Measuring techniques in gasliquid and gasliquidsolid

reactors
Cristophe Boyer, Anne-Marie Duquenne, Gabriel Wild

To cite this version:

Cristophe Boyer, Anne-Marie Duquenne, Gabriel Wild. Measuring techniques in gasliquid
and gasliquidsolid reactors. Chemical Engineering Science, Elsevier, 2002, vol. 57 (n 16), pp.
3185-3215. <10.1016/S0009-2509(02)00193-8>. <hal-01394969>

HAL Id: hal-01394969

https://hal.archives-ouvertes.fr/hal-01394969
Submitted on 10 Nov 2016

HAL is a multi-disciplinary open access Larchive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinee au depot et `a la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publies ou non,
lished or not. The documents may come from emanant des etablissements denseignement et de
teaching and research institutions in France or recherche francais ou etrangers, des laboratoires
abroad, or from public or private research centers. publics ou prives.
Open Archive TOULOUSE Archive Ouverte (OATAO)
OATAO is an open access repository that collects the work of Toulouse researchers and
makes it freely available over the web where possible.

This is an author-deposited version published in : http://oatao.univ-toulouse.fr/

Eprints ID : 16521

To link to this article : DOI : 10.1016/S0009-2509(02)00193-8

URL : http://dx.doi.org/10.1016/S0009-2509(02)00193-8

To cite this version : Boyer, Cristophe and Billet, Anne-Marie and

Wild, Gabriel Measuring techniques in gasliquid and gasliquid
solid reactors. (2002) Chemical Engineering Science, vol. 57 (n
16). pp. 3185-3215. ISSN 0009-2509

Any correspondence concerning this service should be sent to the repository

administrator: [email protected]
 Measuring techniques in gasliquid and gasliquidsolid reactors
Christophe Boyera; , Anne-Marie Duquenneb , Gabriel Wildc
a Centredetudes et de developpement industriels (CEDI) Rene Navarre BP no 3, F 69390, Vernaison France
b Laboratoirede Genie Chimique, ENSIACETCNRS (UMR no 5503) 18 chemin de la Loge, F 31078 Toulouse cedex 04, France
c Laboratoire des Sciences du G enie Chimique, CNRS (UPR no 6811)ENSIC, B.P. 451, F 54001 Nancy, France

Abstract

This article oers an overview of the instrumentation techniques developed for multiphase ow analysis either in gas/liquid or in
gas/liquid/solid reactors. To characterise properly such reactors, experimental data have to be acquired at dierent space scale or time
frequency. The existing multiphase ow metering techniques described give information concerning reactor hydrodynamics such as
pressure, phases holdups, phases velocities, ow regime, size and shape of dispersed inclusions, axial diusion coecients. The measuring
techniques are presented in two groups: the non-intrusive techniques that deliver global, cross-section-averaged or local data, and the
intrusive probes that are dedicated to local measurements. Eventually some examples of multiphase instrumentation development are
reported (trickle-bed and slurry bubble column at semi-industrial scale) in the renery or petrochemical area.
? 2002 Elsevier Science Ltd. All rights reserved.

1. Introduction time resolution of the technique used and the purpose of the
study.
The description and design of gasliquid and gasliquid Measurement techniques can be classied according dif-
solid reactors still relies to a large extent on empirical rules ferent ways. A rst classication distinguishes between
and correlations, which in turn are based on measurements time-averaged and transient measurements and between
made under conditions as relevant as possible to industrial local and global measurements. According to this classi-
practice. This is true for the classical chemical engineering cation, the following types of values or characteristics are
approach, where such quantities as liquid holdup or pressure measured:
drop are predicted via empirical correlations based on data as
numerous and precise as possible. Nevertheless, more mod- Global steady-state characteristics of the reactor (gas or
ern approaches such as computational uid dynamic (CFD) liquid holdup, pressure drop, ow regime, minimum u-
are used to help in the design of multiphase reactors. Even idisation velocity, etc.)
in this case, the physical models used require information Local time-averaged characteristics
on local and transient ow characteristics (e.g. turbulence One-dimensional space: cross-section-averaged values at
characteristics, wake coecients, etc.), since ab initio cal- a given level of a reactor
culations are up to now impossible. Two-dimensional space: local averaged values over a re-
Reliable measuring techniques are therefore needed as actor cross-section (tomographic techniques)
well in academia as in industry for the rational description Three-dimensional space
and the design of multiphase reactors. Depending on the aim Local and transient characteristics of the ow eld.
of the analysis, dierent types of measurements are required
and it is important to maintain an adequacy of space and The local and/or transient character of the measurement
can vary widely, depending on the time and the length
scale.
Corresponding author.
As an example, in case of the design of a trickle bed re-
E-mail addresses: [email protected] (C. Boyer),
[email protected] (A.-M. Duquenne), actor, liquid distribution is of the utmost importance, but in
[email protected] (G. Wild). most cases the ow is fairly steady state. In this case, for
local time averaged, at least 2D measuring techniques of membrane of the sensor. The interpretation of the pressure
the liquid velocity or liquid holdup are needed. In case of drop in terms of static pressure drop and friction pressure
a slurry bubble column, global characteristics are required drop is not always easy: e.g. in case of trickle beds, the pres-
(minimum settling velocity) but also local ones, like the lo- sure drop measured may be the friction pressure drop or the
cal liquid velocity and gas and solid holdup. These mea- total pressure drop, depending on the measuring procedure.
surement needs will be illustrated at the end of this paper by The large scatter observed in pressure drop data in trickle
presenting some examples of instrumentation development beds is partly due to the fact that, quite often, the values re-
in the oil industry. ported are somewhere in-between the friction and the to-
Since the classication between local and global measure- tal pressure drop. In bubble columns, slurry bubble columns
ment is not always possible another classication has been and three-phase uidised beds the friction pressure drop is
preferred in this article. This classication relies more on usually negligible compared with the static pressure drop.
the physical basis of the measurement and nally one can
P


P

distinguish between invasive and non-invasive measuring = g(L L + S S + G G ) + f:
Z Z
techniques.
In the present work, a review of a number of non-invasive (b) Uses of the mean pressure drop. The main use of pres-
and invasive measuring techniques is presented. In both sure measurements in such pieces of equipment is to deter-
cases (non-invasive and invasive) the techniques are mine gas, liquid and/or solid holdup (Joshi, Patil, Ranade, &
grouped according to the physical principle of the measure- Shah, 1990; Wild & Poncin, 1996) in such reactors. In case
ment. In the meantime, the characteristics of the techniques of three-phase reactors, the pressure drop measurement has
according to the rst classication (global or local, time to be combined with another technique to get all the phase
averaged or transient) are indicated as well as orders of holdups.
magnitude of the length and time resolution to be expected. In multiphase reactors, an exact description of the ow
Finally, some examples of the use of measuring techniques behaviour by CFD is usually still impossible and the design
with industrial constraint in the petrochemical and renery of these reactors usually relies on empirical correlations of a
industry are presented. number of characteristics of hydrodynamics, heat and mass
transfer. Usually such correlations are valid at the best for
a given ow regime; dening ways to determine which is
2. Non-invasive techniques the ow regime is therefore an important task. According
to Zahradnk et al. (1997), three main ow regimes can be
There are numerous non-invasive techniques (Table 1) distinguished in bubble columns: homogeneous ow, tran-
available to investigate hydrodynamics of gasliquid and sition ow and established heterogeneous ow; in case of
gasliquidsolid reactors. First, techniques yielding char- small diameter columns, plug ow must be added. In gas
acteristics of the whole reactor or of a part of the reactor liquidsolid uidisation, usually homogeneous ow and het-
(global techniques) will be examined, the characteristics be- erogeneous ow regime are considered. In trickle beds it
ing e.g. the pressure drop, the gas and/or liquid holdup, the is usual to distinguish trickling ow regime (low interac-
ow regime, the bubble size distribution, the mixing char- tion) and dierent forms of high-interaction ow regimes
acteristics of the gas, the liquid or the solid phase. In some, (pulsing ow, bubble ow, spray ow, foamy ow, etc.)
long-term averages are measured, in others uctuations are (Saroha & Nigam, 1996). Time-averaged values of the pres-
analysed. Next, determination techniques of local character- sure drop or the phase holdups can be used to determine ow
istics will be presented (gas and liquid velocity, bubble or regime transitions but give very uncertain results. This tech-
particle size and position). Often, one single technique can nique has however been used in trickle beds (Purwasasmita,
yield more than one characteristic. 1985). The classic drift ux analysis of Wallis (1969) (bub-
ble columns) and of Zuber and Findlay (1965) (airlift re-
2.1. Global techniques actors) are relatively easy to use and give fair precision for
both ow regime transitions in these reactors (see e.g. Vial
2.1.1. Time-averaged pressure drop et al., 2000).
The pressure drop between two levels of a reactor is a very
important variable to know, either because the pressure drop 2.1.2. Measurement and analysis of signal uctuations
is important per se (design of the pumps and compressors), Analysing the uctuations of a signal depending on the
or because it gives information on the holdup of the dierent hydrodynamics is a relatively direct way of determining the
phases or on the ow regime. ow regime in a multiphase reactor. This signal may be a
(a) Measuring technique. It is recommended to measure wall pressure (Matsui, 1984; Glasgow, Erickson, Lee, &
the pressure with sensors ush to the wall; the price of piezo- Patel, 1984; Drahos & Cermak, 1989; Drahos, Zahradnk,
electric pressure sensors decreased in the last 10 years and Puncochar, Fialova, & Bradka, 1991; Drahos, Bradka, &
such equipment is now quite standard. Care must however Puncochar, 1992; Vial et al., 2000; Johnsson, Zijerveld,
be taken that no xed bed solid is directly in contact with the Schouten, van den Bleek, & Leckner, 2000; Kluytmans,
Table 1
Non-invasive techniques

Technique Application Expected results Limitations Spatial resolution Time resolution

Pressure drop GL and GLS Pressure drop, 1D +
liquid holdup, Model required
ow regime Model required
Pressure GL and GLS Flow regime, ++
uctuation turbulence
characteristics
Dynamic gas GL and GLS Bubble size Model required Global technique Steady state
disengagement distribution
technique phase holdup In case of a slurry Global technique Steady state
Tracers GL and GLS Liquid holdup and Eventually sophisticated Global technique Hydrodynamic
mixing characteristics model required, steady state
nding an adequate tracer
Conductimetry GL and GLS Liquid holdup Only electrically 1D +
conductive liquid,
calibration required
Neutron absorption GL and GLS Phase distribution Neutron source required, Path averaged values
-ray absorption safety and price,
X-ray absorption safety and price
Light attenuation GL Gas holdup and Transparent and clean Global +
interfacial area liquid and wall required,
Model required,
low holdup and small
diameter column
Ultrasound attenuation GL Holdup Liquid continuous phase, Path averaged values +
GLS with small diameter column
restrictions
Photography and Mainly GL Bubble/drop size Mainly wall region, ++ +
image analysis shape transparent wall and liquid
Flow regime Low holdup of dispersed
phase,
quantitative image analysis
not always available
Radiography Mainly GL Bubble/drop size and Safety, ++ +
shape, ow regime Low gas holdup
PIV Mainly GL Gas and liquid Low gas holdup ++ +(no velocity
(+ seeding) velocity elds uctuations
observed)
NMR GL and GLS Liquid holdup, Price and size of equipment ++ +
phases velocity,
wetting eciency
Laser Doppler GL Mainly liquid velocity, Transparent liquid, +++ ++
Anemometry and eventually gas relatively low holdup,
phase Doppler velocity and bubble very low holdup,
anemometry size for PDA
Polarography GL and GLS Wall shear stress Support electrolyte needed, +++ ++
(microelectrodes) only aqueous solutions
Tracking of GL and GLS Velocity elds and Heavy calibration required, + +
radioactive particles trajectories use of radioactive particles
- or X-ray tomography GL and GLS Phase holdup Price, high power required +
distribution for large solid holdup
Capacitive or GL and GLS Phase holdup No reliable reconstruction +++
resistive tomography distribution algorithm available
Ultrasonic tomography GL and GLS Phase holdup Low solid and + +
distribution gas holdup
Spatial resolution: + + + : 1 mm; ++ mm; + : cm. Time resolution: + + + : ms; ++ : s; + : minute.
van Wachem, Kuster, & Schouten, 2001), the current
at micro-electrodes imbedded in the wall (Lati, Rode,
Midoux, & Storck, 1992; Lati, Naderifar, & Midoux,
1994a; Lati, Naderifar, Midoux, & Le Mehaute, 1994b)
or in particles (Lesage, 2000) of a trickle bed, conduc-
tivity probes (Briens, Briens, Margaritis, & Hay, 1997b;
Briens, Briens, Hay, Hudson, & Margaritis, 1997a; Maucci,
Briens, Martinuzzi, & Wild, 1999), the sound produced
by the ow in a trickle bed (Kolb, Melli, de Santos, &
Scriven, 1990), in an airlift reactor (Glasgow, Hua, Yiin,
& Erickson, 1992) or in an aerated stirred tank (Boyd &
Varley, 1998). Local probe signal from optical probes (e.g.
Bakshi, Zhong, Jiang, & Fan, 1995), optical transmittance
probes (Kikuchi, Tsutsumi, & Yoshida, 1997), temperature
probes (Thimmapuram, Rao, & Saxena, 1992), hot lm
anemometry (e.g. Magaud, 1999), electrochemical probes Fig. 1. DGD in a bubble column (Camarasa et al., 1999).
(Vial, 2000) can give information of ow regime. Of
course, one has to be careful not to over-interpret uctua-
tions: each measuring technique has an inertia and therefore assumptions made in this technique are the following
a time resolution limit. On the other hand with a very fast (Camarasa et al., 1999):
technique, sampling can be a non-trivial problem, specially
(a) the dispersion is axially homogeneous when the gas
with invasive probes, where bubble-probe interactions may
feed is interrupted,
prove a problem (see Section 3).
(b) there are no bubble-bubble coalescence or break-up
A number of authors have proposed numerical tech-
happening during disengagement,
niques devoted to the interpretation of uctuations in terms
(c) the bubble disengagement of each bubble class is not
of ow regimes: the possibilities are numerous and range
in uenced by the other bubble classes.
from quite classical techniques like statistical (e.g. Drahos
et al., 1991; Letzel, Schouten, Krishna, & van den Bleek, Fig. 1 gives an example of a DGD curve; the liquid height
1997) or spectral (e.g. Bigot, Guyot, Bataille, & Roustan, in a bubble column with the system airwater is represented
1990; Lati et al., 1992) analysis to modern techniques as a function of time (Camarasa et al., 1999). In principle,
with a much more sophisticated mathematical background. it is possible to represent this curve by a number of straight
Such new techniques include fractal analysis (Drahos lines corresponding each to a bubble class with a given
et al., 1992; Briens et al., 1997b; Maucci et al., 1999; Lati bubble rise velocity. However in practice, a bimodal bubble
et al., 1994b; Naderifar, 1995; Luewisuthichat, Tsutsumi, diameter distribution is usually assumed.
& Yoshida, 1995), chaotic analysis (Luewisuthichat, In recent work, Lee, Luo and Fan (1999) checked ex-
Tsutsumi, & Yoshida, 1996; Letzel et al., 1997; Johnsson perimentally the assumptions of DGD in a two-dimensional
et al., 2000; Lin, Juang, Chen, & Chen, 2001; Kluytmans et bubble column by PIV measurements. They show that the
al., 2001), time-frequency analysis (Bakshi et al., 1995; Vial assumptions (b) and (c) of the classical DGD technique are
et al., 2000). A review on these techniques applied to bubble not valid, specially in the coalesced bubble ow (heteroge-
columns was published recently by Vial et al. (2000). neous or transition) regime. The limits of the application of
the DGD technique indicated by these authors have still to
be checked in case of three-dimensional columns.
2.1.3. Dynamic gas disengagement technique (DGD)
The basic principle of this technique is to stop the aer-
ation in a gasliquid or gas-slurry bubble column and to 2.1.4. Tracing techniques
measure the liquid level or the pressure at dierent levels Tracing techniques are used with two aims:
in the reactor as a function of time. This technique can be
used either to determine the global gas and solid holdup determining the mean phase holdup of one of the phases,
in a slurry bubble column (see example in Section 4.2), characterising mixing of a phase.
or to determine the structure of the gas holdup in bubble
columns (amount of gas bubbles separated in classes ac- 2.1.4.1. Tracing of the liquid. Measuring the residence
cording to their size). The latter application has been used time distribution (RTD) of a tracer of the liquid phase is a
for some time (Beinhauer, 1971; cited by Deckwer, 1985). well-known technique to determine the mean liquid holdup
The decrease of the liquid level is observed, using various of a reactor and to characterise the mixing of the liquid
techniques: X-ray (Beinhauer, 1971), visual observation, phase. However, using this technique presents some pitfalls
pressure prole (Daly, Patel, & Bukur, 1992). The basic which shall be presented in the following.
 The theory of RTD is well known (Levenspiel, 1999;
Villermaux, 1995; Shah, Stiegel, & Sharma, 1978; Buham
& Mason, 1993). The tracer is in most cases a salt tracer and
the measuring technique is conductimetry; coloured trac-
ers have also been used, mainly with non-aqueous liquids
(see e.g. Lara Marquez, 1992). Radioactive isotope tracer
techniques present the advantage that non-invasive measure-
ments are possible at dierent levels of the reactor (Blet,
Berne, Chaussy, Perrin, & Schweich, 1999; Pant, Saroha, &
Nigam, 2000). An interesting, if expensive, technique has
been proposed recently by Sakai et al. (2000): these authors
use a non-radioactive tracer which interacts with neutrons.
The advantage is that this technique can be used under high
temperature and pressure conditions (coal liquefaction plant
in the present case). Of course, a neutron source is required.
In Section 4.2 of this paper a refractometry technique is de-
scribed: in this case the tracer is a liquid whose refractive
index is dierent from that of the main liquid.
In the interpretation of a RTD curve, the actual ow be- Fig. 2. Model of an airlift reactor (Dhaouadi et al., 1997).
haviour is represented by a model, with parameters that are
adjusted to t experimental values of the tracer concentra-
tion along time. Depending on the complication of the ow way to proceed is given by Dhaouadi et al. (1997) (Fig. 2):
behaviour, quite simple models with a limited number of an external loop air-lift reactor is considered as consisting
parameters may be used or much more sophisticated ones. in:
In simple cases, a two-parameter model may be sucient
to yield an adequate description of the global ow behaviour a single-phase liquid plug ow in the downcomer,
of a reactor: one of these parameters is usually the mean res- a single-phase liquid CSTR at the bottom junction,
idence time of the liquid (or the linear velocity, or the liquid a liquid-phase plug ow with axial dispersion owing
holdup), the other one being a characteristic of mixing (ax- cocurrently with gas in the riser,
ial dispersion coecient, number of CSTRs, recirculation a gasliquid CSTR and a single liquid-phase CSTR in
rate between stirred cells). If the RTD has been correctly series at the gasliquid separator.
performed, this yields a quite precise determination of the
global liquid holdup. Examples of the application of this The volumes of the dierent parts are estimated indepen-
technique to the determination of the liquid holdup may be dently and the gas holdup in the two gasliquid parts are
found in Larachi, Laurent, Wild, and Midoux, 1991 (gas determined by pressure measurements, so that the only re-
liquid cocurrent ow through a xed bed of non-porous par- maining adjustable parameters are the liquid velocity and
ticles) or in Garca Ochoa, Khalfet, Poncin, & Wild (1997) the axial dispersion coecient in the riser. The tracer mass
(suspended solid bubble column). balances in the dierent parts of the airlift yield a system of
In more complex cases, models with more parameters ODEs and PDEs which has been solved by using the MOD-
have to be used: e.g. in case of mass transfer between the EST software package (Haario, 1994) which is dedicated
owing liquid and stagnant zones, a piston with exchange to this kind of problem. Fig. 3 (Dhaouadi et al., 1997) gives
(PE) or a piston with dispersion and exchange (PDE) may an illustration of the quality of the t obtained in this case.
be used as shown e.g. by Yang, Euzen, and Wild (1990) who Under the same considerations, Leclerc, Claudel, Lintz,
used the model developed by Villermaux and van Swaaij Potier, and Antoine (2000) presented a software package,
(1969) in the case of liquid ow through a bed packed with the aim of which is to interpret in the best possible way RTD
porous particles. Such models contain more adjustable pa- curves in complex systems.
rameters and care must be taken to check the sensitivity of One must keep in mind that the concentrations to be
the agreement between model and experiment with respect taken into account for RTD analysis are mixing cup con-
to the dierent parameters. In case of even more complex re- centrations at the inlet, the outlet and eventually at dierent
actors, with dierent geometric parts, interpretation of RTD levels of the reactor. Unfortunately, most measurement
curves becomes even more tricky. In this case, it is advisable techniques determine volume or column diameter aver-
to split the reactor up in elementary elements (stirred tanks, aged concentrations (through the wall concentrations)
plug ow parts, plug ow with axial dispersion, etc.). The and not mixing cup concentrations. As shown by Briens,
volume of each element, the ow rate, etc. should as far as Margaritis, and Wild (1995), this can lead to quite important
possible be estimated independently of the RTD interpreta- errors in the determination of liquid holdup or liquid resi-
tion to avoid the danger of overtting. An example of this dence time, when the ow behaviour is far from plug ow.
 or by the radiation it emits. A review on such techniques
may be found in Lubbert (1991); the CARPT (Dudukovic,
Devanathan, & Holub, 1991) or RPT (Larachi, Chaouki,
& Kennedy, 1995; Chaouki, Larachi, & Dudukovic, 1997)
techniques which, will be discussed later, are based on the
use of such ow followers, the application of which is re-
stricted to bubble column like reactors.

2.1.4.2. Tracing of the gas-phase. Investigation of

gas-phase dispersion is more dicult than that of liquid:
while there are a number of liquid-phase tracers which can
be considered completely non-volatile, there are no com-
pletely unsoluble gases. Even with the less soluble gases
Fig. 3. RTD curves in an airlift reactor compared with an adjusted model
(e.g. Helium in water), absorption and desorption phenom-
(Dhaouadi et al., 1997) (measurements at three levels of the downcomer).
ena make the interpretation of gas-phase tracing experi-
ments intricate, as shown by Joseph and Shah (1986). It is
therefore necessary to consider simultaneously gasliquid
mass transfer and gas and liquid dispersion. The mixing
of the liquid phase may be investigated also, as shown by
Mawlana, Huang, Wilhelm, and Delmas (1992). It is how-
ever preferable to determine liquid-phase characteristics
independently by liquid tracer experiments.
An interesting tracing technique of the gas proposed by
Lubbert (1991) relies on a stochastic distribution of the
tracer gas with a cross-correlation technique of the outlet
signal with the inlet signal.

2.1.4.3. Tracing of the solid. Tracing of the solid phase

has its importance in uidisation and there are a number
of papers devoted to this subject in gassolid uidisation.
In gasliquidsolid and liquidsolid uidisation, it is of the
utmost importance that the shape, size and density of the
tracer particles be the same as those of the uidised particles.
Three kinds of tracers have been used:
Fig. 4. Comparison of a through the wall concentration with the true Coloured tracers: investigating the behaviour of layers of
RTD in case of a high level of interaction between uid elements particles of dierent colours gives mainly qualitative in-
(Briens et al., 1995).
formation on the mixing of the solid. Furthermore, sep-
arating the particles according to the colours can prove
tiresome. Nevertheless, this technique has been used by
Fig. 4 shows an example proposed by these authors of the Fortin (1984).
dierence between the real RTD curve and a concentration Magnetic tracers: Euzen and Fortin (1987); see also Wild
prole as measured by a through the wall technique (e.g. and Poncin (1996) proposed to impregnate porous cata-
a radioactive tracer concentration measurement) in case of lyst particles with a ferromagnetic substance (e.g. Nickel
a non-ideal ow, with a high level of interaction between oxide which is reduced to Nickel). Two concentric loops
uid elements (Briens et al., 1995). Through the wall tech- are situated around the column. The signal measured in
niques can, however, be used to measure the mixing time the inner loop in presence of an electric current (in the
of the liquid. Another point has to be checked: in a number outer loop) depends on the magnetic properties in the col-
of techniques, the measurement of the liquid-phase concen- umn at the level of the loop. Furthermore, the magnetic
tration is disturbed by the presence of bubbles and the actual properties of the tracer particles can be used to separate
signal has to be ltered: this may not be trivial as exem- them from the inert particles and so use them again. A
plied by a recent paper of Gupta, Al-Dahhan, Dudukovic, similar technique had been used in liquidsolid uidisa-
and Mills (2000). tion in the sixties (Flaschel, Metzdorf, & Renken, 1987).
Instead of using a dissolved liquid-phase tracer, it may be Fluorescent tracers: in case of liquidsolid slurries,
of interest to use a particle tracking technique, i.e. a neutrally Flaschel et al. (1987) proposed to use as a tracer solid
buoyant solid particle is followed by electromagnetic means particles on which a uorescing substance was adsorbed.
 This technique should be applicable also in three-phase increasing function of the specic interfacial area of the dis-
systems. persed phase. The rst applications of this technique were
limited to quite small interfacial areas and required that no
multiple scattering of the light beam by the bubbles take
2.1.5. Conductimetry
place (Calderbank, 1958) and the optical density aL (i.e.
If the liquid in a gasliquid or gasliquidsolid reactor is
the product aL, a being the interfacial area per volume and
electrically conductive and presents a constant ionic concen-
L the optical path length) was limited to quite small val-
tration, the conductivity between two levels of the reactor is
ues. Work done since this period (Kasireddy & Al Taweel,
an increasing function of the liquid holdup. Unfortunately,
1990) has extended the possible values of aL up to much
the conductivity depends not only on the liquid holdup, but
higher values (aLmax = 2500). However, the bubbles have
also on the spatial distribution of the phases. A number of
to be approximately spherical and stringent measures have
mainly empirical laws have been proposed to correlate
to be taken to avoid parasitic light sources, electronic noise,
the liquid saturation L (i.e. the ratio of the liquid holdup to
liquid turbidity, etc.
the liquid holdup in absence of gas) with the conductivity
(c) Ultrasound techniques. The main advantage of
in presence and in absence of gas (Storck, Lati, Barthole,
ultrasound-based measurement techniques, compared to
Laurent, & Charpentier 1986), which can give quite dierent
visible radiation based techniques, is that transparency of
results (as much as 20% dierence between the predictions
the liquid is not required. Some of the ultrasound tech-
of L ). It is therefore advisable to calibrate this technique
niques being invasive, ultrasound techniques will therefore
by another one.
be treated in Section 3.3
This technique has however been used in a number of
elds: Begovich and Watson (1978) investigated the be-
haviour of three-phase uidised beds, Uribe-Salas, Gomez,
2.2. Techniques yielding local characteristics
and Finch (1994) and Shen and Finch (1996) used it in bub-
ble columns.
2.2.1. Visualisation techniques
An elegant variation of the conductimetric technique was
Visualisation techniques yield bubble shape and size.
proposed by Shen and Finch (1996) to determine bubble
They can be classied into photographic (including of
swarm velocity in a bubble column. These authors developed
course video) techniques and radiographic techniques,
a fast conductimetre; the bubbles disturb the conductimetry
Particle Image Velocimetry and techniques based on
signal and the propagation celerity of this perturbation can
NMR.
be interpreted as a measure of the bubble swarm velocity.
(a) Photographic techniques. The aspect of the ow as
Note that in this case it is no longer necessary to know the
seen through the wall is an important feature, and taking
quantitative correlation of the liquid holdup as a function of
pictures through the wall is very frequent in literature (e.g.
the conductivity.
Miyahara, Hamaguchi, Sukeda, & Takahashi, 1986; Lage
& Esposito, 1999). The limitations of this technique are
2.1.6. Radiation attenuation techniques obvious: only the vicinity of the wall can be observed at
When crossing a two-phase mixture, all kinds of radiation high gassing rates; a transparent wall, a transparent liquid
are attenuated. Depending on the kind of radiation used, the are required. In case of high pressure columns, visualisation
attenuation is function of the density along the radiation path requires special windows: Lin, Tsuchiya, and Fan (1998)
or of the interfacial area. use quartz windows that can stand pressures up to 20 MPa.
(a) X-ray, -ray or neutron absorption radiography. The bubbles in most kinds of equipment are not spher-
Gases, liquids and solids have quite dierent absorption ical and can vary widely in shape. Quantitative image
coecients of X-ray or -ray radiation or of neutron beams. analysis can allow to determine, beside the area mean diam-
The measurement of the attenuation of such a radiation eter such quantities as the Feret diameters, the sphericity,
between a source and a receptor on the other side of the etc. (Bendjaballah et al., 1999; Camarasa et al., 1999). In
reactor gives access to the integral of the local mass density three-phase systems, refractive index matching of the solid
distribution along the path of the beam. The basics and the and the liquid can allow to determine bubble size and shape.
limits of the radiography techniques using this principle This technique was used by Peterson, Tankin, and Banko
may be found in Fournier and Jeandey (1993). There are (1984) to investigate three-phase uidised beds. It is how-
no theoretical limits to time resolution; limits are, how- ever quite cumbersome and seldom used.
ever, given by safety and/or price considerations. Space (b) Radiographic techniques. Using more energetic kinds
resolution is limited by the size of the available sensors of radiation to observe bubble forms is an extension of the
( 10 mm in case of -rays, 100 m in case of X rays). photographic technique: e.g. Heindel (2000) uses the Flash
These radiographic techniques are the basis of a number of X-ray Radiography to determine the evolution of bubble
tomographic techniques (see Section 4.1). shape and position. The advantage is that such a technique
(b) Light attenuation. In case of a clean transparent liq- may yield results even in presence of a non-transparent uid;
uid containing bubbles or drops, the light attenuation is an as with the photographic technique, only a single layer of
 Fig. 6. LDA velocity signal in a bubbly system (Mudde, Groen, & van
den Akker, 1998).
Fig. 5. Flow sheet of the PIV system with a particle tracking algorithm
(Reese et al., 1996).

bubbles can be observed reliably. A similar technique has niques may be found in Callaghan (1991). It is based on the
been used by Smith, Gamblin, and Newton (1995). determination voxel by voxel of the quantity of given nuclei
2D or 3D images of the multiphase ow can be obtained (usually protons 1 H); radio-frequency pulses and magnetic
using tomographic techniques (based on radiography), either eld gradient pulses interact with the spins of the nuclei
time averaged or instantaneous. These techniques will be positioned in a static magnetic eld. Reviews on the appli-
considered separately in this article (see Section 2.2.4). cation in the elds of chemical engineering are proposed by
(c) Particle image velocimetry. This technique has be- Gladden (1994) or Chaouki et al. (1997). In the last years,
come quite classical for the determination of velocity elds some applications of this still quite expensive technique
in single-phase ow. A laser sheet is used to illuminate a in the investigation of xed bed reactors have been found
well dened slice of the uid containing seeding particles; in literature (Sederman, Johns, Bramley, Alexander, &
two or more pictures of the sheet are taken at short time Gladden, 1997; Sederman, Johns, Alexander, & Gladden,
intervals. The distances between the positions of seeding 1998; Sharma, Mantle, Gladden, & Winterbottom, 2001).
particles yield the instantaneous velocity eld of the liq-
uid. Dierent variants have been proposed: particle streak 2.2.2. Laser Doppler anemometry and derived techniques
velocimetry (long exposure), particle tracking velocimetry The principle and the main applications of LDA are well
(video recording of particle motion). known and well described, e.g. in the excellent review of
In recent years, a number of authors have used PIV in gas Chaouki et al. (1997): if there is a relative motion between
liquid ow to determine either the liquid velocity eld or the a wave source (laser) and a wave receiver (seeding particle
bubble velocity and size, or both. Reviews on this subject of the liquid or bubble in the liquid), there is a shift of
may be found in Murai et al. (2000); Chaouki et al. (1997) wave frequency; this phenomenon is called the Doppler
or Reese, Mudde, Lee, and Fan (1996). The main problem is eect; measuring this frequency shift by using interference
to identify the particles between the dierent frames. Fig. 5 fringes gives access to the velocity. This technique has been
shows a ow chart of a two-phase PIV system, as proposed applied to the determination of local velocities for some
by Reese et al. (1996). In order to discriminate the dierent time.
phases, a number of possibilities have been proposed: us- LDA yields well-localised values of velocity; the mea-
ing uorescent seeding particles for the liquid and ltering surements are very fast, allowing to determine velocity uc-
the signals (e.g. Broder, Lan, & Sommerfeld, 2000). The tuations. The last years have seen the application of LDA to
same technique combined with refractive index matching the determination of liquid velocities (average values and
has been used by Northrup, Kulp, Angel, and Pinder (1993) uctuations) in bubble columns with quite high gas through-
to determine interstitial velocity elds in a xed bed. It is puts (Mudde, Groen, & van den Akker, 1997; Becker,
also possible to use numerical techniques to exploit images De Bie, & Sweeney, 2000; Vial, Laine, Poncin, Midoux,
obtained with a single CDD camera as shown by Delnoij, & Wild, 2001a; Vial, Poncin, Wild, & Midoux, 2001b;
Kuipers, van Swaaij, and Westerweel (2000); the latter au- Joshi, 2001).
thors propose an original Ensemble correlation technique Fig. 6 shows the velocity signal obtained by Mudde et al.
to extract the relevant signals from the results obtained with (1998) as a function of time in a bubble column. In this g-
a CDD camera. Recently, PIV has been applied to conrm ure, one of the problems of the technique becomes apparent:
the predictions of velocity calculations in visco-elastic liq- during a part of the measuring time, the laser beams do not
uids around rising bubbles (Funfschilling & Li, 2001). reach the place where the liquid velocity is supposed to be
(d) NMR imaging. Nuclear magnetic resonance (NMR) measured, because bubbles interrupt the beam path. In Fig. 7
imaging has been widely used in the eld of medical imaging the RMS of the axial liquid velocity in a bubble column as
and is more and more frequently applied to chemical engi- measured by Vial et al. (2001a, b) is presented. The results
neering problems. The theoretical background of these tech- correspond to two dierent gas distributors.
 Fig. 8. vG; USDA vL; LDA in the riser of an airlift reactor
(Bendjaballah-Lalaoui, 2000).

Fig. 7. RMS of the LDA-measured axial velocity in a bubble column

(Vial et al., 2001a, b)

With coalescing liquids, it is possible to make reliable

measurements at a distance of 0:1 m of the wall at gas
holdup up to 10% (Mudde et al., 1997). At higher gas
velocities, it may happen that the smaller bubbles be not
distinguished from the seeding particles. This can be over-
come by using uorescent seeding particles and by using
a colour lter, suppressing the incident light wavelength,
Fig. 9. PDA equipment (Bauckhage, 1996).
but transparent to the light emitted by uorescence (Broder
et al., 2000). Another pitfall may also be encountered: one
of the basic assumptions of LDA is that the seeding parti- velocity, the slip velocity determined this way looks rea-
cles move exactly with the liquid. However, the distribution sonable, at the higher values, negative slip velocities are
of ne particles in the presence of bubbles may not nec- observed, which does not make any sense in a structured
essarily be homogeneous in the whole of the liquid phase: ow as the one existing in the riser of an airlift. The expla-
over-concentrations of particles in the immediate vicinity nation proposed by Bendjaballah-Lalaoui is that it is prob-
of bubbles have been observed (Fan & Tsuchiya, 1990; able that there is a systematic error in the measurements
Roizard, Poncin, Lapicque, Py, & Midoux, 1999). Liquid ve- of vL; LDA with a dense bubble packing by LDA (overesti-
locities measured by LDA may therefore be contaminated mation) and another systematic error in the bubble veloc-
by bubble velocities. An example of such a contamination ity measurements (the largest and fastest bubbles may have
has been observed by Bendjaballah-Lalaoui (2000) and is been missed).
represented in Figs. 8 and 9. In this gure, the slip velocity In phase Doppler anemometry (PDA), the Laser-Doppler
in the axial direction in the riser of an airlift is represented signal is measured by two sensors; knowing the geometry
as a function of the radius, the parameter being the gas su- of the system, the phase dierence between both signals is
percial velocity. The slip velocity is obtained as the dier- directly related to the size of the particles. Fig. 8 shows the
ence of the time-averaged linear gas velocity vG; USDA (mea- schematic principle of this technique according to Bauckhage
sured by means of a ultrasound re ection Doppler probe) (1996). The light emitted by the laser is divided into
and the time-averaged liquid-velocity vL; LDA measured by two beams with an angle  which are focused on the measu-
LDA. While at the lowest values of the gas supercial ring volume,which, as in LDA can be quite small and where
interference fringes are produced. Two photodetectors are exists near the wall, which enables mass boundary layer de-
placed in a plane situated at an angle from the incoming velopment. It has been used in bubble columns by Magaud,
laser beams and containing the main direction of the ow to Souhar, Wild, and Boisson (2001) and at the wall of trickle
be studied. The photodetectors are collimated on the mea- beds by a number of authors (e.g. Lati et al., 1992; Rode,
suring volume; their angle is 2 . When a spherical bubble Midoux, Lati, & Storck, 1994). Recently Lesage (2000)
crosses the measuring volume, bursts are observed (ampli- adapted this microelectrode technique to measurements at
ed and ltered) by both detectors. The Doppler shift can the surface of solid particles in trickle beds.
be measured as in classical LDA which yields the bubble The advantages of this technique are twofold: it is one of
velocity. The principle of PDA relies on measuring also the the only non-invasive techniques yielding information on the
phase dierence
 between the signals measured by the ow behaviour in the immediate vicinity of the wall; it can
photodetectors, which is proportional to the ratio of the time be miniaturised (microelectrodes) and give also information
lag
t between the interference fringes measured with both on velocity uctuations near the wall. It presents, however,
detectors and the time T between two subsequent interfer- a disadvantage: a conducting liquid is required and the pres-
ence fringes as measured with either one of the detectors: ence of the support electrolyte can change the coalescence
behaviour of the liquid.

t

 = 2 : (1)
T
2.2.4. Radioactive tracking of particles
This phase dierence
 is proportional to the bubble di- Liquid mixing in bubble columns was investigated by the
ameter d according to the following equation: team of Dudukovic (Dudukovic et al., 1991; Chen et al.,
  1999) using single neutrally buoyant spheres containing
1 
d=
; (2) radionucleides emitting -rays. The use of radioactive emit-
2b nc ting tracers is also the basis of the radioactive particle track-
ing (RPT or CARP) technique used by the team of Chaouki
where  is the wave length of the light wave, nc the refrac-
to investigate mixing of the solid in three-phase uidised
tive index of the continuous phase (the liquid) and b a char-
beds (Cassanello, Larachi, Marie, Guy, & Chaouki, 1995;
acteristics of the geometry of the system and (in the case of
Larachi et al., 1995; Chaouki et al., 1997). In the techniques
signals produced by refraction by the dispersed phase), by
proposed by both teams, the rays emitted by the tracer par-
the ratio of the refraction indices of both phases. When the
ticle are followed by a number of detectors around the col-
measured signal corresponds purely to light re ected by the
umn. The reliable and rapid determination of the position of
bubbles, b can be calculated by the following equation:
the particle from the signals obtained by all the detectors is
In case of refracted light, the expression is more compli-
a demanding task since a long and complicated calibration
cated and can be found in Bauckhage (1996) who gives a
procedure is required, but it has been solved by both teams.
thorough review of this measuring technique, its advantages
The results of these techniques are trajectories of the parti-
and problems: most applications concern liquid in gas sys-
cles. These trajectories can be used to conrm CFD calcu-
tems; rst applications to bubble columns have been made
lations or to determine axial dispersion coecients of the
by the team of Sommerfeld (Broder et al., 2000). However,
solid; on the other hand, long-term averages of the velocity
the application is up to now limited to spherical bubbles
yields global circulation patterns.
with diameter below 1:5 mm at low gas holdup.

2.2.3. Polarographic technique 2.2.5. Tomographic techniques

Polarography can be used to measure local wall shear The tomographic techniques have been fully developed
stress. This method is based on the fast electrochemical re- during the last 20 years, rst for medical application and
duction of reagent dissolved in the moving medium (Coeuret more recently for industrial process application. These tech-
and Storck, 1984). The electrochemical system is most often niques are powerful since they give the opportunity to get the
the reduction of hexacyanoferrate III to hexacyanoferrate. phase fraction distribution inside a reactor or a column by
Electrode potential is adjusted so that the reagent concen- a totally non-intrusive measurement. The general principle
tration on electrodes is zero. In these conditions, the mass consists in measuring through the column a physical prop-
transfer on each probe depends only on local hydrodynamic erty that can be related to the phase fraction. Measurements
parameters. A simple model (Reiss & Hanratty, 1963) ex- are performed at dierent angular positions and deliver the
presses local wall shear stress w versus delivered current I average values of the property over the corresponding di-
in a law of the type: ameters. By using a reconstruction algorithm, the acquired
signals are then analysed to provide the phase fraction im-
w = CI 3 : (3) age on a cross-section of the column. Depending on the kind
of physical property used, time and space resolution may
In bubbly ows, the polarography technique is still valid. As dier widely. For tomographic systems that need to have
shown by Souhar (1987), a liquid lm of 20 m thickness a complete rotation of sensors all around the column, the
 X-ray tomographic systems have been applied to dierent
Source y packed beds (Schmitz & Mewes, 1997; Toye, 1996). The
photon uxes being very high with X-ray tube, the detector
sensing area is quite reduced and the spatial resolution can
be as small as 1 1 mm2 . However, such photon rays have
generally a rather low energy level (less than 100 keV) and
their use is limited to low attenuating material or to small
x
column diameter. With -rays, the photons are more pen-
s etrating since their energy can be as high as 1 MeV. Sev-
eral -ray tomographic systems have been devoted to the
investigation of packed beds or bubble columns (Kumar,
Moslemian, & Dudukovic, 1995; Froystein, 1997). In these
Detector types of tomographic systems, the spatial resolution is rather
lower (around 1 1 cm2 ) but remains very interesting. It
must be noted that the measurement time is directly depen-
Fig. 10. Coordinates describing a measured projection.
dent of the source activity and remain high (up to several
hours for a whole measurement in industrial conditions).
The -ray tomographic system must then be used with rather
measurements are not simultaneous and the results are only
stationary ows.
time-averaged results. Dierent kinds of tomographic sys-
(b) Electrical tomographic system. A lot of work has
tems have been developed.
been done to develop tomographic systems based on
(a) Tomography by photon attenuation measurement.
impedance measurements at the wall of a column, a tank
This tomography system is based on the attenuation mea-
or pipe including multiphase ows (Mann et al., 1997;
surement of photon rays such as - or X-ray. This attenuation
Reinecke & Mewes, 1997). These techniques are easy to
being directly proportional to the material density for a given
implement with relatively low cost. Their characteristeric
energy, the resulting image gives the density at each pixel
response dynamics is very fast, they can therefore reliably
and then the phase fraction map. The reconstruction algo-
describe rapid evolutions of ows. Nevertheless, the mea-
rithms are based on the inversion of 2D Fast Fourier Trans-
sured signal is a non-linear function of the phase fractions
form since all the measured attenuations are linear functions
and of the ow conguration: the electric capacity or resis-
of density. The main one is a ltered back-projection al-
tance between two separate elements at the wall depends on
gorithm (Shepp & Logan, 1974; Kak & Slaney, 1987). If
the whole phase repartition in the column, not only on the
f(x; y) is the distribution function of a phase fraction in-
phases present on a straight line between the two elements.
side the column, the measured attenuation corresponding to
The reconstruction calculations are not easy to perform
a projection of attenuation is expressed, for a given angular
and not yet totally satisfying; the spatial resolution remains
position  and a given distance s from column axis (Fig.
very poor unless a priori knowledge is included in the re-
10), by the following relation:
  Z construction procedure. Nevertheless, these tomographic
I techniques give a good information on large unhomogene-
p(s; ) = K ln = f(x; y) dl; (4)
I0 Ls; ity inside a vessel or a column and they are quite adapted
where I0 and I are the photon ux, respectively, measured, to detect ow conguration changes. They have been used
at cylindrical coordinates s and , when the column is lled for bubble columns (Schmitz & Mewes, 2000), three-phase
with one phase (liquid or gas) and when it is crossed by monoliths and trickle beds (Reinecke, 1996; Reinecke &
two-phase ow. K is a constant taking into account the re- Mewes, 1997; Reinecke, Petritsch, Schmitz, & Mewes,
lation between phase fraction and physical attenuation. 1998; Marcandelli, Lamine, Bernard, & Wild, 2000).
If P(S; ) is the one-dimensional Fourier transform of An interesting tomographic system has to be mentioned:
p(s; ), the function f(x; y) can be reconstructed by simply two planes of conducting wire grids have been settled within
taking the inverse Fourier transform of P(S; ): a pipe (Prasser, Bottger, & Zschau, 1998). By analysing the
1
Z Z impedance at all cross points of grids, this system is able
f(x; y) = 2 P(S; )|S|ejS(x cos +y sin ) to detect the gas fraction distribution with both a high fre-
4 0
quency (up to 1000 frame by second) and a good spatial res-
dS d: (5) olution (up to 1000 pixels). However, the system becomes
invasive and is limited to ows in which the continuous
Instead of |S| a lter function can be used. Shepp and Logan
phase is a conductive medium ( 0:5 S=m).
(1974) proposed the following lter function g(S):
   (c) Ultrasonic tomography. Some work has been done
  S 
g(S) = sin ; (6) using ultrasonic transducers to obtain tomographic images
2    through a cross-section: ultrasonic wave propagation de-
where  is the sampling frequency. pends on both the phase fraction and on the phase congu-
Table 2
Classication of invasive techniques

Techniques Applications Expected results Limitations Spatial Time

resolution resolution

Needle probe Mainly gasliquid, but Gas holdup Low solid hold-up
gasliquidsolid also Bubble velocity Not adapted to wall
Bubble chord vicinity
distribution Adapted uid
(conductive liquid for ++ ++
impedance probes)
Size distribution Model required
Multiple tip probes:
Bubble shape reconstruction
needed
Interfacial area
Heat transfer probe Mainly gasliquid, but Gas holdup Low solid holdup
gasliquidsolid also Liquid average velocity Liquid phase
Liquid RMS velocity homogeneous in temperature
Adapted signal ++ ++
analysis required
Not adapted to small
velocities
Ultrasound probe GL and GLS Low gas holdup
Attenuation Gas holdup or ( 20%) + +
technique interfacial area
Doppler Bubble diameter Spherical bubbles += +
technique or bubble velocity Small distance
measurements
Pitot tube GL and GLS Flow direction Moderate gas holdup
or
Pressure drop =+ =+
Liquid velocity

ration ( ow regime, size of dispersed particles). A very in- measuring techniques (Table 2) cannot be avoided; yet, this
teresting work has been done by Warsito, Ohkawa, Kawata, is particularly true for highly turbulent systems, due to two
and Uchida (1999) to combine the measurement of ultra- main reasons:
sonic wave attenuation and propagation velocity to detect
the solid and gas-phase fractions inside a slurry bubble col- In case of nearly industrial operating conditions (partic-
umn. An array of ultrasonic emitters and receivers disposed ular physico-chemical characteristics, opaque walls, high
in parallel conguration at the column wall is rotated all gas holdup, etc.), non-invasive techniques become inef-
around the column. As shown by these authors, at high fre- fective because of walls (as for image analysis technique),
quency the wave propagation velocity does not depend on bubble number density (as for laser Doppler anemometry
gas fraction. The solid fraction is therefore determined from or Particule Image Velocimetry), and non-stationary phe-
the ultrasonic propagation velocity and the gas fraction is nomena: non-invasive techniques usually oer discrete
obtained by interpreting the measurement of ultrasonic wave or time-averaged data, making thus statistical or spectral
attenuation. The authors obtained cross-section images of analysis impossible or cumbersome.
phase fractions using a reconstruction procedure performed non-invasive techniques are often not easy to apply and
with a ltered back projection algorithm. The results are are often quite expensive!
quite interesting but it must be kept in mind that this ap-
The main invasive methods for local measurements ap-
plication is limited to holdup fractions of dispersed phases
peared in the early 60s, before the development of exper-
(solid and gas) below 20%, since wave re ections on gas
imental non-invasive hydrodynamic techniques: searchers
and solid interface rapidly generate a strong attenuation.
involved in experimental work observed quite early the in-
tense non-homogeneous and non-stationary characteristics
3. Invasive techniques of most gas/liquid ows. In the 70s, non-stationary phenom-
ena analysis was conducted through invasive techniques de-
Although optical techniques and tomography are inten- velopment; Jones and Delhaye (1976) wrote one of the rst
sively developed for the analysis of bubble ows, invasive reviews this topic, while Serizawa, Kataoka, and Mishigoshi
 Fig. 11. Optical needle probe (RBI, 1996).

Fig. 12. Double-tip optical probe (RBI, 1996).

(1975) published the rst article (a priori) on complete ex-

perimental investigation of both gas and liquid phases in
totransistor) delivers an analog output signal in proportion
two-phase ows.
to received light intensity.
Impedance needle probes are very similar to optical ones.
3.1. The so-called needle probes They are made of metal wire (stainless steel, nickel, corun-
dum, platinum (e.g. Frohlich et al., 1991; Hogsett & Ishii,
These probes are very thin, sharp ended; they are set- 1997) thus resisting to oxidation. Their diameter ranges
tled to face ow direction, in order to pierce as many bub- from 2:5 mm to 50 m, reaching 12 m in case of surgery
bles as possible. Needle probes are used to study high gas needles (Thang & Davis, 1979; Nakoryakov, Kashinsky, &
holdup systems (Nicol & Davidson, 1988; Utiger, Stuber, Kozenko, 1984). They are embedded within a metal support.
Duquenne, Delmas, & Guy, 1999; Bentifraouine, 1997). In An impedance probe behaves as an electrode (Jones
some cases, they allow a complete description of gas phase & Delhaye, 1976; Liu, 1993). A potentiometer (around
hydrodynamic characteristics. Two main types of needle 100 k
) measures the dierence in voltage between the
probes can be distinguished: probe and the support, the latter being usually grounded
(Hogsett & Ishii, 1997).
optical bre probes (Fig. 11) (b) Double probes. Double probes, or biprobes, need a
resistive or conductive probes, or even more generally tiny distance between the tips (some 0.5 5 mm) (Chabot
impedance probes. et al., 1988; Moujaes & Dougall, 1987; Chen, Zheng, Feng,
& Hofmann, 1998; Kang, Cho, Woo, Kim, & Kim, 2000)
Needle probes are used as single-tip or double-tip systems,
minimising the risk for a bubble which has been pierced
depending on the kind of data expected: single-tip probes
by the front tip, to miss the rear one. A suciently high
lead to gas fraction, and bubbling frequency; double-tip
number of bubbles identied by both bres allows a realistic
probes allow measurements of bubble velocity, time-average
statistical treatment for gas-phase hydrodynamic parameters.
local interfacial area, and mean bubble chord length.
Of course probe tips have to be aligned with bubble rise
direction (Fig. 12).
3.1.1. Principle Some special optical probes (Cartellier & Achard, 1991),
(a) probes and corresponding electronic device. Optical based on needle probe operation principles, have been de-
probes are made in quartz (Jones & Delhaye, 1976), and usu- veloped in laboratories.
ally protected by a protective layer except at the tip (1 cm
long) to allow measurements. The sensing tip can be also probes bent in a U shape tip (Jones & Delhaye, 1976;
made in sapphire for measurement in aggressive environ- Lee & De Lasa, 1987; Yu & Kim, 1991). Such probes
ment (e.g. in a three-phase uidised bed as in Bigot et al., are however extremely fragile and are, to the knowledge
1990). Probe tip diameter is usually smaller than 200 m of the authors, no longer in use,
and sometimes 50 m (Groen, Mudde, & van den Akker, probes with melt extremities, giving a spherically shaped
1995). tip (Moujaes, 1990; Chabot, Lee, Soria, & De Lasa, 1992;
In the case of an optical probe, an infra-red light beam Chabot & De Lasa, 1993).
is guided along the bre to its tip. The bre usually ends
with a thin conical tip to correctly pierce bubbles. Following Some authors proposed multiple (37) point probes
optic laws, this tip transmits the light beam when in liquid (Bruck & Hektor, 1984, Buchholz, Tsepetonides,
medium, or re ects it back (towards the electronic device) Steinemann, & Onken, (1983); Buchholz, Zakrzewsky,
when it is surrounded by gas (Chabot, Farag, & De Lasa, & Schugerl, 1981; Burgess & Calderbank, 1975; Jakobsen,
1998; Choi and Lee, 1990). An opto-electronic device (pho- Svendsen, & Hjarbo, 1992; Svendsen, Luo, Hjarbo, &
 With a double-tip probe, two signals are simultaneously
registered. The cross-correlation function between both sig-
nal leads to the most probable delay needed by a bubble to
migrate between the two tips (Revankar & Ishii, 1992). The
most probable bubble mean axial velocity is then derived
(Roig, Suzanne, & Masbernat, 1998; Moujaes & Dougall,
Fig. 13. Shape of A. Carteliers single bre optical probe. 1987). As mentioned earlier, a bubble can reach the front
tip and miss the rear one; a criteria is therefore required to
associate signal pulses caused by the same bubble on both
Jakobsen, 1998; Wol, 1989; Wol, Briegleb, Bader, tips (Revankar & Ishii, 1992).
Hektor, & Hammer, 1990; Yao, Zheng, Gasche, & To evaluate the specic time-averaged interfacial area,
Hofmann, 1991). A number of such devices are to be found some authors suggest a relation that takes mean bubble ve-
in the review of Lubbert (1991). It is possible with such locity Vb and bubble frequency Nt into account (Kataoka,
probes, at least in principle, to determine the full set of com- Ishii, & Serizawa, 1986; Revankar & Ishii, 1992; Dias,
ponents of the velocity vector, regardless of its direction, and Franca, & Rosa, 2000) in the case of spherical shape bub-
the bubble diameter. There are however practical problems bles (less than 5 mm in diameter) and for quiet bubbling
limiting the use of such devices: the probes rapidly become regime: the product (2Nt V1 ) is consistent with the inter-
b
bulky and so hydrodynamic interaction between bubbles facial area. A correction factor, based on RMS uctuation
and the probe cannot any longer be neglected; furthermore z of axial velocity component, is also used (Kataoka et
the exploitation software is far from easy to develop. al., 1986; Dias et al., 2000). Authors assume z to be
It is worth mentioning that Cartelliers group (Labora- equal to velocity components RMS uctuations in the other
toire des Ecoulements geophysiques et Industriels: LEGI, direction.
Grenoble) developed an original single-bre optical probe; Going back to raw signal bubble pulse width, and using
this probe is constituted of conical and cylindrical parts (Fig. calculated bubble velocity, it is possible to obtain the corre-
13), leading to a specic type of registered signal; this al- sponding chord size along which the probe tip has pierced
lows the simultaneous measurement of several gas-phase the bubble. The local bubble chord length distribution can
characteristics, just like a double probe: local gas holdup then be derived (Yu & Kim, 1991). Its rst momentum is
and interface velocities (Cartellier, Poupot, Chamberod, & the mean chord length C. 
Barrau, 1996; Cartellier, 1992; Barrau, Riviere, Poupot, & Through reverse transform of local bubble chord distribu-
Cartellier, 1999). By detecting the transit time of the bubble tion, it is possible to get the probable corresponding local di-
interface between the probe tip and the end of cylindrical ameter distribution, by assuming an homogeneous bubbling
part, it is possible to intrinsically deduce the bubble veloc- regime and spherical shape for bubbles (Uga, 1972), and by
ity. A specic application of this probe will be described in using a proper statistical law (Clark & Turton, 1988). This
Section 4.1.3. analysis has been extended to ellipsoidal shaped bubbles by
(c) delivered signal. Basic principles for needle probes op- Kamp, Colin, and Fabre (1995); and to heterogeneous bub-
eration being stated, it is easy to understand that the associ- bling regime by Liu, Clark, and Karamavruc (1998). The
ated electronic device delivers a signal varying between two precision is however less satisfactory in the later cases.
bounding levels; these levels depending on the phase sur- It is worth noting that, in the case of an innitely small
rounding the tip of the needle (gas or liquid); bubble-probe probe tip, the mean bubble diameter D is easily deduced
interaction and bubble detection have been extensively de- from the mean bubble chord C (Thang & Davis, 1979;
scribed in literature (Lubbert, 1991; Groen et al., 1995; Moujaes & Dougall, 1987; Frohlich et al., 1991): D = k C. 
Cartellier et al., 1996; Cartellier & Barrau, 1998; Cartellier, If bubble shape is perfectly spherical, k is equal to
1992; Barrau et al., 1999). 1.5; otherwise, it has to be experimentally optimised
In order to correctly describe signal variations between the (Frohlich et al., 1991).
two limiting levels, a high sampling frequency is required:
110 kHz, or even 1 MHz for real time sampling.
The acquired signal is then binarised (Revankar & Ishii, 3.1.3. Technique limitations
1992; Hogsett & Ishii, 1997). Trigger level choice may obey The application of needle probes is limited by some phys-
various criteria (Jones & Delhaye, 1976; Young, Carbonell, ical problems. It is evident, e.g., that, in order to yield reli-
& Ollis, 1991). A frequently used trigger to get rid of noise able results, the probe tip has to be very thin compared with
corresponds to 10% of bubble passage typical signal rise. the mean dimension of the bubble population studied. The
following other drawbacks are encountered: optical needle
3.1.2. Measuring results probes do not operate in all organic liquids because of a
Time-averaged gas holdup and bubbling frequency are too small dierence in refraction index with the gas phase.
directly reached through straight forward signal analysis Organic liquid conductivity is also too small for the use of
(Liu, 1993). impedance probes; in the case of low-conductivity aqueous
 alised by these authors to ows showing random bubbling
regime.

3.2. Heat transfer probes

To describe experimentally the liquid phase within

two-phase ow with moderate gas holdup, hot lm anemom-
etry is more and more often used (this invasive technique
had been developed for single-phase ow for a long time).
Hot lm anemometry, when applied to two-phase ows,
leads to gas fraction and to liquid-phase characteristics:
Fig. 14. Oscillating bubble plume (Salesse et al., 2001).
mean velocity and RMS uctuating velocity.

3.2.1. Basic principles

medium however, some salt may be added to allow reliable (a) interaction between probe and surrounding ow. Hot
measurements. lm anemometry is based on the heat exchanged between an
The following factors intrinsically linked to the measuring electrically heated probe and the surrounding liquid medium.
method may reduce its accuracy: Following King (1914), pioneer of hot probe anemometry,
the convective heat transfer coecient between a probe and
errors due to bubble-probe interaction: piercing hydrody- the surrounding uid depends on the characteristic Reynolds
namics (which depend on probe surface quality among number and can be described by: Nu = A + B Re0:5 .
other parameters) (Hogsett & Ishii, 1997; Cartellier et al., Electric current through the probe, and so hot lm resis-
1996), probe orientation in ow (Groen et al., 1995), bub- tance, are controlled by a Wheatstone bridge associated to
ble shape (for non-spherical bubbles, evaluation of sizes a regulation amplier and a high-pass lter.
and of interfacial area is often questionable (Buchholz (b) calibration. The relation between the probe voltage
et al., 1981); multiple point conductivity probes are then Ew and the liquid velocity is Kings law (Utiger et al., 1999):
to be preferred), E 2 = A + Buen , where, A; B and n are coecients to be
errors due to ow characteristics: turbulence, non-stationary identied; Ue is the apparent uid velocity (whereas the
phenomena (Groen et al., 1995), actual velocity is U , vector), sensed by probe. In the case
statistical bias: in a given area around the probe tip, large of a non-uniform ow, Ue is the normal component of U
bubbles are more likely to be pierced than small ones with regards to the probe.
(Thang & Davis, 1979), The coecients A; B and n of Kings law depend a pri-
errors due to signal treatment: critical trigger choice ori on physical properties (density, viscosity, thermal con-
for bubble identication (Nakoryakov, Kashinsky and ductivity) of the uid near the probe, but also on geomet-
Kozenko, 1984; Serizawa et al., 1975). ric characteristics of the sensor. These parameters may also
vary with the dierence in temperature between the hot lm
3.1.4. Illustration of needle probe application and the uid (Bruuns, 1995).
The recent work of Salesse, Larue de Tournemine, and Calibration of the probe (i.e. determination of the param-
Roig (2001), dedicated to the analysis of an oscillating bub- eters of Kings law) is done by a reference measuring tech-
ble plume, is a spectacular illustration of needle probe ap- nique: Pitot tubes are often chosen (Bruuns, 1995). It has to
plication. be stressed that a calibration curve has to be checked daily
A plume of ne bubbles has been specically generated because sensor characteristics may change rapidly.
within a rectangular vertical tank by means of capillary (c) Probe characteristics. The hot lm probe sensitive
tubes; the typical gas ow rate is 11:6 105 m3 =s (super- element (Fig. 16a) is tiny (25 100 m thickness); it is made
cial velocity: 4:9 mm=s). This plume oscillates in the trans- of a quartz substrate (thermal insulation) recovered with a
verse direction (Fig. 14) with a mean frequency of 0:235 Hz. platinum or nickel lm, and protected from oxidation and
The bubble plume envelope is characterised by means of abrasion by a quartz layer (1 or 2 m thickness).
a series of 10 resistive single-tip probes (regularly spaced In the case of multi-directional ows, two- or three-lm
with 1 cm distance). Each probe delivers a phase function probes are used (Fig. 16b and c).
in a 2500 s long signal which is binarised and plotted versus (d) Bubble ow measurements. Hot lm anemometry has
time (Fig. 15). been applied to bubble ows since the 60 s (Abel & Resch,
An iterative algorithm, derived on the basis of the cor- 1978). Bubble passage on probe generates a deep drop (Fig.
relation between phase function matrix and an assumed 17) in heat exchange and so creates a rupture of signal. This
model, leads to a rather precise representation of bubble is detailed by Bruuns (1995) for the example of a cylindrical
swarm frontiers. This procedure is currently being gener- sensor.
 Fig. 15. Binarized phase function for simultaneous registered signals from 10 mono-tip probes (Salesse et al., 2001).

Fig. 16. Common hot lm probes (Bruuns, 1995).

3.2.2. Measuring results velocity and its RMS uctuations are calculated using Kings
(a) Gas-phase fraction. The rst step of signal treat- law.
ment consists in identication of all bubbles. Several authors Literature reports a few works dedicated to turbu-
applied triggers of the signal magnitude (Delhaye (1969), lent quantities measurements in bubble using hot lm
mentioned by Young et al., 1991; Resch & Leutheusser, anemometry:
1972; Resch, Leutheusser, & Alemu, 1974; Jones & Zuber, the time-averaged local values: the liquid velocity is in-
1978; Toral, 1981; Farrar, 1988; Samways and Bruun, 1992; tegrated over the cross-section and can be compared with
Farrar, Samways, Ali, & Bruun, 1995). the liquid ow rate in the pilot (Serizawa et al., 1975;
It is in fact necessary to combine triggers in both mag- Utiger et al., 1999);
nitude and slope (Wang, Lee, Jones, & Lahey, 1990; Liu from RMS velocity uctuations, turbulent quantities are
& Banko, 1993; Utiger et al., 1999) to recognise bubbles. evaluated and analysed (Serizawa (1974), mentioned by
Time-averaged gas holdup may then be evaluated at probe Jones & Delhaye, 1976; Lance & Bataille, 1983; Magaud,
location. The reliability of this holdup calculation has been 2000). Hogsett and Ishii (1997) even distinguish turbu-
proved in airwater medium by Utiger et al. (1999) using lence due to bubble and turbulence due to wall friction;
an independent validation technique (optic needle probe). by means of a double-sensor probe, two components
(b) Liquid-phase velocity. When the signal drops due of liquid velocity may be simultaneously determined.
to bubble passage have been removed, the remaining Reynolds shear stress can then be deduced (Lance &
part, which corresponds to measurement within the liq- Bataille, 1983; Menzel, In Der Weide, Staudacher, Wein,
uid phase, can be analysed. The time-average local liquid & Onken, 1990);
 Fig. 17. Sample of hot lm probe output signal.

once parts due to bubbles have been eliminated from the Hot lm measurements errors arise from several factors:
signal, one can proceed to the auto-correlation function ow disturbance due to probe, calibration curve uncertainty,
calculation, even if the remaining signal is not continu- signal treatment, signal amplication, signal interpretation
ous. This function is expected to point out a character- in turbulent ow and in low velocity zones, and temperature
istic frequency corresponding to turbulence time scale non-uniformity. The last two points are the most dicult to
(Roig, Larrieu, & Suzanne, 1995). Turbulence and en- evaluate (Fingerson & Freymouth, 1983; Bruuns, 1995).
ergy dissipation power spectra may also be determined, Mass balance, when time-average local liquid velocities
in the same way as with single-phase ow analysis are integrated over a cross-section, has been satised with
(Schugerl, 1982): an error of 10 21% by Young et al. (1991) and of less than
11% by Utiger et al. (1999).
Liquid velocity, however, is in uenced by bubble rise Measurement accuracy is more dicult to estimate for
front and bubble drag. These specic velocity uctuations turbulent parameters; but results are rather reproducable and
are ambiguous: some authors (Roig et al., 1998) think that Lance and Bataille (1983) state that the corresponding error
they are not representative for liquid ow and that they is limited to 10%.
should not contribute to turbulent quantities evaluation.
These authors eliminate the corresponding parts from the
signal. 3.2.4. Similar or derived techniques
Many local techniques exist that are based on heat
transfer principles and are similar to hot lm anemometry
3.2.3. Technique limitations (Young et al., 1991; Bruuns, 1995). Most of them have
Hot lm anemometry can be applied with many media. been specically developed to answer the question of local
Caution must be taken because sensors are fragile: solid reverse ows. Among them, heat pulse probes and split lm
particles with high momentum have to be avoided. What is probes are rather common.
more, the ow studied has to be uniform in temperature. The same kind of probes can be derived using mass
Hot lm anemometry is not a very expensive technique instead of heat transfer: micro-electrodes rely on polaro-
and oers good measurement accuracy (0.1 to 0.2% for graphic technique using a redox system: the most frequent
rigorous experiments; this is as good as laser Doppler reaction is the oxidation of hexacyanoferrate II to hexa-
Anemometry) and a good signal-to-noise ratio. cyanoferrate III. The drawbacks are that the liquid must be a
No limitation in gas holdup has been notied in literature: conductive solution with relatively high support electrolyte
hot lm signals can be useful as long as bubbles can be concentrations, which may induce coalescence inhibition.
distinguished. Experiments on liquid and gas phase through This drawback can be overcome by using the reduction of
hot lm anemometry have been carried under some 20% gas dissolved oxygen on a silver or gold electrode, since this
holdup (Utiger et al., 1999). technique requires a less concentrated support electrolyte
 25 r/R = -1

z=0.10m

20 z=0.50m

z=0.75m

z = 75 cm
gas hold-up (%)

15

z = 50 cm
10

5
z = 10 cm
ai

0
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
r/R (-)

Fig. 18. Gas holdup (%) in an external loop airlift reactor (supercial gas velocity: 0:069 m=s) (Utiger et al., 1999).

(Pauli, Menzel, & Onken, 1989; Pauli, Onken, & propagation in a two-phase medium. These techniques can
Sobolk, 1991; Pannek, Pauli, & Onken, 1994). be divided into two types:
As with hot lm anemometry, the polarographic tech- Some use transmitted wave characteristics (attenuation):
niques needs a signal treatment to discriminate between bub- they lead to gas holdup or to time-averaged local interfacial
bles and liquid (Vial, 2000). This technique however allows area.
the investigation of mass transfer properties (Vial, 2000), Some (Doppler techniques, for instance) use characteris-
and also the measurement of hydrodynamic characteristics tics of the wave that re ected by the bubble surface: magni-
such as liquid velocity and shear stress at the wall (Mitchell tude and frequency of re ected wave are measured and lead
& Hanratty, 1966) or at the surface of a particle (Lesage, to bubble size or to bubble velocity.
2000), two-phase ow properties (Souhar, 1982; Lati et Probes are similar for both kinds of technique. Transmit-
al., 1992,1994a, b; Rode et al., 1994; Magaud, 1999). As tance techniques need an emitting probe and a receiving
with thermal anemometry liquid ow direction can be mea- one, both aiming at the measuring point; US Doppler probes
sured by using probes with three or more microelectrodes generally operate both as emitter and receiver.
(Pauli et al., 1989; Pauli et al., 1991; Pannek et al., 1994).

3.2.5. Hot lm anemometry: example of application 3.3.1. Probe description

Utiger et al. (1999) performed a detailed hydrodynamic The probe consists in a piezo-ceramic disk (less than
map of both phases (waterair) parameters in an exter- 1 mm thick) linked to a damping device included in a metal
nal loop airlift reactor (15 l). Authors used a single sensor tube (Fig. 20) (Stravs, Wahl, von Stokar, & Reilly, 1987;
hot lm probe (DANTEC, 55R11) and a double-tip optical Broring et al., 1991; Bouillard, Alban, Jacques, & Xuereb,
probe. 2001). These probes can be used in opaque, viscous and
Figs. 18 and 19 show radial proles of time-average and even aggressive media; they resist up to 140 C and 20 MPa
uctuating local liquid velocities, drawn for the upper part (Broring et al., 1991).
of the riser column. Such a map of phases properties gives a Acoustic pressure waves generate vibrations within the
good overview of the two-phase ow as well as a diagnostic piezo-ceramic disk; the disk transforms them into a voltage
of peculiar zones. In this work for instance, authors put in signal transmitted to a computer via an electronic device
evidence how, at the bottom of the riser, back ow disturbs constituted by appropriate lters.
gas and liquid distribution. Going up along the riser, gas and These techniques can be applied to US signals ranging
liquid velocities develop into proles of usual shape. between 0.2 and 5 MHz (ultrasound wave eld of frequen-
cies begins at 20 kHz). Ultrasound wave length has indeed
3.3. Ultrasound probes to be chosen signicantly dierent from bubble size to hin-
der resonance phenomena. It is worth noticing that the mea-
Ultrasound techniques for the measurement of bubble suring point locates at focal distance from probe (some 0.2
characteristics are based on the laws of ultrasound wave or 0:3 m and thus does not disturb the ow studied).
 0.6
r/R = -1
z=0.50m

0.5 z=0.75m

0.4
z = 75 cm
Ul (m/s)

0.3
z = 50 cm

0.2

z =10 cm
ai
0.1

0
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
r/R (-)

Fig. 19. Time-averaged liquid velocity (m/s) in an external loop airlift reactor (supercial gas velocity: 0:069 m=s) (Utiger et al., 1999).

vious studies because large frequency band sensors are

used.
Knowing transmittance T , bubble Sauter mean diameter
and sound dispersion coecient S, the local mean inter-
facial area a or the local gas holdup  can be calculated
(Stravs & von Stockar, 1985) by measuring the transmit-
tance T at two dierent frequencies:

8 ln(T ) 6
a= = ; (7)
SL(kdSM =2) dSM

where k = (2=c)f.
(b) Pulse echo technique. Gasliquid interfaces are ef-
fective ultrasound re ectors because acoustic impedance is
Fig. 20. Detailed cross-section view of an ultrasound Doppler probe very dierent in a gas or in a liquid. To analyse re ected
(Broring, Fischer, Korte, Sollinger, & Lubbert, 1991).
wave characteristics, pulsed signals are often used (pulse
echo technique) to lter echo through their transit delays.
3.3.2. Principles and expected data Note that the shorter the pulses, the weaker the echo dis-
(a) Ultra-sound transmittance technique (UTT). When tortion due to the contribution of other parts of the bubble
emitting and receiving probes are separated by a two-phase (Stravs et al., 1987); thus pulse emission frequency ranges
ow, transmittance T is evaluated through the voltages A around 10 kHz.
and A0 measured at the receiver probe in the presence or
absence of bubbles, respectively. re ected magnitude analysis:
T depends on: Focusing an acoustic beam towards a sphere holding at
a distance L from the probe (some 20 mm), one obtains
distance L between the two probes,
a bell shape prole for the echos acoustic magnitude R
bubble size,
(Fig. 21):
bubble projected surface,
On this graph, x co-ordinate stands for the radial position
ultra-sound wave characteristics (frequency f, celerity c
of bubble centre apart from acoustic propagation axis.
in liquid medium).
Probe-bubble distance being known through US pulse mi-
A recent measuring device was developed and used gration time, one can evaluate the size of a bubble hit at
by Bensler (1990) and Boyer (1996). It diers from pre- x = 0 by analysing this prole.
 3.3.3. Ultrasound techniques limitations
Generally speaking, ultrasound techniques allow instan-
taneous measurement of only one characteristic of the dis-
persed phase at a time; furthermore, they are ineective
under high gas holdup conditions (more than 20%), be-
cause the acoustic signal is lost due to repeated re ections
(Broring et al., 1991).
Bubble size evaluation method (Pulse Echo Technique)
can be partially extended to non-spherical bubbles but, in the
case of ellipsoidal bubbles for instance, this leads to a 20%
error in regard with photographic technique measurements
(Stravs et al., 1987).
Note that US Doppler technique is drastically in u-
enced by probe orientation in ow. Furthermore, when a
non-focused acoustic beam is used, spatial resolution and
signal-to-noise ratio become very poor as distance from
probe increases. According to Bouillard et al. (2001), this
technique is intrinsically less precise than Laser Doppler
Anemometry, but also much cheaper.

3.3.4. Ultrasound techniques applications

In the work of Ruer et al. (1995) phase distribution in
an external loop airlift (60 l) equipped with static mixers
and operated with viscous and non-viscous uids (water and
CMC solutions) was investigated. Bubble velocity (vertical
Fig. 21. Echo magnitude prole (Stravs et al., 1987).
and horizontal components) was measured by means of US
Doppler technique.
These authors observe that the distribution of bubble ve-
locity vertical component is wide and complex; velocities
of large and small bubbles are shown to be very dierent;
indeed some very small bubbles, much in uenced by liquid
Receiver
Transmitter ow, ow downwards in the column.
v In horizontal direction, bubble velocity uctuations
do not depend on bubble size anymore; they are due
exclusively to liquid-phase turbulence. The distribu-
tion of bubble velocity horizontal component is wider
bubble
than the corresponding distribution of vertical one (for
the same operating conditions); this observation shows
that turbulence is not isotropic for this ow: the largest
bubbles rise fast and take liquid along with them.
Fig. 22. US Doppler method basic principle for a both emitting and The horizontal component of bubble velocity is drasti-
receiving probe.
cally in uenced by the passage of large and energetic
eddies.

frequency analysis of the re ected wave (Doppler eect):

US Doppler anemometry is based on the dierence in fre- 3.4. Pitot tubes
quency between an emitted US pulse and the correspond-
ing received US pulse that has been re ected by a moving Pressure measurement is a robust and easy way of ow
bubble. This well-known method leads to the local bub- diagnosis. As it has been stated in Section 2, wall pressure
ble velocity distribution (Broring et al., 1991; Yao et al., transducers are widely applied nowadays, but intrusive pres-
1991; Fischer, Broring, & Lubbert, 1992; Ruer, Broring, sure probes, so-called Pitot tubes, are still very useful. For
& Schugerl, 1995). Its principle is shown on Fig. 22. bubbly ows, it may be used with an appropriate procedure.
This technique is well known in the eld of single-phase
The investigated volume of ow is not bigger than a few ows for uid velocity measurements, so the following de-
cubic centimetres. scription will be short.
3.4.1. Principle To evaluate liquid velocity by Pitot tube measurements in
Single Pitot tubes, dedicated to velocity local measure- bubble ow, the mean density of ow, m , has thus to be
ment, should be quite thin (12 mm) (Davis, 1979; Nicol & previously derived: by measuring rst the static pressure or
Davidson, 1988; Chue, 1975) to avoid ow disturbance. Five the gas holdup.
hole Pitot tubes have been described in literature (Bruuns,
1995; Chue, 1975; Ligrani, Singer, & Baun, 1989; Cho & 3.4.3. Limitations of the technique
Becker, 1985). Various shapes are encountered: streamlined, Pitot tube velocity measurements meet similar limitations
spherical or conical; the Pitot tube may even be constituted as measurements made with hot lm probes. Pitot tubes are
of ve parallel tubes. Each nozzle is linked to a manome- thicker and the derived data correspond therefore to a less
ter. The total pressure is read through the central hole; the ne degree of analysis.
other four nozzles, located all around the central one, mea- Pitot tubes may also be dicult to implement but they
sure static pressure. These Pitot tubes allow to measure both are robust and can be applied to hostile environments
the liquid velocity magnitude and direction. (when gasliquid ow includes solid particles). Nicol and
To perform pressure measurements in bubble ow, one Davidson (1988) estimate the error made on liquid-phase
has to hinder bubbles from entering the nozzles. To achieve mean velocity measurement to some 4%.
this, Nicol and Davidson (1988) feed their Pitot tube with As mentioned before a calibration is still needed when
a liquid ow (5 cm=s), making the probe free from the dis- used in bubble ow to take into account of bubble slip ve-
turbance due to bubbles. Caution must be taken, however, locity in the vicinity of Pitot probe.
because this type of apparatus may modify ow structure
around the probe. 3.4.4. Example of technique application
An example of the application of modied Pitot tube is
shown in Section 4.2.
3.4.2. Obtained data
There are two ways to use such Pitot tubes:
4. Application of multiphase metering for extrapolation of
The tubes can be rotated around their own axis to nd gas/liquid processes in the renery and petrochemical area
the precise orientation for which the four external tubes
deliver the same value of pressure; this particular po- In the petrochemical industry area, many processes are
sition reveals the mean direction of the ow. carried out using multiphase reactors since both gasand
The Pitot device can be set vertically in the ow; dif- liquidphase reactants are often involved. The reactions
ferences between the pressure measurements from two being often of catalytic type with a heterogeneous solid sup-
neighbouring holes are then investigated (right-left, ported catalyst, the reactor or absorption column may in-
up-down): volve two- or three-phase ow. These very complex ows
The dierence in pressure from two lateral holes are not easily predicted by commercial CFD codes and, as a
(separated by a vertical distance h) is the dierence consequence, a precise characterisation of the ow by using
in the local hydrostatic pressure, an appropriate multiphase metering system is still needed.
The local value of the apparent density of the During the development procedure, when the scale-up of the
two-phase mixture m is then calculated and leads reactor has to be performed, a global hydrodynamic charac-
to the local gas holdup with a relative error of some terisation is not sucient and local data are required. Fur-
10%. thermore, at this stage, hard constraints are imposed on mea-
The dierence between total pressure and static surement: the mock-up equipment to be investigated may be
pressure is equal to the dynamic pressure, and leads large and the measurement must be handled with real hy-
to the local liquid velocity. drocarbon uids, under pressure and with the real catalyst
According to Davis (1979), dynamic pressure support. As a conclusion such metering problems are quite
measurement from a Pitot tube in bubble ow dicult to solve since they must be precise with a good
gives always a larger value than expected ac- space resolution and must be robust enough to be used in a
cording to the predictive laws for single-phase semi-industrial environment. To illustrate this presentation
liquid ow, provided the Pitot probes size is kept two types of multiphase reactors have been chosen: xed
smaller than bubble mean diameter. A correction bed reactor and slurry bubble column.
factor, taking the disturbing eect of relative mo-
tion of phases around the Pitot tube into account, 4.1. Three-phase xed bed reactors
and ranging between 1 and 1.73, is then used
when measuring the local dynamic pressure. This One of the main problems during the scale-up of such re-
factor however depends also on gas holdup; the actors is to optimise the gas/liquid ow distribution across
maximal value of 1.73 is reached for 60% of gas the reactor catalytic bed section. In many cases, the chemi-
holdup (Davis, 1979). cal reaction involved is exothermic. In order to optimise the
reaction conversion or selectivity, all the catalytic volume
has to be seen by a gas/liquid ow with the same distri-
bution of phases, temperature and pressure. As a matter of
fact, an excess of gas in a catalytic zone can for instance ac-
celerate the reaction and so the local heat generation. This
zone being, in the same time, not refreshed by enough liquid
ow rate, a hot spot can occur, that will dramatically aect
reactor performance. The desirable uniform gas/liquid ow
distribution is all the more dicult to achieve as the reactor
diameter increases. During the development procedure, the
ow distribution has therefore to be correctly determined
on large size mock-up equipment : typically 40 or 60 cm in
diameter.
In a general manner, it is never easy to implement instru- Fig. 23. -ray tomographic system for xed bed reactor.
mentation inside two-phase ow; but with three-phase gas
liquidsolid conditions, the challenge is even more dicult.
The catalytic bed being generally very dense and opaque, no ment with gasliquid ow through the catalytic bed. In a
through-the-wall optical method can be used. To better in- similar way, the catalytic bed porosity can be measured
vestigate the bed, some local sensors have been developed, by dierence between a measurement with dry bed and a
such as e.g. optical probes. These probes give interesting in- measurement with empty column. The reconstructed images
formation, but the measurement volume being very small, a have a resolution of 128 128 pixels.
very large amount of measuring points is necessary to have The system has been tested under both static and dynamic
an idea of the gasliquid ow distribution over the reac- conditions. Among the dierent potential measurement er-
tor cross-section. It is thus necessary to develop techniques rors, the main ones are the statistical error due to random
that determine the ow distribution over the whole reactor emission of photons, the dynamic bias due to gas frac-
section with a good spatial resolution and that are not too tion uctuations and the reconstruction error. The intrinsic
intrusive. In this objective, three dierent techniques have performances of the system have been studied to examine
been developed at the French Institute for Petroleum (IFP). these dierent potential error sources, and the control pa-
The rst one is a -ray tomographic system to detect the liq- rameters, measuring time and scans number, have been op-
uid or gas fraction distribution in a reactor slice; the second timised (Boyer, Fanget, & Legoupil, 2000). The system is
one is a collecting system that allows to measure the liquid able to measure liquid holdup with an absolute accuracy of
and/or gas ows at the bed outlet, and the third one is an op- 3% for typical measuring time less than 2 h for a column
tical probe. The measurement principle and the measuring of 60 cm in diameter. Fig. 24 shows examples of measure-
limitations are fully described in the following sections. ments of the bed porosity, of the gas fraction in a bubble
ow conguration and of the liquid holdup in a trickle ow
conguration. Provided the ow remains stable during the
4.1.1. -ray tomography measuring time, this system gives the gas or liquid ow
The measuring principle of -ray tomography has been distribution with a quite satisfactory spatial resolution. One
described in Section 2.2.4. The major problem to overcome limitation of the system appears when it is used with porous
with xed bed reactor is the strong attenuation coming from particles: since it is impossible to discriminate between the
the solid particles. It is thus necessary to have both a source liquid remaining inside the particles and that remaining be-
with rather high activity and several detectors to minimise tween the particles at rest in drained bed conguration, the
the measuring time. The new tomographic system devel- liquid holdup measured is only the dynamic liquid holdup.
oped at the FIP is composed of a cesium source of 300 mCi
and a series of 32 detectors that are BGO photo-scintillator 4.1.2. Device for liquid collection
transducers. The system can be used with reactors of 40 or The second measuring technique has been developed to
60 cm in diameter (Fig. 23). The -ray has a fan beam ge- determine the liquid ow distribution at the catalyst bed
ometry and the whole system is automatically rotated around outlet. A specic collector tray has been inserted below the
the reactor. At each angular position a photon attenuation trickle bed to sample gas and liquid from 60 sectors of equal
prole is measured. After a complete measurement (360 C), surface (Fig. 25). Each gasliquid stream then ows through
a map of attenuation can be reconstructed using a classical a specic gasliquid separator (Fig. 26). All gas/liquid sep-
reconstruction algorithm, ltered back-projection algorithm arating transparent tubes are connected at their top to al-
(Kak & Slaney, 1987). Since the attenuation is directly pro- low for gas disengagement and to control bottom pressure.
portional to the material density, it is possible to determine Each tube is equipped with restriction valves at the bottom
the liquid or gas holdup by dierence between a measure- to maintain a constant liquid level. These valves are all cal-
ment in dry (or drained) bed conguration and a measure- ibrated with high precision in order to deduce the liquid
 0 600 600
0 600 0
(a) Bed porosity measurement (%) (b) Gas fraction measurement with a (c) Liquid fraction measurement with
liquid flow rate of 56 kg/m2/s and a a liquid flow rate of 1.4 kg/m2/s and a
gas flow rate of 0.76 kg/m2/s gas flow rate of 0.39 kg/m2/s

Fig. 24. Examples of bed porosity measurement (a); gas fraction measurement (b) and liquid retention measurement (c) inside real catalyst bed of 40
and 60 cm in diameter.

Fig. 25. Collector tray at bed outlet.

Fig. 27. Example of liquid ow map measured at the bed outlet.

ow rate from the liquid level, the total pressure drop and
the valve opening section. Fig. 27 shows a typical ow map
obtained with the collecting device with a bad initial liquid
distribution. The spatial resolution is limited because of the
technical complexity and because the diameter of the col-
lecting lines must be large enough to avoid any disturbing
pressure drop. Since this equipment is placed at the bed out-
let, when the liquid distribution has to be measured at dif-
ferent levels it is necessary to vary the catalyst bed height
inside the column.

4.1.3. Single tip optical probe

Finally, once having studied the gasliquid ow distribu-
tion, it can be of interest to characterise the gasliquid inter-
Fig. 26. Gasliquid separator system associated to the collector tray. facial area to have an idea of the mass transfer phenomena
 1.50
1.40
1.30
V 1.20
1.10

Log (VB )
1.00
0.90
0.80
0.70
0.60
0.50
0.40 y = -1.1525x + 1.634
0.30 R = 0.9921
0.20
Tm t 0.10
0.00
Tb 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 1.10 1.20 1.30 1.40

Log ( Tm )

Fig. 28. Measurement principle of the bubble velocity with a single optical probe.

between the phases across the bed. To achieve this objective, The gas-phase ow conguration: it is necessary to char-
a single optical probe technique has been adapted to detect acterise the gas holdup and gas mass ux proles along
the bubble size inside a trickle bed in bubble ow cong- the reactor axis and radius.
uration. The probe is an optical probe as described in the
Concerning the gas ow, the gas holdup and gas ve-
Section 3.1 The measuring principle developed by Cartellier
locity proles are measured using optical probes with a
(1992) consists here in correlating the bubble velocity with
sapphire sensing tip. These probes can be used with solid
the drying time of the probe sensing tip when the bubble
concentrations up to 10%. For higher solid concentrations,
cross it (see Fig. 28). Once knowing the bubble velocity and
the liquid/solid mixture is simulated by using a liquid of
the residence time of a bubble on the probe, it is very simple
same viscosity and same surface tension. The probe used
to determine the bubble chord length. This technique needs
and the signal analysis have been already described in Sec-
a calibration procedure for each probe and gasliquid sys-
tion 3.1 of this paper. The typical proles obtained in a gas
tem used, but only one probe is necessary to determine both
liquid bubble column are described in detail in Schweitzer,
velocity and bubble size. As shown in the previous section,
Bayle, and Gauthier (2001). These gas holdup proles are
it should be noted that this measurement is very sensitive
shown to be close to those measured in gas/solid uidised
to the angle between bubble direction and probe axis. As a
beds with other optical probes that detect the gas holdup by
consequence, to be used inside the bed, a protecting grid has
measuring the light re ected on bubble interface (the sens-
to be included around the probe to create a clean measur-
ing tip does not pierce the bubble).
ing volume and control locally the ow. This grid has been
To detect the total average gas and solid holdups, a dy-
tested inside and outside a catalyst bed (Boyer & Cartellier,
namic measurement of pressure drop is performed at the
1999). The results show that the grid creates locally a small
stop of the unit (Dynamic Gas Disengagement). The gas
gas channeling, but there is no bubble coalescence across
ow is suddenly interrupted: there is rst a disengagement
the grid and the bubble velocity and size measured are sim-
of gas bubbles and then the solid particles settle gently. By
ilar to those measured at the bed outlet. These tests have to
measuring the dierent pressure levels after each step it is
be performed under other ow conditions to complete the
possible to determine the gas and solid fraction, knowing
probe validation but this technique looks very promising.
the liquid and solid density. Fig. 29 shows an example of
pressure measurement during the disengagement procedure.
4.2. Slurry bubble reactor
This type of measurement gives precise and robust results
but it only provides a global information of the phase frac-
In this second type of multiphase ow reactor used in
tion over the whole reactor volume and no time-dependant
the petrochemical area, the reaction takes place in the liquid
information can be obtained.
phase and the gas reactant ows through the liquid phase
In order to determine the liquid-phase agitation, two
from bottom to the top of the reactor. The catalyst support
dierent kinds of local probes have been used in gas
is a solid powder dispersed within the liquid phase. In this
liquid ow conguration with a liquid of same viscosity
type of reactor, several hydrodynamic problems have to be
and surface tension as the liquid/solid mixture. The rst
solved in order to achieve good conversion performance.
ones are used to measure the overall residence time dis-
The main characteristics to be determined are:
tribution by local detection of a tracer concentration for
The liquid-phase agitation: the liquid phase has to be ho- determining both axial and radial diusion coecients. The
mogeneously mixed, but with moderate liquid local ve- second ones are used to measure the local values of liquid
locities to avoid the catalyst phase deterioration. velocity.
 140

135
Solid phase settling
Gas disengagement
130 Separating wall
125
!P in mbar

120

115

110

105

100
900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900
Holes
Time in sec !P

Fig. 29. Pressure drop measurement during the disengagement procedure.

Fig. 31. Modied pitot probe geometry; Warsito et al. (1999).

Si maximum of interference. For this typical wavelength, there

is a direct relation between the cavity width d and the re-
d fractive index n:

3
n= : (8)
d
TiO2
With these probes it is possible to locally detect the tracer
concentration at dierent positions directly inside the reactor
Cavity
without taking account of a sampling line transit time. Since
crossed
the signal is an optical wave, the response time can be as
by liquid
low as 0:1 ms. This kind of probe can be used with dierent
Optical kinds of liquids (hydrocarbon uids or aqueous solutions),
fiber provided the liquid refractive index ranges from 1.3 to 1.5.
In a gas/liquid ow conguration, the probes can be used
unless the measuring cavity is totally dried by the gas.
Finally, for the measurement of the local liquid velocity, a
modied Pitot probe has been developed. This probe is com-
posed of a single tube separated in two dierent sections.
Fig. 30. Geometry of the probe for refractive index measurement. Two holes are perforated following axis at 90 , one in each
tube section (Fig. 31). The two tube sections are connected
to a dierential pressure drop measurement device. When
The axial and radial diusion coecients are determined the liquid ow direction is well known, one hole is oriented
for dierent values of solid concentration and gas ow rate face to the ow and the second hole is oriented tangentially
by performing a residence time analysis at dierent lev- to the ow direction. As said in Section 3.4, the liquid veloc-
els in the slurry bubble column (see Section 2.1.4). The ity can then be deduced from the pressure dierence value,
tracer used is a liquid with a refractive index dierent from once the liquid density is known. When the liquid ow di-
that of the liquid used. The tracer is injected into the con- rection is not known, the measuring tube can be rotated and
tinuous liquid phase and its concentration is measured by by analysing the pressure drop signals at dierent angular
detecting the liquid refractive index with local specic opti- position it is possible to determine the liquid ow direction
cal probes (Graindorge, Laloux, Girault, Martin, & Lefevre, and velocity. This type of probe can be used in gas/liquid
1996). These probes have a very thin sensing tip that in- ow conditions if the liquid phase remains the continuous
cludes a cavity crossed by the liquid phase and delimited by phase. The problem to overcome is to ensure that no gas
two walls (see Fig. 30). A white light beam is sent through bubble is present in the measuring tube by injecting a liquid
an optical bre to this cavity. The rst wall re ects approx- ow through the pressure lines before each measurement
imately 30% of the light signal and the second wall re ects (see Section 3.4). This technique is robust and simple to im-
the other fraction of light that has crossed the liquid phase. plement but the measuring tube remains intrusive and must
By analysing the power spectrum of the resulting light it be quite rigid to ensure that the pressure measurements are
is possible to detect the wavelength  corresponding to the not perturbed by the tube uctuations. This modied Pavlov
tube has been used by Hills (1974) and by Vial (2000) in Bensler, H.P. (1990). Determination de laire interfaciale, du taux de
academia, but seldom under industrial conditions. vide et du diametre moyen de Sauter dans un e coulement a bulles a
partir de lattenuation dun faisceau dultrasons. These de doctorat,
Institut National Polytechnique de Grenoble, France.
Bentifraouine, C. (1997). Hydrodynamique globale, locale et transfert de
5. Conclusion matiere dans un reacteur airlift a boucle externe. These de doctorat,
Institut National Polytechnique de Toulouse, France.
Multiphase ows are very complex because of the diver- Bigot, V., Guyot, I., Bataille, D., & Roustan, M. (1990). Possibilites
sity in nature of the dierent possible phases and of the dif- dutilisation de deux techniques metrologiques, capteurs de pression
ferent possible ow regimes. Their experimental investiga- et bres optiques, pour caracteriser lhydrodynamique dun lit uidise
triphasique. In A. Storck, & G. Wild (Eds.), Recents Progres en Genie
tion is therefore not easy and the most recent non-invasive
des Procedes, Vol. 10 (pp. 143149). Technique et Documentation.
local techniques (Particle Image Velocimetry for instance) Paris: Lavoisier.
are useless in most cases. Multiphase ow analysis requires Blet, V., Berne, P., Chaussy, C., Perrin, S., & Schweich, D. (1999).
dedicated measuring methods to get local or global data. Characterization of a packed column using radioactive tracers.
This paper oers a review of measuring techniques avail- Chemical Engineering Science, 54, 91101.
able to characterise multiphase systems such as the gas Bouillard, J., Alban, B., Jacques, P., & Xuereb, C. (2001). Liquid
ow velocity measurements in stirred tanks by ultra-sound Doppler
liquid or gasliquidsolid reactors that are heavily encoun-
velocimetry. Chemical Engineering Science, 56, 747754.
tered in process engineering area. The very large variety of Boyd, J. W. R., & Varley, J. (1998). Sound measurement as a means
existing techniques puts in evidence the intense worldwide of gas-bubble sizing in aerated agitated tanks. American Institute of
research activity devoted to develop appropriate multiphase Chemical Engineers Journal, 44, 17311739.
ow instrumentation. Boyer, C. (1996). Etude dun procede de mesure des debits dun
For each type of multiphase ow encountered, there is of- e coulement triphasique type eau/huile/gaz. These de doctorat, Institut
National Polytechnique de Grenoble, France.
ten not only one technique to be recommended, as shown by
Boyer, C., & Cartellier, A. (1999). Bubble velocity and size estimation
the examples of application reported in Section 3. The nal using a single optical probe in a gas/liquid ow across a xed bed
choice must take into account the spatial and time resolution reactor. Recents Progres en Genie des Procedes, 13(69), 379386.
required, the measuring constraints (presence of solid, of hy- Boyer, C., Fanget, B., & Legoupil, S. (2000). Development of a
drocarbon uids), the expected range of measuring values new Gamma-Ray Tomographic system to investigate two-phase
(high or low holdups) and the disturbance and inaccuracy gas/liquid ows in trickle bed reactor of large diameter. CHISA2000,
Proceedings Paper no. 588, Prague.
allowed . Several techniques may be simultaneously derived
Briens, L. A., Briens, C. L., Hay, J., Hudson, C., & Margaritis, A. (1997a).
to get complementary data. Hursts analysis to detect minimum uidization and gas maldistribution
in uidized beds. American Institute of Chemical Engineers Journal,
43, 19041908.
References Briens, L. A., Briens, C. L., Margaritis, A., & Hay, J. (1997b). Minimum
liquid uidization velocity in gasliquidsolid uidized beds. American
Bakshi, B. R., Zhong, H., Jiang, P., & Fan, L-S. (1995). Analysis of Institute of Chemical Engineers Journal, 43, 11801189.
ow in gasliquid bubble columns using multi-resolution methods. Briens, C. L., Margaritis, A., & Wild, G. (1995). A new stochastic model
Transactions of the Institution of Chemical Engineers, A: Chemical and measurement errors in residence time distributions of multiphase
Engineering Research and Design, 73, 608614. reactors. Chemical Engineering Science, 50, 279287.
Barrau, E., Riviere, N., Poupot, C., & Cartellier, A. (1999). Single Broder, D., Lan, S., & Sommerfeld, M. (2000). Experimental studies
and double optical probes in airwater two-phase ows: Real time of the hydrodynamics in a bubble column. In R. Kopsel, & P.
signal processing and sensor performance. International Journal of Kuchling (Eds.), Fifth GermanJapanese symposium bubble columns
Multiphase Flow, 25, 229256. (pp. 125 130). Medienzentrum der Bergakademie Freiberg: VDI,
Bauckhage, K. (1996). Gleichzeitige Erfassung von Partikelmerkmalen ISBN 3-86012-109-X 2000.
und Eigenschaften mehrphasiger Stromungen mit Hilfe der Broring, S., Fischer, J., Korte, T., Sollinger, S., & Lubbert, A. (1991).
Phasen-Doppler-Anemometrie. Chemie-Ingenieur-Technik, 68, 253 Flow structure of the dispersed gasphase in real multiphase chemical
266. reactors investigated by a new ultrasound-Doppler technique. Canadian
Becker, S., De Bie, H., & Sweeney, J. (2000). Study on the ow structure Journal of Chemical Engineering, 69, 12471256.
in an aerated at apparatus using laser Doppler anemometry. Chemical
Bruck, F. J., & Hektor, K. (1984). Einsatz der Vierpunkt-Leitfahigkeitsmi-
Engineering and Technology, 23, 222226.
krosonde im Dreiphasen-Fliebett. Chemie-Ingenieur-Technik, 56,
Begovich, J. M., & Watson, J. S. (1978). An electroconductivity method 920922.
for the measurement of axial variation of holdups in three-phase
Bruuns, H. H. (1995). Hot wire aneometry. Principles and signal
uidized beds. American Institute of Chemical Engineers Journal,
analysis. New York: Oxford University Press Inc.
24, 351354.
Buchholz, R., Tsepetonides, J., Steinemann, J., & Onken, U. (1983).
Beinhauer, R. (1971). Dynamische Messungen des relativen Gasgehalts
In uence of gas distribution on interfacial area and mass transfer in
in Blasensaulen mittels Absorption von Rontgenstrahlen. Dissertation
bubble columns. German Chemical Engineering, 6, 105113.
TU Berlin; cited by Deckwer (1985).
Bendjaballah, N., Dhaouadi, H., Poncin, S., Midoux, N., Hornut, J. M., Buchholz, R., Zakrzewsky, W., & Schugerl, K. (1981). Techniques for
& Wild, G. (1999). Hydrodynamics and ow regimes in external loop determining the properties of bubbles in bubble columns. International
airlift reactors. Chemical Engineering Science, 54, 52115221. Chemical Engineering, 21, 180187.
Bendjaballah-Lalaoui, N. (2000). Hydrodynamique globale et locale dans Buham, B. A., & Mason, G. (1993). Hold-up and dispersion: Tracer
un reacteur a gazosiphon a recirculation externe. These de doctorat, residence times, moments and inventory measurements. Chemical
Institut National Polytechnique de Lorraine, Nancy, France. Engineering Science, 48, 38793887.
Burgess, J. M., & Calderbank, P. H. (1975). The measurement of bubble Dhaouadi, H., Poncin, S., Hornut, J.-M., Wild, G., Oinas, P., & Korpijarvi,
parameters in two-phase dispersions II. The structure of sieve tray J. (1997). Mass transfer in an external loop airlift reactor: Experiments
froths. Chemical Engineering Science, 30, 11071121. and modeling. Chemical Engineering Science, 52, 39093917.
Calderbank, P. H. (1958). Physical rate processes in industrial fermenters. Dias, S. G., Franca, F. A., & Rosa, E. S. (2000). Statistical method to
Part I: The interfacial area in gas/liquid contacting with mechanical calculate local interfacial variables in two-phase bubbly ows using
agitation. Transactions of the Institution of Chemical Engineers, 36, intrusive crossing probes. International Journal of Multiphase Flow,
443463. 26, 17971830.
Callaghan, P. T. (1991). Principles of nuclear magnetic resonance Drahos, J., Bradka, F., & Puncochar, M. (1992). Fractal behaviour
microscopy. New York: Oxford University Press. of pressure uctuations in a bubble column. Chemical Engineering
Camarasa, E, Vial, C., Poncin, S., Wild, G., Midoux, N., & Bouillard, Science, 47, 40694075.
J. (1999). In uence of coalescence behaviour of the liquid and of Drahos, J., & Cermak, J. (1989). Diagnostics of gasliquid ow patterns in
gas sparging on hydrodynamics and bubble characteristics in a bubble chemical engineering systems. Chemical Engineering and Processing,
column. Chemical Engineering and Processing, 38, 329344. 26, 147164.
Cartellier, A. (1992). Simultaneous void fraction measurement, bubble Drahos, J., Zahradnk, J., Puncochar, M., Fialova, M., & Bradka,
velocity, and size estimate using a single optical probe in gasliquid F. (1991). Eect of operating conditions of the characteristics of
two-phase ows. Review of Scientic Instruments, 63, 54425453. pressure uctuations in a bubble column. Chemical Engineering and
Cartellier, A., & Achard, J-L. (1991). Local detection probes in uid uid Processing, 29, 107115.
two-phase ows. Review of Scientic Instruments, 62(2), 279303. Dudukovic, M. P., Devanathan, N., & Holub, R. (1991). Multiphase
Cartellier, A., Poupot, C., Chamberod, E., & Barrau, E. (1996). Sondes reactors: Models and experimental verication. Revue de lInstitut
optiques: Innovations sur un capteur classique. La Houille Blanche, Francais du Petrole, 46, 439465.
1/2, 120128. Euzen, J.-P., & Fortin, Y. (1987). Partikelbewegung in einem
Cassanello, M., Larachi, F., Marie, M. N., Guy, C., & Chaouki, J. (1995). Dreiphasen ie bett. Chemie-Ingenieur-Technik, 59, 416419.
Experimental characterization of the solid phase chaotic dynamics Fan, L. S., & Tsuchiya, K. (1990). Bubble wake dynamics
in three-phase uidization. Industrial and Engineering Chemistry in liquids and liquid-solid suspensions. Stoneham, MA, USA:
Research, 34, 29712980. Butterworths-Heinemann Series in Chemical Engineering.
Chabot, J., & De Lasa, H. I. (1993). Gas holdups and bubble Farrar, B. (1988). Hot-lm anemometry in dispersed oilwater
characteristics in a bubble column operated at high temperature. ows. Ph.D. thesis, Department of Mechanical and Manufacturing
Industrial and Engineering Chemistry Research, 32, 25952601. Engineering, University of Bradford, UK.
Chabot, J., Farag, H., & De Lasa, H. (1998). Fluid dynamics of bubble Farrar, B., Samways, A. L., Ali, J., & Bruun, H. H. (1995). A
columns at elevated temperature modelling and investigation with computer based hot-lm technique for two-phase ow measurements.
refractive ber optic sensor. The Chemical Engineering Journal, 70, Measurements in Science and Technology, 6, 15281537.
105113. Fingerson, L. M., & Freymouth, P. (1983). Thermal anemometers. In
Chabot, J., Lee, S. L. P., Soria, A., & De Lasa, H. I. (1992). Interaction R. J. Goldstein (Ed.), Fluid mechanics measurements (pp. 99 154).
between bubbles and ber optic probes in a bubble column. Canadian Washington: Hemisphere.
Journal of Chemical Engineering, 70, 6168. Fischer, J., Broring, S., & Lubbert, A. (1992). Gas-phase properties in
Chaouki, J., Larachi, F., & Dudukovic, M. P. (1997). Noninvasive stirred tank bioreactors. Chemical Engineering Technology, 15, 390
tomographic and velocimetric monitoring of multiphase ows. 394.
Industrial and Engineering Chemistry Research, 36, 44764503. Flaschel, E., Metzdorf, C., & Renken, A. (1987). Methode
Chen, J., Kemoun, A., Al-Dahhan, M. H., Dudukovic, M. P., Lee, zur Messung der axialen Vermischung des Feststos in
D. J., & Fan, L.-S. (1999). Comparative hydrodynamics study in Flussigkeit/Feststo-Wirbelschichten. Chemie-Ingenieur-Technik, 59,
bubble columns using computer-automated radioactive particle tracking 494496.
(CARPT)/computed tomography (CT) and particle image velocimetry Fortin, Y. (1984). Reacteurs a lit uidise triphasique: caracteristiques
(PIV). Chemical Engineering Science, 54, 21992207. hydrodynamiques et melange des particules solides. These de
Chen, Z., Zheng, C., Feng, Y., & Hofmann, H. (1998). Local bubble docteur-ingenieur, Universite de Paris VI, ENSPM, Solaize, France.
behavior in three-phase uidized beds. Canadian Journal of Chemical Fournier, T., & Jeandey, C. (1993). Optimization of an experimental setup
Engineering, 76, 315318. for void fraction determination by the X-ray attenuation technique.
Cho, S. H., & Becker, H. A. (1985). Response of static pressure probes In J.M. Delhaye & G. Cognet (Eds.), Measuring techniques in gas
in turbulent streams. Experiments in uids, 3, 93102. liquid two-phase ows (pp. 199 228). Berlin, Heidelberg, New York,
Choi, K. H., & Lee, W. K. (1990). Comparison of probe methods Tokyo: Springer.
for measurement of bubble properties. Chemical Engineering Frohlich, S., Lotz, M., Larson, B., Lubbert, A., Schugerl, K., &
Communications, 91, 3547. Seekamp, M. (1991). Characterization of a pilot plant airlift tower
Chue, S. H. (1975). Pressure probes for uid measurements. In D. loop reactor: III. Evaluation of local properties of the dispersed gas
Kuchemann (Ed.), Progress in aerospace sciences, Vol. 16 (pp. 147 phase during yeast cultivation and in model media. Biotechnology and
223). Bioengineering, 38, 5664.
Clark, N. N., & Turton, R. (1988). Chord length distributions related to Froystein, T. (1997). Flow imaging by gamma-ray Tomography: Data
bubble size distributions in multiphase ows. International Journal of processing and reconstruction techniques. Frontiers in Industrial
Multiphase Flow, 14, 413424. Process Tomography II, Delft, April 812, 1997.
Daly, J. G., Patel, S. A., & Bukur, D. B. (1992). Measurement of Funfschilling, D., & Li, H.-Z. (2001). Flow of non-Newtonian uids
gas holdups and Sauter mean bubble diameters in bubble column around bubbles: PIV measurements and birefringence visualisation.
reactors by dynamic gas disengagement method. Chemical Engineering Chemical Engineering Science, 56, 11371141.
Science, 47, 36473654. Garca Ochoa, J., Khalfet, R., Poncin, S., & Wild, G. (1997).
Deckwer, W.-D. (1985). Reaktionstechnik in Blasensulena. Frankfurt, Hydrodynamics and Mass Transfer in a suspended solid bubble column
Aarau: Salle+Sauerlander. with polydispersed high density particles. Chemical Engineering
Delnoij, E., Kuipers, J. A. M., van Swaaij, W. P. M., & Westerweel, J. Science, 52, 38273834.
(2000). Measurement of gasliquid two-phase ow in bubble columns Gladden, L. F. (1994). Nuclear magnetic resonance in chemical
using ensemble correlation PIV. Chemical Engineering Science, 55, engineering: Principles and applications. Chemical Engineering
33853395. Science, 49, 33393408.
Glasgow, L. A., Erickson, L. E., Lee, C. H., & Patel, S. A. (1984). Wall Kluytmans, J. H. J., van Wachem, B. G. M., Kuster, B. F. M., Schouten,
pressure uctuations and bubble size distributions at several positions J. C. (2001). Gas hold-up in a slurry bubble column: In uence of
in an airlift fermentor. Chemical Engineering Communications, 29, electrolyte and carbon particles. Industrial and Engineering Chemistry
311336. Research, submitted for publication.
Glasgow, L. A., Hua, J., Yiin, T. Y., & Erickson, L. E. (1992). Acoustic Kolb, W. B., Melli, T. R., de Santos, J. M., & Scriven, L. E. (1990).
studies of interfacial eects in airlift reactors. Chemical Engineering Cocurrent down ow in packed beds. Flow regimes and their acoustic
Communications, 113, 155181. signatures. Industrial and Engineering Chemistry Research, 29, 2380
Graindorge, Ph., Laloux, B., Girault, M., Martin, Ph., & Lefevre, H. C. 2389.
(1996). Metricor 2000: A multiparameter ber-optic sensor instrument. Kumar, S. B., Moslemian, D., & Dudukovic, M. P. (1995). A gamma-ray
Fiber optic and laser sensor XIV, 79 August 1996, Denver, CO. tomographic scanner for imaging voidage distribution in two-phase
Groen, J. S., Mudde, R. F., & van den Akker, H. E. A. (1995). Time ow systems. Flow Measurement Instrumentation, 6, 6173.
dependant behaviour of the ow in a bubble column. Transactions Lage, P. L. C., & Esposito, R. O. (1999). Experimental determination
of the Institution of Chemical Engineers, A: Chemical Engineering of bubble size distributions in bubble columns: Prediction of mean
Research and Design, 73, 615620. bubble diameter and gas hold up. Powder Technology, 101, 142150.
Gupta, P., Al-Dahhan, M. H., Dudukovic, M. P., & Mills, P. L. (2000).
Lance, M., & Bataille, J. (1983). Turbulence in the liquid phase of a
A novel signal ltering methodology for obtaining liquid phase
bubbly airwater ow. Advances in two-phase ow and heat transfer,
tracer responses from conductivity probes. Flow Measurement and
1, 403 427. NATO Specialist Meeting, NATO Advances Science
Instrumentation, 11, 123131.
Institute Series, West Germany.
Haario, H. (1994). MODEST users guide. Helsinki, Finland: Profmath.
Lara Marquez, A. (1992). Etude du transfert de matiere gaz-liquide dans
Heindel, T. J. (2000). Gas ow regime changes in a bubble column lled
les reacteurs a lit xe avec e coulement a cocourant ascendant de
with a bre suspension. Canadian Journal of Chemical Engineering,
gaz et de liquide. These de doctorat, Institut National Polytechnique
78, 10171022.
Hills, J. H. (1974). Radial non-uniformity of velocity and voidage in a de Lorraine, Nancy, France.
bubble column. Transactions of the Institution of Chemical Engineers, Larachi, F., Chaouki, J., & Kennedy, G. (1995). Three dimensional
52, 17. mapping of solids ow elds in multiphase reactors with RPT.
Hogsett, S., & Ishii, M. (1997). Local two-phase ow measurements using American Institute of Chemical Engineers Journal, 41, 439443.
sensor techniques. Nuclear Engineering and Design, 175, 1524. Larachi, F., Laurent, A., Wild, G., & Midoux, N. (1991). Some
Jakobsen, H. A., Svendsen, H. F., & Hjarbo, K. W. (1992). On the experimental liquid saturation results in xed-bed reactors operated
prediction of local ow structures in internal loop and bubble column under elevated pressure in cocurrent up ow and down ow of the gas
reactors using a two- uid model. Computers & Chemical Engineering, and the liquid. Industrial and Engineering Chemistry Research, 30,
175, 55315536. 24042410.
Johnsson, F., Zijerveld, R. C., Schouten, J. C., van den Bleek, & C. Lati, M. A., Naderifar, A., & Midoux, N. (1994a). Stochastic analysis
M., & Leckner, B. (2000). Characterization of uidization regimes by of the local velocity gradient in a trickle-bed reactor. Chemical
time-series analysis of pressure uctuations. International Journal of Engineering Science, 49, 52815289.
Multiphase Flow, 26, 663715. Lati, M. A., Naderifar, A., Midoux, N., & Le Mehaute, A. (1994b).
Jones, O. C., & Delhaye, & J. M. (1976). Transient and statistical Fractal behaviour of local liquid-solid mass transfer uctuations at
measurement techniques for two-phase ows. International Journal the wall of a trickle-bed reactor. Chemical Engineering Science, 49,
of Multiphase Flow, 3, 89. 38233829.
Jones, O. C., & Zuber, N. (1978). Use of a cylindrical hot-lm Lati, M. A., Rode, S., Midoux, N., & Storck, A. (1992). The use of
anemometer for measurement of two-phase void and volume ux microelectrodes for the determination of ow regimes in a trickle-bed
proles in a narrow rectangular channel. American Institute of reactor. Chemical Engineering Science, 47, 19551961.
Chemical Engineers Symposium Series, 74(174), 191204. Leclerc, J.-P., Claudel, S., Lintz, H.-G., Potier, O., & Antoine, B. (2000).
Joseph, S., & Shah, Y. T. (1986). Errors caused by tracer solubility in Theoretical interpretation of residence-time distribution measurements
the measurement of gas phase axial dispersion. Canadian Journal of in industrial processes. Oil and Gas Science and TechnologyRevue
Chemical Engineering, 64, 380386. de lInstitut Francais du Petrole, 55, 159169.
Joshi, J. B. (2001). Computational ow modelling and design of bubble Lee, S. L. P., & De Lasa, H. I. (1987). Phase holdups in three-phase
column reactors. Chemical Engineering Science, 56, 58935933. uidized beds. American Institute of Chemical Engineers Journal,
Joshi, J. B., Patil, T. A., Ranade, V. V., & Shah, Y. T. (1990). 33, 13591370.
Measurement of hydrodynamic parameters in multiphase sparged
Lee, D. J., Luo, X., & Fan, L.-S. (1999). Gas disengagement technique
reactors. Reviews in Chemical Engineering, 6(23), 73227.
in a slurry bubble column operated in the coalesced bubble regime.
Kak, A. C., & Slaney, M. (1987). Computerized tomographic imaging.
Chemical Engineering Science, 54, 22272236.
New York: IEEE Press.
Lesage, F. (2000). Modelisation et experimentation des transferts de
Kamp, A. M. Colin, C., & Fabre, J. (1995). Techniques de mesure par
matiere et de quantite de mouvement dans les reacteurs a lit xe.
sonde optique double en e coulement diphasique a bulles. Cert Onera,
These de doctorat, Institut National Polytechnique de Lorraine, Nancy,
Colloque de Mecanique des Fluides Experimentale, Toulouse, France,
France.
1112 May, 1995.
Kang, Y., Cho, Y. J., Woo, K. J., Kim, K. I., & Kim, S. D. (2000). Bubble Letzel, H. M., Schouten, J. C., Krishna, R., & van den Bleek, C. M.
properties and pressure uctuations in pressurized bubble columns. (1997). Characterisation of regimes and regime transitions in bubble
Chemical Engineering Science, 55, 411419. columns by chaos analysis of pressure signals. Chemical Engineering
Kasireddy, V. K., & Al Taweel, A. M. (1990). An improved light Science, 52, 44474459.
attenuation technique for measuring large interfacial areas. Canadian Levenspiel, O. (1999). Chemical reaction engineering (3rd ed.). New
Journal of Chemical Engineering, 68, 690693. York: Wiley.
Kataoka, I., Ishii, M., & Serizawa, A. (1986). Local formulation and Ligrani, P. M., Singer, B. A., & Baun, L. R. (1989). Miniature ve-hole
measurements of interfacial area concentration in two-phase ow. pressure probe for measurement of three mean velocity components
International Journal of Multiphase Flow, 12, 505529. in low speed ow. Journal of Physics E: Scientic Instruments, 22,
Kikuchi, R., Tsutsumi, A., & Yoshida, K. (1997). Stochastic analysis of 868876.
bubble and particle motions in a 2-D three-phase reactor. Journal of Lin, T.-J., Juang, R.-C., Chen, Y.-C., & Chen, C.-C. (2001). Predictions
Chemical Engineering of Japan, 30, 202209. of ow transitions in a bubble column by chaotic time series analysis
 of pressure uctuation signals. Chemical Engineering Science, 56, Murai, Y., Song, X. Q., Takagi, T., Ishikawa, M., Yamamoto, F., &
10571065. Ohta, J. (2000). Inverse energy cascade structure of turbulence in
Lin, T.-J., Tsuchiya, K., & Fan, L.-S. (1998). Bubble ow characteristics a bubbly ow. PIV measurement and results Japanese Society of
in bubble columns at elevated pressure and temperature. American Mechanical Engineers International Journal Series B, Fluids and
Institute of Chemical Engineers Journal, 44, 545560. Thermal Engineering, 43(2), 188196.
Liu, T. J. (1993). Bubble size and entrance length eects on void Naderifar, A. (1995). Etude experimentale locale et globale du transfert
development in a vertical channel. International Journal of Multiphase de matiere liquide/solide a la paroi dun reacteur a lit xe. These de
Flow, 19, 99113. doctorat, Institut National Polytechnique de Lorraine, Nancy, France.
Liu, W., Clark, N. N., & Karamavruc, A. I. (1998). Relationship Nakoryakov, V. E., Kashinsky, O. N., & Kozenko, B. K. (1984).
between bubble size distributions and chord-length distribution in Electrochemical method for measuring turbulent characteristics of
heterogeneously bubbling systems. Chemical Engineering Science, gasliquid ows. In J.M. Delhaye, & G. Cognet (Eds.), Measuring
53(6), 12671276. techniques in gasliquid two-phase ows (pp. 695 721). Berlin,
Lubbert, A. (1991). Characterization of bioreactors. In H.J. Rehm, G. Heidelberg, New York, Tokyo: Springer.
Reed, A. Puhler, & P. Stadler (Eds.), Biotechnology (2nd ed.). Vol. Nicol, R. S., & Davidson, J. F. (1988). Gas holdup in circulating bubble
4. (pp. 107148). Weinheim, New York: Verlag Chemie. columns. Transactions of the Institution of Chemical Engineers, A:
Luewisuthichat, W., Tsutsumi, A., & Yoshida, K. (1995). Fractal analysis Chemical Engineering Research and Design, 68, 152158.
of particle trajectories in three-phase systems. Transactions of the Northrup, M. A., Kulp, T. S., Angel, S. M., & Pinder, G. F. (1993). Direct
Institution of Chemical Engineers, A: Chemical Engineering Research measurement of interstitial velocity eld variations in a porous medium
and Design, 73, 222227. using uorescent-particle image velocimetry. Chemical Engineering
Luewisuthichat, W., Tsutsumi, A., & Yoshida, K. (1996). Deterministic Science, 48, 1321.
chaos analysis of particle dynamics in three-phase systems. Journal Pannek, S., Pauli, J., & Onken, U. (1994). Determination of local
of Chemical Engineering of Japan, 29, 675682. hydrodynamic parameters in bubble columns by the electrodiusion
Magaud, F. (1999). Contribution experimentale sur letude decoulements method with oxygen as a depolarizer. Journal of Applied
a bulles de type contacteur gazliquide. These de doctorat, Institut Electrochemistry, 24, 666669.
National Polytechnique de Lorraine, Nancy, France. Pant, H. J., Saroha, A. K., & Nigam, K. D. P. (2000). Measurement
Magaud, F., Souhar, M., Wild, G., & Boisson, N. (2001). Experimental of liquid holdup and axial dispersion in trickle bed reactors using
study of bubble columns hydrodynamics. Chemical Engineering radiotracer technique. Nukleonika, 45(4), 235241.
Science, 56, 45974607. Pauli, J., Menzel, T., & Onken, U. (1989). Diusional specic shear-rate
Mann, R., Dickin, F. J., Wang, M., Dyakowski, T., Williams, R. A., measurements in gasliquid two-phase ow. Chemical Engineering
Edwards, R. B., Forrest, A. E., & Holden, P. J. (1997). Application Technology, 12, 374378.
of electrical resistance tomography to interrogate mixing processes at Pauli, J., Onken, U., & Sobolk, V. (1991). Electrodiusional
plant scale. Chemical Engineering Science, 52, 20872097. direction-specic probe for measuring local velocity of aerated aqueous
Marcandelli, C., Lamine, A. S., Bernard, J. R., & Wild, G. (2000). systems. Journal of Applied Electrochemistry, 21, 10731076.
Liquid distribution in trickle-bed reactor. Oil and Gas Science and Peterson, D. A., Tankin, R. S., & Banko, S. G. (1984). Holographic
Technology, 55(4), 407415. measurements of bubble size and velocity in three-phase systems. In
Matsui, G. (1984). Identication of ow regimes in vertical gasliquid J.M. Delhaye & G. Cognet (Eds.), Measuring techniques in gasliquid
two-phase ow using dierential pressure uctuations. International two-phase ows (pp. 121). Berlin, Heidelberg, New York, Tokyo:
Journal of Multiphase Flow, 10, 711720. Springer.
Maucci, E., Briens, C. L., Martinuzzi, R. J., & Wild, G. (1999). Detection Prasser, H.-M., Bottger, A., & Zschau, J. (1998). A new
and characterization of piston ow regime in three-phase uidized electrode-mesh tomograph for gasliquid ows. Flow Measurement
beds. Powder Technology, 103, 243259. and Instrumentation, 9, 111119.
Mawlana, A., Huang, H., Wilhelm, A. M., & Delmas, H. (1992). Purwasasmita, M. (1985). Contribution a letude des reacteurs a lit xe
Transferable tracer analysis for multiphase reactor. Chemical fonctionnant a co-courant vers le bas a fortes vitesses du gaz et du
Engineering Communications, 115, 6782. liquide. These de doctorat, Institut National Polytechnique de Lorraine,
Menzel, T., In Der Weide, T., Staudacher, O., Wein, O., & Onken, U. Nancy, France.
(1990). Reynolds shear stress for modeling of bubble column reactors. Reese, J., Mudde, R. F., Lee, D. J., & Fan, L. -S. (1996). Analysis
Industrial and Engineering Chemistry Research, 29, 988994. of multiphase systems through particle image velocimetry. American
Mitchell, J. E., & Hanratty, T. J. (1966). A study of turbulence at a wall Institute of Chemical Engineers Symposium Series, 92(310), 161
using an electrochemical wall shearstress meter. Journal of Fluid 167.
Mechanics, 26, 199221. Reinecke, N. (1996). Kapazitive Tomographie fur transiente
Miyahara, T., Hamaguchi, M., Sukeda, Y., & Takahashi, T. (1986). Size Mehrphasenstromungen. Dissertation, Fakultat fur Maschinenwesen,
of bubbles and liquid circulation in a bubble column with a draught Universitat Hannover, Germany.
tube and a sieve plate. Canadian Journal of Chemical Engineering, Reinecke, N., & Mewes, D. (1997). Investigation of the two-phase
64, 718725. ow in trickle-bed reactors using capacitance tomography. Chemical
Moujaes, S. F. (1990). Testing a spherical dual-tipped optical ber probe Engineering Science, 52, 21112127.
for local measurements of void fraction and gas velocity in two-phase Reinecke, N., Petritsch, G., Schmitz, D., & Mewes, D. (1998).
ows. Canadian Journal of Chemical Engineering, 68, 504510. Tomographic measurement techniquesVisualisation of multiphase
Moujaes, S., & Dougall, R. S. (1987). Experimental investigation of ow. Chemical Engineering Technology, 21, 718.
cocurrent two-phase ow in a vertical rectangular channel. Canadian Reiss, L. P., & Hanratty, T. J. (1963). An experimental study of the
Journal of Chemical Engineering, 65, 705715. unsteady nature of the viscous sublayer. A.I.Ch.E. Journal, 9, 154
Mudde, R. F., Groen, J. S., & van den Akker, H. E. A. (1997). 160.
Liquid velocity eld in a bubble column: LDA experiments. Chemical Resch, F. J., & Leutheusser, H. J. (1972). Reynolds stress measurements
Engineering Science, 52, 42174224. in hydraulic jumps. Journal of Hydraulic Research, 10(4), 409430.
Mudde, R. F., Groen, J. S., & van den Akker, H. E. A. (1998). Application Resch, F. J., Leutheusser, H. J., & Alemu, S. (1974). Bubbly two-phase
of LDA to bubbly ows. Nuclear Engineering and Design, 184, 329 ow in hydraulic jump. Proceedings of the American society of civil
338. engineers, Journal of Hydraulic Division, 100(HY1), 137149.
Revankar, S. T., & Ishii, M. (1992). Local interfacial area measurement Souhar, M. (1982). Contribution a letude dynamique des e coulements
in bubbly ow. International Journal of Heat and Mass Transfer, diphasiques gazliquide en conduite verticale: cas des regimes a bulles
25, 913925. et a poches. These de doctorat detat, Institut National Polytechnique
Rode, S., Midoux, N., Lati, M. A., & Storck, A. (1994). Hydrodynamics de Lorraine, Nancy, France.
and liquidsolid mass transfer mechanisms in packed beds operating Souhar, M. (1987). Etude experimentale de la turbulence pres de la paroi
in cocurrent gasliquid down ow: An experimental study using en e coulement a bulles. International Journal of Heat Mass Transfer,
electrochemical shear rate sensors. Chemical Engineering Science, 49, 30, 18131823.
13831401. Storck, A., Lati, M. A., Barthole, G., Laurent, A., & Charpentier, J. C.
Roig, V., Larrieu, N., & Suzanne, C. (1995). Turbulent length scales (1986). Electrochemical study of liquidsolid mass transfer in packed
in bubbly mixing layer. International symposium on two-phase ow bed electrodes with upward and downward cocurrent gasliquid ow.
modelling and experimentation, Rome. Journal of Applied. Electrochemistry, 16, 947963.
Roig, V., Suzanne, C., & Masbernat, L. (1998). Experimental investigation Stravs, A. A., & von Stockar, U. (1985). Measurement of interfacial areas
of a turbulent bubbly mixing layer. International Journal of in gasliquid dispersions by ultrasonic pulse transmission. Chemical
Multiphase Flow, 24, 3554. Engineering Science, 40, 11691175.
Roizard, C., Poncin, S., Lapicque, F., Py, X., & Midoux, N. (1999). Stravs, A. A., Wahl, J., von Stokar, U., & Reilly, P. J. (1987).
Behavior of ne particles in the vicinity of a gas bubble in a stagnant Development of an ultrasonic pulse re ection method for measuring
and a moving uid. Chemical Engineering Science, 54, 23172323. relative size distributions of air bubbles in aqueous solutions. Chemical
Ruer, H. M., Broring, S., & Schugerl, K. (1995). Fluid dynamic Engineering Science, 42, 16771987.
characterization of airlift tower loop bioreactors with and without Svendsen, H. F., Luo, H. A., Hjarbo, K. W., & Jakobsen, H. A. (1998).
motionless mixtures. Bioprocess Engineering, 12, 119130. Experimental determination and modelling of bubble-size distributions
Sakai, N., Onozaki, M., Saegusa, H., Ishibashi, H., Hayashi, T., Kobayashi, in bubble columns. Chinese Journal of Chemical Engineering, 6(1),
M., Tachikawa, N., Ishikawa, I., & Morooka, S. (2000). Fluid dynamics 2941.
in coal liquefaction reactors using neutron absorption tracer technique. Thang, N. T., & Davis, M. R. (1979). The structure of bubbly ow
American Institute of Chemical Engineers Journal, 46, 16881693. through Venturis. International Journal of Multiphase Flow, 5, 1737.
Salesse, A., Larue de Tournemine, A., & Roig, V. (2001). Development Thimmapuram, P. R., Rao, N. S., & Saxena, S. C. (1992). Characterization
of bubble cluster detection and identication method. Fifth World of hydrodynamic regimes in a bubble column. Chemical Engineering
conference on experimental heat transfer, uid mechanics and Science, 47, 33553362.
Toral, H. (1981). A study of the hot-wire anemometer for measuring
thermodynamics, Thessaloniki, Greece, September 24 28, 2001.
void fraction in two-phase ow. Journal of Physics E: Science and
Samways, A. L., & Bruun, H. H. (1992). Vertical up- ow of oil/water
Instruments, 14, 822827.
mixtures. Technical Report No. 200, Department of Mechanical and
Toye, D. (1996). Etude de lecoulement ruisselant dans les lits xes par
Manufacturing Engineering, University of Bradford, UK.
tomographie a rayons X. These de Doctorat en Sciences Appliquees,
Saroha, A. K., & Nigam, K. D. P. (1996). Trickle bed reactors. Reviews
Universite de Liege, Belgium.
in Chemical Engineering, 12, 207347.
Uga, T. (1972). Determination of bubble-size distribution in a BWR.
Schmitz, D., & Mewes, D. (2000). Tomographic imaging of transient
Nuclear Engineering and Design, 22, 252261.
multiphase ow in bubble columns. Chemical Engineering Journal,
Uribe-Salas, A., Gomez, C. O., & Finch, J. A. (1994). A conductivity
77, 99104.
technique for gas and solids holdup determination in three-phase
Schugerl, K. (1982). Characterization and performance of single and
reactors. Chemical Engineering Science, 49, 110.
multistage tower reactors with outer loop for cell mass production.
Utiger, M., Stuber, F., Duquenne, A. M., Delmas, H., & Guy, C.
Advances in Biochemical Engineering, 22(93), 93117.
(1999). Local measurements for the study of external loop airlift
Schweitzer, J. M., Bayle, J., & Gauthier, T. (2001). Local gas hold-up hydrodynamics. Canadian Journal of Chemical Engineering, 77, 375
measurements in uidized bed and slurry bubble column. Chemical
382.
Engineering Science, 56, 11031110. Vial, C. (2000). Apport de la mecanique des uides a letude des
Sederman, A. J., Johns, M. L., Alexander, P., & Gladden, L. F. (1998). contacteurs gaz/liquide: Experience et simulation numerique. These de
Structure- ow correlations in packed beds. Chemical Engineering doctorat, Institut National Polytechnique de Lorraine, Nancy, France.
Science, 53, 21172128. Vial, C., Camarasa, E., Poncin, S., Wild, G., Midoux, N., & Bouillard,
Sederman, A. J., Johns, M. L., Bramley, A. S., Alexander, P., & Gladden, J. (2000). Study of hydrodynamic behaviour in bubble columns and
L. F. (1997). Magnetic resonance imaging of liquid ow and pore external loop airlift reactors through analysis of pressure uctuations.
structure within packed beds. Chemical Engineering Science, 52, Chemical Engineering Science, 55, 29572973.
22392250. Vial, C., Laine, R., Poncin, S., Midoux, N., & Wild, G. (2001a).
Serizawa, A., Kataoka, I., & Mishigoshi, I. (1975). Turbulence structure In uence of gas distribution and regime transition on liquid velocity
of airwater bubbly ow. International Journal of Multiphase Flow, and turbulence in a 3-D bubble column. Chemical Engineering Science,
2, 221259. 56, 10851093.
Shah, Y. T., Stiegel, G. J., & Sharma, M. M. (1978). Backmixing in gas Vial, C., Poncin, S., Wild, G., & Midoux, N. (2001b). A simple method
liquid reactors. American Institute of Chemical Engineers Journal, for regime identication and ow characterisation in bubble columns
24, 369400. and airlift reactors. Chemical Engineering and Processing, 40, 35151.
Sharma, S., Mantle, M. D., Gladden, L. F., & Winterbottom, J. M. Villermaux, J. (1995). Genie de la Reaction Chimique (2nd ed.).
(2001). Determination of bed voidage using water substitution and Technique et Documentation. Lavoisier, Paris.
3D magnetic resonance imaging, bed density and pressure drop in Villermaux, J., & van Swaaij, W. P. M. (1969). Modele representatif
packed-bed reactors. Chemical Engineering Science, 56, 587593. de la distribution des temps de sejour dans un reacteur semi-inni a
Shen, G., & Finch, J. A. (1996). Bubble swarm velocity in a column. dispersion axiale avec zones stagnantes. Application a lecoulement
Chemical Engineering Science, 51, 36653674. ruisselant dans des colonnes danneaux Raschig. Chemical Engineering
Shepp, L. A., & Logan, B. F. (1974). The Fourier reconstruction of a Science, 24, 10971111.
head section. IEEE Transcations on Nuclear Science, NS-21, 2143. Wallis, G. B. (1969). One-dimensional two-phase ow. New York:
Smith, G. B., Gamblin, B. R., & Newton, D. (1995). X-ray imaging McGraw-Hill.
of slurry bubble column reactors: The eects of systems pressure Wang, S. K., Lee, S. J., Jones, O. C., & Lahey, R. T. (1990). Statistical
and scale. Transactions of the Institution of Chemical Engineers, A: analysis of turbulent two-phase pipe ow. American Society of
Chemical Engineering Research and Design, 73, 632636. Mechanical EngineersJournal of Fluids Engineering, 112, 8995.
Warsito, M., Ohkawa, N., Kawata, S., & Uchida, S. (1999). with porous particles. Chemical Engineering Science, 45, 3311
Cross-sectional distributions of gas and solid holdups in slurry bubble 3317.
column investigated by ultrasonic computed tomography. Chemical Yao, B. P., Zheng, C., Gasche, H. E., & Hofmann, H. (1991). Bubble
Engineering Science, 54(21), 47114728. behaviour and ow structure of bubble columns. Chemical Engineering
Wild, G., & Poncin, S. (1996). Hydrodynamics of three-phase sparged and Processing, 29, 6575.
reactors. In K.D.P. Nigam, & A. Schumpe (Eds.), Three-phase Young, M. A., Carbonell, R. G., & Ollis, D. F. (1991). Airlift bioreactors:
sparged reactors (pp. 11113). New York: Gordon and Breach Science Analysis of local two-phase hydrodynamics. American Institute of
Publishers, Topics in Chemical Engineering Series. Chemical Engineers Journal, 37, 403428.
Wol, C. (1989). Untersuchungen mit Mehrkanal-Mikrosonden Yu, Y. H., & Kim, S. D. (1991). Bubble properties and local liquid
zum Ein u suspendierter Feststoe auf die Hydrodynamik velocity in the radial direction of cocurrent gasliquid ow. Chemical
in Blasensaulen-Reaktoren fur Chemie und Biotechnologie Engineering Science, 46, 313320.
VDI-Fortschrittsberichte Reihe 3 (Verfahrenstechnik) Nr 176. Zahradnk, J., Fialova, M., Ruzicka, M., Drahos, J., Kastanek, F., &
Dusseldorf, (Germany): VDI-Verlag. Thomas, N. H. (1997). Duality of the gasliquid ow regimes in
Wol, C., Briegleb, F. U., Bader, J., Hektor, K., & Hammer, H. (1990). bubble column reactors. Chemical Engineering Science, 52, 3811
Measurements with multi-point micropobes. Chemical Engineering 3826.
Technology, 13, 172184. Zuber, N., & Findlay, J. A. (1965). Average volumetric concentration
Yang, X. L., Euzen, J. P., & Wild, G. (1990). Residence time distribution in two-phase ow systems. Journal of Heat Transfer, 87, 453
of the liquid in gasliquid cocurrent up ow xed bed reactors 468.