Experimental Thermal and Fluid Science 39 (2012) 79–89

Contents lists available at SciVerse ScienceDirect

Experimental Thermal and Fluid Science

journal homepage: www.elsevier .com/locate /et fs

Experimental studies on bubble characteristics for R134a–DMF bubble absorber

M. Suresh, A. Mani ⇑Refrigeration & Airconditioning Laboratory, Department of Mechanical Engineering, Indian Institute of Technology Madras, India

a r t i c l e i n f o

Article history:Received 15 December 2010Received in revised form 8 January 2012Accepted 9 January 2012Available online 16 January 2012

Keywords:AbsorptionR134a–DMFBubble absorberBubble characteristicsCorrelation

0894-1777/$ - see front matter � 2012 Elsevier Inc. Adoi:10.1016/j.expthermflusci.2012.01.011

⇑ Corresponding author. Tel.: +91 44 22574666; faxE-mail address: mania@iitm.ac.in (A. Mani).

a b s t r a c t

Experimental investigations have been carried out to visualize bubble behavior and effect of gas flow rateand liquid concentration on bubble characteristics of Tetrafluoro ethane (R134a) in liquid R134a–Dimethyl Formamide (DMF) solution in a glass absorber. Bubble behavior was studied in still as wellas flowing solution. At lower gas flow rates, bubbles tend to be spherical due to the dominance of surfacetension. At higher gas flow rates, bubbles tend to be hemispherical due to the dominance of inertia force.Bubble diameter during detachment increases with increase in gas flow rates. Also, the bubble diameterinitially increases as the solution concentration increases up to 0.075 kg kg�1 and then starts decreasingas the solution concentration increases further. At lower solution concentrations, effect of absorptiondriving potential (xeq � xl) is dominant than that of surface tension, due to this bubble diameter increases.When the solution concentration is increased further, effect of surface tension is dominant resulting indecrease in bubble diameter. Bubble diameter measured from the experiments is compared with thatof the numerical model developed by the authors and other literature experimental correlations and itis found that the agreement is good. A correlation for bubble diameter during detachment is presentedfrom the experimental studies, which can be useful for the design of R134a–DMF bubble absorber.

� 2012 Elsevier Inc. All rights reserved.

1. Introduction

Many environment friendly fluid combinations have been sug-gested by number of investigators in order to overcome some ofthe limitations of well known working pairs viz., ammonia–waterand lithium bromide–water for the vapor absorption refrigerationsystems (VARSs). Though R22-organic solvent based absorptionrefrigeration systems have been extensively studied by Fatouh[1], Karthikeyan et al. [2] and Sujatha et al. [3,11–13], HydroChlo-roFluoro Carbons (HCFCs) along with ChloroFluoro Carbons (CFCs),are also covered by Montreal and other International Protocols andare being phased out. So environment friendly R134a based VARSare being investigated. Nezu et al. [4] examined the possibility oftesting R134a as a refrigerant in VARS with various organic sol-vents and showed that the R134a–Dimethyl Acetamide (DMA)and the R134a–Dimethyl Formamide (DMF) systems are consid-ered attractive as the working-fluid pairs for the absorption refrig-eration system than other R134a-absorbent systems. Yokozeki [5]studied theoretical performance of various refrigerant–absorbentpairs in a VARS by the use of equations of state. Of these, R134a–DMF and R134a–DMA systems exhibit better performance, com-pared to other R134a-absorbent systems. Also the circulation ratiois less and COP is more for R134a–DMF system compared toR134a–DMA system. Mani [6] carried out experimental studies

ll rights reserved.

: +91 44 22570509.

on R134a–DMF based compact vapor absorption refrigeration sys-tem with plate heat exchangers and reported that this systemcould be very competitive for applications ranging from �10 �Cto 10 �C, with heat source temperature in the range of 80–90 �Cand with cooling water as coolant for absorber and condenser inthe temperature range of 20–35 �C.

Absorber is considered as one of the crucial components in thevapor absorption refrigeration system. Kang et al. [7] carried outanalytical investigation of falling film and bubble type absorbersand found that absorption rate of the bubble type absorber wasfound to be always higher than that of the falling film mode. Bub-ble type absorber provides better heat and mass transfer coeffi-cients, also good wettability and mixing between the liquid andthe vapor [7].

Absorption process is characterized by simultaneous heat andmass transfer phenomena. These mechanisms, though compli-cated, influence the system performance significantly. Elperinand Fominyk [8] studied combined heat and mass transfer mecha-nisms at all stages of bubble growth and rise in a bubble absorber,which can be useful in design calculations of gas–liquid absorbers.Lee et al. [9] performed both the numerical and experimental anal-yses in the absorption process of a bubble absorber. Numericalmodel in these studies can be used for the optimum design ofabsorber. Merrill and Perez-Blanco [10] developed an analyticalmodel to predict bubble dynamics in binary sub-cooled solutions.This model improves the understanding of bubble absorptiondynamics.

Nomenclature

A nozzle cross-section area (m2)Bu Buoyancy numberCv specific heat at constant volume (kJ kg�1 K�1)Cp specific heat at constant pressure (kJ kg�1 K�1)C local thermal capability of binary mixture (kJ kmol�1)d diameter (m)Dx mass diffusion coefficient (m2 s�1)Fr Froude numberg gravitational acceleration (m s�2)h specific enthalpy (kJ kg�1, kJ kmol�1)jr reduced mass transfer rateK thermal conductivity (W m�1 K�1)m gas mass flow rate (kg s�1)n number of gas injection nozzlesp pressure (bar, kPa, N m�2)r radial coordinate (m)R bubble radius (m)Re Reynolds numbert time (s)T temperature (K)v velocity (m s�1)V volume (m3)We Weber number

x liquid molar fraction (kmol kmol�1)X liquid mass fraction (kg kg�1)y gas molar fraction (kmol kmol�1)

Greek symbolsl dynamic viscosity (Pa s)r surface tension (N m�1)q density (kg m�3)d increment1 absorber boundary surface

Subscriptsb bubbleeq equilibriumE equivalentg gash headerl liquidn nozzleo orificeo,l liquid based on orifice diametero,v vapor based on orifice diameterv vapor

80 M. Suresh, A. Mani / Experimental Thermal and Fluid Science 39 (2012) 79–89

Sujatha et al. [11,12] carried out numerical analysis in a verticaltubular bubble absorber working with R22 as refrigerant and fiveorganic fluids namely Dimethyl Formamide (DMF), Dimethyl Acet-amide (DMA), Dimethyl Ether of Tetraethylene Glycol (DMETEG),Dimethyl Ether of Diethylene Glycol (DMEDEG) and N-Methyl Pyr-rolidone (NMP) as absorbents. This model is validated by compar-ing with the results available in the literature. Based on theseresults, a correlation for mass transfer coefficient has beensuggested for the vertical tubular bubble absorber. Sujatha et al.[13] have also carried out experimental studies on a vertical tubu-lar bubble absorber working with R22–DMF. Experimental pres-sure drop, heat transfer coefficient and mass transfer coefficientare compared with the results obtained from the numerical model.

Kang et al. [14] developed a model for a bubble absorber with aplate type heat exchanger by considering the combined heat andmass transfer analysis in both liquid and vapor regions. All geomet-ric variables such as distance between two plates, number ofplates, and width of the plates could be selected optimally for gi-ven thermal conditions by the developed design model for ammo-nia–water combination.

Staicovici [15–17] used non-equilibrium phenomenologicaltheory to evaluate gas–liquid interaction.

The design of bubble absorber, based on non-equilibrium ther-modynamics could be suited to a modern compact plate type con-struction and offer better absorption efficiency and minimumpressure loss on the gas side. Suresh and Mani [18] developed anumerical model on bubble dynamics, heat and mass characteris-tics of R134a/DMF based bubble absorber using phenomenologicaltheory and validated by comparing with the results available inliterature.

Kang et al. [19] developed an experimental correlation of masstransfer coefficient for ammonia–water bubble absorption. Kanget al. [20] also developed a correlation for initial bubble diameter,which can be used to calculate the interfacial area in the design ofammonia–water bubble absorber.

Present experimental work is carried out to study bubblebehavior and effect of gas flow rate and liquid concentration on

bubble characteristics for R134a–DMF bubble absorber and com-pare with numerical model developed by authors [18].

2. Bubble dynamics

2.1. Effect of gas flow, nozzle size and shape on bubble characteristics

Many studies have been done and reported in the literatureabout bubble formation from a nozzle. The effect of nozzle diame-ter, gas flow rate and liquid properties such as surface tension, li-quid viscosity, liquid and vapor density on bubble formation,growth and detachment have been considered. Number of investi-gators have developed models and correlations to determine thebubble diameter. Kumar and Kuloor [21] developed theoretical ap-proaches to observe the process of bubble formation. They re-ported that bubble forms in two stages. The first stage comprisesof pure expansion when the bubble base remains attached to thenozzle tip. The various forces taken into account in this stage arebuoyancy, surface tension, inertia and viscous drag. The first stageends when the upward buoyancy becomes equal to the sum of thedownward forces. The end of the first stage marks the beginning ofthe second stage. During this stage the upward buoyancy forceexceeds the sum of the downward forces. As a result, the base ofthe expanding bubble moves away from the nozzle tip. The bubbleis assumed to detach when the base of the bubble has moved a dis-tance equal to the radius of the bubble at the end of the first stage.Chakraborty et al. [22] developed a numerical simulation to ana-lyse the problem of bubble formation in quiescent and co-flowingliquids. The investigation showed that the bubble formationprocess is divided into two stages, namely the expansion stage, fol-lowed by the collapse stage, starting with the formation of a neckat the orifice tip. For bubble formation in a co-flowing liquid, thebubbles are approximately spherical only in the early stages of for-mation, eventually becoming non-spherical prior to bubble detach-ment. The dynamics of the bubble breakup process are faster in aco-flowing liquid than that in a quiescent liquid and thus create

R134a vapour in

Cooling water out

R134a-DMF liquid solution in

Cooling water in

R134a-DMF liquid solution out

Bubble

Fig. 1. Schematic diagram representing bubble absorber model.

M. Suresh, A. Mani / Experimental Thermal and Fluid Science 39 (2012) 79–89 81

a smaller bubble size. The liquid-to-gas mean velocity ratio con-trols the bubble size and bubbling time. When this ratio increases,the detached bubble volume and the bubble formation time de-crease. Terasaka and Tsuge [23] conducted experiments and mea-sured bubble volumes and shapes formed from a constant-flownozzle submerged in liquids. They also developed a non-sphericalbubble formation model under constant-flow conditions and thecalculated volumes and shapes of bubbles agreed with the experi-mental results. Terasaka et al. [24] also experimentally investi-gated the effects of liquid velocity, nozzle diameter, gas chamberand gas flow rate on the volumes, shapes and growth rates of bub-bles formed at a nozzle submerged in a co-flowing liquid. Tsugeet al. [25] studied the effect of system pressure, orifice diameter,gas flow rate and gas chamber volume on the volumes and shapesof bubbles formed at a single orifice under highly pressurized con-ditions and reported that bubble volume increased with decreasingsystem pressure, increasing nozzle diameter and increasing gasflow rate.

Akita and Yoshida [26] measured bubble size by using photo-graphic method for fluid pairs of water–air, water–oxygen, gly-col–air and methanol–air and presented that the only factorsaffecting the initial bubble size were the orifice diameter and gasvelocity flowing through the orifice. They also reported that thebubble size was independent of the properties such as surface ten-sion, liquid viscosity, liquid and vapor density. Kumar et al. [27]and Bhavaraju et al. [28] proposed correlations of bubble size asfunctions of orifice diameter. Bhavaraju et al. reported that bubblediameter was dependent on gas flow rate, but it was essentiallyindependent of orifice diameter even though they proposed corre-lations of bubble size as functions of orifice diameter. Kang et al.[20] reported that bubble diameter is significantly influenced byorifice diameter for ammonia–water bubble absorption. Theyfound from the experiments that the departing bubbles becamehemispherical as the vapor velocity increased. Shape and size ofa nozzle influence the bubble diameter during detachment. Tsugeet al. [29,30] investigated the effects of nozzle shape and operatingparameters on bubble formation mechanism from the verticallydownward nozzle and reported that the bubble size is influencedby the nozzle tip edge angle, nozzle diameter and gas flow rate.At the same gas flow rate, the bubble size decreases with decreas-ing edge angle. Also, as the edge angle decreases, the bubbles tendto more easily detach from the slanted direction of nozzle tip.When smooth tip nozzle is used, bubble diameter during detach-ment is larger than that from sharp tip and smaller than that fromblunt tip. However, bubble detaches easily from blunt tip andsharp tip nozzles compared to smooth tip nozzle.

3. Numerical model

The authors [18] developed a numerical model to study thebubble dynamics and absorption of the R134a gas bubble in liquidR134a–Dimethyl Formamide (DMF) solution by applying phenom-enological theory of heat and mass transfer. Equations, initial aswell as boundary conditions, solution and some results are brieflydiscussed in this section.

3.1. Model description

As shown in Fig. 1, R134a vapor coming from the evaporator ofthe vapor absorption refrigeration system is allowed to passthrough a header, at a flow rate of mh. Vapor in the header haspressure pg,h, temperature Tg and volume Vh. From the header,the vapor is distributed and injected into R134a–DMF liquid solu-tion at the flow rate of mb through nozzles (diameter d) andabsorbed. Pressure, temperature and mole fraction of the vapor in-

jected into the liquid are pg, Tg and y respectively. Pressure, tem-perature and mole fraction of the solution at the gas–liquidinterface and at the absorber boundary surface are respectivelypl, Tl and x and p1, T1 and x1.

3.2. Model equations

Problem is formulated from the dynamics of a single bubble,namely a spherical symmetry flow and uniformity of gas pressurein the bubble. For a spherically symmetric extension of a gas bub-ble in an infinite viscous fluid and for homobaric approximation(gas pressure is uniform in the bubble), only the motion of outerliquid is of interest. Moreover, the following assumptions are con-sidered for the liquid and gas medium: (a) the liquid gravitationalforces are neglected; liquid is Newtonian, viscous and incompress-ible fluid; (b) the refrigerant vapor in the header and bubble is con-sidered as an ideal gas; as the bubble evolves the temperature ofrefrigerant vapor in the header and bubble remains constant anddoes not involve in the heat transfer process; the gas–liquid inter-action is an evolutional phenomenon, deemed to be an opensystem.

In the liquid phase, continuity, Navier–Stokes, heat and masstransfer equations are considered. In the gas phase, First law of ther-modynamics and Bernoulli’s equation are considered. In the gas–li-quid interface, equations of motion, energy equation, gas massbalance and liquid mass balance equations are considered. The fol-lowing equations are formulated and solved for modeling the bubbleabsorption.

-10

0

10

20

30

40

50

60

70

80

90

100

1.010.0100.01000.0

Bubble growth time, s

B

ubbl

e ra

dius

, mm

Rat

e of

cha

nge

of b

ubbl

e ra

dius

, x 1

0-1

, mm

s-1

-10

0

10

20

30

40

50

60

70

80

90

100

Seco

nd d

eriv

ativ

e of

bub

ble

radi

us x

10

-1, m

s-2

Bubble radiusRate of change of bubble radiusSecond derivative of bubble radius

Gas inlet temperature = 0°CGas mass fraction = 0.99 Solution inlet temperature = 35°CSolution pressure = 291.3 kPaSolution initial concentration = 0.2 kgkg-1

Fig. 2. Bubble dynamics of vapor during absorption process [18].

82 M. Suresh, A. Mani / Experimental Thermal and Fluid Science 39 (2012) 79–89

@2R@t2 ¼ pg;h �

m2b

A2qg

� p1 �2rR

!1qlþ @R

@t

� �2

j2rql

qg� 1

!" #1

Rð1� jrÞ

ð1Þ

where jr ¼pg

qlcv;g Tg, mb ¼ A

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiqg jpg;h � pg j

q@pg;h

@t¼ mh � nmb½ � hg

Vhð2Þ

@x@tþ @R@t

Rr

� �2@x@r¼ 1

r2

@

@rDxr2 @x

@r

� �ð3Þ

qlCpl@Tl

@tþ @R@t

Rr

� �2@Tl

@r

!¼ 1

r2

@

@rKr2 @Tl

@r

� �ð4Þ

Eqs. (1) and (2) are solved to determine the bubble radius R, rate

of change in bubble radius@R@t

, second derivative of bubble radius

@2R@t2 , gas pressure in the header pg,h. Eqs. (3) and (4) are solved to

find the distribution of liquid concentration and temperature frombubble interface, along the radial direction.

3.3. Initial and boundary conditions

To complete the model formulation, initial and boundary condi-tions must be specified. The initial conditions are adopted asfollows:

At t ¼ 0; r P R :

Rð0Þ ¼ R0;@R@tð0Þ ¼ v r¼Rð0Þ ¼ 0; Tlð0; rÞ ¼ T1; xð0; rÞ ¼ x1

pg;hð0Þ ¼ p1 þ2rR0þ Dpg;t0

; pgð0Þ ¼ p1 þ2rR0

; plð0; rÞ ¼ p1

The boundary conditions are specified at the interface and farfrom the interface or absorber boundary surface.

At interface; t > 0; r ¼ R :

dx ¼ ðy� xÞjr þ Dx

@x@r@R@t

�� �� ; dTl ¼1

cp;ljrC þ

K @Tl@r

ql@R@t

�� �� !

At absorber boundary surface; t P 0 :

pl ¼ p1; x ¼ x1; Tl ¼ T1

When the absorber is provided with a cooling system, coolant tem-perature can be chosen at the surface.

3.4. Numerical solution

The non-linear ordinary differential Eqs. (1) and (2) are solvedby means of initial value problem solvers using explicit Runge–Kutta method with 4th order accuracy. In this method, the bubbleparameters are computed at a particular time step using the solu-tion arrived at the immediately preceding time step as an input.The partial differential Eqs. (3) and (4) are solved by using finitedifference methods with fully implicit technique. The time deriva-tive is approximated by forward differencing and the spatialderivatives are approximated by central differencing. First-orderaccuracy in time and second-order accuracy in space are achieved.Fully implicit techniques are unconditionally stable, without anyconstraints on choices of time step or grid spacing. An adaptive

computational grid must be employed to handle the movingliquid–vapor interface.

The necessary properties for R134a–DMF solution namelyliquid density, surface tension, diffusion coefficient, thermal con-ductivity, specific heat, etc., have been calculated as functions ofliquid temperature, at each time step, using experimental correla-tions from [31] with 3–5% error. The liquid enthalpy has been cal-culated from the equations of Nezu et al. [4]. The density andenthalpy of R134a gas have been calculated from the equationsof Yokozeki [5].

Then the model is simulated for R134a/DMF. The followingrange of input data is used for the model analysis.

Absorption pressure

163–412 kPa Gas mass flow rate 0.91–2.11 g s�1

Gas inlet temperature

�10 to 10 �C Gas mole fraction 0.99 Liquid inlet temperature 17–60 �C Initial liquid mass fraction 0.2–0.4 kg kg�1

Fig. 2 shows the numerical solution for the bubble dynamicsduring absorption. The bubble radius continuously increases tillthe bubble interface reaches equilibrium conditions and then,replaced by next new bubble.

The actual mass transfer rate is calculated as

jrql@V@t

� �ð5Þ

The coupled heat transfer rate is calculated as

Cjrql@V@t

� �ð6Þ

In Fig. 3, bubble gas pressure and header gas pressure continu-ously increase, since bubble feed rate is higher than the actualabsorption rate. When heat and mass transfer rates slow down,both pressures start decreasing.

4. Experimental setup

Fig. 4 shows the schematic diagram of experimental setup. Thisconsists of a glass bubble absorber, strong and weak solution tanks,solution pump, cooling water simulator, instrumentation andvalves. Tables 1 and 2 show the specifications of main componentsand instruments in the setup, respectively. Glass absorber consistsof two concentric tubes. DMF solution is pumped from weak solu-tion tank, through the bottom of inner tube by a solution pump.

Solutionpump

Weak solution

tank

Solution charging

line

Strong solution

tank

R134a gas collection tank

NRV1

L

L P

Online density meter

S1

Ice box

NV1

HSV2

HSV3

HSV4

HSV6 HSV5

Gas charging

lineHSV7

HSV

HSV10 HSV

HSV8

Computer with Data Acquisition syst

R134a refrigerant line

R134a-DMF Solution line

Water line

Hand shut-off

Gate valve

Needle valve

S

L

P

T

Pressure gauge

Temperature g

Level gauge

Flow meter

Non-Return va

Mass flow conM

Fig. 4. Schematic diagram of glass bub

290

295

300

305

310

315

320

325

330

335

340

345

1.010.0100.01000.0

Bubble growth time, s

Bub

ble

pres

sure

, kP

a

290

295

300

305

310

315

320

325

330

335

340

345

Hea

der

pres

sure

, kP

a

Bubble pressureHeader pressure

Gas inlet temperature = 0°°CGas mass fraction = 0.99Solution inlet temperature = 35°°CSolution pressure = 291.3 kPaSolution initial concentration = 0.2 kgkg-1

Fig. 3. Variation of header and bubble pressure during absorption process [18].

M. Suresh, A. Mani / Experimental Thermal and Fluid Science 39 (2012) 79–89 83

R134a gas is supplied from a storage cylinder through a mass flowcontroller unit and injected through a nozzle installed at the bot-tom of inner tube. Strong DMF solution is collected in the strongsolution tank at the top of the absorber. Cooling water is supplied,by cooling water simulator, through the absorber annulus in coun-ter flow direction to the solution and gas. Cooling water simulatorconsists of a R22 based vapor compression refrigeration (VCR) cir-cuit of 3.4 TR capacity, a cooling water tank insulated with ex-panded polyethylene (EPE) sheets, electric heaters, pump, flowmeter, PT100 sensor, PID temperature controller, contactor, pipingand valves. VCR circuit consists of a hermetically sealed reciprocat-ing compressor, an air cooled condenser, a thermostatic expansionvalve and a cooling coil.

This system also includes various measuring devices such astemperature sensors, pressure sensors, flow meters, mass flowcontroller and online density meter fitted at suitable locations as

2 T2

MP1 T1 Pressure

regulatingvalve

T9

P3

T3T4

T5

T6

T7

T8

T11

T10

T12

HSV1

9

11

R134a gascylinderem

High speed camera

S2

GV1

GV2

Waterpump

Video camerarecorder

Glassabsorber

valve

auge

lve

troller unit

Cooling water simulator

ble absorber experimental setup.

Table 1Specifications of main components.

Glass absorberType Tube in tubeFluid circuit Tube side solutionAnnulus Cooling waterMaterial GlassInside tube ID: 33 mm; OD: 37 mmOutside tube ID: 46 mm; OD: 50 mmTube length 1000 mm

NozzleType TubeMaterial CopperInner diameter 2.2 mmOuter diameter 4.0 mm

Solution pumpType Metering pumpCapacity 50 lphDischarge pressure 40 barPlunger diameter 25 mmStrokes per minute 100

Table 2Specifications of main instruments.

Online density meterTube inner diameter 6.6 mmTube length (inlet to outlet) 500 mmMeasuring range 0–3 g/m3

Measuring repeatability 5 � 10�5 g/m3

Accuracy in the adjusted range 2 � 10�4 g/m3, 0.1%Temperature range 0–100 �CPressure range 0–50 barFlow rate (water) 100–500 l/h

Mass flow controllerFlow rate 10–2500 ccpmResponse time <2 s within ±2% of set pointAccuracy Within 0.05% of full scaleLinearity Within 0.5% of full scaleRepeatability Within 0.2% of full scaleControl valve type Normally closed solenoidInput power +15 V DC at 25 mA; �15 V DC at 180 mAPower consumption 31 W (Max.)Input command signal 0–5 V DCOutput indication 0–5 V DC

Pressure transducersType PiezoresistiveMeasuring range 0–10 bar absoluteTemperature �20 to 80 �CInput 8–28 V DCOutput 0–5 V DC (3 wires)Accuracy ±1% of full scaleResponse time Maximum:250 msMaterial SS316L

Flow metersType Glass rotameterFlow rate Minimum:0.2 lpm, maximum: 2.0 lpmOperating pressure 4 bar (abs.)End fitting material SS316Float material SS316Tube material Borosilicate glassTube taper length 250 mmAccuracy ±1% on F.S.D

CameraModel Nikon D90Color representation sRGBFocal length 85 mm, 105 mmF-Number F/5.3, F/5.6, F/11, F/22Exposure time 1/50 s to 1/800 sMetering mode PatternExposure compensation 0–5 step

84 M. Suresh, A. Mani / Experimental Thermal and Fluid Science 39 (2012) 79–89

shown in Fig. 4. All these measuring instruments are pre-cali-brated. Twelve numbers of calibrated copper-constantan thermo-couples are used as temperature sensors with a measurement

uncertainty up to ±0.5 �C. Three numbers of piezo-electric typepressure transducers are used as pressure sensors with a measure-ment uncertainty up to ±1.2%. Glass rotameters are used tomeasure the flow of solution and cooling water with a measure-ment uncertainty up to ±2.5%. Mass flow controller unit is usedto measure the volume flow rate of gas with a measurement uncer-tainty of ±1%. An online density meter is used to measure the den-sity of strong and weak solutions with a measurement uncertaintyof ±0.1%. Concentrations of strong and weak solutions are evalu-ated from the measured density values using HBT (Hankinson–Bro-bst–Thomson) equation used by Reid and others [31]. Readingsfrom all these instruments and sensors are monitored continuouslyby connecting them to a data acquisition system and a computer.

5. Experimental procedure

Initially DMF solution is charged into the weak solution tank.Then it is pumped through inner tube of glass absorber by solutionpump. Solution is collected in the strong solution tank at the top ofthe absorber. Solution is returned to the weak solution tankthrough a needle valve after each set of experimentation. Initialconcentration of the solution at the absorber inlet is measuredby online density meter. The pump is switched off after some timeand solution remains still in the inner tube of absorber. Solution in-let pressure and temperature remain constant at 1.1 bar and 30 �C,respectively. Then R134a gas is injected from a storage cylinderthrough the nozzle at the bottom of the absorber. Gas flow rateis measured by mass flow controller unit. During this process, a vi-deo camera recorder and a high speed still camera with shutterspeed variations up to 1/800 s are used to visualize the bubblebehavior. After photographic study, solution is pumped to strongsolution tank and is not allowed into the weak solution tank byclosing the needle valve. Absorber is filled with fresh solution fromweak solution tank where initial concentration is maintained. Thisprocess is repeated for various gas flow rates and photographicstudy is carried out. After completion of experimentation, the solu-tion in the strong solution tank is transferred into the weak solu-tion tank by opening the needle valve. Then solution initialconcentration is increased by injecting required quantity ofR134a gas through a charging line in weak solution tank andmonitored by online density meter. Then, gas injection process tothe absorber and photographic study are repeated for various gasflow rates and various initial solution concentrations. Entire pro-cess mentioned above, is carried out in still solution in the absor-ber. The above experimentation is repeated by maintaining thesolution in the absorber with flowing solution using the pumpcontinuously.

Bubble photographs are taken along with reference objects(measured by a calibrated scale) inside and beside the bubble ab-sorber, which are in parallel with the absorber. The bubbles arephotographed from three sides of the glass absorber. However,additional visualization in the direction perpendicular to the cur-rent direction of flow visualization requires the camera to be posi-tioned at the top or bottom of glass bubble absorber. This posespractical difficulties. So, additional flow visualization could notbe carried out, which may lead to occurrence of errors in the esti-mation of bubble diameter. These photos are uploaded andzoomed in Adobe Photoshop CS2 software version 9.0. Bubbleshape either spherical or hemispherical is divided into no. of seg-ments with different radii. Circles are drawn corresponding torespective radii and superimposed on the zoomed-in profile ofbubble image uploaded in Photoshop software.

Equivalent volume of sphere ðVEÞ ¼p6 ðd

31 þ d3

2 þ d33 þ � � � d

3i Þ

ið7Þ

M. Suresh, A. Mani / Experimental Thermal and Fluid Science 39 (2012) 79–89 85

Bubble diameter ðdbÞ ¼6VE

p

� �13

ð8Þ

The uncertainties are related to fixing the center of the actualbubble profile outline and using the measure tool to calculate theradii of different segments. The variation of profile center pointsduring no. of repeated measurements lies with 2–3% accuracy withrespect to maximum bubble radius measured during experimenta-tion. So the maximum uncertainty in the measurement ofminimum bubble radius varies from 4.49% to 6.97%. The abovemethod of estimating bubble diameter is based on the assumptionthat the bubble is a sphere. Hence, while estimating bubble diame-ter, errors can arise due to location of the bubble center as well asany distortion of the bubble surface. Though distortion is an aberra-tion which deforms the image in round pipes, in the present workthe camera was positioned at all three sides in such a way that im-age distortion was minimized with effective use of zoom and main-taining optimum distance between the camera and glass absorber.In addition, diameter of nozzle in the image was also measuredusing Photoshop software and compared with its actual diameter.This proportion was used as calibration scale inside the glass absor-ber, in further reducing the errors in bubble diameter estimation.

(a) Bubble formation

(b) Bubble growth

(c) Bubble detachment Gas flow rate = 0.0006 m3h-1

0.0012 m3h-1

(I) (II)

2.2

4

Fig. 5. Visualization of R134a–DMF bubbl

Image processing techniques have been used to read the colorvalue of pixels in the image, to adjust color balance, brightness,contrast and image sharpness of the available quality of the bubbleimages without any compromise on the accuracy of measurement.

6. Results and discussion

Experimentation was conducted in glass bubble absorber sys-tem by varying the operating parameters viz., gas flow rate from0.0006 m3 h�1 to 0.006 m3 h�1, solution initial concentration from0.01 kg kg�1 to 0.15 kg kg�1. Solution flow rate was maintained at0.0125 m3 h�1 in flowing condition. Solution inlet pressure wasmaintained constant at 1.1 bar in still conditions and 1.2 bar inflowing conditions. Solution temperature was kept constant at30 �C at the inlet.

Fig. 5 shows the visualization of R134a–DMF bubble absorptionprocess at various stages in still solution with an initial concentra-tion equal to 0.01 kg kg�1 at different gas flow rates.

Earlier, Staicovici [16] conducted experiments by injecting asingle gaseous ammonia bubble into the NH3/H2O solution at nor-mal pressure, ambient temperature and small molar fractions.

(a) Gas flume

(b) Gas flume

0.0036 m3h-1 0.03 m3h-1

(III) (IV)

(c) Gas flume

e absorption process in still solution.

86 M. Suresh, A. Mani / Experimental Thermal and Fluid Science 39 (2012) 79–89

Based upon the numerous observations recorded, the author ex-plained that at low and continuous gas mass flow rates (10�5–10�4 g s�1), bubble dynamics takes place in two stages: periodicbubble growth to a maximum volume and collapse to a zero vol-ume without detaching itself from the nozzle. At high and contin-uous gas mass flow rates, bubble collapse does not occur. Instead ahigh frequency of bubble growth from a minimum volume to amaximum volume without detachment from the nozzle occurs.

However, after observing bubble visualization of R134a–DMFbubble absorption, it is noted from Fig. 5 that at lower gas flowrates in the range of 0.0006–0.0036 m3 h�1, the bubble is formedat the tip of the nozzle, it continuously grows to a maximum vol-ume and then it detaches from the nozzle and finally absorbed intothe solution. It does not collapse at the nozzle itself. The detach-ment can be clearly seen in the pictures as in Fig. 5. Whereas athigher flow rates (>0.006 m3 h�1), as mentioned by Staicovici[16], bubble detachment and collapse could not be observed.

Staicovici [16] studied a single bubble which was almost sta-tionary at the tip of the nozzle (almost zero flow rate). So, due tothe lack of inertial force, the bubble reached a maximum volume

(a) Bubble formation

(b) Bubble growth

(c) Bubble detachment Gas flow rate = 0.0006 m3h-1

0.0012 m3h-1

(I) (II)

Fig. 6. Visualization of R134a–DMF bubble

and collapsed at the tip without detaching itself from the nozzle.In the present work, the flow rate is considerable compared to thatof Staicovici’s work. So, due to substantial inertial force, the bubbleis detached from the nozzle and absorbed into the solution withoutcollapsing at the tip.

Fig. 6 shows the visualization of R134a–DMF bubble absorptionprocess at various stages in flowing solution at 0.0125 m3 h�1 withan initial concentration equal to 0.01 kg kg�1 at different gas flowrates. It is observed that the bubble diameter during detachmentis 20–35% more than that formed in still solution. It is also seenthat at lower gas flow rates and in still solution, bubbles tend tobe spherical due to the dominance of surface tension force.Whereas, in flowing solution, the bubbles tend to be hemisphericaldue to the dominance of inertial force, at lower as well as highergas flow rates.

Fig. 7 compares the experimental bubble diameter in still solu-tion with that of numerical model, for various gas flow rates.Agreement is good with a maximum deviation of ±5%. Fig. 8 depictsvariation of bubble diameter during detachment in still solution aswell as flowing solution for various gas flow rates. In both cases,

(a) Gas flume

(b) Gas flume

(c) Gas flume

0.0036 m3h-1 0.015 m3h-1

(III) (IV)

absorption process in flowing solution.

0

1

2

3

4

5

6

7

0 0.001 0.002 0.003 0.004 0.005 0.006 0.007

Bub

ble

diam

eter

dur

ing

deta

chm

ent,

mm

Gas flow rate, m3h-1

Numerical model

Experimental

Gas inlet pressure = 600 kPaGas inlet temperature = 32 oCSolution pressure = 110 kPaSolution temperature = 30 oCSolution concentration = 0.01 kgkg -1

Nozzle diameter = 2.2 mm

5% deviation

Fig. 7. Comparison between experimental and numerical model bubble diameter.

0

4

8

12

16

0 0.001 0.002 0.003 0.004 0.005 0.006 0.007

Bub

ble

diam

eter

dur

ing

deta

chm

ent,

mm

Gas flow rate, m3h-1

Numerical model

Experimental

Gas inlet pressure = 600 kPaGas inlet temperature = 32 oCSolution flow rate = 0.0125 m3h-1

Solution inlet pressure = 110 kPaSolution inlet temperature = 30 oCSolution inlet concentration = 0.01 kgkg-1

Nozzle diameter = 2.2 mm Flowing solution

Still solution

5% deviation

12% deviation

Fig. 8. Effect of gas flow rate on bubble diameter during detachment.

0

2

4

6

8

0 0.05 0.1 0.15 0.2

Bub

ble

diam

eter

dur

ing

deta

chm

ent,

mm

Solution initial concentration, kgkg-1

Numerical model

Experimental

Gas inlet pressure = 600 kPaGas inlet temperature = 32 oCSolution inlet pressure = 110 kPaSolution inlet temperature = 30 oCNozzle diameter = 2.2 mm

Gas flow rate = 0.0048 m3h-1

Gas flow rate = 0.0012m3h-13% deviation

5% deviation

Fig. 9. Effect of solution concentration on bubble diameter during detachment.

0

2

4

6

8

0 0.05 0.1 0.15 0.2

Solution initial concentration, kgkg-1

Bub

ble

diam

eter

dur

ing

deta

chm

ent,

mm

Gas flow rate = 0.0006 m hGas flow rate = 0.0012 m hGas flow rate = 0.0024 m hGas flow rate = 0.0036 m hGas flow rate = 0.0048 m hGas flow rate = 0.006 m h

Gas inlet pressure = 600 kPa

Gas inlet temperature = 32 Solution inlet pressure = 110 kPa

Solution inlet temperature = 30 Nozzle diameter = 2.2 m

-1

-1 -1

-1

-1

-1

3

33

3

3

3

oC

oCm

Fig. 10. Effect of solution initial concentration on bubble diameter during detach-ment for different gas flow rates in still solution.

M. Suresh, A. Mani / Experimental Thermal and Fluid Science 39 (2012) 79–89 87

the bubble diameter increases as the gas flow rate increases. This issimilar to the trend observed in the numerical model developed bythe authors [18] to study bubble dynamics using phenomenologi-cal theory. Fig. 9 compares the experimental bubble diameter instill solution with that of numerical model, for various solution ini-tial concentrations. Fig. 10 shows the variation of bubble diameterduring detachment with respect to solution initial concentration atvarious gas flow rates. The bubble diameter initially increases asthe solution concentrations increases up to 0.075 kg kg�1 and thenstarts decreasing as the solution concentration increases further.As the solution concentration increases both absorption drivingpotential (xeq � xl) as well as surface tension decrease. Decreasein absorption driving potential tends to increase the bubble diam-eter whereas decrease in surface tension tends to decrease thebubble diameter. However, at lower solution concentrations, theeffect of absorption driving potential is dominant than that of sur-face tension, due to which the bubble diameter increases. Whenthe solution concentration is increased further, the effect of surfacetension is dominant resulting in decrease in bubble diameter.

7. Correlation for bubble diameter during detachment

Based on the above experiments, a correlation for bubble diam-eter during detachment has been developed as a function ofReynolds number, Weber number and Buoyancy number.

db

dn¼ c1Rec2

v Wec3v Buc4

v Xglc5 ð9Þ

where

Rev ¼ inertial force=viscous force ¼ qvvndn

lvð10Þ

Wev ¼ inertial force=surface tension ¼ qvv2ndn

rð11Þ

Buv ¼ inertial force=buoyancy force ¼ qvv2n

gdnðql � qvÞð12Þ

Xgl ¼ Xv � Xl ð13Þ

The generic form of the correlation includes inertial force, vis-cous force, surface tension, buoyancy force and absorption poten-tial. All three non-dimensional numbers are defined based on thenozzle diameter (dn) and the departing bubble velocity from thenozzle (vn), which is calculated from the inlet gas flow rate andnozzle cross section area (An). The correlation for bubble diameterhas been developed based on nozzle diameter since it is an impor-tant influencing factor. So nozzle diameter is used as length scale innon-dimensional numbers.

Kang et al. [20] compared their experimental data on the ratioof bubble diameter to nozzle diameter for ammonia–water bubbleabsorption, with Akita and Yoshida’s correlation [26] and Bhavara-ju’s correlation [28]. Bhavaraju’s correlation overestimated theirdata by about 30–40%, whereas Akita and Yoshida’s correlationdeviated within ±20%.

Present experimental data are also compared with both theabove correlations and same deviations are observed. Fig. 11 showsa comparison of present experimental data with Bhavaraju’s corre-lation. The correlation underestimates the present experimental

1.0

1.5

2.0

2.5

3.0

3.5

4.0

5.20.25.1

Correlation-ratio of bubble diameter to nozzle diameter

Exp

erim

enta

l dat

a-ra

tio

of b

ubbl

e Bhavaraju et al. (1978)Present experimental data

+ 35%

- 35%

Bhavaraju's correlations

db = (6 do /g ) 1/3

db/do = 3.23Reo,l-0.1Fro,v

0.21

diam

eter

to

nozz

le d

iam

eter

Fig. 11. Comparison between experimental data and Bhavaraju’s correlation [28].

0.8

1.3

1.8

2.3

2.8

3.3

3.8

0.8 1.3 1.8 2.3 2.8 3.3

Correlation-ratio of bubble diameter to nozzle diameter

Exp

erim

enta

l dat

a-ra

tio

of b

ubbl

e di

amet

er t

o no

zzle

dia

met

er

Akita and Yoshida (1974)

Present experimental data + 25%

- 25%

Akita and Yoshida correlation

db/do = 1.88(Vo / gdo ))1/3

Fig. 12. Comparison between experimental data and Akita and Yoshida’s correla-tion [26].

1.0

1.5

2.0

2.5

3.0

3.5

4.0

1.0 1.5 2.0 2.5 3.0 3.5

Correlation-ratio of bubble diameter to nozzle diameter

Exp

erim

enta

l dat

a-ra

tio

of b

ubbl

e

diam

eter

to

nozz

le d

iam

eter

Solution initial concentration = 0.01 kgkg

Solution initial concentration = 0.05 kgkg

Solution initial concentration = 0.10 kgkg

Solution initial concentration = 0.15 kgkg

+ 12%

- 12%

-1

-1

-1

-1

db/do = 3.277(Rev)0.0226(We v)

0.0925 (Buv)0.0542(xgl)0.108

Fig. 13. Comparison between experimental and regressed correlation data.

88 M. Suresh, A. Mani / Experimental Thermal and Fluid Science 39 (2012) 79–89

data by about 30–40%. Fig. 12 shows the comparison with Akitaand Yoshida’s correlation. It also underestimates the presentexperimental data by about 25%.

Akita and Yoshida reported that the bubble size was indepen-dent of the properties such as surface tension, liquid viscosity,liquid and vapor density. They presented that orifice diameterand gas velocity were the factors affecting the bubble size. Bhavar-aju et al. reported that bubble diameter was dependent on gas flowrate, but it was essentially independent of orifice diameter eventhough they proposed correlations of bubble size as functions oforifice diameter. Kang et al. [20] presented experimental correla-tion for bubble diameter using ammonia–water as working fluidand reported that bubble diameter is significantly influenced byorifice diameter for ammonia–water bubble absorption. Theauthors of present work agree with Kang et al. and have presenteda correlation for bubble diameter as function of vapor Reynoldsnumber, Buoyancy number, Weber number and concentration dif-ference between gas and solution for R134a–DMF as working fluid.It is also observed that nozzle diameter, gas flow rate and liquidproperties influence bubble diameter. This is the reason for thedeviation of bubble diameter correlation from Akita and Yoshida’scorrelation and Bhavaraju’s correlation.

Fig. 13 shows the experimental correlation of the bubble diam-eter during detachment from the present study. A regressiontechnique using Generalized Autoregressive Conditional Heter-oskedasticity (GARCH) tool box in MATLAB is employed to findout the coefficients in the generic form of the correlation in Eq.(1). The following correlation has been developed with R134a–DMF with ±12% error band.

db

dn¼ 3:277ðRevÞ0:0226ðWevÞ0:0925ðBuvÞ0:0542ðXglÞ0:108 ð14Þ

The above correlation is valid in the range

for Rev ¼ 200—2000; Wev ¼ 0:0025—0:25; Buv

¼ 0:0025—0:25; Xgl ¼ 0:84—0:98

It can be seen that the bubble diameter increases as Reynoldsnumber, Weber number and Buoyancy number increase, since allthree non-dimensional parameters are functions of bubble velocityfrom nozzle which is directly proportional to inlet gas flow rate.Also the bubble diameter increases as the concentration differencebetween gas and solution increases.

8. Conclusions

Experimental investigations have been carried out on a glassbubble absorber to visualize bubble behavior and effect of gas flowrate and liquid concentration on bubble characteristics of R134a inDMF. Bubble behavior has been studied in still as well as flowingsolution. The following conclusions are drawn based on the presentstudy.

– Bubble behavior was visualized using photographic studies. Atlower gas flow rates in still solution, bubbles tend to be spher-ical due to the dominance of surface tension. At lower as well ashigher gas flow rates in flowing solution, the bubbles tend to behemispherical due to the dominance of inertial force.

– Gas flow rate and solution concentration influence the bubblecharacteristics significantly during absorption process.

– Bubble diameter during detachment and the ratio of bubblediameter to nozzle diameter increase with the increase in gasflow rates.

– Bubble diameter during detachment and the ratio of bubblediameter to nozzle diameter initially increases as the solutionconcentrations goes up to 0.075 kg kg�1 and then starts decreas-ing as the solution concentration increases further.

– Bubble diameter measured from the experiments is comparedwith numerical model and it is found that agreement is goodwith a maximum deviation of ±5%.

– Akita and Yoshida’s correlation and Bhavaraju’s correlationunderestimate the bubble diameter during detachment forR134a–DMF by 25–40%. A correlation for bubble diameter dur-ing detachment has been presented from the experimentalstudies with ±12% error band.

References

[1] M. Fatouh A.M. Ghattas, Studies on HCFC based vapour absorptionrefrigeration systems suitable for low potential heat sources, PhD thesis,Indian Institute of Technology Madras, 1994.

M. Suresh, A. Mani / Experimental Thermal and Fluid Science 39 (2012) 79–89 89

[2] G. Karthikeyan, A. Mani, S. Srinivasa Murthy, Performance comparison ofdifferent working fluids with transfer tank operated vapour absorptionrefrigeration system with transfer tank, Journal of Renewable Energy 6(1995) 835–837.

[3] K.S. Sujatha, Studies on absorber and transfer tank of the vapour absorptionrefrigeration system, PhD thesis, Indian Institute of Technology Madras, 1997.

[4] Y. Nezu, N. Hisada, T. Ishiyama, K. Watanabe, Thermodynamic properties ofworking-fluid pairs with R-134a for absorption refrigeration system. In:Natural Working-Fluids 2002, IIR Gustav Lorentzen Conf. 5th., China,September 17–20, 2002, pp. 446–453.

[5] A. Yokozeki, Theoretical performances of various refrigerant–absorbent pairsin a vapour absorption refrigeration cycle by the use of equation of state,Applied Energy 80 (2005) 383–399.

[6] A. Mani, Studies on compact bubble absorber of the vapour absorptionrefrigeration system, A Report Submitted Department of Science andTechnology, Government of India, 2009.

[7] Y.T. Kang, A. Akisawa, T. Kashiwagi, Analytical investigation of two differentabsorption modes: falling film and Bubble types, International Journal ofRefrigeration 23 (6) (2000) 430–443.

[8] T. Elperin, A. Fominyk, Four stages of the simultaneous mass and heat transferduring bubble formation and rise in a bubbly absorber, Chemical EngineeringScience 58 (2003) 3555–3564.

[9] Jae-Cheol Lee, Ki-Bong Lee, Byung-Hee Chun, Chan Ho Lee, Jong Joo Ha, SungHyun Kim, A study on numerical simulations and experiments for masstransfer in bubble mode absorber of ammonia and water, International Journalof Refrigeration 26 (5) (2003) 551–558.

[10] T.L. Merrill, H. Perez-Blanco, Combined heat and mass transfer during bubbleabsorption in binary solutions, International Journal of Heat and Mass Transfer40 (3) (1997) 589–603.

[11] K.S. Sujatha, A. Mani, S. Srinivasa Murthy, Finite element analysis of a bubbleabsorber, International Journal of Numerical Methods for Heat & Fluid Flow 7(7) (1997) 737–750.

[12] K.S. Sujatha, A. Mani, S. Srinivasa Murthy, Analysis of a bubble absorberworking with R22 and five organic absorbents, Heat and Mass Transfer 32(1997) 255–259.

[13] K.S. Sujatha, A. Mani, S. Srinivasa Murthy, Experiments on a bubble absorber,International Communications in Heat and Mass Transfer 26 (7) (1999) 975–984.

[14] Y.T. Kang, R.N. Christensen, T. Kashiwagi, Ammonia–water bubble absorberwith a plate heat exchanger, ASHRAE Transactions 104 (1) (1998) 1–11.

[15] M.D. Staicovici, A phenemenological theory of polycomponent interactions innon-ideal mixtures. Application to NH3/H2O and other working pairs,International Journal of Refrigeration 23 (2) (2000) 153–167.

[16] M.D. Staicovici, A non-equilibrium phenomenological theory of the mass andheat transfer in physical and chemical interactions. Part I – theory and its

application to NH3/H2O and other working systems, International Journal ofHeat and Mass Transfer 43 (22) (2000) 4153–4173.

[17] M.D. Staicovici, A non-equilibrium phenomenological theory of the mass andheat transfer in physical and chemical interactions. Part II – modeling of theNH3/H2O bubble absorption, analytical study of absorption and experiments,International Journal of Heat and Mass Transfer 43 (22) (2000) 4175–4188.

[18] M. Suresh, A. Mani, Heat and mass transfer studies on R134a bubble absorberin R134a/DMF based on phenomenological theory, International Journal ofHeat and Mass Transfer 53 (13–14) (2010) 2813–2825.

[19] Y.T. Kang, T. Nagano, T. Kashiwagi, Mass transfer correlation of NH3–H2Obubble absorption, International Journal of Refrigeration 25 (7) (2002) 878–886.

[20] Y.T. Kang, T. Nagano, T. Kashiwagi, Visualization of bubble behavior andbubble diameter correlation for NH3–H2O bubble absorption, InternationalJournal of Refrigeration 25 (1) (2002) 127–135.

[21] R. Kumar, N.R. Kuloor, The formation of bubbles and drops, Advances inChemical Engineering 8 (1970) 256–368.

[22] I. Chakraborty, G. Biswas, P.S. Ghoshdastidar, Bubble generation in quiescentand co-flowing liquids, International Journal of Heat and Mass Transfer 54(2011) 4673–4688.

[23] Koichi Terasaka, Hideki Tsuge, Bubble formation under constant flowconditions, Chemical Engineering Science 48 (19) (1993) 3417–3422.

[24] Koichi Terasaka, Hideki Tsuge, Hirokazu Matsue, Bubble formation incocurrently upward flowing liquid, The Canadian Journal of ChemicalEngineering 77 (3) (1999) 458–464.

[25] Hideki Tsuge, Yoshito Nakajima, Koichi Terasaka, Behaviour of bubbles formedfrom a submerged orifice under high system pressure, Chemical EngineeringScience 47 (13–14) (1992) 3273–3280.

[26] K. Akita, F. Yoshida, Bubble size, interfacial area and liquid phase mass transfercoefficient in bubble columns, Industrial & Engineering Chemistry ProcessDesign and Development 13 (1) (1974) 84–91.

[27] A. Kumar, D.E. Degaleesan, G.S. Laddha, H.E. Hoelscher, Bubble swarmcharacteristics in bubble columns, Canadian Journal of Chemical Engineers54 (1976) 503–508.

[28] S.M. Bhavaraju, T.W.F. Russell, H.W. Blanch, The design of gas sparged devicesfor viscous liquid systems, AIChE Journal 24 (3) (1978) 454–466.

[29] Hideki Tsuge, Yusuke Tezuka, Masae Mitsudani, Bubble formation mechanismfrom downward nozzle—effect of nozzle shape and operating parameters,Chemical Engineering Science 61 (2006) 3290–3298.

[30] Hideki Tsuge, Yusuke Tezuka, Masae Mitsudani, The effect of nozzle shape onbubble formation from a downward nozzle, Chemical Engineering Technology29 (9) (2006) 1097–1101.

[31] Robert C. Reid, John M. Prausnitz, Bruce E. Poling, The Properties of Gases andLiquids, fourth ed., McGraw-Hill Book Company, 1989.