# Air-lift pumps characteristics under two-phase flow conditions

**ABSTRACT** Air-lift pumps are finding increasing use where pump reliability and low maintenance are required, where corrosive, abrasive, or radioactive fluids in nuclear applications must be handled and when a compressed air is readily available as a source of a renewable energy for water pumping applications. The objective of the present study is to evaluate the performance of a pump under predetermined operating conditions and to optimize the related parameters. For this purpose, an air-lift pump was designed and tested. Experiments were performed for nine submergence ratios, and three risers of different lengths with different air injection pressures. Moreover, the pump was tested under different two-phase flow patterns. A theoretical model is proposed in this study taking into account the flow patterns at the best efficiency range where the pump is operated. The present results showed that the pump capacity and efficiency are functions of the air mass flow rate, submergence ratio, and riser pipe length. The best efficiency range of the air-lift pumps operation was found to be in the slug and slug-churn flow regimes. The proposed model has been compared with experimental data and the most cited models available. The proposed model is in good agreement with experimental results and found to predict the liquid volumetric flux for different flow patterns including bubbly, slug and churn flow patterns.

**4**Bookmarks

**·**

**437**Views

- [Show abstract] [Hide abstract]

**ABSTRACT:**Airlift Systems (ALS) are widely used in various industrial applications. As the main part of the flow through ALS's upriser pipe, is formed by gas-liquid flow, the analysis of such systems will be accompanied by problems of two-phase flow modelling. Several effective variables are involved in ALS; thereupon comprehensive method is needed to consider these parameters. Exergy analysis can be considered as a simple solution for the realisation of the preferred domain of ALS's operation. Here, this method has been proposed to examine the performance of ALS. Based on thermodynamic principles, an analytical model has been implemented in each phase and the respective experimental data have been collected from the test rig. A new efficiency definition for ALS has been proposed and compared with the existing definitions available in the literature. Finally, flow availability and entropy generation have been estimated by this method in the ALS.International Journal of Exergy 01/2011; 8(4):407 - 424. · 0.85 Impact Factor - SourceAvailable from: Essam Moustafa Wahba[Show abstract] [Hide abstract]

**ABSTRACT:**The present study investigates a hierarchy of models for predicting the performance of air-lift pumps. Investigated models range from simplified one-dimensional analytical models to large eddy simulation (LES). Numerical results from LES and from two differ-ent analytical models are validated against experimental data available from the air-lift pump research program at Alexandria University. Present LES employs the volume of fluid (VOF) method to model the multiphase flow in the riser pipe. In general, LES is shown to provide fairly accurate predictions for the air-lift pump performance. Moreover, numerical flow patterns in the riser pipe are in good qualitative and quantitative agree-ment with their corresponding experimental patterns and with flow pattern maps avail-able in the literature. On the other hand, analytical models are shown to provide results that are of surprisingly comparable accuracy to LES in terms of predicting the pump per-formance curve. However, due to the steady one-dimensional nature of these models, they are incapable of providing information about the different flow patterns developing in the riser pipe and the transient nature of the pumping process. [DOI: 10.1115/1.4027473]Journal of Fluids Engineering 11/2014; 136(11):111301. · 0.94 Impact Factor - SourceAvailable from: ocean.kisti.re.kr[Show abstract] [Hide abstract]

**ABSTRACT:**An airlift pump can be used to pump liquids and sediments within itself, which cannot easily be pumped up by a conventional method, by using the airlift effect. This characteristic of the airlift pump can be exploited in a DCFC (Direct Carbon Fuel Cell) so that molten fuel with high temperature may be carried or transported. The basic characteristics of airlift are investigated. A simple system is constructed, where the reservoir is filled with water, a tube is inserted, and air is supplied from the bottom of the tube. Then, water is lifted and its flow rate is measured. Bubble patterns in the tube are observed in a range of air flow rates with the parameters of the tube diameter and submergence ratio, leading to four distinct regimes. The pumping performance is predicted, and the correlation between the supplied gas flow rate and the induced flow rate of water is found.Transactions of the Korean Society of Mechanical Engineers B 09/2013; 37(9).

Page 1

Air-lift pumps characteristics under two-phase flow conditions

Sadek Z. Kassaba, Hamdy. A. Kandila, Hassan A. Wardaa, Wael H. Ahmedb,*

aMechanical Engineering Department, Faculty of Engineering, Alexandria University Alexandria, Egypt

bNuclear Safety Solution Ltd., AMEC, 700 University Avenue, Toronto, Ontario, Canada M5G 1X6

a r t i c l ei n f o

Article history:

Received 31 March 2008

Received in revised form 22 September

2008

Accepted 23 September 2008

Available online xxxx

Keywords:

Air-lift pumps

Two-phase flow

Vertical pipe

a b s t r a c t

Air-lift pumps are finding increasing use where pump reliability and low maintenance are required,

where corrosive, abrasive, or radioactive fluids in nuclear applications must be handled and when a com-

pressed air is readily available as a source of a renewable energy for water pumping applications. The

objective of the present study is to evaluate the performance of a pump under predetermined operating

conditions and to optimize the related parameters. For this purpose, an air-lift pump was designed and

tested. Experiments were performed for nine submergence ratios, and three risers of different lengths

with different air injection pressures. Moreover, the pump was tested under different two-phase flow

patterns. A theoretical model is proposed in this study taking into account the flow patterns at the best

efficiency range where the pump is operated. The present results showed that the pump capacity and effi-

ciency are functions of the air mass flow rate, submergence ratio, and riser pipe length. The best efficiency

range of the air-lift pumps operation was found to be in the slug and slug-churn flow regimes. The pro-

posed model has been compared with experimental data and the most cited models available. The pro-

posed model is in good agreement with experimental results and found to predict the liquid volumetric

flux for different flow patterns including bubbly, slug and churn flow patterns.

? 2008 Elsevier Inc. All rights reserved.

1. Introduction

The great focus towards the renewable energy for water pump-

ing applications brought the attention to revisit the analysis of the

air-lift pumps operated in two-phase flow. As the pneumatic trans-

mission wind pumps operate on the principle of compressed air by

using a small industrial air compressor to drive an air-lift pump or

pneumatic displacement pumps. The main advantage of this meth-

od is that there is no mechanical transmission from the windmill to

the pump, which avoids water hammer and other related dynamic

problems. The pump can operate slowly even while the windmill is

running rapidly with no dynamic problem. Other advantages are

its simplicity and low maintenance. However, this technology is

still under development and will require intensive field testing be-

fore it can be commercialized.

Air-lift pump is a device for raising liquids or mixtures of liquids

and solids through a vertical pipe, partially submerged in the li-

quid, by means of compressed air introduced into the pipe near

the lower end. The air then returns up in a discharge pipe carrying

the liquid with it. The pump works by ‘‘aerating” the liquid in the

discharge pipe. The added air lowers the specific gravity of the fluid

mixture. Since it is lighter than the surrounding liquid, it is pushed

upwards.

The principles of air-lift pumping were understood since about

1882, but practical use of air-lift did not appear until around the

beginning of the twentieth century (Bergeles 1949). In comparison

with other pumps, the particular merit of the air-lift pump is the

mechanical simplicity. Moreover, they can be used in a corrosive

environment, and are easy to use in irregularly shaped wells where

other deep well pumps do not fit. Thus, theoretically, the mainte-

nance of this kind of pumps has a lower cost and higher reliability.

There is a wide use of the air-lift pumps in many applications such

as in under water explorations or for rising of coarse particle sus-

pensions (Stenning and Martin 1968), dredging of river estuaries

and harbors, and sludge extraction in sewage treatment plant

(Storch 1975).

The present study is concerned with the applications of air-lift

pumps in pumping liquids. In this case, the flow in the pump riser

is a two-phase flow. The flow of the two phases in the riser of an

air-lift pump is a direct application of the upward flow in vertical

round tubes. The common flow patterns for vertical upward

flow are changing as the mass quantity is increased (Taitel et al.

1980).

Sharma and Sachdeva (1976) studied the factors that affect the

performance of big diameter air-lift pumps operating in shallow

depths. They related the pump performance to the type of flow

pattern in the riser. In addition, DeCachard and Delhaye (1996)

showed experimentally that the dominant flow pattern in the

practical operating range of a small diameter air-lift pump is slug

0142-727X/$ - see front matter ? 2008 Elsevier Inc. All rights reserved.

doi:10.1016/j.ijheatfluidflow.2008.09.002

* Corresponding author. Tel.: +1 6135846814; fax: +1 6135849497.

E-mail addresses: wael.ahmed@amec.ca, wael.ahmed@gmail.com (W.H. Ahmed).

International Journal of Heat and Fluid Flow xxx (2008) xxx–xxx

Contents lists available at ScienceDirect

International Journal of Heat and Fluid Flow

journal homepage: www.elsevier.com/locate/ijhff

ARTICLE IN PRESS

Please cite this article in press as: Kassab, S.Z. et al., Air-lift pumps characteristics under two-phase flow conditions, Int. J. Heat Fluid Flow

(2008), doi:10.1016/j.ijheatfluidflow.2008.09.002

Page 2

flow. A few studies are available on the effect of the liquid proper-

ties such as surface tension and viscosity and pipe geometry on the

air-lift pump performance. For example, Khalil and Mansour

(1990) carried out experimental work to study the effect of intro-

ducing a surfactant in the pumped liquid. They proved that the

use of a surfactant in small concentration always increases the

capacity and efficiency of the pump. Further, they found that the

pump performance is a function of the volumetric air-flow rate,

submergence ratio (Hs/L) as shown in Fig. 1, air-supply pressure

and surfactant concentration. Iguchi and Terauchi (2001) studied

the effect of the pipe-wall wettability on the transition among bub-

bly and slug flow regimes in air–water two-phase flow in a vertical

pipe. They changed the advancing contact angle of an acrylic pipe

by coating its inner wall by a hydrophilic substance or liquid

paraffin. Some differences were observed between pipe walls of

different wettability. Flow pattern maps were developed for differ-

ent pipe-wall wettability. Also, Furukawa and Fukano (2001) inves-

tigated experimentally the effects of the liquid viscosity on the

flow patterns of upward air–liquid two-phase flow in a vertical

pipe. They used three different liquids including water and glycol

solutions. They proposed flow pattern maps for each liquid viscos-

ity. It was found that flow pattern transitions strongly depend on

the liquid viscosity. Recently, Hitoshi et al. (2003) found experi-

mentally that the gas-injection point has a big effect on discharged

water. They concluded that, as gas-injection point increases above

45 pipe diameters, the discharged water decreases.

Although the geometry of the pump is very simple, the theoret-

ical study of its performance is very complicated. The efforts to ex-

plain and study its performance started very early in the last

century. Among the early classical theories of the air-lift pump,

are those proposed by Harris, Lorenz, Gibson, and Swindin, as men-

tioned by Stapanoff (1929). Harris considered the force of buoy-

ancy of the air bubbles as the motive force of the pump. He

analyzed the motion of the bubble, and obtained the relation be-

tween the size of the bubble, slip (relative velocity of the air bubble

with respect to the water), and the head produced. Lorenz, in

developing his theory, wrote Bernoulli’s equation for a differential

head corresponding to a given flow in the discharge pipe, introduc-

ing variable pressure, density of mixture, and integrating between

the head limits and thus he obtained the relation between the vari-

ables involved. Stapanoff (1929) used the thermodynamics theory

in studying the effect of the submergence, the diameter of the riser

pipe, air to water ratio, climate, and introducing compressed air

above the surface of the water in the well, on the efficiency of

the air-lift pump. He found that treating the air-lift pump thermo-

dynamically has a definite advantage in explaining many points of

its operation.

More than three decades later, a theoretical treatment of the

air-lift pump, based on the theory of slug flow, was presented by

Nicklin (1963). He studied the effects of different parameters

including; diameter, length, pressure at the top of riser tube, sub-

mergence ratio, and water volumetric flow rate, on the air-lift-

pump efficiency. He found that, by neglecting the entrance effects

and assuming slug flow in the riser tube, the performance of the

air-lift pump may be obtained based on two-phase slug flow re-

gime. He used a definition of the efficiency of the pump as the work

done in lifting the liquid, divided by the work done by the air as it

expands isothermally. The theory presented by Nicklin was ex-

tended by Reinemann et al. (1986) taking into account the effect

Nomenclature

A

b

D

f

g

Hs

Hd

K

L

Ls

P

Q

Sr

s

U

V

pipe cross-sectional area, m2

wetted perimeter of pipe

pipe diameter, m

friction factor

gravitational acceleration, m/s2

static depth of water, m

static lift

friction parameter

pipe length, m

pipe suction length, m

pressure, N/m2

volume flow rate (discharge), m3/s

Sr= Hs/L, submergence ratio

slip ratio

superficial velocity, m/s

velocity, m/s

W

q

g

s

e

Re

weight, N

density, kg/m3

efficiency,%

wall shear stress, N/m2

pipe roughness

Reynolds number

Subscripts

1

2

a

gs

g

L

ls

S

entering injection zone

leaving injection zone

atmospheric

gas as a single phase

gas

liquid

liquid as a single phase

solid

8

10

11

9

13

12

3

16

6

15

14

7

5

4

1

Hs

L

Ls

2

Fig. 1. A schematic diagram of the experimental setup. 1.Riser, 2. Down comer, 3.

Overhead collecting tank, 4. Scale, 5. Drain, 6.Water feeding tank, 7. Over flow pipe,

8. Compressor, 9. Regulator, 10. Pressure gage, 11. Thermometer, 12. Air-jacket, 13.

Below meter, 14. Feeding water line, 15. Control valve, 16. Float.

2

S.Z. Kassab et al./International Journal of Heat and Fluid Flow xxx (2008) xxx–xxx

ARTICLE IN PRESS

Please cite this article in press as: Kassab, S.Z. et al., Air-lift pumps characteristics under two-phase flow conditions, Int. J. Heat Fluid Flow

(2008), doi:10.1016/j.ijheatfluidflow.2008.09.002

Page 3

of surface tension on the bubble velocity. Another analytical study

of air-lift pump performance was presented by Stenning and Mar-

tin (1968). They used the continuity and momentum equations,

assuming one-dimensional flow in the pump riser, and used the re-

sults of two-phase flow research to solve the governing equations.

They found that one-dimensional-flow theory forms a good basis

for the performance analysis of air-lift pumps. Also, Clark and Da-

bolt (1986) introduced a general design equation for air-lift pumps

operating in slug flow regime by integrating the differential

momentum equation over the whole pump length. The model

was validated by the operation curves plotted from their experi-

mental results. They indicated also that the analysis presented by

Nicklin (1963) was accurate only in the design of short pumps,

since there is no provision for variation in gas density over the tube

length. In addition, they concluded that, the frictional pressure loss

becomes significant for small riser diameter of about 10 mm.

The present study reviewed the previous models, dealt with

the air-lift pump performance when operating in two-phase flow

regime. A modified version of these models is developed. The re-

sults of the proposed modified model are then compared with the

experimental measurements of Kassab et al. (2007) and other

available models. In order to study the performance of the air-lift

pump under different flow patterns the air-flow rate range was

extended in the present study to cover different flow patterns

including bubbly, slug, churn and annular flow conditions. In

addition, pipes with different lengths were used to study the ef-

fect of the riser-pipe length, at the same submergence ratio, on

the pump performance. Experiments were performed for nine

submergence ratios, and three lengths of the pipe riser with dif-

ferent air injection pressures. Moreover, the different flow re-

gimes and the transition of the flow patterns were observed and

recorded and the results are discussed taking flow patterns into

consideration.

2. Experimental setup

The experimental setup used in the present study is schemati-

cally shown in Fig. 1. It consists of a vertical transparent pipe (1),

of 3.75 m length and 25.4 mm inner diameter, and a down-comer

(2) of 30 mm inner diameter. The riser pipe is divided into three

sections to allow studying the effect of changing the length of

the riser pipe. The upper end of the riser is connected to an over-

head-collecting tank (3) where the air escapes to atmosphere and

water flow rate is measured according to the water level in the

tank using a calibrated scale (4). The overhead tank is designed

to absorb the water surface fluctuations and damp free vortices

and thus provides accurate flow rate measurements. Water may

be directed through a pipe (5) to the drain. The movable water sup-

ply tank (6) is kept at a constant water head by overflowing the

water through a pipe (7). The tank may also be moved up or down

to change the submergence ratio. All pipes and tanks are made of

transparent material for visibility of the flow structure.

Air is supplied to the air injection system from a central air

compressor station. The station consists of a 55 kW Ingersoll Rand

screw compressor (8) delivering 8.2 m3of free air per minute at a

maximum pressure of 8 bar, through a mass refrigeration dryer

and filtration system, and an air reservoir of 3 m3capacity. Air-

flows from the air reservoir through a 25.4 mm diameter pipeline

to an on/off valve, then to a pressure-reducing valve (regulator)

(9), where the pressure is reduced to the desired working pressure

(1 ? 104–2.7 ? 105Pa) to cover the required experimental range.

Air is then injected into the riser at a constant pressure that can

be measured by the pressure gage (10). A mercury thermometer

(11) is used for measuring the upstream air temperature. Then a

constant air mass flow rate passes through an air jacket (12)

around the vertical pipe using an air injector. The volume of air

is measured using a calibrated below-meter (13). The air injector

consists of 56 small holes of 3 mm diameter uniformly distributed

around the pipe perimeter in seven rows and eight columns to in-

sure uniform feed of the air into the pipe at the mixing section,

which is 20 cm above the lower end of the pipe.

Various submergence ratios (from 0.2 to 0.75) were investi-

gated in the present study. This range of submergence ratios was

obtained in increments approximately 0.1, that covers most indus-

trial applications where the air-lift pump is used. For each submer-

gence ratio, the air-flow rate was varied and the corresponding

flow rate of water was measured. In order to obtain specified and

planed measurements, a specific operating procedure was followed

for each run.

3. Results and discussion

3.1. Water flow rate

The results of lifting water in the riser tube of an air-lift pump,

at various values of air mass flow rates corresponding to different

values of air injection pressures are presented. Fig. 2 shows the

water flow rate as a function of the air-flow rate at a submergence

ratio of 0.4. Using flow visualization, Fig. 3, and the experimental

results, it was noticed that for low values of air mass flow rate from

0 to 1 kg/h depends on the submergence ratio, no water is lifted

due to the buoyant force exerted by the air bubbles is not enough

to raise any water. The total quantity of air penetrates the water

column without lifting any water. Fig. 3a shows that the flow re-

gime is totally bubbly for very small values of air-flow rate. As

the air mass flow rate is increased, a train of air slugs starts to de-

velop in the pipe. It consists of some small slugs of lengths from 10

to 15 cm distributed along the pipe length. The biggest bubble is

located at the upper part of the pipe while the smallest one is lo-

cated at the lower part of the pipe. The train of air slugs moves

slowly upwards leaving the pipe without any bubbles for a few

seconds. The distribution is then repeated after a few seconds. This

is because the air takes a few seconds to accumulate in the lower

end of the pipe. It was observed that the air pressure at injection

point fluctuates by 1 or 2 kPa around a mean value. This is due

to the accumulation of air to form a slug that is strong enough to

penetrate the water column and move upwards. When the air

mass flow rate is increased slightly over 0.229 kg/h, the water flow

starts, and the flow picture, Fig. 3b, is similar to the case of the

penetrating column except that, the slugs become taller. The

04812 1620

Air mass flow rate (kg/hr)

0

1000

2000

3000

Water mass flow rate (kg/hr)

a

bc, de,f

g,h

Bubbly flow

Bubbly-Slug

Slug flow

Slug-Churn flow

Annular flow

Submergence ratio = 0.4

Fig. 2. Variation of air mass flow rate with water mass flow rate at submergence

ratio = 0.4 (a–g represent the sequence of photos presented in Fig. 3).

S.Z. Kassab et al./International Journal of Heat and Fluid Flow xxx (2008) xxx–xxx

3

ARTICLE IN PRESS

Please cite this article in press as: Kassab, S.Z. et al., Air-lift pumps characteristics under two-phase flow conditions, Int. J. Heat Fluid Flow

(2008), doi:10.1016/j.ijheatfluidflow.2008.09.002

Page 4

structure of the slug flow in the pipe can be described by a se-

quence of a big air slug followed by a water slug that contains some

small air bubbles and then smaller air and water slugs. The cycle is

then repeated as shown in Figs. 3c–d. This observation agrees with

the experimental work by Sekoguchi et al. (1981). This sequence of

slug formation may be due to the time taken to charge enough

quantity of air capable of lifting a certain amount of water. This ex-

plains the oscillation of the air injection pressure. The oscillation

and the instability of the pump were studied by Hjalmars (1973).

He found that, for a large air-lift pump, when the static lift was in-

creased to about 30-40 pipe diameters, the instability of the air-lift

pump starts. However, when the value of the static lift reached a

critical value, instability sets in with a periodic and increasing var-

iation of water discharge around its stationary value. The reason of

these oscillations was also discussed by Sekoguchi et al. (1981).

They referred this behavior to the compressibility of the gas in

the gas–liquid two-phase mixture and reversal of water flow from

the two-phase mixing section to the head tank. In the present

study a very small variation was observed, and hence the instabil-

ity effects were neglected.

The reason for the transition from bubbly to slug flow is that, as

the gas flow is increased, the bubbles get closer together and colli-

sion occurs. Therefore, some of the collisions lead to coalescence of

bubbles and eventually to the formation of slugs. For a submer-

gence ratio of 0.75, the lifting of water becomes noticeable at a va-

lue of air-flow rate of approximately 1 kg/h. Any slight increase in

the airflow rate beyond that value causes the water flow rate to in-

crease rapidly. It is noted that in the region where the water mass

flow rate is increasing (from 403 to 642 kg/h), the flow regime is

mostly slug-churn flow. In addition, the maximum flow rate of

water (642 kg/h) occurs when the flow pattern is slug-churn flow.

As can be seen in Fig. 2, the water mass flow rate increases as

the air mass flow rate increased, until it reaches a maximum value.

It was noticed that, in the region in which the water flow rate in-

creases, the flow pattern changed from slug to slug-churn flow as

shown in Fig. 3e–f. This can be explained as follows: as the gas slug

rises through the liquid, the direction of gas velocity inside the slug

is upwards, while the water velocity direction in the thin film

around the air slug is usually downwards, so the flow is counter

current. At some critical value of air mass flow rate the gas velocity

will suddenly disrupt the liquid film (the film will flood) and there-

fore the slug flow will break down to give churn flow with pulsat-

ing, highly unstable pattern, as suggested by Nicklin (1963). It is

noted that the region of the slug-churn flow is the main region

where the pump should be operating. The quantity of lifted water

remains almost constant for a small range of air-flow rates (from

5.3 kg/h to 6.2 kg/h). Physically, the maximum water flow rate is

reached when the frictional pressure drop caused by further addi-

tion of air exceeds the buoyancy effect of the additional air, as was

explained by Reinemann et al. (1986). Further increase in the air-

flow causes a slight decrease in the water flow rate to a value of

(51 kg/h), and the flow regime changes from slug-churn to annular

flow type as shown in Fig. 3g. This transition is because the gas

velocity becomes high enough to support the liquid as a film on

the tube wall, and also because the pressure drop exceeds the

buoyancy effect.

Fig. 3h shows the annular flow pattern and also how water is

lifted through the pipe riser. A small amount of water is trans-

ported as a liquid film on the tube walls, while another part forms

small droplets of water injected in the pipe core upward. Measure-

ments were performed at different values of submergence ratios,

and the results are shown in Fig. 4. It is clear that the performance

curves of the air-lift pump are shifted upward while the submer-

gence ratio is increased. All the performance curves at different

Fig. 3. Photos present the sequence of the flow patterns: (a) Bubbly, (b) bubbly to slug, (c and d) slug, (e and f) slug-churn, (g and h) annular.

048121620

Air mass flow rate (kg/hr)

0

500

1000

1500

2000

2500

3000

Water mass flow rate (kg/hr)

Sr=0.2

Sr=0.227

Sr=0.3

Sr=0.4

Sr=0.484

Sr=0.57

Sr=0.67

Sr=0.75

Fitting Lines

Fig. 4. Variation of water mass flow rate with air mass flow rate at different values

of submergence ratio.

4

S.Z. Kassab et al./International Journal of Heat and Fluid Flow xxx (2008) xxx–xxx

ARTICLE IN PRESS

Please cite this article in press as: Kassab, S.Z. et al., Air-lift pumps characteristics under two-phase flow conditions, Int. J. Heat Fluid Flow

(2008), doi:10.1016/j.ijheatfluidflow.2008.09.002

Page 5

values of submergence ratio have similar trend. Also, during the

experiments, it was noticed that the flow pattern changes by the

same sequence as indicated earlier (for the 0.4 submergence-ratio

case). In addition, Fig. 4 shows that, for a fixed value of air mass

flow rate, the water flow rate increases with the increase of the

submergence ratio.

3.2. The pump efficiency

The definition of the air-lift pumps efficiency is given by Nicklin

(1963) as

g ¼qgQLðL ? HsÞ

PaQaLnpin

Pa

where QLis the water discharge, Qais the volumetric flow rate of air,

Pinthe injection pressure of air, q is the liquid density and Pais the

atmospheric pressure.

Pump efficiency versus air mass flow rate, for a submergence

ratio of 0.4, is shown in Fig. 5a. As the air mass flow rate is in-

creased, the efficiency increases rapidly from 0 to reach a maxi-

mum value of 32.4% at an air mass flow equals 1.5 kg/h, and

then, it tends to decrease as the air mass flow rate is increased.

The efficiency curves at different values of submergence ratio

are presented in Fig. 5b. They have similar trend as the one pre-

sented in Fig. 5a for submergence ratio of 0.4. In addition the best

efficiency range is in the first region of the performance curve,

which is a slug flow region. The maximum efficiency increased to

44% when the submergence ratio is increased to 0.75 as presented

in Fig. 5b. Comparing the water mass flow rate results presented in

Fig. 4 with the efficiency results presented in Fig. 5, it is important

to notice that, the maximum efficiency does not occur at the max-

imum water mass flow rate for all values of submergence ratio.

3.3. Effect of riser length

The effect of riser length was also studied in the present study.

This was done by fixing the submergence ratio and varying the sta-

tic lift by changing the riser length. Experiments were performed

for submergence ratio values of 0.4, 0.484, 0.5, 0.57, 0.67, and

0.74, and different total pipe lengths of; 175 cm, 275 cm, and

375 cm. The performance curves of the pump are presented for

each case in Fig. 6. While the effect of the variation of the static lift

on pump efficiency for submergence ratio equal to 0.57, is shown

in Fig. 7. It is clear from these figures that, for any riser length,

there is a slight effect of the total pipe length on the performance

pattern, in the slug flow region. However, the maximum water

mass flow rate decreases when the static lift is decreased, and

the performance curve shifts down by a small amount for the

remaining regions. Moreover, for submergence ratios lower than

0.3, it was noticed that, there was no water lifted if the static lift

is increased to 82.5 cm (32 pipe diameters). This is because air

was not able to carry the water column to a distance equal to

the static lift, and air penetrates the water column without lifting

an accountable mass of water. The results presented in Figs. 6

and 7 show that not only the submergence ratio affects the air-lift

pump performance but also the magnitude of the riser length.

3.4. Flow pattern map

Fig. 8a shows the distribution of the test data on the flow pat-

tern map proposed by Taitel et al. (1980). It can be seen that the

regions occupied by the experimental data points agree with the

observations. Some experimental measurements from the high

efficiency range (g > 20%) at different submergence ratios are

mapped on the flow-pattern map as shown in Fig. 8b. The best effi-

ciency range where the efficiency exceeds 20% lies totally in the

slug or slug-churn flow regions. This may explain why most analyt-

ical studies were based on the slug or slug-churn flow patterns.

4. Modeling the air-lift pump performance

The problem considered is the prediction of the liquid mass

flow rate as a function of the air mass flow rate. The geometric

parameters (L, Hs, Ls, and D), the pressure conditions (Pa, Pin), and

the fluid properties are given as input data to the theoretical model

equations. Where L is the riser-pipe length, Hsis the static head of

water, Lsis the length of the suction part of the pipe, D is the pipe

diameter, Pais the atmospheric pressure, and Pinis the injection

pressure to the riser pipe. As the basic performance data of the

air-lift pump is computed, secondary results such as the efficiency

may easily be determined.

4.1. Clark and Dabolt model

The general design equation for air-lift pumps operating in the

slug flow regime that was developed by Clark and Dabolt (1986) is

taken as a preliminary investigation model of the pump perfor-

mance. A computer program was developed for the equations de-

rived by Clark and Dabolt (1986), using the momentum balance.

Derivation is based on the assumption that the two-phase flow re-

mains within the slug flow mode and the flow is one-dimensional.

048121620

Air mass flow rate (kg/hr)

Air mass flow rate (kg/hr)

0

20

40

Efficency (%)

Efficency (%)

Spline smoothing

Submergence ratio= 0.4

048121620

10

30

50

Submergence ratio (Sr)

Sr=0.74

Sr=0.67

Sr=0.57

Sr=0.4

Sr=0.277

a

b

Fig. 5. Variation of pump efficiency with air mass flow rate at: (a) submergence

ratio = 0.4 (b) different values of submergence ratio.

S.Z. Kassab et al./International Journal of Heat and Fluid Flow xxx (2008) xxx–xxx

5

ARTICLE IN PRESS

Please cite this article in press as: Kassab, S.Z. et al., Air-lift pumps characteristics under two-phase flow conditions, Int. J. Heat Fluid Flow

(2008), doi:10.1016/j.ijheatfluidflow.2008.09.002

Page 6

The theoretical predictions using Clark and Dabolt model to-

gether with the corresponding experimental results of Kassab

et al. (2001) are presented in Fig. 9 for a submergence ratio of

0.4. The agreement between these two results is quite reasonable

up to an air mass flow rate of 3.4 kg/h. The theoretical model does

not predict the experimental data for air mass flow rates higher

than 3.4 kg/h. The results obtained for different submergence ra-

tios (not shown), were also compared with those predicted by

the model proposed by Clark and Dabolt (1986). The comparison

showed that the model is suitable only for the first region of pump

performance, where the flow regime is slug. From the previous

comparison and in order to extend the operating range of the

air-lift pump, it is essential to develop a model which is more gen-

eral than that proposed by Clark and Dabolt (1986).

4.2. A modified model for the performance of the air-lift pumps

As discussed earlier in the literature review, the direct approach

to study the air-lift pump performance employs the momentum

equation and the continuity equation, assuming one-dimensional

flow. This assumption is valid for the practical operating range of

the air-lift pumps as concluded by Clark and Dabolt (1986). Con-

048 121620

0

1000

2000

3000

048121620

0

1000

2000

3000

04812 16 20

0

1000

2000

3000

048121620

0

1000

2000

3000

048 121620

Air mass flow rate (kg/hr)

Air mass flow rate (kg/hr)

Air mass flow rate (kg/hr) Air mass flow rate (kg/hr)

Air mass flow rate (kg/hr) Air mass flow rate (kg/hr)

0

1000

2000

3000

water mass flow rate (kg/hr)

water mass flow rate (kg/hr)

water mass flow rate (kg/hr)

water mass flow rate (kg/hr)

water mass flow rate (kg/hr)

water mass flow rate (kg/hr)

048121620

0

1000

2000

3000

a

b

c

f

e

d

Submergence ratio= 0.5

Static lift= 187.5 cm

Static lift= 137.5 cm

Static lift= 87.5 cm

Submergence ratio= 0.4

Static lift=225 cm

Static lift=165 cm

Static lift=105 cm

Submergence ratio=0.57

Static lift =161.25 cm

Static lift =118.25 cm

Static lift =75.25 cm

Submergence Ratio=0.484

Static Lift=193.5 cm

Static Lift=142 cm

Submergence ratio=0.67

Static lift= 123.75 cm

Static lift=90.75 cm

Submergence ratio=0.74

Static lift =97.5 cm

Static lift =71.5 cm

Fig. 6. Effect of static lift on the air-lift pump performance at various values of submergence ratio.

6

S.Z. Kassab et al./International Journal of Heat and Fluid Flow xxx (2008) xxx–xxx

ARTICLE IN PRESS

Please cite this article in press as: Kassab, S.Z. et al., Air-lift pumps characteristics under two-phase flow conditions, Int. J. Heat Fluid Flow

(2008), doi:10.1016/j.ijheatfluidflow.2008.09.002

Page 7

sider a vertical pipe partly full of liquid, and let the base of the pipe

have a reference height of zero as shown in Fig. 10. Since the tube is

filled to a static head of Hs, then the static pressure, Po, at the base

of the pipe is given by Bernoulli’s equation as follows:

Po¼ PaþqLgHs?1

where qLis the liquid density, Pais the atmospheric pressure and V1

is the water velocity at the inlet section.

Neglecting the density changes of the air, the continuity equa-

tion can be written as follows:

2qLV2

1

ð1Þ

AV2¼ Qgþ QL¼ Qgþ AV1

where V2 is the mixture velocity of air and water leaving the

injector.

Dividing all terms of Eq. (2) by QL= AV1, gives

V2¼ V1 1 þQg

QL

ð2Þ

??

ð3Þ

Neglecting the air mass flow rate compared to the liquid mass flow

rate, the continuity equation can be written as follows:

q2AV2¼ qLAV1

So,

ð4Þ

q2¼ qL

V1

V2

ð5Þ

Substituting Eq. (3) in Eq. (5), we obtain

qL

1 þ

The momentum equation applied to the injector as a control vol-

ume, neglecting the wall friction, is given by

q2¼

Qg

QL

??

ð6Þ

P2¼ Po?qLV1ðV2? V1Þ

From Eq. (3) in (7), then

ð7Þ

P2¼ Po?qLV1Qg

Hence, combining (1) and (8), gives

A

ð8Þ

P2¼ PaþqLgHs?1

Neglecting momentum changes caused by the flow adjustment

after the mixer, the momentum equation for the upper portion of

2qLV2

1?qLV1Qg

A

ð9Þ

the pump can be written as suggested by Stenning and Martin

(1968) in the form:

P2? Pa¼ sLb

wheres is the average wall shear stress, b is the wetted perimeter of

the pipe, and W is the total weight of the gas and liquid in the pipe.

An expression for the average shear stress, s, was suggested by

Griffith and Wallis (1961) as follows:

?

where f is the friction factor assuming that the water alone flows

through the pipe.The weight of the fluid in the pipe equals the total

weight of liquid plus gas, which can be obtained as follows:

AþW

A

ð10Þ

s ¼ fqL

QL

A

?2

ð1 þQg

QLÞð11Þ

W ¼ LðqLALþqgAgÞð12Þ

048 121620

Air mass flow rate (kg/hr)

10

30

50

Efficency (%)

Submergence Ratio=0.57

Static Lift=161.25 cm

Static Lift=118.25 cm

Static Lift=75.25 cm

Fig. 7. Variation of the pump efficiency with the static lift at submergence

ratio = 0.7.

0.00.11.010.0100.0

0.00.11.010.0100.0

Gas superfitial velocity (m/s)

Liquid superfitial velocity (m/s)

Present observation

Slug flow

Slug-Churn flow

Annular flow

Slug

Slug or Churn

Annular

Gas superfitial velocity (m/s)

0.00

0.01

0.10

1.00

10.00

0.00

0.01

0.10

1.00

10.00

Liquid superfitial velocity (m/s)

Present observation

Slug flow

Slug-Churn flow

Slug

Slug or Churn

Annular

Fig. 8. Comparison between the results of the present study and Taitel et al. (1980)

flow-pattern map: (a) distribution of the experimental data (b) distribution of the

best efficiency points.

S.Z. Kassab et al./International Journal of Heat and Fluid Flow xxx (2008) xxx–xxx

7

ARTICLE IN PRESS

Please cite this article in press as: Kassab, S.Z. et al., Air-lift pumps characteristics under two-phase flow conditions, Int. J. Heat Fluid Flow

(2008), doi:10.1016/j.ijheatfluidflow.2008.09.002

Page 8

where ALis the area for the liquid phase, and Agis the area for the

gas phase.

A ¼ ALþ Ag

Qg¼ AgVg

QL¼ AlVL¼ AV1

Substituting (13)–(15) in (12) and neglecting the density of gas with

respect to the liquid density, we obtain:

ð13Þ

ð14Þ

ð15Þ

W ¼ L

qLA

1 þ

Qg

sQL

??

ð16Þ

where s is the slip ratio which equals:

s ¼Vg

VL

ð17Þ

where Vg and VL are the actual velocities of gas and liquid,

respectively.

Substituting from Eqs. (11) and (16) in Eq. (10), we get:

P2¼ Paþ4fL

DqLV2

11 þQg

QL

??

þqL

L

1 þ

Qg

sQL

??

ð18Þ

This equation was obtained by Stenning and Martin (1968) and it

can be written as follows

?

Hs

L?

1

1 þ

Qg

sQL

?? ¼V2

1

2gL

ðK þ 1Þ þ ðK þ 2ÞQg

QL

?

ð19Þ

where K is the friction factor which is given by

K ¼4fL

Stenning and Martin (1968) used the above equations in their

analytical model but they fixed the values of the slip ratio (s),

and the friction factor (K). Physically, the slip ratio changes as

the water mass flow rate and air mass flow rate changes. Also,

the friction factor changes with changing the flow conditions.

In the present work, the slip ratio is considered as a function of

the water and air mass flow rates as expressed by Griffith and Wal-

lis (1961) for slug flow in the form:

ffiffiffiffiffiffi

Also, the friction factor is obtained using Colebrook equation as

listed by Haaland (1983), where the friction factor, f, may be ob-

tained by solving the following equation:

D

ð20Þ

s ¼ 1:2 þ 0:2Qg

QL

þ0:35

gD

p

V1

ð21Þ

1ffiffiffi

where e is the pipe roughness and Re is the Reynolds number.

Finally, the pump efficiency is calculated in the present study

using the definition given by Nicklin (1963) as follows:

g ¼qgQLðL ? HsÞ

PaQalnpin

Pa

f

p ¼ ?2:0loge=D

3:7þ2:51

Re

ffiffiffi

f

p

!

ð22Þ

ð23Þ

where QLis the water discharge, Qais the volumetric flow rate of air,

Pin is the injection pressure of air, and Pa is the atmospheric

pressure.

The modified model is obtained by inserting Eqs. (21) and (22)

in Eq. (19). Using this modified model, a computer program was

developed in order to investigate the air-lift pump performance

over an extended range of the pump operation. A calculation pro-

cedure to obtain the results using the proposed model as follows:

(1) The geometrical parameters of L, D, pipe roughness e, and

the water density q and viscosity l, are known. Then for a

known air inlet pressure the inlet air mass flow rate is

assigned.

(2) Select a static head Hs for a certain submergence ratio.

(3) Assume a value of water mass flow rate.

(4) Compute the coefficient of the friction f from Colebrook Eq.

(22), also calculate the slip ratio ‘‘s” from Eq. (21).

(5) Calculate the value of friction factor ‘‘K” from Eq. (20).

(6) Calculate the value of the left hand side and the right hand

side of Eq. (19).

(7) Repeat steps 3–6 until the total difference between the left

hand side and the right hand side of (19) becomes less than

0.001.

5. Comparison with the experimental data

In order to evaluate the validity of the results obtained using the

proposed model a comparison with the experimental results of

12840

Air mass flow rate (kg/hr)

0

500

1000

1500

Water mass flow rate (kg/hr)

Submergence Ratio = o.4

Clark & Dabolt (1986)

Experimental data

slug flow

Fig. 9. Comparison between the results of Clark and Dabolt (1986) and the

experimental results of Kassab et al. (2001).

(1)

(2)

Water

AirAir

Hs

L

Disch

Free Surface

Air Jacket

Fig. 10. Model for analysis of airlift pump.

8

S.Z. Kassab et al./International Journal of Heat and Fluid Flow xxx (2008) xxx–xxx

ARTICLE IN PRESS

Please cite this article in press as: Kassab, S.Z. et al., Air-lift pumps characteristics under two-phase flow conditions, Int. J. Heat Fluid Flow

(2008), doi:10.1016/j.ijheatfluidflow.2008.09.002

Page 9

Kassab et al. (2001) was performed. This comparison is shown in

Fig. 11a for a submergence ratio of 0.484. It is clear from this figure

that the agreement between the results of the proposed model and

the experimental data is good over a range of air mass flow rates

up to 8.3 kg/h while there is a large deviation for the remainder

of the pump performance curve. This is due to the transition to

the annular flow regime that is not taken into account in the pres-

ent model.

More comparisons between the results of the proposed model

and the experimental results obtained by Kassab et al. (2001), for

some other submergence ratios, are shown in Figs. 11b–d. In gen-

eral, good agreement is obtained for all presented submergence ra-

tios. In addition, comparing the results presented in Fig. 9 with

those in Fig. 11, one notices that the agreement of the present

modified model with the experimental data is better than that of

Clark and Dabolt (1986).

The average deviation based on root mean square values be-

tween the results of the present model and the experimental re-

sults is about 15%, which is acceptable if the model is used to

investigate the performance in practical applications. It appears

that the one-dimensional theory forms a good basis for the perfor-

mance analysis of air-lift pumps.

The influence of the riser pipe length on the airlift performance

was also studied using the modified model. Fig. 12 shows the effect

of the riser pipe length on the air-lift pump performance in both

theoretical and experimental results. It is noted that, the model

proposed in the present study senses well the change of riser pipe

length.

The efficiency of the air-lift pump is computed using the pres-

ent model based on the definition presented by Nicklin (1963).

The obtained results are presented in Fig. 13 and are compared

with the experimental results of Kassab et al. (2001). Good agree-

ment between the theoretical and experimental results is achieved.

0

4

8121620

Air mass flow rate (kg/hr)

(a) Submergence ratio = 0.484

Air mass flow rate (kg/hr)

(b) Submergence ratio = 0.57

0

500

1000

1500

0

500

1000

1500

Submergence ratio=0.484

experimental data

Proposed model

Annular flow regime

12840

Air mass flow rate (kg/hr)

(d) Submergence ratio = 0.74

12840

Air mass flow rate (kg/hr)

(c) Submergence ratio = 0.67

12840

water mass flow rate (kg/hr)

water mass flow rate (kg/hr)

Submergence ratio=0.57

Theortical modle

Experimental data

water mass flow rate (kg/hr)

0

1000

2000

3000

water mass flow rate (kg/hr)

0

1000

2000

3000

Submergence ratio=0.67

Theoritecal model

Experimental data

Submergence ratio=0.74

Theortical model

Experimental data

Fig. 11. Comparisons between the proposed model and the experimental results of Kassab et al. (2001).

048 12

Air mass flow rate (Kg/hr)

0

500

1000

1500

Water mass flow rate (Kg/hr)

Submergence ratio=0.484

Theortical model (L=3.75m)

Theortical model (L=2.75m)

Experimental data (L=3.75m)

Experimental data (L=2.75m)

Fig. 12. Effect of the riser pipe length on the airlift-pump performance.

S.Z. Kassab et al./International Journal of Heat and Fluid Flow xxx (2008) xxx–xxx

9

ARTICLE IN PRESS

Please cite this article in press as: Kassab, S.Z. et al., Air-lift pumps characteristics under two-phase flow conditions, Int. J. Heat Fluid Flow

(2008), doi:10.1016/j.ijheatfluidflow.2008.09.002

Page 10

5.1. Comparison with other models

Fig. 14 shows a comparison between the results of the proposed

model and the corresponding results obtained using Clark and Da-

bolt (1986) as well as Stenning and Martin (1968) models. It is

clear from the comparison that the proposed model gives better

agreement with the experimental data than especially in the best

efficiency range while it is over predicting the experimental data

in the annular flow pattern region. The range of model applicability

is taken as acceptable error between the experimental data and the

model predictions. In the present experimental range (air mass

flow rate between 0 and 12 kg/h), the model found to be predict

the water mass flow rate within 15% based on the root mean

square. This difference found to increase as the flow pattern

changes to annular flow. Also, the proposed model tends to predict

the point where the pump starts to operate much better than the

other two models.

6. Uncertainty analysis

The uncertainty analysis was performed according to the multi-

variate Taylor Series method and summarized in Table 1

"

UR¼

oR

oX1UX1

??2

þ

oR

oX2UX2

??2

þ ??? þ

oR

oXJUXJ

??2

#1=2

ð24Þ

where UXis the uncertainty in the measured variable X. Also, the

uncertainty for a measuring instrument is calculated from:

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

where eois the interpolation error, ecis the instrument error

eo¼ ?1

For example the procedures of calculating the uncertainty of the

measured variable are as follows:

UX¼

e2

oþ e2

c

q

ð25Þ

2? Re solution

ð26Þ

7. Concluding remarks

The following concluding remarks can be deduced from the

present study:

? As the submerged ratio increases, the maximum efficiency of the

pump increases at the same air-flow rate.

? The air-lift pump lifted maximum amount of liquid if operated

in the slug or slug-churn regimes.

? The maximum efficiency does not occur at the maximum water

mass flow rate.

? The best efficiency points always located in the slug or slug-

churn flow patterns.

? For the same submergence ratio, varying the length of the riser

pipe affects the air-lift pump performance.

? The one-dimensional model proposed in the present study can

predict the air-lift pump performance and it can be used in

the design of air-lift pumps for different flow patterns including

bubbly, slug and churn flow patterns.

? The proposed model gives a good agreement with the experi-

mental results within the practical range of operation of the

air-lift pumps.

References

Bergeles, A.E., 1949. Flow of gas–liquid mixture. Chem. Eng., 104–106.

Clark, N.N., Dabolt, R.J., 1986. A general design equation for air-lift pumps operating

in slug flow. AICHE J. 32, 56–64.

DeCachard, F., Delhaye, J.M., 1996. A slug-churn flow model for small diameter air

lift pumps. Int. J. Multiphase Flow 22, 627–649.

Furukawa, T., Fukano, T., 2001. Effects of liquid viscosity on flow patterns in

vertical upward gas–liquid two-phase flow. Int. J. Multiphase Flow 27, 1109–

1126.

Griffith, P., Wallis, G.B., 1961. Two-phase slug flow. J. Heat Transfer, Trans. ASME 83,

307–320.

Haaland, S.E., 1983. Simple and explicit formulas for the friction factor in turbulent

flow. J. Fluids Eng. 103, 89–90.

Hitoshi, F., Satoshi, O., Hirohiko, T., 2003. Operation performance of a small air-

lift pump for conveying solid particles. J. Energy Resour. Technol. 125, 17–

25.

Hjalmars, S., 1973. The origin of instability in air lift pumps. J. Appl. Mech. Trans.

ASME 95, 399–404.

Iguchi, M., Terauchi, Y., 2001. Boundaries among bubbly and slug flow regimes in

air–water two-phase flows in vertical pipe of poor wettability. Int. J. Multiphase

Flow 27, 729–735.

Kassab, S.Z., Kandil, H.A., Warda, H.A., Ahmed, W.H., 2001. Performance of an air lift

pump operating in two-phase flow. In: Proceedings of ICFDP7: The Seventh

International Congress on Fluid Dynamics & Propulsion, December 18–20, 2001,

Cairo, Egypt, Paper No. ICFDP7-2001004.

04812

Air mass flow rate (kg/hr)

0

20

40

Submergence ratio=0.48

Experime t n al data

Theoretical model

%

Fig. 13. Comparison between the calculated efficiency and that obtained experi-

mentally by Kassab et al. (2001).

2 1840

Air mass flow rate (kg/hr)

0

500

1000

1500

Water mass flow rate (kg/hr)

Submergence Ratio= 0.4

Experiment al data

Clar k&Dabolt(1986)

Stenning&Mar tin(1968)

Propose d model

Fig. 14. Comparison between the results of various models and the experimental

results of Kassab et al. (2001).

Table 1

Uncertainty of the measured variables

QuantityPercentage uncertainty (%)

Air mass flow rate

Water mass flow rate

Pressure

Temperature

2.5

4

3

2.5

10

S.Z. Kassab et al./International Journal of Heat and Fluid Flow xxx (2008) xxx–xxx

ARTICLE IN PRESS

Please cite this article in press as: Kassab, S.Z. et al., Air-lift pumps characteristics under two-phase flow conditions, Int. J. Heat Fluid Flow

(2008), doi:10.1016/j.ijheatfluidflow.2008.09.002

Page 11

Kassab, S.Z., Kandil, H.A., Warda, H.A., Ahmed, W.H., 2007. Experimental and

analytical investigations of airlift pumps operating in three-phase flow. Chem.

Eng. J. 131, 273–281.

Khalil, M.F., Mansour, H., 1990. Improvement of the performance of an air lift pump

by means of surfactants, Presented at the Sixth International Symposium of

Heat and Mass Transfer, Miami, FL, USA.

Nicklin, D.J., 1963. The air lift pump theory and optimization. Trans. Inst. Chem. Eng.

41, 29–39.

Reinemann, D.J., Patrlange, J.Y., Timmons, M.B., 1986. Theory of small diameter air

lift pump. Int. J. Multiphase Flow 16, 337–355.

Sekoguchi, K., Matsumura, K., Fukano, T., 1981. Characteristics of flow fluctuation in

natural circulation air lift pum. JSME 14, 1960–1966.

Sharma, N.D., Sachdeva, M.M., 1976. An air lift pump performance study. AICHE J.

32, 61–64.

Stapanoff, A.J., 1929. Thermodynamic theory of air-lift pump. ASME 51, 49–55.

Stenning, A.H., Martin, C.B., 1968. An analytical and experimental study of air lift

pump performance. J. Eng. Power, Trans. ASME 90, 106–110.

Storch. B., 1975. Extraction of sludges by pneumatic pumping. In: Second

Symposium on Jet Pumps and Ejectors and Gas Lift Techniques, Churchill

College, Cambridge, England, G4, pp. 51–60.

Taitel, Y., Bornea, D., Dukler, A.E., 1980. Modeling flow pattern transition for steady

upward gas–liquid flow in vertical tubes. AICHE J 26, 345–354.

S.Z. Kassab et al./International Journal of Heat and Fluid Flow xxx (2008) xxx–xxx

11

ARTICLE IN PRESS

Please cite this article in press as: Kassab, S.Z. et al., Air-lift pumps characteristics under two-phase flow conditions, Int. J. Heat Fluid Flow

(2008), doi:10.1016/j.ijheatfluidflow.2008.09.002