Professional Documents
Culture Documents
a b s t r a c t
Experimental results of measurements of the bubble and slug lengths in Taylor (slug) flow are presented. The exper-
iments were carried out using 3 different straight microchannels (microreactor with square cross-section made of
polydimethyloxosilane (PDMS); microreactor with circular cross-section made of glass; microreactor with rectangu-
lar cross-section made of polyethylene terephthalate modified by glycol (PETg)) and 4 different liquids (water, ethanol
propanol and heptane). The results have been compared with the available literature correlations. It is concluded,
that the values obtained from the correlation proposed by Laborie et al. [Laborie, S., Cabassud, C., Durant-Bourlier, L.,
Laine, J.M., 1999. Characterization of gas–liquid two-phase flow inside capillaries. Chem Eng Sci 54, 5723–5735] do not
agree with the results of measurements, while the agreement of these results with the predictions obtained using the
correlation proposed by Qian and Lawal [Qian, D., Lawal, A., 2006. Numerical study on gas and liquid slugs for Taylor
flow in a T-junction microchannel. Chem Eng Sci 61, 7609–7625] is good. New, corrected values of the pre-exponential
constant and the exponents in the Qian and Lawal [Qian, D., Lawal, A., 2006. Numerical study on gas and liquid slugs
for Taylor flow in a T-junction microchannel. Chem Eng Sci 61, 7609–7625] correlation are proposed.
© 2009 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
∗
Corresponding author. Tel.: +48 22 234 63 19.
E-mail address: sobieszuk@ichip.pw.edu.pl (P. Sobieszuk).
Received 12 December 2008; Received in revised form 5 July 2009; Accepted 7 July 2009
0263-8762/$ – see front matter © 2009 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.cherd.2009.07.007
264 chemical engineering research and design 8 8 ( 2 0 1 0 ) 263–269
where from:
The flow pattern depends on the liquid and gas flow rates,
on the properties of the fluids, channel geometry and its LL ∼d−2.81 (4)
material. Several diagrams, called flow maps, have been pro-
posed in the literature, to depict regions in which a given It follows from these correlations, that LG should be
flow pattern occurs (e.g. Jayawardena et al., 1997; Triplett et approximately inversely proportional to d while LL should be
al., 1999; Waelchli and von Rohr, 2006). An example of such a approximately inversely proportional to the third power of d.
map is shown in Fig. 1. There were also numerous efforts to In another study, Qian and Lawal (2006) performed numer-
develop an universal flow map, valid for different microchan- ical simulations of the Taylor flow, using commercial CFD
nels and different gas–liquid systems. Although these efforts package FLUENT (release 6.1.22.2003). In this package the
had limited success, they all confirm, that the slug (Taylor) volume of fluid (VOF) model is used to simulate two-phase
flow occupies a large part of any flow map, and is obtainable fluid–fluid flows. As a result of 148 numerical simulations for
in the most practically interesting ranges of gas and liquid flow channel widths ranging from 0.25 to 1 mm, they proposed the
following correlations:
(LG + LL ) −1.05
= 1.637ε−0.893
G (1 − εG ) Re−0.075 Ca−0.0687 (5)
d
LG −1.05
= 1.637ε0.107
G (1 − εG ) Re−0.075 Ca−0.0687 (6)
d
LL −0.05
= 1.637ε−0.893
G (1 − εG ) Re−0.075 Ca−0.0687 (7)
d
al., 2005; van Baten and Krishna, 2005). In Taylor flow, if the
liquid exhibits good wettability of the channel wall, the wall is
covered by a thin liquid layer, and the gas bubbles are sliding
over this “lubricating” layer (Schwartz et al., 1986; Bretherton,
1961). The interfacial gas–liquid area is therefore composed of
two parts: the lateral film part and the perpendicular cap part.
Pohorecki (2007) proposed a criterion for the activity of the lat-
eral part of the interfacial area in the mass transfer processes.
The criterion also includes bubble length in a microchannel.
A comprehensive paper on the hydrodynamics of bubble
formation in microbubble columns with single and multiple
channels was published by Haverkamp et al. (2006). Fol-
lowing an earlier paper by Haverkamp (2002), they report
a distribution of bubble size, observed both in single and
multiple channel columns—a phenomenon only occasionally
observed in our work. A very interesting study on CFD simula-
tions of hydrodynamics of slug flow in curved microchannels Fig. 2 – PDMS microreactor, Y shaped.
was published by Kumar et al. (2007). The authors deter-
mined numerically (using the control volume finite difference
method, CVFDM) the gas bubble and liquid slug length for both
straight and curved microchannels, and found good agree-
ment of their results for straight microchannels with Qian
and Lawal (2006) correlations. Bubble and slug lengths were
measurement experimentally (using photographic method) by
Warnier et al. (2008). The authors concentrated on the thick-
ness of the liquid film surrounding a bubble, and did not
report direct information on bubble and slug lengths. Compu-
tational modelling and results of experimental measurements
of the slug size were also reported for liquid–liquid slug flow by
Kashid et al. (2007), and Kashid and Agar (2007). The results for
liquid–liquid flow cannot, however, be directly compared with
those for gas–liquid flow. It should be pointed out, that both in
numerical simulations and experiments performed using the
photographic method, the bubble and slug lengths are usually
measured along the channel axis. Only in a few papers (e.g. Fig. 3 – Glass microreactor, Y shaped.
Heiszwolf et al., 2001) these parameters were measured using
the conductivity method, which makes the results difficult to (PETg) by micromachining method and is shown in Fig. 4.
compare directly with those obtained from the photographs. Four liquids were used for the investigations, namely: water,
In this paper we report experimental values of gas bubble propanol and ethanol (PETg microreactor), ethanol and hep-
and liquid slug lengths, obtained using 3 different microreac- tane (PDMS microreactor), ethanol (glass microreactor). The
tors and 4 different liquids, with nitrogen as the gas phase. physico-chemical properties of the liquids used are shown in
Table 2.
2. Experiments The experimental set-up is shown in Fig. 5. The gas (nitro-
gen) was supplied from a cylinder through a reducing valve
Three microreactors were used. The microreactors were
designed for visual investigations of the flow in two-phase
Table 2 – Physico-chemical properties of the liquid
gas–liquid system. Table 1 shows shapes, dimensions and employed.
materials of microreactors. The first microreactor with square
Density Viscosity Surface tension
cross-section was built using polydimethyloxosilane (PDMS)
(kg m−3 ) (Pa s) (N m−1 )
by photolithography method and it is shown in Fig. 2. The
second microreactor with circular cross-section was made of Ethanol 790 0.00119 0.0225
Heptane 685 0.00041 0.0203
glass (three capillaries were joined together) and is shown in
Propanol 804 0.00220 0.0239
Fig. 3. The third microreactor with rectangular cross-section
Water 998 0.00103 0.0729
was built of polyethylene terephthalate modified by glycol
266 chemical engineering research and design 8 8 ( 2 0 1 0 ) 263–269
this reason the results obtained for the PET microreactor have
been excluded from the further analysis.
The good agreement between our results and the Qian and
Lawal (2006) correlation convinced us, that the form of this
correlation is suitable for the description of Taylor flow in
microchannels. Further support of this form of correlation fol-
Fig. 8 – Comparison of the experimental results of bubble lows from the “switching” mechanism, proposed by Pohorecki
length and those calculated from Qian and Lawal (2006) and and Kula (2008). This form of correlation was thus used to
Laborie et al. (1999) correlations for PDMS and glass describe our results. Table 3 shows the values of the pre-
microreactor. exponential constants and exponents obtained for the results
from the PDMS and the glass microreactors. The correlations
were obtained using the least squares method.
For the PDMS reactor and heptane–nitrogen system, we
obtained the correlation:
LG −0.946
= 1.301ε0.086
G (1 − εG ) Re−0.142 Ca−0.152 (8)
d
LG −1.009
= 1.298ε0.106
G (1 − εG ) Re−0.009 Ca0.013 (9)
d
LG −1.076
= 1.306ε0.028
G (1 − εG ) Re−0.143 Ca−0.186 (10)
d
Qian and Lawal correlation (6) 1.637 0.107 −1.05 −0.075 −0.0687
PDMS reactor (heptane–nitrogen) 1.301 0.086 −0.946 −0.142 −0.152
PDMS reactor (ethanol–nitrogen) 1.298 0.106 −1.009 −0.009 0.013
Glass reactor (ethanol–nitrogen) 1.306 0.028 −1.076 −0.143 −0.186
Proposed general correlation 1.302 0.073 −1.010 −0.098 −0.108
We = Re · Ca
LG −1.01
= 1.302ε0.073
G (1 − εG ) Re−0.098 Ca−0.108 (11)
d