understanding of the behaviour of the small-scale facility, we per-formed experiments at atmospheric conditions with nitrogen and pure water, using several surfactants. We show that the foaming performance is mainly determined by the hydrodynamics at the bubble formation and the physical-chemical properties of the surfactant.
Chapter 9, which is the final chapter, consists of two parts. The first part presents the overall conclusions of the experiments in the intermediate-scale flow loop. The second part presents some ideas for the modelling of gas-liquid flow with surfactants, using the experimental results generated in the intermediate-scale flow loop and in the small-scale tests. The second part also discusses how small-scale experiments can be used to evaluate the surfactant performance.
Note that, except for this first chapter and for the final chapter, all chapters are con-ference or journal papers. The paper in chapter 4 was presented at an international conference, the other papers have been submitted to peer-reviewed international
jour-nals (with some papers already published). Therefore, there is some overlap between the different chapters, especially in the introductory parts and in the description of the experimental setup.
Flow Visualisation
∗
In this work, the influence of surfactants on air-water flow was studied by performing ex-periments in a 12 metre long, 50 mm inner diameter, vertical pipe at ambient conditions. High-speed visualisation of the flow shows that the morphology of the air-water interface deter-mines the formation of foam. The foam subsequently alters the flow morphology significantly. In annular flow, the foam suppresses the roll waves, and a foamy crest is formed on the ripple waves. In the churn flow regime, the flooding waves and the downwards motion of the liquid film are suppressed by the foam. The foam is transported in foam waves moving upwards superposed on an almost stagnant foam substrate at the pipe wall. Foam thus effectively reduces the superficial gas velocity at which the transition from annular to churn flow occurs. These experiments make more clear how surfactants can postpone liquid loading in vertical pipes, such as in gas wells.
2.1
Introduction
Surfactants are molecules with a hydrophilic head group and a hydrophobic tail and are used in a wide range of applications. For example, surfactants are used to create emulsions, as wetting agents, and to create foam (see e.g. (Farn, 2006)). This latter function is of particular interest to the gas industry, as it is known from experience that it allows for a longer, stable production from a gas well.
This latter function is of particular interest to the gas industry, as it is known from experience that it allows for a longer, stable production from a gas well. In gas wells, both gas and liquid (water and/or gas condensate) are produced. This leads to a multiphase flow inside the well tubing that connects the reservoir with the surface. At the preferred operating conditions, at high gas velocities, the gas is able to drag along the liquid to the surface and the flow pattern in the well is annular dispersed. As the reservoir pressure declines, the gas flow rate decreases until the gas is no longer able to bring the liquid to the surface. Consequently, the liquid will start to accumulate at the bottom of the well. The additional liquid creates a hydrostatic pressure on the gas reservoir, severely limiting, or even prohibiting, gas production. This is called liquid loading and it is closely related to the changes of the flow pattern inside the gas well (see e.g. Lea et al. (2008)).
∗Published as A. T. van Nimwegen, L. M. Portela, and R. A. W. M. Henkes. The effect of surfactants
on air-water annular and churn flow in vertical pipes. Part 1: Morphology of the air-water interface. International Journal of Multiphase Flow, 71:133 – 145, 2014b
From field experience in the gas industry, it is known that the injection of surfactants at the bottom of the well postpones liquid loading, allowing additional gas production over some additional years (Lea et al., 2008), before closure of the well. However, the knowledge and understanding of the effect of surfactants on the gas-liquid pipe-flow is still poor.
There exists a long history of observation and visualisation of two-phase flow in pipes. For instance, visualisation was used to classify flow patterns for vertical flow, as summarised by Taitel et al. (1980). The goal of this work is to gain qualitative understanding of the influence of surfactants on upward air-water flow in a vertical pipe at ambient conditions using flow visualisation. More specifically, images from a high speed camera were used to characterise the changes in flow pattern morphology due to surfactants.
In the remainder of the paper, first the flow patterns in air- water flow are discussed in more detail. Next, a section is devoted to the behaviour of surfactants in aqueous solutions and the foam that is created from these solutions; previous work relevant for the current research is presented. After the description of the experimental setup, the results of flow visualisation are discussed, showing the changes in flow pattern due to surfactants.
2.2
Flow patterns in upward vertical air-water flow
Depending on the superficial liquid velocity, usl, and on the superficial gas velocity,
usg, different flow patterns occur in vertical pipes. On the left of figure 2.1, the flow
pattern map for air-water flow in a 50 mm inner diameter vertical pipe at ambient conditions is shown, based on the work by Taitel et al. (1980). In the present work, superficial gas velocities between 6.4 m/s and 45 m/s are considered, while the superficial liquid velocity is varied between 2 and 40 mm/s. This operational region is indicated by the rectangle in figure 4.1 and corresponds to the annular and churn flow regimes, which are shown in schematic drawings on the left of figure 2.1. At high gas velocities, co-current annular flow is obtained. In this flow pattern, the liquid is contained in a thin film at the wall and in droplets entrained in the gas core, and both the liquid film and the droplets continuously move upwards. On the interface between the liquid film and the gas core three types of waves can be distin-guished: ripple waves, roll waves and ephemeral waves. Ripple waves are capillary waves, with a small amplitude compared to the wavelength (Asali and Hanratty, 1993). Roll waves are much larger and stretch along the entire pipe circumference, remaining coherent while travelling large distances in the axial direction (Belt et al.,
2010). Roll waves occur only above a certain critical usl. At even larger superficial
liquid velocities, ephemeral waves are present. These are still much larger than the ripple waves, but do not span the entire pipe circumference (Wolf et al., 1996). When the superficial gas velocity is reduced in the annular flow regime, the roll
waves grow in size, causing the formation of large ligaments. At even lower usg, the
