ORIGINAL

Characterizing fluid dynamics in a bubble column aimedfor the determination of reactive mass transfer

Péter Kováts1 & Dominique Thévenin1& Katharina Zähringer1

Received: 19 October 2016 /Accepted: 21 August 2017 /Published online: 4 September 2017# Springer-Verlag GmbH Germany 2017

Abstract Bubble column reactors are multiphase reactorsthat are used in many process engineering applications. Inthese reactors a gas phase comes into contact with a fluidphase to initiate or support reactions. The transport processfrom the gas to the liquid phase is often the limiting factor.Characterizing this process is therefore essential for the opti-mization of multiphase reactors. For a better understanding ofthe transfer mechanisms and subsequent chemical reactions, alaboratory-scale bubble column reactor was investigated.First, to characterize the flow field in the reactor, two differentmethods have been applied. The shadowgraphy technique isused for the characterisation of the bubbles (bubble diameter,velocity, shape or position) for various process conditions.This technique is based on particle recognition with backlightillumination, combined with particle tracking velocimetry(PTV). The bubble trajectories in the column can also be ob-tained in this manner. Secondly, the liquid phase flow has beenanalysed by particle image velocimetry (PIV). The combina-tion of both methods, delivering relevant informationconcerning disperse (bubbles) and continuous (liquid) phases,leads to a complete fluid dynamical characterization of thereactor, which is the pre-condition for the analysis of masstransfer between both phases.

Keywords Bubble column reactor . CO2 . PIV .

Shadowgraphy . Two-phase flow . Two-tracer-LIF

Nomenclature

de [mm] equivalent bubble diameterdB [mm] equivalent mean bubble diameterUB [m/s] absolute rising velocity of single bubblesUB,mean [m/s] absolute, mean rising velocity of bubblesfB [1/s] bubble formation frequencyVB [mm3] bubble mean volumeAB [mm2] bubble mean surface areaVL [m/s] mean liquid velocity

Dimensionless numbers

CB bubble centricityEo Eötvös numberFrb Froude number around a bubbleFrc Froude number in the columnMo Morton numberReb Reynolds number around a bubbleRec Reynolds number in the columnWeb Weber number around a bubbleWec Weber number in the column

1 Introduction

Bubble columns are widely used in the industry, for examplein bioprocesses or in water treatment. Usually, they areemployed for a chemical process, where the dispersed gascomes into reaction with the liquid phase through gas-liquidmass transfer. Quite often, the underlying transport processesare the limiting factor controlling reaction rate. Therefore,characterizing these processes is essential to optimize multi-phase reactors. Gas-liquid mass transfer is very strongly influ-enced by the flow field in the column; as a consequence,

* Péter Kovátspeter.kovats@ovgu.de

1 Laboratory of Fluid Dynamics and Technical Flows,Otto-von-Guericke-Universität Magdeburg, Universitätsplatz 2,D-39106 Magdeburg, Germany

Heat Mass Transfer (2018) 54:453–461DOI 10.1007/s00231-017-2142-0

understanding fluid dynamics of the dispersed and continuousphases is the first step toward process optimization.

Although bubble motion and behaviour have beenwidely studied, most of these studies considered the be-haviour of a single bubble [1–12]. Some experimentshave been reported in bubble swarms, but the bubbleswere either not uniform [13, 14] or have been generatedby one single inlet nozzle [1, 15, 16]. Additionally, mostpublications considered rectangular bubble columns, farfrom practical configurations [17–21]. Other experimentshave been carried out in cylindrical bubble columns [22,23] using Laser Doppler Anemometry, which is a pointmeasurement technique, and therefore is unable to inves-tigate simultaneously the whole flow field in the bubblecolumn at the same time. Though these studies havedelivered important information, it is therefore necessaryto go one step further.

For this purpose, the present study considers interactingbubble swarms in a cylindrical tank, generated by four iden-tical nozzles. In a first step, the relationship between bubblediameter and bubble velocity has been investigated in thecolumn for different water qualities and compared to the liter-ature. The shadowgraphy technique combined with ParticleTracking Velocimetry (PTV) was used to identify the bubblesand measure relevant bubble characteristics, like diameter,shape, position, velocity and trajectories. For the liquid phase,

Particle Image Velocimetry (PIV) was used to examine thesurrounding hydrodynamics.

Since these measurements ultimately aim at characterizingthe mass transfer from the gas to the liquid, the method fordetermining the absorbed gas concentrations in the liquid isfirst presented, together with the experimental setup. The de-scription of all other experimental results is then proposed.Since these results are useful to validate numerical models, astructured experimental database has been established, whereall results can be downloaded [24].

2 Experimental setup

2.1 Bubble column reactor

A bubble column with a diameter of 0.14 m and a heightof 0.8 m was employed for all experiments (number 4 inFigs. 2 and 3) [25, 26]. This bubble column was madefrom acrylic glass with a thickness of 4 mm and wassurrounded by a rectangular acrylic box. This box containsde-ionized water for a better refractive index matching.Preliminary experiments have shown that this gives betterresults, than to match the index of acrylic glass. The bot-tom of the bubble column can be changed from a singlenozzle gas outlet to a bottom with 4 in-line nozzles, whichare spaced by 2.2 cm. The nozzles are stainless steel cap-illaries with an inner diameter of 0.25 mm. Through thesenozzles, bubbles with a diameter of 2.5–3 mm are pro-duced. In all cases – except for filling height and waterquality measurements – the column was filled up with11.5 l de-ionised water. For the measurements with differ-ent pH values, NaOH was used to increase the pH value inthe column and HCl or CO2 gas to decrease it. The CO2

gas was supplied from a pressurised bottle to the initiallystagnant liquid phase. All the measurements were made atatmospheric pressure and room temperature.

The first reaction to be investigated in this reactor is themass transfer from CO2 bubbles into an initially basic(pH = 9) water-NaOH liquid phase. During this reaction,

a b cFig. 1 a raw image from 1stcamera imaging the pH-traceruranine; b raw image from 2ndcamera imaging the inert tracerpyridine 2; c treated image aftercalibration showing the pH. Allthree images show a region from100 to 320 mm from the bottomof the column

Fig. 2 Experimental setup for shadowgraphy

454 Heat Mass Transfer (2018) 54:453–461

the pH value is decreasing. In order to track the pH, the two-tracer laser-induced fluorescence technique (2 T–LIF) will beused [27]. For this purpose, the reactor is initially filled witha de-ionised water and NaOH mixture (pH = 9) that containsalso two fluorescent dyes. One of these dyes is changing itsfluorescence intensity with the pH (uranine), the other stayspassive (pyridine 2). With the help of this second dye, re-flections on the bubble surface, bubble shadows and the lasersheet inhomogeneities can be reduced drastically (see Fig. 1from left to right).

2.2 Shadowgraphy setup

For the shadowgraphy measurements, an Imager pro HS 4 MCCD camera (number 1. in Fig. 2) was used with a2016 × 2016 pixel resolution and 1279 Hz maximal frame rateat maximal resolution. The camera was equipped with a Carl-Zeiss Makro Planar T*2/50 mm ZF-I lens. To illuminate theinvestigation area, two Dedocool CoolH halogen lights wereused (number 2 in Fig. 2). To avoid damage of the CCD,because of the high illumination intensity, the light was diffusedon a white background (number 3 in Fig. 2). The images weretaken with a frame rate of 0.1 kHz in the lower 0.3 m of thebubble column. The experimental images were acquired andprocessed with DaVis software (LaVision). With this method,the influence of different parameters (gas flow rate, water qual-ity, pH and fill height) was investigated on the bubble flow.

2.3 PIV setup

For the PIV measurements one Imager Intense CCD cam-era (number 1 in Fig. 3) was applied with a 1376 × 1040pixel resolution and 10 Hz maximal frame rate. The cam-era was equipped with a Nikon AF Micro Nikkor 60 mmf/2.8D lens. As tracer particles, polymethyl methacrylate(PMMA) Rhodamin B particles were used with a meandiameter of 10 μm.

The particles were excited by a double-pulsed Nd:YAGlaser (Spectra Physics) at 532 nm (number 2 in Fig. 3).

The produced laser beam was expanded by a light sheetoptics (number 3 in Fig. 3). To supress the laser light andreflections on the bubbles surfaces a 537 nm longpassfilter was mounted onto the camera lens. The geometricalpositions were calibrated with a 3D calibration target. Theexperimental images were acquired with 6.6 Hz indouble-frame mode. The delay time between the twoframes was 5 ms. The flow rate of the CO2 gas was setup through a variable-area flow-meter to 8 l/h. Forpostprocessing of the raw images DaVis 8.2 (LaVision)has been used. For the vector calculation a cross-correlation (multi-pass, decreasing size) PIV algorithmwas used, with an interrogation window size from64 × 64 pixels to 32 × 32 pixels. To remove false vectorsand refine the vector field, especially in the vicinity andshadows of the bubbles, vector postprocessing was neces-sary, as described later in detail.

3 Postprocessing and results

3.1 Shadowgraphy

During one measurement, four series of 250 images weretaken for each parameter, which are listed in Table 1.Preliminary experiments showed, that this number is suffi-cient to get statistical convergence (Fig. 4).

For bubble size detection, a common shadowgraphy imageprocessingwas applied (Fig. 5). From the raw images (Fig. 5a)a previously recorded background was first removed (Fig. 5b)to enhance contrast. Then, the high intensity areas are recog-nized for each bubble (Fig. 5c) and their diameters are calcu-lated (Fig. 5d), as well as coordinates and centricityparameters.

For the bubble velocity detection a double-frame recordingwas made from the time series. Using an additional PTV al-gorithm [28], it was then possible to follow in the secondframe the movement of the bubbles that have been recognisedin the first frame. The produced vectors show the movement

Fig. 3 Experimental setup forPIV and measurement windows

Heat Mass Transfer (2018) 54:453–461 455

of each bubble and according to this, the bubble velocity (Fig.5d). With the help of these vectors, a trajectory can then bereconstructed for each bubble (Fig. 6).

As it can be seen in Fig. 7, showing a long-time acquisition(2.5 s), the bubbles follow a three-dimensional helical path, asexpected for this diameter, corresponding to region three in theclassification of [29].Mendelson divided themean diameters intofour regions as follows: Region 1: de< 0.7mm; region 2: 0.7 < d-e < 1.4 mm; region 3: 1.4 < de < 6 mm; region 4: de > 6 mm.

The considered bubbles are in the third region, with anaverage diameter of 2.2–3 mm (see Table 1 later). In thisregion the bubbles are no longer completely spherical, as ob-served also in the present experiments.

In order to check possible modifications in bubble behav-iour when varying process conditions, those have been mod-ified systematically, once at a time. First, the influence offilling height on bubble diameter and velocity was deter-mined. The images were taken for five different filling vol-umes of 3, 5, 7, 9 and 11.72 l (maximum filling volume). Nonoticeable differences could be observed between the fivecases. The average bubble diameter remained constant at2.4 mm and the average velocity was always 0.33 m/s.

Then, the influence of water quality was investigated. Theapplied water was always clear, not contaminated, but in dif-ferent qualities: 1) tap water, 2) de-ionised water, 3) distilledwater and 4) bi-distilled water. The results showed, that thewater quality did not influence at all bubble behaviour in ourcase. The average diameter and velocity were the same asmeasured previously for the different filling.

The next investigated parameter was the pH, since masstransfer determination requires a change in pH using two-tracer LIF. An acidic, a nearly neutral and a basic water solu-tion were tested. As expected, the CO2 dissolved best in the

basic liquid, so that at pH 12.5 the measured average bubblediameter was only 2.3 mm, compared to 2.4 (neutral case) and2.5 mm (acidic case). In contrast to the diameter, the averagevelocity showed nomeasurable differences. This result is con-firmed by the diagram shown for instance in [30] and also laterin Fig. 10, where the difference in the average velocity forbubble sizes between 2.3 and 2.5 mm is extremely small.

The last parameter varied was the gas flow rate. Differentvalues were set by a variable-area flow-meter (Yokogawa).The lowest investigated flow rate was 3.5 l/h, because this isthe lowest flow rate, where all the four nozzles still producedbubbles. Figures 8 and 9 show that the flow rate has a largeimpact on bubble size and velocity. The higher the flow rate,the larger is the bubble diameter, but also the broader is thebubble size distribution (BSD).

Concerning the average velocity, the effect is naturallyinverse. The smaller bubbles induced by a low flow rateshow higher velocities with a narrower distribution (Fig. 9).

Summarizing these results, it appears that only the gasflow rate has a pronounced influence on the velocity andsize of the bubbles in the investigated bubble column.The value of the pH does influence bubble size, but onlyin a very slight manner (less than 10% relative differ-ence) and has thus no noticeably impact on bubble

Fig. 6 Exemplary bubble trajectories, coloured with instantaneousvertical velocity component

a b c d

2 mm 1

2

Fig. 5 Shadowgraphy processing with two exemplary bubbles (1 and 2). For the better visibility the double frame recording was overlapped in oneimage. (a) – raw images, (b) – images after subtraction the background, (c) – bubble recognition, (d) – bubble diameter and calculated trajectories

Fig. 4 Bubble size distribution as function of averaged image numbers

456 Heat Mass Transfer (2018) 54:453–461

velocity. All other operating parameters do not influencethe bubble sizes and velocities at all (Table 1).

In the literature many correlations have been presented inan effort to predict bubble velocity as a function of bubblediameter (for instance [29, 31–33]). Fig. 10 shows the com-parison of the present experimental data for CO2 bubbles withthe velocities predicted with different correlations. Overall,the mean bubble diameter varied between 1.4 and 5 mm inour bubble column, with a standard deviation smaller than0.04 m/s for the respective bubble diameters. An excellentagreement is obtained for this particular region with all pub-lished correlations.

Concerning the shape of the curve, the best fit is ob-tained with the correlation from Rodríguez [31]. Thiscorrelation takes into account the viscous and surfacetension effects on the bubble rising velocity.

UB ¼ 1ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

1

V2T1

þ 1

V2T2

s ð1Þ

where VT1 is the bubble rising velocity when viscous effectsare important and when surface tension effects are

considerable, VT2 is the corresponding velocity.

VT pot ¼ 1

36

Δρgd2eμL

ð2Þ

VT1 ¼ VT pot 1þ 0:73667gdeð Þ1=2VT pot

" #1=2

ð3Þ

VT2 ¼ 3σρLde

þ gdeΔρ2ρL

� �1=2

ð4Þ

3.2 Liquid dynamics

Since the field of view of the camera covers only about onefourth of the overall bubble column height, four measurementwindows (Fig. 3W1-W4) were chosen along the column axis.In each window, 1000 double images were acquired by PIV,delivering 1000 individual, instantaneous velocity fields.

After the first processing step, the bubbles and theirshadows are clearly visible, due to a wealth of spurious

ba Fig. 7 Bubble trajectories infront view (a) and side view (b),coloured with instantaneousvertical velocity component

Fig. 9 Bubble velocities as function of gas flow rateFig. 8 Bubble size distribution (BSD) as function of gas flow rate

Heat Mass Transfer (2018) 54:453–461 457

Fig. 10 Comparison betweenpredictions from literature andown experimental data for CO2

bubble velocity as function ofbubble diameter

b a c

Fig. 11 Vertical velocity component (m/s) for one snapshot without anyvector postprocessing (a) and after postprocessing (b). The image Bb^ isreprinted on the right with an adapted colour scale showing only

physically meaningful values (c). The raw particle image is used asbackground for the location of the bubbles and of the shadows. Thesevelocity fields have been obtained at the W2 measurement window

Table 1 Measurement resultsshowing mean bubble diameters(mm) and mean bubble verticalvelocities (m/s)

Average bubblediameter

Average bubblevertical velocity

Filling volume (l) 11.72 2.4 0.33

9 2.3 0.33

7 2.4 0.33

5 2.4 0.34

3 2.3 0.34

pH 12.48, raised with NaOH 2.3 0.33

6 2.3 0.33

3.8, decreased with HCl 2.4 0.33

4, decreased with CO2 2.5 0.33

Gas flow rate (l/h) 3.5 2.3 0.33

5 2.5 0.33

7.5 2.9 0.31

10 3.1 0.31

Water quality 1× distilled 2.4 0.31

2× distilled 2.3 0.33

de-ionized 2.3 0.33

tap water 2.3 0.33

458 Heat Mass Transfer (2018) 54:453–461

vectors (Fig. 11 left, deep red and blue regions).Superimposing the raw image with the visible bubbles onthe PIV velocity field, it is clear that the optical distortioncomes from the bubbles; the laser sheet coming from the rightin these pictures, there are no tracer particles at all (within thebubbles), or the illumination is too weak (in the shadows ofthe bubbles). Thus, these regions cannot be processed appro-priately using a standard cross-correlation. To get rid of theseimperfections, it is necessary to use further, dedicated postprocessing steps.

For this purpose, an allowable vector range filter was firstemployed according to the generated Vx/Vy scatter plot. Then,with the help of a median filter, which computes the medianvector from the 8 neighbouring interrogation areas and thencompares the vector in the center with this median vector and

its deviation from the neighbouring vectors, the highest (but insome cases wrong) correlation peak was replaced by the sec-ond, third or fourth highest correlation peak, until a sufficientagreement is obtained within the neighbourhood. The currentvector position was disabled if none of the four peaks were inthe allowed range of the median vector. These few disabledareas were afterwards replaced with an interpolated valuefrom the 8 neighbours.

The result obtained in this manner can be seen exem-plari ly in Fig. 11 (center), once more with thesuperimposed raw image to locate the bubbles and theirshadows. For a better readability, the same velocity fieldis presented again with a different colour scale (nowadapted to the real maximum and minimum values) onthe right side of Fig. 11. In this figure, the vertical ve-locity component is represented for the second measure-ment window (Fig. 3). Regions with maximum instanta-neous liquid velocities of about 9 cm/s (but still

a

b

c

d

e

Vy

[m/s]

Fig. 13 Mean vertical velocitycomponent of the liquid (m/s) forthe full height of the column,combining the results of all fourmeasurement windows (a) andmean vertical velocity profiles(crosses), together with standarddeviation (dashed) in the high of30 mm (b), 200 mm (c), 400 mm(d) and 650 mm (e)

Fig. 12 Averaged vertical velocity component of the liquid phase for thesecond measurement window

Fig. 14 Flow field near the surface with velocity vectors

Heat Mass Transfer (2018) 54:453–461 459

noticeably lower than the mean bubble velocity, seeagain Fig. 10) can be recognized in the central part ofthe column. Near the walls, regions with negative veloc-ities appear, which are part of a descending flow (recir-culation loop of the liquid).

To obtain an average flow field, 1000 individual ve-locity fields were averaged after post processing for eachposition (see Fig. 12 for the second measurementwindow). As expected, the averaged images show analmost symmetric velocity field in the bubble column,with a large ascending part around the centre of the col-umn and a thin descending zone near the column walls.Average peak values are obviously noticeably lower thanthe instantaneous peak values, here below 6 cm/s.

Combining the results of all four measurement windows(obtained separately) a full view of the liquid flow field withinthe column can be obtained (Fig. 13).

At the bottom of the image, the four bubble nozzlesare clearly visible (Fig.13 a and b). Within the first0.3 m, as long as the bubble swarm is still concentrated,the mean vertical velocity reaches a maximum valueslightly below 6 cm/s (Fig.13 b and c). Above 0.3 m,the bubbles start to spread sideways and finally occupythe whole width of the column, the process finishing at aheight around 0.45 m. As a consequence of bubblespreading, the overall mean velocity is decreasing slight-ly (Fig. 13 d). In the top window, a clear and relativelylarge backflow near the reactor walls exists (Fig. 13 aand e), showing the recirculation loop of the liquid afterthe bubbles leave through the free surface.

This becomes even clearer when plotting the velocity vec-tors on top of the velocity magnitude (Fig. 14).

Many parameters are essential to characterize the processesoccurring in a bubble column. The most important ones aresummarized in Table 2, using the notations described in thenomenclature.

All these values, as well as the complete data setscontaining bubble diameters, bubble and liquid velocityfields, as well as the ongoing further LIF measurementresults are freely available through an online database,accessible under (http://www.uni-magdeburg.de/isut/LSS/spp1740). The authors do hope that this will be usefulas a source of information to provide validation dataand comparisons for theoretical and numerical studiesperformed in other groups.

4 Conclusions

In this paper, different optical measurement techniques havebeen combined to characterize in detail the fluid dynamics of abubble column reactor. Shadowgraphy combined with PTVwas used to investigate the bubble behaviour, while PIV tech-nique was applied to characterize the liquid flow.

The bubble behaviour has been examined as a function ofseveral process parameters. Only the gas flow rate inducedsignificant changes in bubble size and velocity. Increasingthe gas flow rate, large bubbles are generated, but travellingat lower velocities. Since bubble size and velocity directlyimpact mass transfer and mixing, it is clear that the gas flowrate is an essential parameter for this study. The obtained mea-surement data fit very well published correlations. Because ofthe three dimensional motion of the bubbles, planar shadow-graphy cannot resolve completely bubble movement. Toavoid this problem, a stereo-shadowgraphy setup with twocameras will be used in the future.

Using PIV, the flow field in the bubble reactor could beinvestigated in detail, considering four vertical lines of bubbleswarms. Combining four windows, a full description of thehydrodynamics in the whole column was obtained, showingmuch lower velocities compared to the bubbles, with a clearrecirculation loop near the reactor walls.

Under the considered conditions, the problem of bubbleshadows interacting with the laser sheet has finally beensolved using a dedicated post processing procedure. To im-prove further signal-to-noise ratio, applying a second lasersheet from the other side would be advantageous.

After this hydrodynamic characterization, the analysis ofmass transfer from the gas bubbles to the liquid is now beingdone, using 2-Tracer-LIF to reduce the optical influence of thebubbles on the measurement results.

This experimental study is part of a Priority Program of theDFG (SPP 1740 BReactive Bubbly Flows^). In the frame ofthe program a virtual data bank has been created [24] to shareall relevant data with the scientific community, supportingvalidation of numerical simulations and comparisons with re-sults from other groups. Such validations are currently in workand will be published soon.

Acknowledgements This work has been financially supported by theGerman research foundation (DFG) in the framework of the SPP 1740BReactive Bubbly Flows^ under project number ZA-527/1-1.

Table 2 Main parameters of theinvestigated bubble column de

[mm]

dB[mm]

CB

[−]UB

[m s−1]

UB,mean

[m s−1]

fB[s−1]

VB[mm3]

AB

[mm2]

VL

[m s−1]

1.44–6.3 2.6 0.443 0.044–0.56 0.31 145–160 9.2 21.24 0.045

Reb Rec Web Wec Frb Frc Eo Mo

803 6368 3.4 3.94 3.78 0.015 0.91 2.55E-11

460 Heat Mass Transfer (2018) 54:453–461

The authors would also like to thank the students Niklas Brandt and TimAndré Kulbeik for their help in doing the experiments. The help of C. Kisowand S. Herbst for building the bubble column is gratefully acknowledged.

References

1. Besbes S, El Hajem M, Ben Aissia H, Champagne JY, Jay J (2015)PIV measurements and Eulerian–Lagrangian simulations of the un-steady gas–liquid flow in a needle sparger rectangular bubble column.ChemEng Sci 126:560–572. https://doi.org/10.1016/j.ces.2014.12.046

2. Dietrich N, Francois J, JimenezM, Cockx A, Guiraud P, Hébrard G(2015) Fast measurements of the gas-liquid diffusion coefficient inthe gaussian wake of a spherical bubble. Chem Eng Technol 38(5):941–946. https://doi.org/10.1002/ceat.201400471

3. Hosoda S, Tryggvason G, Hosokawa S, Tomiyama A (2015)Dissolution of single carbon dioxide bubbles in a vertical pipe. JChem Eng Jpn 48(6):418–426. https://doi.org/10.1252/jcej.14we241

4. Huang J, Saito T (2015) Influence of bubble-surface contaminationon instantaneous mass transfer. Chem Eng Technol 38(11):1947–1954. https://doi.org/10.1002/ceat.201500056

5. Huang J, Saito T (2016) Influences of gas-liquid interface contam-ination on bubble motions, bubble wakes, and instantaneous masstransfer. Chem Eng Sci. https://doi.org/10.1016/j.ces.2016.05.013

6. Jimenez M, Dietrich N, Grace JR, Hébrard G (2014) Oxygen masstransfer and hydrodynamic behaviour in wastewater: determinationof local impact of surfactants by visualization techniques. WaterRes 58:111–121. https://doi.org/10.1016/j.watres.2014.03.065

7. JimenezM, Dietrich N, Hébrard G (2013)Mass transfer in the wake ofnon-spherical air bubbles quantified by quenching of fluorescence.ChemEng Sci 100:160–171. https://doi.org/10.1016/j.ces.2013.01.036

8. Nagami Y, Saito T (2013) Measurement of modulation induced byinteraction between bubble motion and liquid-phase motion in thedecaying turbulence formed by an oscillating-grid. Particuology11(2):158–169. https://doi.org/10.1016/j.partic.2012.06.012

9. Rollbusch P, Bothe M, Becker M, Ludwig M, Grünewald M, SchlüterM, Franke R (2015) Bubble columns operated under industrially rele-vant conditions - current understanding of design parameters. ChemEng Sci 126:660–678. https://doi.org/10.1016/j.ces.2014.11.061

10. Saito T, Toriu M (2015) Effects of a bubble and the surroundingliquid motions on the instantaneous mass transfer across the gas-liquid interface. Chem Eng J 265:164–175. https://doi.org/10.1016/j.cej.2014.12.039

11. Valiorgue P, Souzy N, Hajem ME, Hadid HB, Simoëns S (2013)Concentration measurement in the wake of a free rising bubbleusing planar laser-induced fluorescence (PLIF) with a calibrationtaking into account fluorescence extinction variations. Exp Fluids54(4). https://doi.org/10.1007/s00348-013-1501-y

12. Wegener M, Paul N, Kraume M (2014) Fluid dynamics andmass transfer at single droplets in liquid/liquid systems. Int JHeat Mass Transf 71:475–495. https://doi.org/10.1016/j.ijheatmasstransfer.2013.12.024

13. SatheMJ,Mathpati CS, Deshpande SS, Khan Z, Ekambara K, JoshiJB (2011) Investigation of flow structures and transport phenomenain bubble columns using particle image velocimetry and miniaturepressure sensors. Chem Eng Sci 66(14):3087–3107. https://doi.org/10.1016/j.ces.2011.04.002

14. Sathe MJ, Thaker IH, Strand TE, Joshi JB (2010) Advanced PIV/LIF and shadowgraphy system to visualize flow structure in two-phase bubbly flows. Chem Eng Sci 65(8):2431–2442. https://doi.org/10.1016/j.ces.2009.11.014

15. Liu ZL, Zheng Y, Jia L, Zhang QK (2005) Study of bubble inducedflow structure using PIV. Chem Eng Sci 60(13):3537–3552. https://doi.org/10.1016/j.ces.2004.03.049

16. Sathe M, Joshi J, Evans G (2013) Characterization of turbulence inrectangular bubble column. Chem Eng Sci 100:52–68. https://doi.org/10.1016/j.ces.2013.01.004

17. Bitog JPP, Lee IB, Oh HM, Hong SW, Seo IH, Kwon KS (2014)Optimised hydrodynamic parameters for the design ofphotobioreactors using computational fluid dynamics and experi-mental validation. Biosyst Eng 122:42–61. https://doi.org/10.1016/j.biosystemseng.2014.03.006

18. Dietrich N, Loubière K, Jimenez M, Hébrard G, Gourdon C (2013)A new direct technique for visualizing and measuring gas-liquidmass transfer around bubbles moving in a straight millimetricsquare channel. Chem Eng Sci 100:172–182. https://doi.org/10.1016/j.ces.2013.03.041

19. Haggerty R, Argerich A, Martí E (2008) Development of a Bsmart^tracer for the assessment ofmicrobiological activity and sediment-waterinteraction in natural waters: the resazurin-resorufin system. WaterResour Res 44(4):W00D01.https://doi.org/10.1029/2007WR006670

20. Sanchez-Forero DI, Costa KK, Amaral RL, Taranto OP, Mori M(2015) Influence of liquid viscosity and solid load in a three-phasebubble column using stereoscopic particle image velocimetry (stereo-PIV). Chem Eng Trans 43:1579–1584. https://doi.org/10.3303/CET1543264

21. Zhou X, Doup B, Sun X (2013) Measurements of liquid-phaseturbulence in gas-liquid two-phase flows using particle imagevelocimetry. Meas Sci Technol 24 (12):art.No. 125303. doi:https://doi.org/10.1088/0957-0233/24/12/125303

22. Becker S, De Bie H, Sweeney J (1999) Dynamic flow behaviour inbubble columns. Chem Eng Sci 54(21):4929–4935. https://doi.org/10.1016/S0009-2509(99)00214-6

23. Pfleger D, Becker S (2001) Modelling and simulation of the dy-namic flow behaviour in a bubble column. Chem Eng Sci 56(4):1737–1747. https://doi.org/10.1016/S0009-2509(00)00403-6

24. Reactive Bubbly Flows - Bubble Column Reactor Database.(2015). http://www.uni-magdeburg.de/isut/LSS/spp1740/

25. Kováts P, Thévenin D, Zähringer K (2017) Investigation of masstransfer and hydrodynamics in a model bubble column. Chem EngTechnol. https://doi.org/10.1002/ceat.201600679

26. Zähringer K, Wagner L-M, Kováts P, Thévenin D (2014)Experimental characterization of the mass transfer from gas to liq-uids in a two-phase bubble column. Paper presented at the 7thinternational berlin workshop on transport phenomena with mov-ing boundaries and more, berlin, Germany, 30th-31st October 2014

27. Lehwald A, Thévenin D, Zähringer K (2010) Quantifying macro-mixing and micro-mixing in a static mixer using two-tracer laser-induced fluorescence. Exp Fluids 48(5):823–836. https://doi.org/10.1007/s00348-009-0769-4

28. Bordas R, Roloff C, Thevenin D, Shaw RA (2013) Experimentaldetermination of droplet collision rates in turbulence. New J Phys15:045010. https://doi.org/10.1088/1367-2630/15/4/045010

29. Mendelson HD (1967) The prediction of bubble terminal velocitiesfrom wave theory. AICHE J 13:250–253. https://doi.org/10.1002/aic.690130213

30. Clift R, Grace JR, Weber ME (1978) Bubbles, drops, and particles.Academic Press, New York

31. Baz-Rodriguez S, Aguilar-Corona A, Soria A (2012) Rising veloc-ity for single bubbles in pure liquids. Rev Mex Ing Quim 11(2):269–278

32. Fan LSTK (1990) Bubble wake dynamics in liquids and liquid-solid suspensions. Butterworth-Heinemann, Massachusetts, USA

33. Lehrer IH (1976) A rational terminal velocity equation for bubblesand drops at intermediate and high Reynolds numbers. J Chem EngJpn 9:237–240

Heat Mass Transfer (2018) 54:453–461 461