Embed Size (px)
Citation preview
Chemical Engineering Science 92 (2013) 190–197
Contents lists available at SciVerse ScienceDirect
Chemical Engineering Science
0009-25
http://d
n Corrnn Cor
E-m
karin.sc
journal homepage: www.elsevier.com/locate/ces
Droplet break-up mechanism in premix emulsification usingpacked beds
Akmal Nazir n, Remko Marcel Boom, Karin Schroen nn
Wageningen University, Food Process Engineering Group, Bomenweg 2, 6703 HD Wageningen, The Netherlands
H I G H L I G H T S
c Oil-in-water emulsions were successfully prepared using packed beds.c Interstitial void size and flow velocity determined the droplet break-up.c Droplet break-up either dominated by constriction (Reo40) or inertia (Re440).
a r t i c l e i n f o
Article history:
Received 12 November 2012
Received in revised form
7 January 2013
Accepted 11 January 2013Available online 19 January 2013
Keywords:
Emulsion
Premix membrane emulsification
Homogenization
Packed bed
Porous media
Droplet break-up mechanism
09/$ - see front matter & 2013 Elsevier Ltd. A
x.doi.org/10.1016/j.ces.2013.01.021
esponding author. Tel.: þ31 317 482240; fax
responding author. Tel.: þ31 317 482231; fa
ail addresses: [email protected], akmal.naz
[email protected] (K. Schroen).
a b s t r a c t
Some emulsification techniques based on microstructures are known for the monodispersity of
produced droplets, however, they lack in scalability. The techniques that are able to produce emulsions
in larger amounts do not usually produce monodispersed droplets. We here report on a specific
technique that has the potential to combine the best of both worlds: premix emulsification using a
packed bed of differently sized glass beads (55, 65, 78 and 90 mm) supported by a metal sieve. The
production of oil-in-water emulsions was targeted, and the process conditions especially related to
internal structure of the porous media like interstitial void size and bed height were investigated.
The Reynolds number, Re, was used to characterize the flow inside the packed bed consisting of
asymmetric pores following a tortuous path inside the porous media. The void size and the flow
velocity determined the droplet break-up. Two droplet break-up mechanisms were identified: either
dominated by constriction (Reo40) or inertia (Re440). Droplets below 5 mm (droplet to void size
ratioE0.2) could easily be produced; having relatively narrow droplet size distribution (droplet
spanE0.75). The measured fluxes were comparable to the highest reported flux values for premix
membrane emulsification studies. Statistically significant scaling relations were established for the
studied process conditions.
& 2013 Elsevier Ltd. All rights reserved.
1. Introduction
Emulsions are a part of many aspects of daily life and havegained a lot of interest especially related to their desired dropletsize (distribution). Food, chemical, cosmetics and pharmaceuticalindustries are among the main sectors in which emulsions havefound numerous applications (Chappat, 1994). Smaller dropletsizes and appreciable droplet uniformity is generally required fora good structural stability of a product. To further prolong theshelf-life it is also possible to adjust physical properties, likeviscosity and density, of the dispersed and the continuous phases
ll rights reserved.
: þ31 317 482237.
x: þ31 317 482237.
[email protected] (A. Nazir),
that will prevent or at least delay (the onset of) creaming, sedi-mentation or coalescence over time.
In the last decades, several microstructured emulsificationsystems have been proposed with the intent of having bettercontrol on the droplet size and droplet size distribution, com-pared to the conventional energy intensive equipment. Amongthem, membrane emulsification is prominent, either in cross-flow(Nakashima et al., 1991) or dead-end (premix) mode (Suzukiet al., 1998). In cross-flow emulsification, the dispersed phase ispushed through the membrane pores into the flowing continuousphase on the other side of the membrane. This technique isknown to yield high droplet monodispersity at low dispersedphase fractions. To produce emulsions at higher dispersed phasefractions, premix membrane emulsification in which a coarseemulsion is pushed through the membrane to get a fine emulsion,is an interesting alternative given its low energy input. Inaddition, a number of microfluidic devices have been proposed,
A. Nazir et al. / Chemical Engineering Science 92 (2013) 190–197 191
such as microchannels (Kawakatsu et al., 1997), T- (Thorsen et al.,2001) and Y-junctions (Steegmans et al., 2009), flow focusingdevices (Anna et al., 2003), and edge-based droplet generation(EDGE) devices (Van Dijke et al., 2010). Each technique has itsown pros and cons, but up- or out-scaling of these devices is still amajor challenge.
To evaluate the potential of an emulsification process, inaddition to droplet size (distribution), other factors have to beconsidered like production rate, energy efficiency, and ease ofdesign etc. Several publications report on premix emulsificationusing different membranes and under different operating condi-tions. The production rate is quite high in premix emulsification,and the emulsions obtained are relatively monodispersed(Altenbach-Rehm et al., 2002; Suzuki et al., 1996; Vladisavljevicet al., 2004). The mean droplet diameter can be tuned through themembrane pore size and the operating parameters especially thetransmembrane pressure. However, its sensitivity to depth foulingof the membrane, and in tandem with this, the inaccessibility ofthe membrane pores to cleaning agents is regarded as the largestdisadvantage of premix emulsification, which in many cases isdecisive.
Application of a packed bed of small glass beads (supported bya metal sieve) results in a ‘‘dynamic membrane’’ that is morpho-logically similar to the conventional membranes (Van der Zwanet al., 2008). The system has a great advantage that the particlescan be easily cleaned by disintegrating the bed, and then the bedcan be formed anew. Van der Zwan et al. (2008) compared theirresults with this system to those of Vladisavljevic et al. (2006)with membranes, and found that the results were in goodagreement with a low dispersed phase volume fraction of hex-adecane in an aqueous phase. Hence, this system can be used as amodel for premix membrane emulsification.
In this investigation, we took the packed bed system muchfurther and investigated glass beads of different sizes focusing onthe internal structure of the bed. The droplet break-up mechan-ism was explained in terms of a change in flow behavior insidethe porous media, and existing scaling relations were successfullymodified to relate all operating conditions to the droplet sizeproduced.
2. Experimental
2.1. O/W premix preparation
An O/W premix emulsion consisting of 5% n-hexadecane (99%for synthesis, MERCK) in Milli-Q water having 0.5% v/v Tween-20(for synthesis, MERCK) was used. For each experiment, 300 ml ofcoarse premix was prepared with an ultra-turrax homogenizer(IKAs T-18 basic) operated at 3500 rpm for 671 min. This led toreproducible starting emulsions for our experiments havingSauter mean droplet diameter, d32, typically around 30 mm anda droplet span, d, of 0.9.
2.2. Emulsification setup
The emulsification setup is shown in Fig. 1. The pressure vesselcontaining the premix was connected to a module (a Plexiglascolumn) having a packed bed of glass beads on top of a supportsieve (described below) with an effective surface area of 1.43 cm2.The support sieve was held in place by two rubber o-rings (aboveand below) at the bottom junction of the column. To properly wetthe system with the continuous phase, a small amount ofcontinuous phase was introduced inside the module. Then itwas turned upside down for few times and placed vertically tolet the glass beads settle down in the form of a bed. The emulsion
vessel (containing the premix) was pressurized with nitrogenkeeping the valve connected to the column opened.
The emulsification was started by opening the outlet valve ofthe module and homogenized emulsion was collected in a beakerplaced on an electrical balance for digitally recording the increasein mass every second. The flux, J, across the packed bed wascalculated from the mass flow rate, fm, using the relation:
J¼fm
reA, ð1Þ
where re is the emulsion density, and A is the effective surfacearea of the packed bed. The emulsification process was repeatedup to five cycles.
The coarse premix and all homogenized emulsion sampleswere analyzed for droplet size (distribution) with a laser diffrac-tion particle size analyzer (Mastersizer 2000, Malvern Instru-ments Ltd., UK). The machine takes three readings, and theaverage values of droplet size and droplet size distribution (span)were determined and used in the analysis.
2.3. Support sieve
Nickel sieves (kind gift of Stork Veco BV, the Netherlands)having straight-through rectangular pores were used as a supportfor the glass beads. A sieve having an average pore size of11.6 mm�331 mm was used that was thick enough (350 mm) toprovide a good support to the bed and withstand the appliedpressure. The SEM images of top and bottom view of the sieve areshown in Fig. 2.
2.4. Packed bed
Hydrophilic glass beads (100HFL, Pneumix SMG-AF) havingdiameters between 30–200 mm were used in this study. Threemetal test sieves of pore sizes 90, 75 and 63 mm were put on topof each other on an electrical shaker (JEL Engelsmann AG,Ludwigshafen, Germany) with the sieve having the largest poresize at the top and the one with the smallest pore size at thebottom. Four different size fractions of the glass beads wereobtained after sieving the stock beads. The particle size (distribu-tion) of the fractions obtained was also analyzed with a laserdiffraction particle size analyzer (Mastersizer 2000, MalvernInstruments Ltd., UK). Fig. 3 shows the particle size distributionof each fraction. The span of the distributions in all the cases wasaround 0.6570.01.
The particle (rp) and bulk (rb) densities of each fraction weremeasured in water and in air, respectively, and subsequently theporosity, E(¼1–rb/rp), was calculated. For all the fractions theporosity was around 0.4.
The capillary model for fixed beds proposed by Comiti andRenaud (1989) was used to determine the structural properties ofthe porous media like interstitial void and tortuosity. This modelassumes the packed beds to consist of a bundle of identicalcylindrical tortuous pores, and this concept has been adoptedby various researchers (Kiss et al., 2011; Legrand et al., 2001;Morancais et al., 1999) to characterize the emulsification processin static mixers.
The interstitial void diameter, dv, was defined as
dv ¼4E
Avð1�EÞ, ð2Þ
where Av is the dynamic specific surface area, a ratio of wettedsurface area to volume of solid, and is related to glass bead
Fig. 1. Schematic representation of experimental setup.
200 µm100 µm
Fig. 2. SEM image of the support sieve used in this study: (left) top view, plane
surface and (right) bottom view, structured surface (Nazir et al., 2011).
0
5
10
15
20
25
30
000100101
Volu
me
[%]
Particle size [µm]
Fig. 3. Particle size distribution of different fractions of glass beads: (J) 55, (D) 65,
(&) 78 and (B) 90 mm.
A. Nazir et al. / Chemical Engineering Science 92 (2013) 190–197192
diameter, db, by
Av ¼6
db: ð3Þ
The bed tortuosity, t, was calculated as
t¼ 1þq lnð1=EÞ, ð4Þ
where q¼0.41 for tightly packed spheres. The average interstitialvoid velocity, vv, was defined as
vv ¼votE
, ð5Þ
where vo is the superficial velocity equal to J/E. The flow inside thepacked bed was characterized using the Reynolds number, Re,which is a ratio of the inertial to the viscous forces defined as
Re¼revvdv
Ze
, ð6Þ
where Ze is the emulsion viscosity.The energy density, EV, defined as energy input per unit volume
of emulsion is related to pressure drop, DP, through
EV ¼P
fV
¼DPfV
fV
¼DP, ð7Þ
where P is the power input and fV is the volume flow rate. Formore than one pass the energy densities of all the passes arecumulative.
Even though droplet break-up may occur when no glass beadsare used and only the metal sieve is present, this is only true forvery high pore velocities (Nazir et al., 2011) which were notattained in this investigation. Thus, break-up by the sieve can besafely disregarded. The droplet break-up is expected to occur onlydue to the packed bed. Contrary to Van der Zwan et al. (2008)who also considered the metal sieve, we here took only thepressure drop over the bed for droplet size analysis, as wepreviously established that the pressure drop over the membranewe use is extremely small. The pressure drop over a bed of glassbeads of diameter db at a bed height of H was calculated using the
A. Nazir et al. / Chemical Engineering Science 92 (2013) 190–197 193
Ergun equation:
DP¼150ð1�EÞ2ZevoH
E3d2b
þ1:75ð1�EÞrev2
oH
E3db: ð8Þ
3. Results and discussion
3.1. Effect of bead size and bed height
The bead size and bead packing arrangement are directlyrelated to the size of the interstitial voids between the beads.These interconnected interstitial voids can be seen as intercon-nected asymmetric capillaries that follow an irregular paththrough the packed bed, somewhat comparable to pores inconventional membranes; especially to those in ceramic mem-branes which are prepared by sintering a packed bed of individualceramic particles similar to those used in our packed bed. The sizeof the beads constituting the bed, thus, is an important parameterto be considered for understanding the emulsification processusing a packed bed.
The experiments were carried out using glass beads of differ-ent average diameters i.e., 55, 65, 78 and 90 mm at a constant bedheight of 2.5 mm with an applied pressure of 200 kPa. The dropletdiameter obtained using different glass beads was made dimen-sionless by dividing the droplet diameter by the respective voiddiameter (Eq. (2)). These values are plotted in Fig. 4 against thevoid size. A small droplet to void size ratio was obtained for55 mm glass beads, which then increased for 65 mm glass beads.A further increase in void size again resulted in a smaller ratio.
It is remarkable that the droplet size is considerably smallerthan the interstitial void size irrespective of the applied transmem-brane pressure; the droplet to void size ratio is as low as 0.2. Invarious premix emulsification studies, droplet to pore size ratiosover a range from 0.2 to 4.1 have been reported (see Table A.2 in
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0 10 20 30 40 50
d 32/dv[-]
dv [µm]
Fig. 4. Effect of interstitial void diameter, dv, on dimensionless droplet diam
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0 5 10 15 20 25
d 32/d v
[-]
H [mm]
Fig. 5. Effect of bed height, H, on dimensionless droplet diameter,
Nazir et al., 2010); thus in this work we are at the low end of thisspectrum. In cross flow membrane emulsification the droplet size istypically 3 to 10 times the pore size (Katoh et al., 1996; Mine et al.,1996; Peng and Williams, 1998; Williams et al., 1998), andnarrower pores are needed to produce the same droplet size.
The largest droplet size reduction took place till the third pass,and after this pass there was only minor decrease in the dropletsize. The span values depended strongly on the void size, anddecreased with smaller void sizes. The span decreased withrepeated passes through the bed, especially in case of smaller voidsize. However, for larger void sizes a small increase in span wasobserved which means that given the size of the droplets, thisparticle size is no longer effective in breaking up these droplets,while smaller particles are still effective in doing so. This wasinvestigated further by varying the bed height for one particle size.
Emulsification was carried out with 78 mm beads using differ-ent bed heights from 1–20 mm at an applied pressure of 200 kPa.In Fig. 5, the dimensionless droplet diameter and droplet span areplotted against the bed height used. The results show an increasein the droplet size with the bed height. The increase in dropletsize is attributed to a decrease in void velocity with increasingbed height, resulting in less shear force on the droplets, as wasalso demonstrated by Van der Zwan et al. (2006) who usedmicrofluidic systems to vary the ‘bed height’. Moreover, thechances that droplets meet inside the packed bed increase withincreasing bed height, possibly leading to coalescence.
The span values decreased strongly with the increasing bedheight up to a certain limit, after which it remained almost constant.Also here the span seemed to increase very slightly with increasingnumber of passes, but this effect is small. The observed effects aremost probably a combination of break-up and coalescence thatultimately cancel out upon reaching an average droplet size.
We further investigated the effect of the flow velocity insidethe porous media on the final droplet size and uniformity usingthe Reynolds number that characterizes the flow inside the
0.6
0.7
0.8
0.9
1.0
1.1
0 10 20 30 40 50
[-]
dv [µm]
��
eter, d32/dv, and droplet span, d: (&) 1st, (D) 3rd pass and (B) 5th pass.
0.6
0.7
0.8
0.9
1
1.1
1.2
0 5 10 15 20 25
��[-]
H [mm]
d32/dv, and droplet span, d: (&) 1st, (D) 3rd and (B) 5th pass.
0.0
0.1
0.2
0.3
0.4
0.5
d 32/d v
[-]
Re[-]
0.4
0.6
0.8
1.0
1.2
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80
��[-]
Re[-]
Decreasing dvIncreasing H
Decreasing dvIncreasing H
Fig. 6. Dimensionless droplet diameter, d32/dv, and droplet span, d, as a function of Reynolds number, Re: (D) void size varied, dv, and (J) bed height varied, H.
Table 1Droplet size and flux obtained after different passes through the 55 mm glass
beads at different applied pressures.
Applied
pressure
[kPa]
Pass through the packed bed
1 3 5
d32
[mm]
J
[m3 m�2 h�1]
d32
[mm]
J
[m3 m�2 h�1]
d32
[mm]
J
[m3 m�2 h�1]
50 14.1 30.0 8.0 34.9 6.5 33.5
100 11.0 157.9 5.9 99.5 5.3 128.6
150 8.9 247.4 5.6 241.9 5.2 207.8
200 8.2 279.8 5.3 291.9 4.8 309.3
250 8.1 388.5 5.2 303.2 4.5 340.1
300 7.3 433.6 4.8 456.6 4.3 482.0
400 7.5 505.3 4.7 585.8 4.2 653.0
500 5.6 636.8 4.3 763.4 3.6 723.1
0
5
10
15
20
25
0.1 1 10 100
Volu
me
[%]
Dropet diamter [µm]
Fig. 7. Droplet size distribution as a function of applied pressure: (&) 50, (þ) 100,
(D) 200, (� ) 300, (B) 400 and (J) 500 kPa.
A. Nazir et al. / Chemical Engineering Science 92 (2013) 190–197194
porous media. We here show two series of experiments, one inwhich the bed height was varied (yielding different hydrody-namic resistances and therefore yielding different velocities andReynolds numbers) and one in which different sized beads wereused in the bed. The final droplet to void size ratio and the spanobtained after the fifth pass of each experiment are plottedagainst the respective Reynolds number (Fig. 6).
Both data series start off separately at relatively low Reynoldsnumber but then merge at Reynolds number above 40. At lowReynolds numbers, the void size has a greater influence on thedroplet size than the velocity, indicating that the constrictionneeded for spontaneous droplet snap-off is here dominant in thesize reduction; this is corroborated by the reduction of the span.With increasing Reynolds number both curves are the same,which indicates that the break-up mechanism becomes similar(a decreasing bed height or an increasing void size) and this is dueto dominance of inertial effects (i.e., droplet break-up due to localshear forces). This inertial droplet break-up region is characterizedby a decrease in droplet uniformity.
This establishes that the internal structure of a packed bed isimportant in determining the droplet size and uniformity. Withsmall interstitial voids, the constrictions mainly determine thedroplet snap-off, which is the mechanism found in microchanneldevices. However, contrary to microchannel devices the shearstress could also be relevant to droplet disruption to some extentdepending upon the interstitial void flow velocities. With largervoid sizes and sufficient flow, the local shear forces are so large,that these determine the break-up of the droplets. This is similarto the mechanisms found in T- and Y-junctions, and in flow focusingand co-flow microdevices.
As more uniform droplets were produced in spontaneous dropletforming region at low Re, further experiments were carried outusing 55 mm glass beads at different flow velocities. The subsequentdroplet size (distribution) is discussed in the next section.
3.2. Droplet break-up in spontaneous droplet formation dominated
region
The droplet size and the corresponding flux obtained afterdifferent passes through the bed using different applied pressuresare shown in Table 1. The droplet size decreases with increasingapplied pressure as would be expected at increased shear on thedroplets inside the voids as a result of higher flow velocities andalso due to an increased number of active voids. At all appliedpressures and especially at elevated pressures, the droplet sizereduction was highest after the first pass. Significant furtherreduction was found up till the third pass after which only minorreduction was found. The flux values are reasonably high and ofcourse have direct dependency on the pressure applied. Theobtained fluxes are comparable to the highest reported flux values
for premix membrane emulsification (while the size distributionremains much narrower as discussed in the next section) andaround 100–1000 higher than those with cross flow membraneemulsification (Nazir et al., 2010). There is no clear trend of the fluxover the different passes, which we think is due to the inherentvariability in the internal structure of a packed bed that may existeven among repeated emulsification passes in a single experiment.
The droplet size distributions after the final pass through thebed are shown in Fig. 7. The distribution is wide at lowerpressures (r100 kPa) and becomes narrower (a span valuetypically around 0.75) with a moderate increase in pressure. Afurther increase in pressure (4400 kPa) resulted again in a widerdistribution and in the formation of a small second peak with
10 10 10 10 10 10 10 102 3 4 5 6 7 8 910
10
10
10
-1
0
1
2
d 32[µ
m]
EV [J m-3]
Fig. 9. Sauter mean droplet diameter, d32, as a function of energy density, EV, for
various emulsification devices: (þ) grooved microchannel (Sugiura et al., 2000),
(B) straight-through microchannel (Kobayashi and Nakajima, 2002; Kobayashi
et al., 2002), (� ) EDGE emulsification (van Dijke et al., 2009), (&) Y-junction
(Steegmans, 2009), (J) premix emulsification using 55 mm glass beads (this
research), (K) cross-flow membrane emulsification (Lambrich and Schubert,
2005) (’) flat valve homogenizer (Lambrich and Schubert, 2005), (D) orifice valve
(Lambrich and Schubert, 2005), and (m) microfluidizer (Lambrich and Schubert,
2005).
A. Nazir et al. / Chemical Engineering Science 92 (2013) 190–197 195
dropletso1 mm. These might be regarded as satellite dropletsproduced as a result of increased void velocity at higher pressures(Rayleigh instability), and are in agreement with droplet beak-upby flow (either shear or extensional).
3.3. Scaling relations
The droplet disruption in a continuous emulsification methodcan be correlated to the energy density, which is the amount ofenergy applied per unit volume of emulsion in the dispersingzone of an emulsification device (Karbstein and Schubert, 1995).In case of emulsification using a packed bed of glass beads, theenergy density is related to the pressure drop across the bedthrough the relation given in Eq. (7) and is cumulative over all thepasses. For the relation of the energy density, EV (bar) to theSauter mean droplet size, d32 (mm), the following equation isproposed:
d32
dv¼ a E�bV
H
D
� �gð9Þ
where a, b and g are fit parameters. This is a modified form of theequation that was originally derived by Van der Zwan et al. (2008)for packed beds. As energy density is cumulative over all thepasses, it also covers the effect of the number of passes. Further,to characterize the emulsification process using differently sizedglass beads, the droplet to void size ratio, d32/dv, is used here. Thedv is the calculated interstitial void diameter based on the glassbead size, db, as discussed in the theory section. The H/D is theheight to diameter ratio of the packed bed.
The droplet size was fitted to the energy density and otherprocess related parameters like void size and bed height for eachpass through the bed. In Table 2, the calculated values of the fitparameters are shown along with their standard deviations. Oneshould bear in mind that the particle sizes, bed heights, andpressures are all incorporated in the data shown in Fig. 8
Table 2Values, standard deviations, and correlation coefficients of fit parameters of Eq.
(9).
Fit parameters Values and standard
deviations
Correlation coefficients
a b g
a 0.69370.034 1.00 0.88 0.70
b 0.17270.017 0.88 1.00 0.49
g 0.35170.020 0.70 0.49 1.00
0
5
10
15
20
25
0 5 10 15 20 25
d 32(E
xper
imen
tal)
[µm
]
d32 (Eq. 9) [µm]
Fig. 8. Experimental values of Sauter mean diameter, d32, plotted against the
result of Eq. (9) using fit parameters given in Table 2. See Table 3 for explanation of
the symbols.
(Table 3). The experimental and modeled data show reasonableagreement, especially given the large variations in process con-ditions that were explored. As a result of internal structure of theporous media the process is comparable regarding energy usage,irrespective of the specific break-up mechanism, and can bedescribed by Eq. (9). As shown in Table 2, the standard deviationsof the fit parameters are less than 10% and they are not correlatedwith each other.
The Sauter mean diameter of the droplet produced using glassbeads and the corresponding energy input is plotted in Fig. 9, andcompared with various microstructured and conventional emul-sification devices from literature. Obviously, each device has itsown energy requirement principally depending upon the specificdroplet break-up mechanism taking place. Compared to emulsi-fication techniques as Y-junction and flat valve homogenizers, theenergy requirement of the packed bed system is less. Apart fromthe energy density comparison relative to the droplet sizeproduced, this plot also yields information on the range of dropletsizes that a technique may give. This is especially true in case ofmicrofluidic devices (e.g., microchannel an EDGE emulsification)where there is a narrow working region for stable dropletformation. Although it is useful to know the specific energyrequirement of a certain process, we should keep in mind thatwe cannot rank different processes by only looking at their energyefficiencies. In addition, for example, droplet size (distribution),production rate, and the robustness of a system behaves underdifferent product properties, etc., are very important.
The results discussed in this paper show that the emulsifica-tion using a packed bed of glass beads is a technique withpotential for the controlled production of emulsions. The systemoffers more flexibility (in terms of desired droplet size andproduction rate) as the internal structure of the packed bed canbe easily regulated using differently sized particles. In thatrespect the packed bed (a dynamic membrane) outperforms theclassical membranes. Compared to microstructured emulsifica-tion systems, the up-scaling is much easier as the packed bed areacan be increased using larger support sieves or by constructing
Table 3Symbols used in Fig. 8.
Symbols Process conditions Symbols Process conditions
Bead size [mm] Bed height [mm] Applied pressure [kPa] Bead size [mm] Bed height [mm] Applied pressure [kPa]
55 2.5 50 � 65 2.5 200
55 2.5 100 78 1 200
B 55 2.5 150 n 78 2.5 200
55 2.5 200 78 5 200
þ 55 2.5 250 D 78 10 200
55 2.5 300 78 15 200
J 55 2.5 400 & 78 20 200
55 2.5 500 90 2.5 200
A. Nazir et al. / Chemical Engineering Science 92 (2013) 190–197196
several modules in parallel. The sieves that we used can bestandardly produced in sizes of41 m2; dispersing glass beadsover the sieve is not such a big issue since they can be depositedonto the sieve by filtering a dispersion, and re-dispersion can beachieved by back pulsing the column. As the research willprogress in this area, we expect that the production of evensmaller droplet size using smaller glass beads or other materialswill become realistic.
4. Conclusions
Oil-in-water emulsions were prepared by premix emulsifi-cation using a packed bed of differently sized glass beadssupported by a metal sieve. The process conditions especiallyrelated to internal structure of the porous media like interstitialvoid size and bed height were investigated, and both werefound responsible for droplet break-up. Two different dropletbreak-up mechanisms were identified: either dominated byconstriction (Reo40) or inertia (Re440). Droplets below5 mm (droplet to void size ratioE0.2) could easily be producedusing void sizes that were considerably larger, and the mea-sured fluxes were comparable to the highest reported fluxvalues for premix emulsification studies. An appreciable dro-plet size distribution (droplet spanE0.75) was obtained over apressure range of 200–400 kPa. Scaling relations were estab-lished based on the studied process conditions, and a reason-ably good agreement was observed between the experimentaland modeled data, making these relations suitable for designstudies.
Nomenclature
A effective surface area of packed bed [m2]Av dynamic specific surface area [m2 m�3]D packed bed diameter [m]d32 Sauter mean droplet diameter [m]db glass bead diameter [m]dv interstitial void diameter [m]EV energy density [J m�3]H packed bed height [m]J flux [m3 m�2 s�1
¼m s�1]P power input [W]q Eq. (4) constant [dimensionless]Re Reynolds number [dimensionless]vo superficial velocity [m s�1]vv interstitial void velocity [m s�1]
Greek letters
DP pressure drop [Pa]a, b and g Eq. (9) fit parameters [dimensionless]d droplet span [dimensionless]E porosity [dimensionless]Ze emulsion viscosity [Pa s]rb bulk density [kg m�3]re emulsion density [kg m�3]rp particle density [kg m�3]t bed tortuosity [dimensionless]fm mass flow rate [g s�1]fV volume flow rate [m3 s�1]
Acknowledgments
The authors thank to Higher Education Commission of Paki-stan and NanoNextNL for their financial support for this research.
References
Altenbach-Rehm, J., Suzuki, K., Schubert, H., 2002. Production of O/W Emulsionswith Narrow Droplet Size Distribution by Repeated Premix MembraneEmulsification, Paper Presented in 3rd World Congress on Emulsions, Lyon,France.
Anna, S.L., Bontoux, N., Stone, H.A., 2003. Formation of dispersions using ‘‘flowfocusing’’ in microchannels. Appl. Phys. Lett. 82, 364–366.
Chappat, M., 1994. Some applications of emulsions. Colloids Surf. A: Physicochem.Eng. Aspects 91, 57–77.
Comiti, J., Renaud, M., 1989. A new model for determining mean structureparameters of fixed beds from pressure drop measurements: application tobeds packed with parallelepipedal particles. Chem. Eng. Sci. 44, 1539–1545.
Karbstein, H., Schubert, H., 1995. Developments in the continuous mechanicalproduction of oil-in-water macro-emulsions. Chem. Eng. Process. 34, 205–211.
Katoh, R., Asano, Y., Furuya, A., Sotoyama, K., Tomita, M., 1996. Preparation of foodemulsions using a membrane emulsification system. J. Membr. Sci. 113,131–135.
Kawakatsu, T., Kikuchi, Y., Nakajima, M., 1997. Regular-sized cell creation inmicrochannel emulsification by visual microprocessing method. J. Am. OilChem. Soc. 74, 317–321.
Kiss, N., Brenn, G., Pucher, H., Wieser, J., Scheler, S., Jennewein, H., Suzzi, D.,Khinast, J., 2011. Formation of O/W emulsions by static mixers for pharma-ceutical applications. Chem. Eng. Sci. 66, 5084–5094.
Kobayashi, I., Nakajima, M., 2002. Effect of emulsifiers on the preparation of food-grade oil-in-water emulsions using a straight-through extrusion filter. Eur. J.Lipid Sci. Technol. 104, 720–727.
Kobayashi, I., Nakajima, M., Chun, K., Kikuchi, Y., Fujita, H., 2002. Silicon array ofelongated through-holes for monodisperse emulsion droplets. AIChE J. 48,1639–1644.
Lambrich, U., Schubert, H., 2005. Emulsification using microporous systems. J.Membr. Sci. 257, 76–84.
Legrand, J., Moranc-ais, P., Carnelle, G., 2001. Liquid–liquid dispersion in an SMX-Sulzer static mixer. Chem. Eng. Res. Des. 79, 949–956.
A. Nazir et al. / Chemical Engineering Science 92 (2013) 190–197 197
Mine, Y., Shimizu, M., Nakashima, T., 1996. Preparation and stabilization of simpleand multiple emulsions using a microporous glass membrane. Colloids Surf.B: Biointerfaces 6, 261–268.
Morancais, P., Hirech, K., Carnelle, G., Legrand, J., 1999. Friction factor in staticmixer and determination of geometric parameters of SMX sulzer mixers.Chem. Eng. Commun. 171, 77–93.
Nakashima, T., Shimizu, M., Kukizaki, M., 1991. Membrane emulsification bymicroporous glass. Key Eng. Mater. 61–62, 513–516.
Nazir, A., Schroen, K., Boom, R., 2010. Premix emulsification: a review. J. Membr.Sci. 362, 1–11.
Nazir, A., Schroen, K., Boom, R., 2011. High-throughput premix membraneemulsification using nickel sieves having straight-through pores. J. Membr.Sci. 383, 116–123.
Peng, S.J., Williams, R.A., 1998. Controlled production of emulsions using acrossflow membrane: part I: droplet formation from a single pore. Chem.Eng. Res. Des. 76, 894–901.
Steegmans, M.L.J., 2009. Emulsification in Microfluidic Y- and T-junctions. Ph.D.Thesis. Wageningen University, The Netherlands.
Steegmans, M.L.J., Schroen, K.G.P.H., Boom, R.M., 2009. Characterization of emul-sification at flat microchannel Y junctions. Langmuir 25, 3396–3401.
Sugiura, S., Nakajima, M., Tong, J., Nabetani, H., Seki, M., 2000. Preparation ofmonodispersed solid lipid microspheres using a microchannel emulsificationtechnique. J. Colloid Interface Sci. 227, 95–103.
Suzuki, K., Fujiki, I., Hagura, Y., 1998. Preparation of corn oil/water and water/cornoil emulsions using PTFE membranes. Food Sci. Technol. Int. 4, 164–167.
Suzuki, K., Shuto, I., Hagura, Y., 1996. Characteristics of the membrane emulsifica-tion method combined with preliminary emulsification for preparing cornoil-in-water emulsions. Food Sci. Technol. Int. Tokyo 2, 43–47.
Thorsen, T., Roberts, R.W., Arnold, F.H., Quake, S.R., 2001. Dynamic patternformation in a vesicle-generating microfluidic device. Phys. Rev. Lett. 86,4163–4166.
Van der Zwan, E., Schroen, K., van Dijke, K., Boom, R., 2006. Visualization of dropletbreak-up in pre-mix membrane emulsification using microfluidic devices.Colloids Surf. A: Physicochem. Eng. Aspects 277, 223–229.
Van der Zwan, E.A., Schroen, C.G.P.H., Boom, R.M., 2008. Premix membraneemulsification by using a packed layer of glass beads. AIChE J. 54, 2190–2197.
Van Dijke, K., De Ruiter, R., Schroen, K., Boom, R., 2010. The mechanism of dropletformation in microfluidic EDGE systems. Soft Matter 6, 321–330.
van Dijke, K., Veldhuis, G., Schroen, K., Boom, R., 2009. Parallelized edge-baseddroplet generation (EDGE) devices. Lab Chip 9, 2824–2830.
Vladisavljevic, G.T., Shimizu, M., Nakashima, T., 2004. Preparation of monodispersemultiple emulsions at high production rates by multi-stage premix membraneemulsification. J. Membr. Sci. 244, 97–106.
Vladisavljevic, G.T., Surh, J., McClements, J.D., 2006. Effect of emulsifier type ondroplet disruption in repeated Shirasu porous glass membrane homogeniza-tion. Langmuir 22, 4526–4533.
Williams, R.A., Peng, S.J., Wheeler, D.A., Morley, N.C., Taylor, D., Whalley, M.,Houldsworth, D.W., 1998. Controlled production of emulsions using a cross-flow membrane part II: industrial scale manufacture. Chem. Eng. Res. Des. 76,902–910.
