Flow Measurement and Instrumentation 24 (2012) 36–42

Contents lists available at SciVerse ScienceDirect

Flow Measurement and Instrumentation

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

Measurements and characterization of bubble dynamics in capillary two-phaseflows by a micro double-tip conductivity probeZhou Yuan, Zhang Qingyong, Liu Hui ∗, Lei ZhigangState Key Laboratory of Chemical Resource Engineering, Beijing University of Chemical Technology, Beijing 100029, China

a r t i c l e i n f o

Article history:Received 30 August 2011Received in revised form25 February 2012Accepted 22 March 2012

Keywords:Double-tip conductivity probeVertical capillaryTaylor flowHydrodynamicsCCD

a b s t r a c t

With a particular focus on the characterization of gas–liquid Taylor flow in individual channels ofmonolithic beds, a micro double-tip conductivity probe was developed for the measurements andcharacterization of bubble dynamics in capillary two-phase flows. The Taylor flow hydrodynamics invertical capillaries with a circular cross section of 2.98 mm in hydraulic diameter was investigatedincluding flow regime, bubble rise velocity, and liquid slug length in a wide range of gas and liquidsuperficial velocities. It is demonstrated that the micro double-tip conductivity probe method is suitablefor identifying flow regimes in small-scale capillaries and measuring bubble dynamic parameters of theTaylor flow. Furthermore, variation trends of bubble rise velocity and frequency, and liquid slug lengthwith varying gas and liquid superficial velocities in the Taylor flow regime are demonstrated. Based onthe experimental data, correlations for prediction of the above parameters were obtained.

© 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Multiphasemonolithic reactors (MMRs) are a new type of struc-tured reactors offering superior advantages including low pressuredrop, high geometric surface area, good mass transfer, and me-chanical integrity [1–3], compared with the traditional multiphasereactors such as trickle beds and slurry bubble columns. The advan-tages mainly stem from the unique multi-channel configuration ofthemonolithic catalyst bed,which consists of a bundle of separatedchannels (in parallel) with slit sizes of 1–3 mm. The multiplicityof identical single channels in a monolithic bed should allow foreasier scale-up and operation of such reactors. Hence MMRs arepromising reactor packings suitable for catalytic gas–liquid–solidreactions encountered extensively in chemical, petrochemical, bio-chemical, and environmental processes [4–6]. In such applications,understanding the gas–liquid hydrodynamics in the channels is animportant issue because it is closely related to the mass transfer[7–9] and reaction performances [10] of the processes.

Non-invasivemeasuring techniques have been used tomeasurethe two phase flow parameters on a single capillary scale [11–14]and a monolithic bed scale [15–18] operating in the gas–liquidco-current down-flow mode. Essentially, these techniques can becategorized into the conventional camera imaging method andmore sophisticated techniques such as computed tomography(CT) and magnetic resonance imaging (MRI) tomography. Liuet al. [14] utilized a high speed camera to measure hydrodynamic

∗ Corresponding author. Tel.: +86 10 64433695.E-mail address: hliu@mail.buct.edu.cn (H. Liu).

0955-5986/$ – see front matter© 2012 Elsevier Ltd. All rights reserved.doi:10.1016/j.flowmeasinst.2012.03.001

parameters of the two-phase Taylor flow in vertical capillaries,including especially the flow regime, bubble rise velocity, andliquid slug length. Such a method is easy for a visualizing andimaging analysis of the gas–liquid flow behavior, but merelysuitable for transparent systems. For an opaque monolithic bed,CT and MRI are more suitable. Al-Dahhan and coworkers [15,16]have used CT to characterize the gas–liquid distribution overthe cross-section of monolithic beds; however the technique ishard to determine the slug lengths and bubble velocities. TheMRI technique was adopted by Gladden and coworkers [17] andHeibel et al. [18] to study the gas–liquid distribution in multiphasesystems including monolithic beds. Because of the need of highintensity magnetic fields, applying MRI is extremely expensive.

In view of the limitationsmentioned above, herewe investigatethe extended use of the double-tip conductivity probe, whichhas been proved to be an efficient tool for measuring bubbledynamic parameters in bubbly flows [19,20], to the gas–liquidflows in MMRs. In this work, a double-tip conductivity probeis inserted into a capillary to measure the bubble parameters.The validation of the measuring method is made by comparingthe results from the double-tip conductivity probe and the CCDimagingmethod. Furthermore, hydrodynamic characteristics, suchas the gas holdup, bubble velocity, bubble frequency, and liquidslug length in the Taylor flow regime, are also reported.

2. Experimental system

A schematic representation of the experimental setup isdepicted in Fig. 1. It consists of a liquid system (1), a gas system(2), a capillary test section (3), a measurement system (4) anda data recording and analysis system (5). All experiments were

Y. Zhou et al. / Flow Measurement and Instrumentation 24 (2012) 36–42 37

Notations

dc Capillary hydraulic diameter (m)fB Bubble frequency (s−1)g Gravitational constant (m/s2)Lslug Liquid slug length (m)Re′

G Gas phase Reynolds number, UGdcρL/µLUG Superficial gas velocity (m/s)UL Superficial liquid velocity (m/s)UTP Flooding gas velocity (m/s)VB Bubble rise velocity (m/s)

Greek Letters

εG Gas holdupµG Gas viscosity (mPa s)µL Liquid viscosity (mPa s)ρG Gas density (kg/m3)ρL Liquid density (kg/m3)σL Surface tension (N/m)ψslug Dimensionless liquid slug length, Lslug/dc

Dimensionless Groups

Ca Capillary number, µLUTP/σLEo Eötvös number, (ρL − ρG)dc2g/σ

Subscripts

con Conductivity probeCCD CCD system

conducted at ambient temperature and pressure, with nitrogenand tap water as the gas and liquid phases, respectively. Gasand liquid were fed co-currently to the bottom of the capillarythrough a 3mm-diameter polyvinyl chloride (PVC) tee connection.During experiments, compressed nitrogen was fed through apre-calibrated float-type gas flowmeter to the tee connection. Amanually operated control valveswas used to regulate the gas flowrate. Liquidwas delivered from an elevated 10-L storage vessel intoa pre-calibrated float-type flowmeter; gravity provided the drivingforce for the liquid flow. The gas–liquid flow arrangement allowedfor an independent alteration of the gas and liquid flow rates. Thetest section is a glass capillary of 0.804 m high and 2.98 mm indiameter. After the capillary test section, liquid was dischargedfrom the capillary into a disengagement zone, and gas was givenout into the air. Two special fixation parts were designed andmadefor fixing the capillary position and easily inserting the measuringinstruments; one is an inlet gas–liquid mixing part at the bottomof the capillary, and the other is an outlet extending part at the top,which includes the gas–liquid disengagement zone.

Total pressure drop was recorded using a differential pressuretransducer (MDM495, China), with a measuring range of −6 to1 kPa. System flow stability was assured before eachmeasurementonce the output signals of the pressure drop transducer weresteady.

3. Measuring methods for bubble dynamics

The double-tip conductivity probe, as shown in Fig. 2, wasmadeof 0.1 mm-diameter platinum wire (1) with a sharply finishedtip of a very short conductive distance of 0.5 mm, coated withepoxy resin and connected with a cupreous lead (2). Then itwas encased in a 1.0 mm-diameter glass capillary (3), outside ofwhich a stainless steel tube of 2.0 mm-diameter (4) was mounted.Insulating glue (5) was injected into the spacing between the tube

Fig. 1. Schematic representation of the experimental setup.

Fig. 2. Conductivity probe. 1. Pt probe φ 0.1 mm; 2. cupreous lead φ 0.5 mm;3. glass capillary φ 1.0 mm. 4. stainless steel tube φ 2.0 mm; 5. glue seal.

Fig. 3. Electric circuit for conductivity probe. 1. conductivity probe; 2. resistance;3. DC electrical source; 4. A/D transform card; 5.personal computer.

and the capillary to seal the two ends. A reference electrode, madeof cupreous lead, was attached to the inside wall surface oppositeto the two probe tips. It is noted that only 1.5 mm length ofsharpened conductivity probe tip with 0.1 mm in diameter wasexposed into the capillary to minimize the influence of probe onthe fluid flow.

The electric circuit for the signal processing system is shownin Fig. 3. DC voltage at 24 V was applied to the probe. After theresistance transformed the electrical signals within the range of4–20mA into voltage signals, the voltage signalswere transformedinto digital signals by a A/D transform card (Double-Nuo AC6111,made in China), and then the signals were received by thecomputer for real-time recording with the output signals of proberanging from 1 to 5 V.

The difference in conductivities of gas phase and liquid phasewas detected by the two probe tips; for the bubble the voltage

38 Y. Zhou et al. / Flow Measurement and Instrumentation 24 (2012) 36–42

Fig. 4. Typical probe signal and square wave processing of double-tip conductivity probe output signal.

Fig. 5. Determination of hydrodynamic parameters obtained by conductivity probemethod.

of output signal was low, and was high for the liquid phase.The output signal from the probe was transformed into a squarewave in the wave-reforming section, and finally the reformedsignals were processed with a personal computer (PC). Fig. 4shows the typical probe signals and the corresponding squarewave processing in the Taylor flow regime, where red and blacklines represent double-tip conductivity probe output signals of upstream and down stream, respectively.

In the experiments, the distance between the two tipswas fixedto be L = 0.025m, which is shorter than the length of any unit cellunder the operating conditions. In every experiment the samplingfrequency was 1000 Hz with a measuring period of around 20 s.

The individual bubble velocity was calculated from the lagtime between the two up- and down-stream signals, 1ti andthe distance between the two tips, L, i.e., VBi = L/1ti; seeFig. 5. The average bubble velocity VB was calculated by VB =1NB

NBi=1 VBi, where NB is the bubble number. The gas holdup, εG,

was determined from the upstream probe signal as the time ratioof the duration time for the bubbles and the total sampling period,i.e., εG =

1Ti/T , in which1Ti is the duration time experienced

by the bubble i at the tip, and T is themeasuring period. The bubblefrequency, fB, was determined by counting the bubble numberspassing during the measuring period. With known total passagetime T and the bubble number NB, the bubble frequency wasobtained by fB = NB/T .

For validation of the conductivity probe method, a CCD camera(Sony DCR-TRV38E) was positioned midway along the capillaryheight with its focus being adjusted in such a way that it captured

rising bubbles and liquid slugs at varying gas and liquid velocities.After the steady state was achieved, movies were made for atime span of around 20 s, and then analyzed using image-analysissoftware. Given the capture rate of 30 fps, the bubble frequency(fB), which is defined as the number of bubbles that traverse agiven point in the capillary per unit time, and the bubble risingvelocity (VB), which is obtained by registering the time required fora gas bubble to pass a known distance along the capillary height,were determined. Gas holdup (εG), which is the ratio of the bubblelength to the unit cell length, can be obtained directly from images.

The reported data from the present experiments repeated wellwith a relative error of ±3.5% in average.

4. Results and discussion

4.1. Verification of double-tip conductivity probe method

First, we used the double-tip conductivity probe method toidentify the gas–liquid flow regimes and compared the resultswith those from the CCD imaging method. Typical images andconductivity output signals of flow regimes are shown in Fig. 6. Fivedistinct flow patterns were observed as follows.

The dispersed bubbly flow (Fig. 6(a)) typically occurs atrelatively high liquid velocities and low gas velocities and ischaracterized by the presence of fast rising bubbleswith diametersmuch less than the capillary diameter. The bubbles are observed tobe spherical in shape and dispersed at random in the liquid, so theoutput signals of the conductivity probe are ruleless and the clearinterface between bubble and liquid can hardly be identified.

In the bubble train flow regime (Fig. 6(b)), bubbles take spher-ical or spheroidal shapes with diameters equal to the capillary di-ameter. The bubbles pass the capillary one by one like a train, andconsequently the conductivity signals are well-regulated.

The Taylor flow (Fig. 6(c)) consists of gas bubbles with lengthgreater than the capillary diameter that move along the capillaryseparated from each other by liquid slugs. Depending on the gasand liquid flow rates and properties, the bubbles often have hemi-spherically shaped tops and flattened tails. Conductivity outputsignals show well the characters of the Taylor flow regime withclearly distinguished phase interface.

The churn flow (Fig. 6(d)) occurs at very high gas velocities. Itconsists of very long gas bubbles and relatively small liquid slugs.Because of the high gas velocity, a wave or ripple motion is oftenobserved in the liquid slug. The conductivity signals capture thecharacters of the flow.

Y. Zhou et al. / Flow Measurement and Instrumentation 24 (2012) 36–42 39

(a) Dispersed bubbly flow regime. (b) Bubble train flow regime.

(c) Taylor flow regime. (d) Churn flow regime.

(e) Annual flow regime.

Fig. 6. Comparison of flow regimes between CCD method and conductivity probe method. (a) Dispersed bubbly flow regime, (b) bubble train flow regime, (c) Taylor flowregime, (d) churn flow regime, and (e) annual flow regime.

The annular flow (Fig. 6(e)) occurs at excessively high gasvelocities and very low liquid velocities. Here, a continuous gasphase is present in the central core of the capillary with the liquidphase being displaced to form a very thin liquid film between thecapillary wall and the gas phase, so the output signals keep at lowvoltage.

From the discussion above, we see the conductivity probegrasps the essential features of the five flow patterns occurringin capillary two phase flows. However, for the practical use

of the conductivity probe, i.e., to determine quantitatively thebubble parameters like bubble velocities, we have to confineourselves to the measurements of the Taylor flow, because aquantitative phase identification or a bi-value treatment of theoutput signal is difficult to made for the other flow patterns.Therefore, the validation of bubble dynamic measurement by thepresent probe is confined to the measurement results of bubbledynamic parameters in the Taylor flow from the conductivity probemethod and CCD method. Fig. 7 shows a comparison of bubble

40 Y. Zhou et al. / Flow Measurement and Instrumentation 24 (2012) 36–42

(a) Bubble frequency. (b) Bubble rising velocity.

(c) Gas holdup.

Fig. 7. Comparison of hydrodynamic parameters between CCD method and conductivity probe method. (a) Bubble frequency, (b) bubble rising velocity, (c) gas holdup.

frequency (Fig. 7(a)), bubble rising velocity (Fig. 7(b)) and gasholdup (Fig. 7(c)) obtained by the two methods; the two methodsmatch well to each other, and the deviations are about ±5%, ±7%and ±7%, respectively. In Figs. 7(b) and (c), the values obtained byconductivity probe are larger than those by CCD. The reason forthis is that at high gas and liquid velocities, the time resolutionof the CCD camera adopted in this work is not high enough tocapture the fast dynamics of bubbles. On the contrary, the samplingfrequency of the conductivity probe could be high enough andtimely responses to the dynamics of bubbles, especially at highgas and liquid velocities. In what follows, we use the conductivityprobe to investigate the bubble dynamics in the Taylor flow in awide range of gas and liquid velocities.

4.2. Bubble dynamic characteristics and correlations in Taylor flow

Fig. 8 shows the gas holdup εG increases with increasing UGand decreasing UL; at higher UG(>0.05 m/s), εG increases slowlywithUL. For estimation, a comparison between the gas holdup dataobtained and the following relation

εG =UG

UG + UL=

UG

UTP(1)

was made; see Fig. 9. It is noted the calculated values are mainlywithin ±20% of the measured values.

Liquid slug length is an important hydrodynamic parameterthat has been reported to have a very significant effect ongas–liquid mass transfer in capillaries. The value of Lslug decreases

Fig. 8. Variation of gas holdup, εG with superficial gas velocity, UG and superficialliquid velocity, UL .

with an increase of UG and increases with an increase of UL; seeFig. 10. At lower UG(>0.05 m/s), Lslug decreases sharply and thetrend turns very slowly with increasing UG.

Kreutzer [21] described the slug length data of Heiszwolfet al. [22] in 200 cpsi monoliths using the following correlation

ψslug =εL

−0.00141 − 1.556ε2L Ln(εL). (2)

Y. Zhou et al. / Flow Measurement and Instrumentation 24 (2012) 36–42 41

Fig. 9. Comparison of experimental gas holdup with predicted values from Eq. (1).

Fig. 10. Variation of slug length Lslug with superficial gas velocity,UG and superficialliquid velocity, UL .

Liu et al. [14] proposed an empirical correlation to evaluate theliquid slug length in vertical capillaries

UTPLslug

= 0.088Re0.72G Re0.19L . (3)

In this work, the experimental data of liquid slug length werecorrelated and the following correlation was obtained:

ψslug = 309.427 ×

Re′

G

Eo

−0.819

. (4)

Fig. 11 shows a comparison of the liquid slug lengths obtainedby our experiments, Eq. (4), and the predicted slug lengths usingEqs. (2), (3). Remarkably, an enormous amount of scatters observedfor the literature correlations. This could be mostly attributable tothe dependency of the liquid slug length on the inlet configurationof capillaries [14].

As shown in Fig. 12, the bubble velocity VB increases withincreasing UG and decreases with increasing UL. At higherUG(>0.05 m/s), VB increases slowly with UL, and the same trendcan be seen with the change of UL. Many approaches have beenproposed in the literature to estimating bubble rise velocities ina capillary. To overcome difficulties associated with estimatingthe bubble diameter, as well as accounting for the effect ofliquid properties on bubble rise velocity, the following practicalrelationship was developed by Liu et al. [14].

VB

UTP=

11 − 0.61Ca0.33

. (5)

Fig. 11. Comparison of experimental dimensionless liquid slug lengths withpredicted values by various correlations.

Fig. 12. Variation of bubble rise velocity VB with superficial gas velocity, UG andsuperficial liquid velocity, UL .

Fig. 13. Comparison of bubble rise velocity between results of conductivitymethodand predictions.

Ca is the capillary number, which is defined as µLUTP/σL. Fig. 13shows plots of experimental results of conductivity method, aswell as predictions obtained using Eq. (5). The agreement betweenthe data and the calculated is satisfactory.

5. Conclusions

A micro double-tip conductivity probe was developed forthe measurements and characterization of bubble dynamics in

42 Y. Zhou et al. / Flow Measurement and Instrumentation 24 (2012) 36–42

capillary two-phase flows and used to investigate the Taylor flowhydrodynamics in vertical capillaries with circular cross sectionof 2.98 mm in hydraulic diameter, including flow regime, bubblerise velocity, and liquid slug length in a wide range of gas andliquid superficial velocities. Based on the work performed and thecomparison presented, the following major conclusions can bedrawn:

(1) The micro double-tip conductivity probe method is usable andaccurate for identifying flow regimes in small-scale capillariesand measuring bubble dynamic parameters of the Taylor flowwith the advantages of simple making, low cost, easy use andhigh accuracy.

(2) Variation trends of bubble rise velocity and frequency, andliquid slug length with varying gas and liquid superficialvelocities in the Taylor flow regime are demonstrated based onthe data obtained, and correlations for prediction of the aboveparameters were obtained.

Finally, it should be stressed thatwhile the present conductivityprobe method has been demonstrated to be quite suitable for themeasurements in a single capillary, still the present work opensthe question of how it is used in opaque monolithic beds. Giventhe easy positioning of the probe, it is possible to install a numberof the probes across the section of the bed and thereby measurethe bubble dynamics in individual channels of the bed.

Acknowledgments

We acknowledge financial support from the State Key Devel-opment Program for Basic Research (2006CB2020504), and theNational Nature Science Foundation of China under Grant Nos.20821004 and 21076008, China.

References

[1] Edvinsson RK, Cybulski A. A comparison between the monolithic reactor andtrickle bed reactor for liquid phase hydrogenations. Catalysis Today 1995;24(1–2):173–9.

[2] Nijhuis TA, Kreutzer MT, Roumijn ACJ, Kapteijn F, Moulijn J. Monolithiccatalysts as efficient three phase reactors. Chemical Engineering Science 2001;56(3):823–9.

[3] Roy S, Heibel AK, LiuW, Boger T. Design ofmonolithic catalysts formulti-phasereactions. Chemical Engineering Science 2004;59(5):957–66.

[4] Boger T, Zieverink MMP, Kreutzer MT, Kapteijn F, Moulijn J, Addiego WP.Monolithic catalysts as an alternative to slurry systems: hydrogenationof edible Oil. Industrial and Engineering Chemistry Research 2004;43(10):2337–44.

[5] Edvinsson R, Irandoust S. Hydrodesulfurization of dibenzothiophene in amonolithic catalyst reactor. Industrial and Engineering Chemistry Research1993;32(2):391–5.

[6] Klinghoffer AA, Cerro RL, Abraham MA. Influence of flow properties on theperformance of the monolith froth reactor for catalytic wet oxidation of aceticacid. Industrial and Engineering Chemistry Research 1998;37(4):1203–10.

[7] Bercic G, Pintar A. The fole of gas bubbles and liquid slug lengths on masstransport in the Taylor flow through capillaries. Chemical Engineering Science1997;52(21–22):3709–19.

[8] Kreutzer MT, Du P, Heiszwolf JJ, Kapteijn F, Moulijn JA. Mass transfercharacteristics of three-phase monolith reactors. Chemical EngineeringScience 2001;56(21–22):6015–23.

[9] Irandoust S, Andersson B. Mass-transfer and liquid-phase reactions in asegmented two-phase flowmonolithic catalyst reactor. Chemical EngineeringScience 1988;43(8):1983–8.

[10] Bercic G. Influence of operating conditions on the observed reaction rate in thesingle channel monolith reactor. Catalysis Today 2001;69:147–52.

[11] Thulasidas TC, Abraham MA, Cerro RL. Bubble-train flow in capillaries ofcircular and square cross-section. Chemical Engineering Science 1995;50(2):183–99.

[12] Thulasidas TC, Abraham MA, Cerro RL. Dispersion during bubble-train flow incapillaries. Chemical Engineering Science 1999;54(1):61–76.

[13] Laborie S, Cabassud C, Durand-Bourlier L, Laine JM. Characterisation ofgas–liquid two-phase flow inside capillaries. Chemical Engineering Science1999;54(23):5723–35.

[14] Liu H, Vandu CO, Krishna R. Hydrodynamics of Taylor flow in verticalcapillaries: flow regimes, bubble rise velocity, liquid slug length, and pressuredrop. Industrial and Engineering Chemistry Research 2005;44:4884–97.

[15] Al-Dahhan MH, Kemoun A, Cartolano AR. Phase distribution in an upflowmonolith reactor using computed tomography. AIChE Journal 2006;52(2):745–53.

[16] Bauer T, Roy S, Lange R, Al-Dahhan M. Liquid saturation and gas–liquiddistribution in multiphase monolithic reactors. Chemical Engineering Science2005;60(11):3101–6.

[17] Mantle MD, Sederman AJ, Gladden LF, Raymahasay S, Winterbottom JM,Stitt EH, Dynamic MRI. visualization of two-phase flow in a ceramic monolith.AIChE Journal 2002;48(4):909–12.

[18] Heibel AK, Vergeldt FJ, As H, Kapteijn F, Moulijn J, Boger T. Gas and liquiddistribution in the monolith film flow reactor. AIChE Journal 2003;49(12):3007–17.

[19] Lu S, Chen X, Yu G, Yu Z. Measurement of the bubble parameters in bubblecolumnby conductivity probe. Chemical Reaction Engineering and Technology2003;19(4):344–51.

[20] Xie W, Neethling SJ, Cilliers JJ. A novel approach for estimating the averagebubble size for foams flowing in vertical columns. Chemical EngineeringScience 2004;59(1):81–6.

[21] Kreutzer MT. Hydrodynamics of Taylor flow in capillaries and monolithreactors. Delft, The Netherlands: Delft University Press; 2003.

[22] Heiszwolf JJ, Kreutzer MT, Van den Eiijnden MG, Kapteijn F, Moulijn JA. Gas-liquid mass transfer of aqueous Taylor flow in monoliths. Catalysis Today2001;69:51–5.