Vous êtes sur la page 1sur 13

The Effect of Surfactants on Vertical

Air/Water Flow for Prevention of


Liquid Loading
A. T. van Nimwegen and L. M. Portela, Delft University of Technology, and R. A. W. M. Henkes, Delft University of
Technology and Shell Projects and Technology

Summary the flow in the well tubing to the total pressure losses (which
From field experience in the gas industry, it is known that includes, for example, the flow from the reservoir to the
inject- ing surfactants at the bottom of a gas well can prevent bottom hole) is relatively small, and the additional production
liquid load- ing. To better understand how the selection of the caused by the surfactants will be less than for deeper wells.
surfactant influences the deliquification performance, Even though sur- factants are often applied as a deliquification
laboratory experi- ments of air/water flow at atmospheric technique, under- standing of the influence of these surfactants
conditions were per- formed, in which two different surfactants on the multiphase flow in the well is still poor.
(a pure surfactant, sodium dodecyl sulfate, and a commercial Nodal analysis, in which the gas well is considered to be a
surfactant blend) were added to the water. In the experiments, sys- tem of two components in series, is often used to
a high-speed camera was used to visualize the flow, and understand liq- uid loading of gas wells. The first of the two
pressure-gradient measurements were performed. Both components is the flow of gas from the reservoir to the bottom
surfactants increase the pressure gradient at high gas-flow rates of the well tubing. The difference between the pressure in the
and decrease the pressure gradient at low gas- flow rates. The reservoir and the pres- sure in the bottom hole increases with
minimum in the pressure gradient moves to lower gas-flow increasing gas-flow rate. Therefore, as shown in Fig. 1, the
rates with increasing surfactant concentration. This is related to reservoir curve (also called the inflow-performance curve)
the transition between annular flow and churn flow, which is shows a decreasing bottomhole pres- sure as the gas-flow rate
shifted to lower gas-flow rates because of the formation of an increases.
almost stagnant foam substrate at the wall of the pipe. At high The second of the two components is the flow of gas (and
surfactant concentration, it appears that the churn flow re- liq- uid) from the bottom hole to the surface through the well
gime is no longer present at all and that there is a direct tubing. At large gas-flow rates, in which liquid loading does
transition from annular flow to slug flow. The results also not occur, the difference between the bottomhole pressure and
show that the crit- ical micelle concentration, the equilibrium the surface pressure increases with increasing gas-flow rate. At
surface tension, the dynamic surface tension, and the surface low gas-flow rates, because of liquid-loading problems, the
elasticity are poor pre- dictors of the effect of the surfactant on difference between the bottomhole pressure and the surface
the flow. pressure increases with decreasing gas-flow rate. In between,
there is a minimum in the pressure difference. The bottomhole
pressure as a function of the gas-flow rate in the well tubing is
given by the tubing-perform- ance curve (TPC), which is also
Introduction shown in Fig. 1. Production can only occur when the reservoir
In gas wells, both gas and some liquid are produced (the liquid curve and the TPC intersect. Because operation in the liquid-
in the form of water and gas condensate). When the reservoir loading regime (at gas-flow rates below the minimum in the
pres- sure is high, resulting in a high gas-production rate, the TPC) is unstable, one should maintain operation at gas-flow
gas veloc- ity in the production tubing is large enough to drag rates above this minimum.
the liquids to the surface. The flow regime in the pipe is When the surfactants are injected at the bottom of the well,
annular-dispersed; the liquid is contained in a liquid film at the the liquid will start to foam, thereby changing the
wall of the tubing and in entrained droplets in the gas core. hydrodynamics of the flow in the production tubing. This leads
When the reservoir pressure declines over field life, the gas to a change in the TPC, which creates stable operating points
velocity in the tubing decreases and the gas can no longer drag at lower reservoir pressure. To investigate this change in
the liquid along to the surface. Subse- quently, the liquid hydrodynamics, we previ- ously performed laboratory
accumulates at the bottom of the well, causing an increased experiments with air/water flow with and without surfactants
hydrostatic pressure gradient along the well, and thus an in a vertical pipe at atmospheric pressure and at room
increased backpressure on the reservoir. This can severely temperature (Van Nimwegen et al. 2015a, b). The results show
limit or even completely stop the gas production (Lea et al. that surfactants decrease the pressure gradient at low gas-flow
2003). rates. The minimum in the TPC can even become a plateau,
From experience in the gas industry, it is known that the with an almost constant pressure gradient across a range of
injec- tion of surfactants at the bottom of the well can prevent gas-flow rates. The reduction of the pressure gradient by the
liquid loading (Campbell et al. 2001). One can find many surfactants at low gas-flow rates occurs because the flow
examples of successful deliquification of gas wells by use of becomes more regular, which causes a decrease in the
surfactants in the literature (Jelinek and Schramm 2005; hydrostatic pressure gradient.
Schinagl et al. 2007; Bremner et al. 2010). In a recent However, our previous work considers only one surfactant.
questionnaire among gas-well operators in The Netherlands To evaluate the importance of surfactant selection, in this work
(Yavuz 2011), it was determined that the use of surfactants is we will investigate the effect of two different surfactants on
the most commonly used deliquifica- tion technique. One can the flow. The effect will be shown qualitatively with
also apply surfactants in shallow gas wells, although for these visualization using a high-speed camera and quantitatively
wells more deliquification techniques, including subsurface with measurements of the pressure gradient. The results will
pumps, are economically viable (Reinicke et al. 2007). provide insight into the impor- tance of surfactant selection for
Furthermore, for shallow wells, the contribution of gas-well deliquification.

Copyright V
C 2016 Society of Petroleum Engineers
Air/Water Flow in Vertical Pipes
In air/water pipe flow, the flow morphology is strongly
This paper (SPE 164095) was accepted for presentation at the SPE International
Symposium on Oilfield Chemistry, The Woodlands, Texas, USA, 8–10 April 2013, and
dependent on the gas- and liquid-flow rates. The flow
revised for publication. Original manuscript received for review 17 September 2013. Revised morphology has been
manuscript received for review 3 April 2015. Paper peer approved 9 June 2015.

488 April 2016 SPE


Journal
cial velocities are equal to the volumetric flow rate divided by
the pipe cross-section area. For single-phase flow, the pressure
gradi- ent increases with increasing gas flow rate. For air-water
flow, at large gas flow rates, in the annular flow regime, the
Tubing performance pressure gra- dient increases with increasing usg. At low usg, in
Bottomhole Pressure curve (TPC) the churn flow regime, the pressure gradient increases with
decreasing gas flow rate. The minimum in the pressure
gradient, between these two regimes, is closely related to the
Stable operating point transition between churn and an- nular flow, which is also
indicated in Fig. 3.
One can understand the behavior of the pressure gradient
from a balance of forces on the gas core along a length dx,
which yields
Unstable @p
Reservoir —ag — agqgg — Fi ¼ 0: .............................................. ð1Þ
operating points @x
curve
Reservoir curve
In this equation, a indicates the holdup, q the density, and p
(declined pressure) the pressure. Furthermore, g is the gravitational acceleration,
Gas Flow Rate and the subscripts g and l indicate the gas and the liquid phase,
respec- tively. Fi is the interfacial friction per unit volume
exerted by the liquid phase on the gas, and can be rewritten in
Fig. 1—Schematic showing the TPC and the inflow performance analogy with gas flow in a pipe with a rough wall:
curve.
classified into different flow patterns, for horizontal (Taitel 4ðD — 2dÞ2 q gðug — uiÞ
2
4 qgu2 g
and Duckler 1976), vertical (Taitel et al. 1980), and inclined Fi ¼ fi = D fi 2 : . . . . . . . ð2Þ
flow (Barnea 1987). In this work, we focus on vertical flow, D2 2
for which
the four commonly defined flow patterns are presented in Fig. In this equation, ug is the actual velocity of the gas, ui is the
2. velocity at the gas-liquid interface, D is the pipe diameter, d is
At large gas flow rates, there is an annular flow, in which the thickness of the liquid film at the wall, and fi is the friction
gas is the continuous phase, and liquid is present in a film at factor of the interface. Note that we neglected the forces from
the wall, and in droplets in the gas core. In the annular flow entrained droplets compared with the forces from the liquid
regime, larger roll waves and small ripple waves are present at film at the wall. Because ug is usually much larger than ui, the
the liquid inter- face. As the gas flow rate decreases, the gas is latter can often be neglected. Furthermore, d is usually much
unable to drag the liquid upward continuously, which marks smaller than D, such that it can also be neglected. At large gas
the transition to the churn flow regime. In the churn flow flow rates, in the annular flow regime, the flow morphology is
regime, the liquid film flows downward intermittently, and the relatively regular, and fi is relatively small; the pressure
morphology of the flow becomes more complex, as waves on gradient is mostly proportional to u2. In the churn flow
the liquid film grow in size and large liquid ligaments are regime, the flow morphology becomes increasingly irregular
entrained by the gas. Large, aerated waves, called flooding asg usg decreases, leading to a large increase in fi and in the
waves, appear that cause the upward trans- port of the liquid pressure gradient with decreasing usg.
(Hewitt and Jayanti 1993). Liquid loading is related to the We can also perform a balance of forces over the entire
transition between annular flow and churn flow. pipe cross section along a section of length dx, assuming that
At even lower gas flow rates, the liquid becomes the the entire wall of the pipe is covered with the liquid film (i.e.,
continu- ous phase, resulting in a slug flow, in which liquid there is no contact between the gas and the pipe wall) from
slugs are alter- nated with long Taylor bubbles. At even lower which we obtain
liquid flow rates, the gas is contained in small bubbles rising @p
upward in the liquid phase; this is the bubbly flow regime. — — alqlg — agqgg — Fl ¼ 0:......................................ð3Þ
The morphology of the flow has a large effect on the @x
pressure gradient. The pressure gradient for both single-phase In Eq. 3, Fl is the frictional force between the liquid and the
air flow and for air-water flow in a 50-mm diameter vertical wall per unit volume. Eq. 3 shows that the total pressure
pipe is shown in Fig. 3 as a function of the superficial gas gradient consists of a hydrostatic component al qlg and a
velocity (usg). Superfi- wall frictional

Bubbly Flow Slug Flow Churn Flow Annular Flow

Fig. 2—Overview of the different flow patterns occurring in vertical gas/liquid pipe flow. The gas-flow rate increases from left to
right.

April 2016 SPE 489


Journal
are aggregates of surfactants in which the hydrophilic headgroups
Churn flow Annular flow Single-phase air flow shelter the hydrophobic tails.
1200
Air/water flow
As shown in the Turner criterion, in which the velocity of
Pressure Drop (Pa/m) the onset of flow reversal only scales with the one-fourth
1000 power of the surface tension (c1=4), this reduction of the
Pressure-drop minimum surface tension is not sufficient to explain the difference in the
800 behavior of the flow as a result of the addition of surfactants.
Furthermore, because the largest structures at the air/water
interface are inertial waves and not capillary waves, a change
600 in surface tension is not expected to cause large changes in the
interfacial morphology.
400
The surface tension of a surfactant solution is not merely a
static value, as is the case for pure liquids. When a new
gas/liquid interface is created, the surface tension is equal to
200 that of water. Because surfactants have a hydrophobic and a
hydrophilic part, they will preferentially adsorb at the air/liquid
interface. There- fore, as time progresses, more surfactants
5 10 15 20 25 30 35 40 diffuse to the surface, in which they are adsorbed at the air-
liquid interface. As a conse-
usg (m/s)
quence, the surface tension decreases until equilibrium is reached
Fig. 3—The measured TPC for single-phase air flow and for air/ between the interface and the bulk liquid. This development of
water flow without surfactants; the transition between annular the surface tension in time is called the dynamic surface tension.
flow and churn flow is indicated by the dashed line. These dynamic effects at the interface also lead to a surface
elasticity (Miller et al. 2001). When a gas/liquid interface of a
sur- factant solution at equilibrium with the bulk solution is
component Fl. From this perspective, in the annular flow expanded, the surfactant density at the interface decreases,
regime, the pressure gradient increases with increasing usg as a leading to an increase of the surface tension. Over time,
result of the increasing wall friction, because the liquid at the because of the adsorp- tion of surfactants at the interface, the
wall moves upward faster. In the churn flow regime, it is the surface tension returns to its original value. These processes at
hydrostatic com- ponent of the pressure gradient that increases the interface lead to a sur- face dilational elasticity and a
sharply with decreasing usg. surface dilational viscosity. In mea- surement of these surface
Because of its relevance to liquid loading, the annular-flow- rheological properties, the area of the interface is varied
to- churn-flow transition was studied extensively. The Turner sinusoidally with a frequency x, and the sur- face tension is
crite- rion (Turner et al. 1969), often used to predict the measured in time. The (complex) surface dilational modulus is
loading point of gas wells, determines the critical gas velocity given by
of the onset of liq- uid loading uc from the reversal of the
largest droplets in the flow: dc
" #1=4 E¼ : ............................. ð6Þ
40ðql — qg Þcg . . . . . . . . . . . . . . . . . . . . ð4Þ
d ln A
uc ¼ :
qgCD Because of the dynamic behavior of the interface, in general,
the surface tension c(t) and the interfacial area A(t) are out of
Here, ql and qg are the liquid and gas density, respectively; c is phase, leading to complex values of E. Furthermore, E depends
the surface tension; g is the gravitational acceleration; and CD on x: In the low-frequency limit, the variation of the surface
is the drag coefficient of the droplets, assumed to be equal to area is slower than the diffusion of surfactant molecules from
0.44 for these large droplets. The largest droplet size is the bulk to the surface, and the surface tension is
assumed to cor- respond to a Weber number of 30: approximately constant. At large x, there is no time for the
surfactants to diffuse to the surface within an oscillation
q gug2d period, and E will be largest. In this last case, the behavior of
We ¼ ¼ 30: ..........................................................ð5Þ the interface is fully elastic, and E will be real. In general, the
c modulus of E represents the dilational elas- ticity. The
dilational viscosity is related to the phase difference between
For our setup, the Turner criterion predicts the onset of the interfacial area and the surface-tension oscillations. As
liquid loading at a gas velocity of 15 m/s, when the largest shown by Myrvold and Hansen (1998), the dilational elasticity
droplets in the flow are 8 mm in diameter. This velocity is given by
corresponds to the transition between annular and churn flow c
in our setup. However,
Van’t Westende (2007) found from experiments in the same jEj ¼ a
; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ð7Þ
flow loop, as used in the present study, that such large droplets Aa=Ao
do not
occur in the gas core, and furthermore, that it is the liquid film
that starts to reverse at the transition to churn flow. However, where ca and Aa are the surface tension and interfacial-area
new results show that near the gas/liquid interface much-larger ampli- tudes, respectively, and Ao is the average jinterfacial

drop- lets exist. The interplay between these larger droplets and area. At zero concentration (for pure water), the surface tension
the liq- uid film is essential to understanding the onset of liquid is always the same regardless of changes in the interfacial jarea,
j
loading (Khosla et al. 2013). To better understand the and E 0. At large surfactant concentrations, diffusion is very
influence of the sur- factant on the interfacial morphology of quick, and E is low. In between, there is a maximum in the
the liquid film, in this work, high-speed movies of the flow are elasticity, in which there are sufficient surfactants to obtain
made. changes in surface ten- sion, but diffusion is not yet so fast that
deviations from the equi- librium surface tension are quickly
restored to the equilibrium. A model of this behavior was
Surfactants and Foaming developed by Lucassen and van den Tempel (1972).
Surfactants are molecules with a hydrophobic and a The most important effect of these surfactants is that the
hydrophilic part, and therefore they preferentially adsorb at the dynamic effects at the interface outlined in the preceding allow
air/water inter- face. In the process, they decrease the surface the creation of a stable foam, thereby significantly reducing the
tension of the water, usually by up to a factor of 2 to 3 effective density of the liquid phase. The generation of foam
(DeGennes et al. 2004). The equilibrium surface tension requires the agitation of the surfactant solution, such that air
decreases with increasing surfactant concentration, until the bub- bles are entrained within the liquid. In the literature,
critical micelle concentration is reached. Above this several meth- ods of agitation are known. In laboratory tests in
concentration, the surface and bulk concentrations of single the gas industry, often a foam column is used, in which gas
surfactant molecules (monomers) no longer increase. Instead, is sparged through a
additional surfactants that are added form micelles, which

490 April 2016 SPE


Journal
Water to
tank/drain Air

30D

30D

10D
P

40D
P

120D
Annular
water inlet

Water from
tank/tap 20D

Air from
compressor

Fig. 4—A photograph of the setup (left) and a schematic of the setup (right) showing the location of the pressure sensors and the
high-speed camera.

frit at the bottom of the column. The foam carries liquid to the influence of both the surfactant type and the surfactant
top of the column, and the amount of liquid that can be concentra- tion on the hydrodynamics of the air/water flow.
transported in this way is assumed to be a measure of the
unloading efficiency of the surfactant (Nguyen 2009).
Another way of creating foam is through the Ross-Miles Experimental Setup
test. Here, a volume of surfactant solution is sprayed through a In this section, the flow loop and the experiments used for the
nozzle on top of another volume of the solution. The amount sur- factant characterization are described.
of foam generated in this process is recorded (ASTM 2007). In
our setup, the foam is agitated by the hydrodynamics of the Flow Loop. The experimental setup consists of a 12-m-long
flow. The agita- tion causes air to be entrained into the liquid ver- tical pipe with a 50-mm internal diameter that is operated
phase through the formation of bubbles, droplets, and at ambi- ent conditions. The pipe is made of Perspex, making it
ligaments. Because the agita- tion is caused by the interactions transparent and allowing visualization of the flow with a high-
between the air and the water, it is dependent on Fi, the speed camera placed outside the setup. A photograph and a
interfacial friction. As shown in Eq. 1, the interfacial friction is schematic of the setup are presented in Fig. 4.
proportional to the pressure gradient (there is a factor ag The air flow is regulated through a mass-flow controller
difference). Because in gas wells, the same flow pat- terns (M1W D6383) with a range of 0 to 5,000 L/min and an
occur, as in our experiments (Lea et al. 2003), we expect Fi, accuracy of 2% of the measured value. The air is introduced at
and therefore the agitation of the liquid phase, to be similar in the bottom of the pipe, and water is injected through an
actual gas wells and in our experiments. annulus 1 m from the bottom. The water flow rate is regulated
The effectiveness of a surfactant as a foaming agent is through a valve (Badger RC 200 with a size of 1/4 in.). With a
depend- ent on the properties of the interfaces of the surfactant proportional-integral-drive controller, the valve is connected to
solution, such as the decrease of the equilibrium surface two turbine flowmeters (Equ- flow PFA), with ranges of 0.2 to
tension, the diffu- sion characteristic of the surfactants or the 2 L/min and 2 to 20 L/min and an accuracy of 2% of the
dynamic surface tension (Rosen and Hua 1988; Rosen et al. measured value.
1991), the elastic properties of the interfaces caused by the In the experiments without surfactants, the injected water
surfactants, and the effect of the sur- factants on the disjoining comes directly from the tap. After the water passes through the
pressure (Myers 2005). In this work, the equilibrium surface setup, it is disposed of through the drain. When experiments
tension, the dynamic surface tension, and the surface elasticity with surfactants are performed, first, a 50-L surfactant solution
of the surfactant solutions are measured. is pre- pared in a tank placed near the setup. Subsequently, the
From previous measurements in our flow loop, we found water is pumped from the tank, through the flow control, and
that the size of the structures in the morphology of the liquid into the setup. After passing through the setup, the water flows
film and the amount of foam created increase with decreasing back to the tank (i.e., the surfactant solution is recirculated).
gas flow rate and increasing liquid flow rate. The foam is able Note that the water is not foaming when it is injected into the
to suppress the largest structures of the flow (i.e., the roll setup; all foam shown in the visualization results is created
waves in annular flow and the flooding waves in churn flow). through the hydrodynamics of the multiphase flow inside the
It is also able to decrease the entrainment of droplets from the pipe. At the end of the pipe, a hose with a large (160 mm)
liquid film. Effectively, the foam is able to shift the transition diameter is attached, preventing the foam from flowing out
between annular and churn flow to lower gas velocities. from the top of the setup along with the air. At the end of this
Correspondingly, the pressure gradient at low gas-flow rates is hose, the air is released into the atmosphere.
decreased (Van Nimwegen et al. 2015a, b). Three pressure sensors (type UNIK 5000 by General
To verify how this behavior is affected by the surfactant Electric with premium accuracy, a range of 70-mbar gauge,
that is used, in this work, experiments with two different and a fre- quency response of approximately 1 kHz) are placed
surfactants are performed. For each surfactant, different along the
concentrations are con- sidered. The goal of the research is to
observe and quantify the

April 2016 SPE 491


Journal
1000 W Lamps

Camera (1000 fps)


Transparent sheet

Fig. 5—Setup used in the recording of the high-speed movies.

flow loop, at 6, 8, and 10 m from the water injection. The


results for the pressure gradient presented in this paper are Fig. 6—Maximum-bubble-pressure tensiometer used to mea-
from the pres- sure sensors at 8 and 10 m from the water sure the dynamic surface tension.
injection. The third sen- sor is used to determine the flow
development. Results show that the air/water flow is developed
at 6 m from the water injection, whereas flow with surfactants used. This is the most-studied surfactant in literature (Myers
takes a longer distance to develop at higher gas velocities. 2005), and it is chosen in the present study because it has a
When the gas velocity is below 20 m/s, which is the range in differ- ent dynamic surface tension than the commercial
which liquid loading occurs and in which we are most foamer, as is shown in the next section. Note that, for the
interested, the flow with surfactants is essentially devel- oped dynamic and equilib- rium surface-tension experiments for the
6 m from the water injection (Van Nimwegen et al. 2015a). commercial foamer and for SDS, solutions were made using
Visualization of the flow is performed with an Olympus I- demineralized water. In the flow-loop experiments for the
Speed 2 camera, capable of making movies at 1,000 frames/sec commercial foamer, solutions with tap water are made,
at its max- imum resolution of 800 600. The camera is placed whereas for SDS, solutions with demineral- ized water are
×
at a distance of approximately 1 m from the pipe, 8.5 m above considered, because SDS is incompatible with the water
the water injec- tion. Two 1,000-W lamps are used to chemistry of tap water. More information on this is pre- sented
illuminate the pipe from the back. A diffusive sheet is placed in a separate section.
between the lamps and the pipe to avoid glare. A schematic of To measure the equilibrium surface tension of the
this setup is presented in Fig. 5. surfactants, a du Noüy ring tensiometer is used. This
tensiometer determines the force that the liquid exerts on a
Surfactant Characterization. In this work, two surfactants are platinum ring as it is pulled upward through the air/liquid
used. The first is a proprietary surfactant product (Trifoam 820 interface. Because the area of the liquid interface increases
Block; Oilchem GmbH, Dessau-Roßlau, Germany), which will when it is pulled upward by the ring and it takes time for
be referred to as the “commercial foamer” in the remainder of surfactants to migrate to the surface, one should pull the ring
this pa- per. This product has been used successfully to upward very slowly to ensure that dynamic effects do not
deliquify gas wells in the field and was selected as a reference affect the measurements. The technique is described in more
case, showing the behavior of a surfactant with the correct detail by Zuidema and Waters (1941). For the measurements
properties to deliquify a gas well. Besides a surfactant, the with the commercial foamer, a Krüss educational
main additives in this product are a corrosion inhibitor and a tensiometer with an accuracy of 63 mN/m was used. For
freezing-point depressurizer. Concentra- tions of the SDS, a Krüss K9 tensiometer was used with an accuracy of 61
commercial foamer always indicate the concentration of the mN/m.
total product, and not just of its surfactant components. The dynamic surface tension (DST) is measured with a
As a second surfactant, SDS (sodium dodecyl sulfate, Sinter- face BPA-1S maximum-bubble-pressure tensiometer,
BioUl- tra, 20% solution in H 2O; Sigma-Aldrich, St. Louis, shown in Fig. 6. This device creates bubbles by blowing air
USA) is though a capil- lary. The surface tension at the bubble
interface is obtained

75 75
70 70
Surface Tension (mN/m)

Surface Tension (mN/m)

65 65

60 60

55 55

50 50

45 45

40 40

35 35

30 30
101 102 103 101 102 103 104
Concentration (ppm) Concentration (ppm)

Fig. 7—Equilibrium surface tension of the commercial foamer (left) and SDS (right), measured with a du Noüy ring tensiometer.

492 April 2016 SPE


Journal
75 75

70 70

65 65

60 60
 (mN/m)

 (mN/m)
55 55

500 ppm 300 ppm


50 50
1000 ppm 590 ppm
2000 ppm 890 ppm
45 45
3000 ppm 1180 ppm

40 40
10−3 10−2 10−1 100 101 102 10−3 10−2 10−1 100 101 102
t (s) t (s)

Fig. 8—DST of the commercial foamer (left) and SDS (right) for different concentrations.

through measurement of the Laplace pressure, when the bubble not reached for either of the surfactants within the
is a hemisphere with a radius equal to that of the capillary. measurement time. But, the most-striking aspect is that at the
Varying the gas-flow rate through the capillary changes the shortest lifetime considered here, 1 millisecond, the surface
lifetime of the interface: At higher flow rates, it takes less time tension of the SDS so- lution is already significantly lower
before the bubble is spherical. In this way, one can measure the than that of water. For the commercial foamer, this is not the
development of the surface tension in time. In this work, case. This difference in behav- ior between the two surfactants
interface lifetimes between 1 millisecond and 10 seconds are is the reason for the selection of SDS as a second surfactant.
considered. This technique is explained in more detail by Fig. 9 shows the measurements of the surface elasticity of
Fainerman et al. (2004). the commercial foamer and SDS. The measurement range in
The surface elasticity is measured with a Sinterface PAT-1 the flow- loop experiments is indicated in the figures by the
drop-profile tensiometer. In this device, a droplet with a dashed lines. From the left figure for the commercial foamer,
volume of 12 lm is created at the end of a capillary. one can observe that all flow-loop measurements in this work
Subsequently, a pho- tograph of the droplet is made every half are performed at concentrations above the maximum in the
a second, and the surface tension of the air-liquid interface is surface elasticity; for increasing concentration, the surface
determined from the shape of the droplet. After the surface elasticity decreases. The val- ues of the elasticity are larger
tension has reached its equilibrium value, the volume of the than for SDS, and the frequency de- pendence is also larger.
droplet is oscillated with an amplitude of 7% of the droplet For SDS, the measurement range in the flow loop is mostly at
volume, at frequencies between 10 mHz and 50 mhz. Note that concentrations below the maximum in the surface elasticity:
the frequency of the oscillation should be signifi- cantly lower For increasing concentration, the surface elas- ticity increases.
than the frequency at which the surface tension is measured to Furthermore, the value of the surface elasticity is lower for
accurately evaluate the development of the surface tension. SDS than for the commercial foamer. This might be related to
From the amplitude of the surface-tension variation, finally, the lower DST for SDS at low timescales; deviations of the
the surface elasticity is determined, according to Eq. 7. A surface tension from equilibrium are quickly decreased. Table 1
general overview of this technique is presented in Loglio et al. gives an overview of the different surfactant concentrations in
(2001) and in Ravera et al. (2010). the experiments.
Results of the Surfactant Characterization. Fig. 7 shows the
equilibrium surface tension of the commercial foamer and the Pressure-Gradient Measurements
equilibrium surface tension of SDS as a function of
concentration (given in ppm). For the commercial foamer, the In this section, results of the pressure-gradient measurements
equilibrium sur- face tension decreases to approximately 35 are given for the surfactants and concentrations summarized in
mN/m when the con- centration is increased. The critical Table
micelle concentration (CMC), above which the surface tension 1. Each tubing-performance curve (TPC) was measured by
no longer decreases with increas- ing concentration, is decreasing the superficial gas velocity from a maximum value
approximately 500 ppm. For SDS, a similar decrease in surface of approximately 38 m/s until 1–7 m/s, depending on the
tension is obtained. However, the CMC is sig- nificantly higher measure- ment. Unlike for gas wells, in which the bottomhole
and is equal to approximately 1700 ppm. Note that there is a pressure is measured, in the TPCs for the experimental setup,
dip in the surface tension of SDS near the CMC. This is caused the average pressure gradient is measured as a function of the
by impurities of the SDS, mostly consisting of dodecanol, superficial gas velocity. During these measurements, the
which can have a significant effect on the surface ac- tivity superficial liquid veloc- ity is fixed at usl 10 mm/s unless
(Fainerman et al. 2010). indicated otherwise. First, the effect of the surfactant
The results for the DST of the two surfactants are shown in concentration is¼ investigated. Subse- quently, TPCs were
Fig. 8. For the commercial foamer, the equilibrium surface ten- obtained at a constant concentration for three liquid-flow rates.
sion no longer changes for concentrations above 500 ppm.
How- ever, the figure shows that the DST still changes Effect of Concentration. Commercial Foamer. In the left
significantly above this concentration. At 3,000 ppm, the graph of Fig. 10, the measured TPCs are given for the
surface tension at a surface lifetime of 10 seconds is more than commercial foamer at different concentrations. The results
20 mN/m lower than at the CMC. Note that, at a surface show that the sur- factant increases the average pressure
lifetime of 10 seconds, the equilibrium surface tension is not gradient at high gas-flow rates. This increase becomes larger
yet reached for the concentra- tions shown here. If the when the surfactant concentra- tion is increased. Fig. 11 shows
measurements had been continued, for each of the that at these high surfactant con- centrations, not only the
concentrations, a surface tension of 35 mN/m would eventually average pressure gradient, but also the root-mean-square (rms)
have been reached. value of the pressure-gradient fluctuations is increased. As
For SDS, several concentrations below the CMC are consid- stated in earlier work (Van Nimwegen et al. 2015b), this is
ered. The concentrations are chosen on the basis of the because of the increase of the interfacial friction caused by the
behavior in the flow loop, discussed in the next sections. The foam forming on the interface of the liquid film.
results at the highest concentration show that the equilibrium
surface tension is

April 2016 SPE 493


Journal
25 25
10 mHz 10 mHz
50 mHz 50 mHz

Surface Elasticity (mN/m)


Surface Elasticity (mN/m)
20 20

15 15

10 10

5 5

0 0
500 1000 2000 4000 100 1000 10000
Concentration (ppm) Concentration (ppm)

Fig. 9—Surface elasticity of the commercial foamer (left) and of SDS (right) as a function of concentration for two different frequen-
cies of the interfacial-area oscillation. The dashed lines indicate the measurement range in the flow-loop measurements. The
measurements for SDS are taken from Kawale (2012).

Commercial Foamer SDS


Concentration (ppm) Concentration (× CMC) Concentration (ppm) Concentration (× CMC)
500 1 200 0.11
1,000 2 390 0.22
2,000 4 590 0.33
3,000 6 790 0.44
Table 1—Overview of the surfactant concentrations used in the large-scale flow-loop experiments.

1600 1600
air/water
1400 1400 200 ppm
390 ppm
1200 1200 590 ppm
890 ppm
1000 1000
dp/dx (Pa/m)

dp/dx (Pa/m)

800 800

600 600

400 air/water 400


500 ppm
1000 ppm
200 200
2000 ppm
3000 ppm
0 0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
usg (m/s) usg (m/s)

Fig. 10—TPCs (with the average pressure gradient on the vertical axis) for the commercial foamer (left) and SDS (right) at different
concentrations. For comparison, the TPC for tap water without surfactants is also given. In all measurements, usl ¼ 10 mm/s.

400 400
air/water
rms Pressure−Gradient Fluctuations (Pa/m)

rms Pressure−Gradient Fluctuations (Pa/m)

500 ppm
350 350 air/water
1000 ppm
2000 ppm 200 ppm
300 3000 ppm 300 390 ppm
590 ppm
250 250 890 ppm

200 200

150 150

100 100

50 50

0 0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
usg (m/s) usg (m/s)

Fig. 11—The rms pressure fluctuations of the commercial foamer (left) and SDS (right) as a function of the gas-flow rate, for differ-
ent concentrations at usl ¼ 10 mm/s.

494 April 2016 SPE


Journal
The most-interesting effect of the surfactant occurs at low pressure gradient of water (approximately 10,000 Pa/m) at
gas flow rates. At the critical micelle concentration (CMC) usg ¼
0 m/s.
(500 ppm), the minimum of the pressure-gradient shifts In the right graph in Fig. 11, the rms of the pressure-
slightly toward lower superficial gas velocities, whereas the gradient fluctuations in the SDS measurements is given. As
pressure gradient at this minimum is slightly increased. More was also the case for the commercial foamer, the pressure-
importantly, at usg < 15 m/s, the pressure gradient is decreased gradient fluctuations increase for high gas-flow rates when the
by the surfactants. The effect of the surfactant becomes surfactant is added. At low gas-flow rates, the surfactants
stronger at a concentration of 1,000 ppm, in which the pressure reduce the pressure-gradient fluctuations; the reduction is
gradient is essentially constant for superficial gas velocities largest at a concentration of 390 ppm. This is the
between 7 and 21 m/s. In a gas well, such behavior would concentration at which there is a constant pressure gradient for
make it possible to produce at much-lower gas flow rates. The a large range of usg. Such behavior was also observed for the
decrease in the total pressure gradient is solely caused by a commercial foamer, in which a concentration of 1,000 ppm led
decrease in hydrostatic head, whereas the frictional pressure to both a constant pressure gradient for a large range of usg and
gradi- ent always increases when surfactants are added (Van the lowest pressure-gradient fluctuations. Note that for the
Nimwegen et al. 2015b). highest concentrations of SDS, 590 and 890 ppm, the pressure-
When the surfactant concentration is increased above gradient fluctuations are relatively high at the minimum of the
1,000 ppm, the pressure gradient between usg¼ 7 and 21 m/s pressure gradient (usg 2.2 and 1.1 m/s, respectively), whereas
increases with increasing surfactant concentration. Note that a for the other concentrations, the pressure-gradient fluctu-
local minimum in the pressure gradient forms at ump ations are low at the minimum.¼
¼ 21 m/s,At
which the pressure gradient for air/water flow is minimum.
in
Overall, the results for SDS and the commercial foamer are
gas flow rates below ump, the transition to churn flow leads to very similar: Only the surfactant concentration at which these
an increased agitation of the liquid phase, and therefore to results occur differs between the surfactants. To model the
increased formation of foam. The additional foam increases fi flow, we would like to know the effect of the surfactants at a
and therefore the total pressure gradient. At even lower certain concentration. However, the results for Trifoam and
concentrations, the lower density of the foam allows the foam SDS show that (1) the absolute concentration is not a good
to be dragged upward more easily than water, keeping the flow predictor of the effect of the surfactants in the flow loop and
very regular compared with air/water flow. Therefore, fi does (2) the concentration relative to the CMC is also not a good
not increase that much at lower usg, and the pressure gradient predictor of the effect of the surfactants. Furthermore, the
decreases again. Note that at a surfactant concentration of equilibrium surface tension at the lower concentrations for
3,000 ppm, the pressure gradient still decreases at superficial SDS is lower than for the commercial foamer, whereas results
gas velocities below 6.5 m/s, indicating a minimum at an even in the large-scale setup are similar, indi- cating that (3) the
lower gas flow rate. Fig. 11 shows that the rms fluctuations of equilibrium surface tension is also not a good predictor of the
the pressure gradient at low gas flow rates, caused by the effect of the surfactants in the flow loop. Further- more, for
churning of the liquid, are also decreased signifi- cantly after SDS, the surface elasticity increases with increasing
the injection of surfactants. However, above a concen- tration concentration in the measurement range, whereas, for the com-
of 1000 ppm, the pressure-gradient fluctuations are not mercial foamer, the surface elasticity decreases with increasing
suppressed further. concentration; therefore, (4) the surface elasticity is not a good
Sodium Dodecyl Sulfate (SDS). In the right graph in Fig. 10, predictor of the effect of the surfactant either. Furthermore, the
the TPCs for SDS are shown for the concentrations listed in behavior of the dynamic surface tension for SDS is different
from that for the commercial foamer, with a faster reduction at
Table short timescales and a smaller reduction at long timescales for
1. Qualitatively, the results for SDS and the commercial SDS. Therefore, (5) the dynamic surface tension also does not
foamer are very similar. Both surfactants increase the pressure seem to be a good predictor of the effect of the surfactants in
gradient at high gas-flow rates, whereas at low gas-flow rates, the setup at low usg.
the pressure gradient is decreased. For both surfactants, in a
range of concen- trations, the pressure gradient is almost
constant for usg 7 to 20 m/s. Note, however, ¼ that a Effect of Liquid-Flow Rate. In Fig. 12, the effect of the liquid
significantly lower concentration of SDS is required (390 ppm, flow rate on the pressure gradient is illustrated. The behavior
or 22% of the CMC) to reach this con- stant pressure gradient of the two surfactants is quite similar at large gas velocities. At
than for the commercial foamer (1,000 ppm, or 200% of the large usg, the pressure gradient increases with increasing usl. In
CMC). This shows that both the abso- lute concentration and the an- nular flow regime in air-water flow, similar to for the
the concentration relative to the CMC are bad predictors of the flow with surfactants at high usg, the pressure gradient is
effect of the surfactant on the flow. dependent on the liquid flow rate. A larger usl leads to a thicker
There is one major difference between the results for SDS liquid film with larger waves, which increases fi, leading to an
and the results for the commercial foamer: At large gas-flow increase in the pres- sure gradient. In the churn flow regime,
rates, the pressure gradient for SDS is much larger than for the the air-water flow is irreg- ular regardless of the liquid flow
commercial foamer. This indicates a rougher interface of the rate, and the pressure gradient is independent of the liquid flow
film, which could be caused by increased foam formation. This rate (Van Nimwegen et al. 2012).
could be related to the lower values of the DST at short time For the flow with surfactants, the average pressure gradient
scales for SDS. At large usg, entrainment of air might only is strongly dependent on the liquid flow rate even for low usg,
occur for very short times, requiring fast diffusion of surfactant indi- cating that the surfactants shift the transition between
to the bubble surface to gen- erate foam. churn and annular flow to lower usg. Because at low usg, the
Because, for SDS, measurements were made up to lower usg pressure gradi- ent is independent of usl for air-water flow, but
than for the commercial foamer, the results more clearly show not for flow with surfactants, the surfactants are more effective
that the minimum in the pressure gradient shifts to lower usg at reducing the pres- sure gradient at low usl. The results for the
for increasing concentrations. At the largest concentration commercial foamer also show that the pressure gradient
of 890 ppm, the pressure gradient is lowest at usg 1 m/s, and increases for low usg at usl 40 mm/s, indicating the
with a value of 350 Pa/m is lower than for = air/water flow at surfactants are less able to shift the churn-annular transition at
ump. This shows that the minimum in the pressure gradient is high¼ usl.
shifted from usg 21 to 1.1 m/s, a decrease of approximately a
factor 20. Also for the other concentrations, the pressure SDS Measurements in Tap Water. Initially, we performed all
¼
gradient is lower at usg 1.1 m/s than at usg 2.1 m/s. In the next measurements with SDS in tap-water solutions. However,
section, the flow visualization is used to give an explanation of when measuring the same TPC three times with the same
this¼phenomenon. Note that¼ all TPCs that are considered in this solution, the reproducibility at low usg was poor, as shown in
work, even though the pressure gradient differs a great deal at Fig. 13. The most noteworthy about these results is that the
the lowest gas flow rates considered here, will all converge surfactants become more effective as the surfactant solution
to the hydrostatic ages. After studying the

April 2016 SPE 495


Journal
1600 1600
usl = 2 mm/s
usl = 2 mm/s
1400 usl = 10 mm/s 1400
usl = 10 mm/s
usl = 40 mm/s
1200 1200 usl = 40 mm/s

1000 1000
dp/dx (Pa/m)

dp/dx (Pa/m)
800 800

600 600

400 400

200 200

0 0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
usg (m/s) usg (m/s)

Fig. 12—The TPCs at different superficial liquid velocities for the commercial foamer at a concentration of 1,000 ppm (left) and for
SDS at a concentration of 590 ppm (right).

SDS solution, it was found that SDS does not dissolve of the measurements is indicated by the error bars. The figure
properly in tap water, because the dodecylsulfate anion shows that the measurement reproducibility is not affected by the
precipitates with cati- ons present in the tap water. Because the surfac- tants. Note, however, that the surfactants degrade in time
pump increases the tem- perature of the solution when the because of the shear forces in the pump and in the setup, and by
setup is in operation, some of the precipitated surfactants action of bacte- ria. Therefore, each TPC is measured with a fresh
dissolve again, thereby increasing the concentration of solution.
dissolved surfactants. Although, qualitatively, the results are
not much affected by the water chemistry, the SDS
concentration required for a certain pressure-gradient reduction Flow Visualization. To better understand the behavior of the
at low usg is larger when tap water is used than when pressure gradient of the two surfactants, flow visualization was
demineralized water is used. Furthermore, with demineralized performed for air-water flow—for the commercial foamer at a
water, the surfac- tant solutions are better defined. Therefore, concentration of 1000 ppm and for SDS at concentrations of
demineralized water is used in all measurements for SDS. 390 and 790 ppm. Snapshots of the visualization are shown in
These results indicate that the water chemistry can have a large Fig. 15 for usg 21.5, 16.1, 10.7, and 6.4 m/s. The air/water flow
effect on the surfactant per- formance. In gas wells, in which is rele-¼vant not just as a reference for comparison: Because the
the amount of dissolved salts is much greater than for tap liquid does not foam when introduced into the setup, it is
water, the compatibility of the surfac- tant with the brine representative of the flow just after injection, in which the
should always be confirmed. When the surfac- tants are foam is formed.
compatible with the water chemistry, the results for the TPC In air/water flow at the largest considered superficial gas
are qualitatively similar. ve- locity of 21.5 m/s, approximately the minimum of the
pressure gradient, there is a cocurrent annular flow. The liquid
film moves upward continuously, and ripple waves and a roll
Measurement Reproducibility. To evaluate the reproducibility of wave are visible in the image. When decreasing the superficial
the measurements, the TPCs of air-water flow and of the gas velocity to
commercial foamer at a concentration of 500 ppm were measured 16.1 m/s, the liquid film becomes almost stagnant, whereas the
five times each (Van Nimwegen et al. 2015a). In Fig. 14, the waves on the gas/liquid interface grow. Below 15 m/s, in
standard deviation which the transition to churn flow occurs, the liquid film
starts to move

1600
1000
Measurement 1 1400
900
Measurement 2
800 Measurement 3 air/water
1200
500 ppm
700
1000
dp/dx (Pa/m)
dp/dx (Pa/m)

600
800
500

400 600

300 400
200
200
100
0
0 6 10 15 20 25 30 35 40
0 5 10 15 20 25 30 35 40 usg (m/s)
usg (m/s)
Fig. 14—Comparison of the reproducibility of TPS measure-
Fig. 13—Three subsequent measurements with the same SDS ments for air/water flow and for flow with 500 ppm of commer-
solution, showing bas reproducibility at low usg. The concen- cial surfactant. The error bars indicate the standard deviation of
tration of SDS is 590 ppm and usl ¼ 2 mm/s. the five performed measurements.

496 April 2016 SPE


Journal
water

commercial
foamer
1000 ppm

SDS
390 ppm

SDS
790 ppm

usg = 6.4 m/s usg = 10.7 m/s usg = 16.1 m/s usg = 21.5 m/s

Fig. 15—Overview of the flow visualization, showing air/water flow without surfactants, with 1,000 ppm of the commercial foamer
added and with 590 and 1,180 ppm SDS added. In all images, usl ¼ 10 mm/s.

downward intermittently. Large flooding waves appear, and flow rate further shows that the substrate starts to move
the amount of entrainment increases. Overall, the agitation of downward and the foam waves move in a more irregular way.
the liq- uid is much more violent. Similar to for the results of the pressure gradient, the
The results for the commercial foamer show that at visualiza- tion results for 1,000 ppm of the commercial
usg ¼
21.5 m/s, with the smallest waves in the air/water flow and foamer and 390 ppm of SDS are very similar. The largest
therefore the lowest agitation of the liquid, least foam forms. difference is that at usg 6.4 m/s the flow with 1,000 ppm of
The foam moves upward in waves of varying sizes—larger the ¼commercial foamer is already churning, whereas the flow
waves moving upward faster. The foam increases the effective with an SDS concentration of 390 ppm is still regular. The
interfacial roughness somewhat, leading to a small increase in pressure gradient for SDS does not increase until usg 4 m/s,
the pressure gradient. As the gas velocity is lowered and the which also indicates≤the flow should still be regular at usg 6.4
agitation of the liquid phase increases, more foam forms until m/s.
at 10.7 m/s an almost stagnant foam film is created, which we ¼
When the concentration of SDS is increased to 790 ppm,
call the substrate. Because at this usg the film no longer moves espe- cially at the two largest gas-flow rates in the figure, the
downward, this means that the foam has lowered the amount of foam increases. This larger amount of foam can lead
superficial gas velocity of the transition between annular and to a larger interfacial roughness, which leads to a larger fi and a
churn flow. Some foam waves move upward superposed on larger pres- sure gradient. At lower gas velocities (usg 10.7 and
the substrate. Decreasing the gas- 6.4 m/s), the results are more similar ¼ to the lower
concentration, although the

April 2016 SPE 497


Journal
SDS 390 ppm

SDS 790 ppm


usg = 2.1 m/s usg = 4.3 m/s usg = 2.1 m/s usg = 4.3 m/s

Fig. 16—Comparison of the flow visualization at two different SDS concentrations and at low gas-flow rates. Again, usl ¼ 10 mm/s.

foam appears more dense, probably because the foam film is circumference. No waves or other features are visible on the
thicker, and the bubble size appears to be somewhat smaller. foam film. The second and third images show the front of the
To investigate the transition to more-irregular flow for SDS, slug mov- ing upward. In the fourth image, the slug occupies
snapshots of the flow at superficial gas velocities below 6.4 the entire field of view, and the last image shows the back of
m/s are shown in Fig. 16. At the lowest concentration shown in the ¼slug. The slug flow also explains the large pressure
the figure, the morphology of the foam film does become fluctuations found at usg 1.1 m/s for an SDS concentration
irregular at usg 4.3 m/s and usg 2.1 m/s. The foam at the wall of 790 ppm. When there is a slug in between the pressure
¼ to move downward
starts ¼ faster between subsequent waves, sensors, the pressure gradient is much larger than when there is
and the upward-moving waves become larger and more no slug. Slug flow, though less regular, also occurs for the
irregular, causing the increase of the pressure oscillations. At a lower surfactant concentrations. This is the reason for the
surfactant concentra- tion of 790 ppm, only at usg 2.1 m/s decrease in the average
¼ pressure gradient
¼ when going from usg
some of the foam film starts to move downward. The foam 2.1 m/s to usg 1.1 m/s observed for all surfactant
film becomes thicker, block-¼ing more light, and the upward concentrations.
moving waves can only be seen as vague shadows. The rigidity
of the foam film indicates a high foam viscosity, which also Future Work
indicates a low liquid content (Weaire and Hutzler 1999); the The conditions in our laboratory setup are very different from
density of the liquid is decreased more here than in other those in actual wells: The diameter of the tubing, the pressure,
experiments, which, from Turner’s criterion, also leads to the the temperature, and the gas and liquid composition are
largest decrease in the gas-flow rate of the onset of liquid different in actual wells. Despite these differences, we think
loading. that the results shown in this work are qualitatively similar to
For the highest SDS concentration of 790 ppm, the those obtained in an actual gas well. These different
pressure conditions, however, will have an effect on the quantitative
gradient at a superficial gas velocity of 1.1 m/s was very low. properties of the flow in the tubing; for instance, a larger
To investigate the corresponding flow morphology, flow pressure leads to a larger density of the gas, which means that
visualization was performed at usg 1.1 m/s. It shows that the at the same gas velocity, the gas exerts more force on the
low pressure gradient¼is caused by a flow-pattern transition to liquid. The minimum of the tubing-performance curve would
slug flow. This means that the churn-flow pattern did not occur therefore move to lower superficial gas velocities as the
at all at this large concentration. We verified that such a pressure increases. With respect to the tubing size, we already
transition also occurs for the commercial foamer. An example have extended our current work by performing experiments in
of slug flow is presented in Fig. 17. The first image shows the pipes of different diameters. From these latter experimental
flow between slugs—a film of foam that is slowly moving results, we hope to make a mechanistic model of the flow,
downward along the wall. Compared with larger gas-flow taking into account the scaling of the diameter. We will also
rates, the bubble size has increased some- what, and the layer try to find proper scaling relations for the pressure and the
of foam is less thick. It moves downward very regularly, with temperature that can be implemented in the model.
the foam moving equally fast along the entire pipe

Fig. 17—Illustration of a slug moving upward in the flow with the commercial foamer at a concentration of 3,000 ppm. usg ¼ 1.1 m/s
and usl ¼ 10 mm/s.

498 April 2016 SPE


Journal
A complication of modeling the flow with surfactants com- at the wall, and slugs are gently moving upward. Further
pared with flow without surfactants is that it is unknown how research into this flow pattern should be performed to
much foam is formed from the liquid (i.e., what the determine at which gas- and liquid-flow rates it will occur.
foamability of the liquid is). As shown in the Results section, These results give more insight into the prevention of liquid
none of the properties of the surfactant solution that we have loading in gas wells by surfactants and also give initial insight
measured so far can predict the effect of the surfactant solution into the differences in the flow that are caused by choosing a
(i.e., the foamabil- ity) in the setup. Furthermore, it is known dif- ferent surfactant. These results can be used as a foundation
that the components of the brine can have a significant effect for a future mechanistic model of gas/liquid/foam flow.
on the critical micelle con- centration and on the solubility of
the surfactant (Corrin and Har- kins 1947; Zhou and Hao Nomenclature
2011), which would, in turn, have an effect on the foamability.
Furthermore, the hydrocarbons present in the liquid can have a A ¼ interfacial area in dilational-rheology measurements, m2
large effect on the foamability (Willis et al. 2008), and the Aa ¼ amplitude of the interfacial-area oscillation in
foamability is affected by the temperature and pressure. To dilational- rheology measurements, m2
model the flow, an appropriate experiment to evaluate this A0 ¼ average interfacial-area oscillation in dilational-
foamability should be designed. In the gas industry, already rheology measurements, m2
small-scale experiments are used to evaluate the foam- ability, CD ¼ drag coefficient, -
in which actual field fluids are used (Yang et al. 2007). d ¼ droplet diameter, m
In our future work, we hope to develop a small-scale
foamabil- ity test that can predict the amount of foam formed E ¼ surface-dilational modulus, mN/m
in our setup for the aqueous solutions that we use. The results fi ¼ interfacial-friction coefficient, -
3
of the foamabil- ity test will then be used as an input to our Fi ¼ gas/liquid interfacial force per unit volume, N/m
mechanistic model. When applying this model for actual wells, Fl ¼ force exerted by the wall on the liquid per unit volume,
the small-scale setup should be used at the actual field N/m3
conditions to find the proper input parameters to the flow
model. g ¼ gravitational acceleration, m/s2
p ¼ pressure, Pa
uc ¼ critical gas velocity, determined by the Turner
Conclusion criterion, m/s
From field experience in gas production, it is known that ug ¼ velocity of the gas, m/s
surfac- tants can prevent liquid loading in gas wells. To gain ui ¼ velocity of the interface, m/s
understand- ing of how these surfactants affect the flow in the ump ¼ superficial gas velocity corresponding to the minimum
production tubing, laboratory experiments with air/water flow in dp/dx in air/water flow, m/s
at atmospheric pressure and at room temperature were usg ¼ superficial gas velocity, m/s
performed. In earlier work, the influence of a commercial usl ¼ superficial liquid velocity,
foamer on this flow was investigated. In the current paper, m/s ¼We Weber
additional experiments were performed with a second number, -
surfactant, sodium dodecyl sulfate (SDS), to investigate the x ¼ axial coordinate, m
importance of the surfactant selection.
For both surfactants, the hydrodynamics of the flow lead to ag ¼ gas holdup, m3/m3
the formation of foam, which results in an increase of the al ¼ liquid holdup, m3/m3
volume and the viscosity of the liquid phase and a decrease in c ¼ surface tension, mN/m
the density of the liquid phase. At large gas-flow rates, the ca ¼ amplitude of the surface-tension oscillation in
increased volume leads to foam waves that increase the dilational- rheology measurements
interfacial friction factor, causing an increase of the pressure qg ¼ gas density, kg/m3
gradient. At low gas-flow rates, the lower density of the foam
allows it to be carried upward more easily by the air, which ql ¼ liquid density, kg/m3
leads to a more regular flow, shifts the transition between
annular and churn flow to lower superficial gas velocities, and
results in a significantly smaller hydrostatic pressure gradient. Acknowledgments
The minimum in the pressure gradient, in which the two effects The authors would like to thank Grace Curtis-Holsgrove at the
outlined previously are balanced, moves to lower gas flow Shell Technology Centre in Amsterdam for the measurements
rates as the surfactant concentration is increased. At moderate of the equilibrium surface tension of SDS, and of the dynamic
concentrations (1,000 ppm for the commercial foamer and 390 surface tension (DST) of both the commercial foamer and
ppm for SDS), the minimum becomes a plateau, in which the SDS. We would like to thank Durgesh Kawale for performing
pressure gradient is constant for a range of gas velocities. the surface-elasticity measurements for SDS. Furthermore, we
Quantitatively, results for the two surfactants are similar, would like to thank Johan Sjöblum, professor from the
although the optimum surfactant concentration for reducing the Ugelstadt laboratory at the Nor- wegian University of Science
pressure gradient at a certain usg is not the same for the two and Technology, for allowing us to perform surface-elasticity
sur- factants. Furthermore, the critical micelle concentration measurements in his laboratory. We gratefully acknowledge
(CMC), the equilibrium surface tension, the dynamic surface Sébastien Simon for his help in running the surface-elasticity
tension, and the surface elasticity do not seem to be good measurements. The project is funded by NAM, a Dutch
predictors of the dose rate required to decrease the pressure subsidiary of Shell and ExxonMobil. The authors would like to
gradient. The results for SDS and the commercial surfactant thank Gert de Vries, Kees Veeken, Ewout Biezen, and Ruud
can be similar, while each of these properties of the surfactant Trompert from NAM for the valuable discussions.
solutions is different. For instance, for SDS, more foam is
formed at a quarter of the CMC, than for the commercial
foamer at twice the CMC. Also, although the equilibrium
surface tension is lower for the commercial foamer than for References
SDS in every case, more foam is formed in the SDS ASTM. 2007. Standard test method for foaming properties of surface-
experiments. active agents. D1173 U 07. Pennsylvania, USA: ASTM.
Measurements of SDS at a concentration of 890 ppm and Barnea, D. 1987. A Unified Model for Predicting Flow-Pattern
for the commercial surfactant at a concentration of 3,000 ppm Transitions for the Whole Range of Pipe Inclinations. International
show a slug flow at a superficial gas velocity of approximately J. Multiphase Flow 13 (1): 1–12. http://dx.doi.org/10.1016/0301-
1 m/s. In this slug flow regime, the average pressure gradient 9322(87)90002-4.
has a mini- mum; it is even lower than the lowest pressure Bremner, B., Murphy, L. M., Affinito, R. J. et al. 2010. The First
gradient measured for air/water flow. The slug flow shows a Through- Tubing Capillary String Installation in the UK Sector of
very regular foam film the North Sea. Paper presented at the SPE/CoTA Coiled Tubing
and Well Interven- tion Conference and Exhibition, The
Woodlands, Texas, USA, 23–24 March. SPE-130672-MS.
http://dx.doi.org/10.2118/130672-MS.

April 2016 SPE 499


Journal
Campbell, S., Ramachandran, S, and Bartrip, K. 2001. Corrosion the Offshore Europe meeting, Aberdeen, 4–7 September. SPE-
Inhibi- tion/Foamer Combination Treatment to Enhance Gas 108380- MS. http://dx.doi.org/10.2118/108380-MS.
Production. Pa- per presented at the SPE Production and Taitel, Y. and Duckler, A. E. 1976. A Model for Predicting Flow
Operations Symposium, Oklahoma City, Oklahoma, USA, 24–27 Regime Transitions in Horizontal and Near Horizontal Gas-Liquid
March. SPE-67325-MS. http://dx.doi.org/10.2118/67325-MS. Flows. AlChE J. 22 (1): 47–55.
Corrin, M. L. and Harkins, W. D. 1947. The Effect of Salts on the http://dx.doi.org/10.1002/aic.690220105.
Critical Concentration for the Formation of Micelles in Colloidal Taitel, Y., Bornea, D., and Dukler, A. 1980. Modelling Flow Pattern
Electrolytes Tran- sitions for Steady Upward Gas-Liquid Flow in Vertical
1. J. Am. Chem. Soc. American Chemical Society 69 (3): 683–688. Tubes. AIChE J. 26 (3): 345–354.
http://dx.doi.org/10.1021/ja01195a065. http://dx.doi.org/10.1002/aic.690260304.
DeGennes, P-G., Brochard-Wyart, F., and Quere, D. 2004. Capillarity Turner, R. G., Hubbard, M. G., and Dukler, A. E. 1969. Analysis and
and Wetting Phenomena. New York: Springer. Pre- diction of Minimum Flow Rate for the Continuous Removal of
Fainerman, V. B., Makievski, A. V., and Miller, R. 2004. Accurate Liquids From Gas Wells. J Pet Technol 21 (11): 1475–1482. SPE-
Analy- sis of the Bubble Formation Process in Maximum Bubble 2198-PA. http://dx.doi.org/10.2118/2198-PA.
Pressure Tensiometry. Review of Scientific Instruments 75 (1): 213. Van Nimwegen, A. T., Portela, L. M., and Henkes, R. A. W. M. 2012.
http:// dx.doi.org/10.1063/1.1632996. A First Look at the Hydrodynamics of Air- Water-Foam Flow for
Fainerman, V.B., Lylyk, S. V., Aksenenko, E. V. et al. 2010. Surface Gas Well Deliquification. In 8th North American Conference on
Ten- sion Isotherms, Adsorption Dynamics and Dilational Visco- Multi- phase Technology, 367–379. Banff, Alberta, Canada: BHR
Elasticity of Sodium Dodecyl Sulphate Solutions. Colloids and Group.
Surfaces A: Physicochemical and Engineering Aspects 354 (1–3): Van Nimwegen, A. T, Portela, L. M,, and Henkes, R. A. W. M. 2015a.
8–15. http:// dx.doi.org/10.1016/j.colsurfa.2009.02.022. The Effect of Surfactants on Air-Water Annular and Churn Flow in
Hewitt, G. F. and Jayanti, S. 1993. To Churn or Not to Churn. Interna- Vertical Pipes. Part 2: Liquid Holdup and Pressure Gradient
tional J. Multiphase Flow 19 (3): 527–529. Dynam- ics. International J. Multiphase Flow 71: 146–158.
http://dx.doi.org/10.1016/ http://dx.doi.org/ 10.1016/j.ijmultiphaseflow.2014.03.007.
0301-9322(93)90065-3. Van Nimwegen, A. T., Portela, L. M., and Henkes, R. A. W. M.
Jelinek, W. and Schramm, L. L. 2005. Improved Production From 2015b. The Effect of Surfactants on Air–Water Annular and Churn
Mature Gas Wells by Introducing Surfactants Into Wells. Paper Flow in Vertical Pipes. Part 1: Morphology of the Air–Water
presented at the International Petroleum Technology Conference, Interface. Inter- national J. Multiphase Flow 71: 133–145.
Doha, 21–23 No- vember. IPTC-11028-MS. http://dx.doi.org/10.1016/ j.ijmultiphaseflow.2014.03.008
Kawale, D. 2012. Characterization of Surfactants by Dilational Rheology: Van’t Westende, J. M. C. 2007. Droplets in Annular-Dispersed Gas-liquid
Influence on Wetness of Rising Foams. Trondheim. Pipe-flows, dissertation. Delft University of Technology, (January 14,
Khosla, V., Kalter, R., and Portela, L. M. 2013. Visual Investigation of 2008).
the Gas-Liquid Interface in Annular Pipe Flow. Weaire, D. and Hutzler, S. 1999. The Physics of Foams. Oxford:
Lea, J., Nickens, H. V., and Wells, M. 2003. Gas Well Deliquification. Oxford University Press.
Willis, M. J., Horsup, D. I., and Nguyen, D. T. 2008. Chemical
Oxford: Gulf Professional Publishing. Foamers for Gas Well Deliquification. Paper presented at the SPE
Loglio, G., Pandolfini, P., Miller, R. et al. 2001. Drop and Bubble Asia Pacific Oil and Gas Conference and Exhibition, Perth, 20–22
Shape Analysis as a Tool for Dilational Rheologocal Studies of October. SPE- 115633-MS. http://dx.doi.org/10.2118/115633-MS.
Interfacial Layers. In Novel Methods to Study Interfacial Layers, Yang, J., Jovancicevic, V., and Ramachandran, S. 2007. Foam for Gas
439–483, ed. D. Möbius and R. Miller. Elsevier Science B.V. Well Deliquification. Colloids and Surfaces A: Physicochemical
Lucassen, J. and van den Tempel, M. 1972. Dynamic Measurements of and Engineering Aspects 309 (1–3): 177–181.
Dilational Properties of a Liquid Interface. Chemical Eng. Sci. 27: http://dx.doi.org/10.1016/ j.colsurfa.2006.10.011.
1283–1291. http://dx.doi.org/10.1016/0009-2509(72)80104-0.
Miller, R., Makievski, A. V., and Fainerman, V. B. 2001. Dynamics of Yavuz, F. 2011. Overview of Deliquification Measures in the Netherlands
Adsorption From Solutions. In Studies in Interface Science—Volume —Result of GWD Questionnaire. Presented at the European Gas
13, 287–399, ed. V. B. Fainerman, D. Möbius, and R. Miller. Well Deliquification Conference, Groningen, The Netherlands, 28–
Amster- dam: Elsevier B.V. 29 September.
Myers, D. 2005. Surfactant Science and Technology, third edition. Zhou, X. and Hao, J. 2011. Solubility of NaBr, NaCl, and KBr in
New York: John Wiley & Sons. Surfac- tant Aqueous Solutions. J. Chemical & Eng. Data 56 (4):
Myrvold, R. and Hansen, F. 1998. Surface Elasticity and Viscosity 951–955. http://dx.doi.org/10.1021/je100905g.
From Oscillating Bubbles Measured by Automatic Axisymmetric
Drop Shape Analysis. J. Colloid Interface Sci. 207 (1): 97–105.
Nguyen, D. T. 2009. Fundamental Surfactant Properties of Foamers
for Increasing Gas Production. Petrol. Sci. & Technol. 27 (7): 733– Andreas T. van Nimwegen is a post-doctoral-degree
745. http://dx.doi.org/10.1080/10916-460802105658. researcher at Delft University of Technology. He just completed
Ravera, F., Loglio, G., and Kovalchuk, V. I. 2010. Interfacial his PhD degree research at the same university on the effect
Dilational Rheology by Oscillating Bubble/Drop Methods. Current of surfactants on gas/liquid pipe flows.
Opinion in Colloid & Interface Sci. 15 (4): 217–228.
http://dx.doi.org/10.1016/ j.cocis.2010.04.001. Lu´ıs M. Portela is assistant professor at Delft University of Tech-
Reinicke, K. M., Albrecht, I., and Thurow, M. 2007. Sustaining nology. He teaches courses on fluid mechanics, transport phe-
Produc- tion in Water Producing Gas Wells. Paper presented at the nomena, and multiphase flow. Portela’s main research
2nd Euro- pean Gas Well Deliquification Conference, Groningen, interests are in fluid mechanics and, in particular, multiphase
The Netherlands, 24–26 September. flow. He holds a PhD degree in mechanical engineering from
Rosen, M. J. and Hua, X. Y. 1988. Dynamic Surface Tension of Stanford University.
Aqueous Surfactant Solutions. 1. Basic Parameters. J. Colloid and
Interface Sci. 124 (2): 652–659. http://dx.doi.org/10.1016/0021- Rudolphus A. W. M. Henkes is Team Lead Multiphase Flow and
9797(88)90203-2. the Principal Technical Expert Fluid Flow for Shell Projects and
Rosen, M. J., Hua, X.Y., and Zhu, Z. H. 1991. Dynamic Surface Technology. He is also Professor Multiphase Flow at Delft Uni-
Tension of Aqueous Surfactant Solutions: IV. Relationship to versity of Technology. Henkes holds a PhD degree from that
Foaming. Surfac- tants in Solution 11: 315–327. university in fluid flow and heat transfer. He has more than 25
http://dx.doi.org/10.1007/978-1-4615- 3836-3_22. years of experience in fluid flow and more than 15 years of ex-
Schinagl, W., Green, S., Hodds, A. et al. 2007. Most Successful Batch perience in oil and gas production. Henkes was involved in
Application of Surfactant in North Sea Gas Wells. Paper presented the multiphase flow design of many of Shell’s major projects.
at
Over the years, he has published journal papers on heat trans-
fer (particularly, natural convection), boundary layers, and
multiphase flow.

500 April 2016 SPE


Journal

Vous aimerez peut-être aussi