Die Rendon CK 1998

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

734

Ind. Eng. Chem. Res. 1998, 37, 734-738

Loop Venturi ReactorsA Feasible Alternative to Stirred Tank


Reactors?
Laurent L. van Dierendonck*
RUG, Nijenborg, 9747 AG Groningen, The Netherlands

Jindr ich Zahradnk


Institute of Chemical Process Fundamentals, Academy of Sciences of the Czech Republic, 165 02 Prague 6,
Czech Republic

Va clav Linek
Department of Chemical Engineering, Prague Institute of Chemical Technology, 166 28 Prague 6,
Czech Republic

During the past decennia, the stirred tank reactor, especially for gas-liquid operations, has
received much attention. However, proper design of turbine-stirred gas-liquid reactors on an
industrial scale can still be difficult to make. On the large scale, the removal of heat may become
a limiting factor. Installation of additional cooling coils into the reactor vessel makes the design
problems even more complex. The development of loop Venturi designs offers some new solutions
to the scale-up questions, especially when high pressures are involved. Recent publications on
this type of reactor demonstrate fast mixing (including micromixing), high mass-transfer rate,
and an independently designed heat exchanger in the circulation loop of the reactor. The gas
loop ensuring complete gas mixing represents an additional favorable feature. On the whole,
the loop Venturi reactors can be viewed as an efficient alternative to the stirred tank reactors,
offering easier scale-up. This conclusion is supported by the less dependency of the mass-transfer
rate and mixing on the reactor scale.
Introduction
To design a gas-liquid reactor for commercial processes, the designer requires a lot of information about
the system involved. However, most of the information
can be collected on a small scale (bench or semi-technical
scale). In the selection process of a gas-liquid contactor
type, the stirred tank reactor with a standard turbine
stirrer (Rushton) is the most common choice (Figure 1).
This type of reactor possesses a wide application area
in chemical- and bio-process industry and, accordingly,
has been receiving extensive coverage in the literature,
including most notably classic textbooks by Uhl and
Gray (1966), Nagata (1975), and Oldshue (1983). A
thorough analysis of gas contacting with liquids in
stirred vessels has been presented by Joshi et al. (1982),
Mann (1983), and more recently by Tatterson (1991).
In spite of routine industrial use of stirred tank reactors,
their design and scale-up still pose many questions and
rely to a considerable extent on particular experience
and know-how of designers. In the industrial practice,
the prevailing approach is still scaling-up by empirical
testing with the actual process through a series of scaleup stages. Westerterp and co-workers (1963) performed
in the sixties comprehensive studies both on the physical and chemical behavior of the stirred tank reactors.
Many publications of the group contributed to grant a
citation price and improved understanding of the basic
design. Many investigators came up with their solutions for the scale-up. An astonishing number of papers
in the literature, dealing with the aspects of stirring and
the scale-up, confuses designers. There has been little
agreement about the approach to the scale-up among

Figure 1. Scheme of a standard geometry vessel with a Rushton


type turbine stirrer.

various authors. Essentially, reports on investigations


performed on a large scale have been missing. The
main question faced by the designer has been keeping
either the power input (Pg/VL) or the stirrer speed
constant. For the latter case, an effective stirrer speed
was defined as a criterion for scale-up (UL ) (N - N0)D2/T). The problem is complicated by the effect of
increasing superficial gas velocity (g/F ) Ug) with the
reactor scale. In this respect, flooding of the stirrer
occurs if Ug g 0.05 m s-1. Smith and Warmoeskerken
(1985) defined the operation window for the turbine
stirrer preventing gas overloading. It has been known
that the extent of liquid mixing decreases with scale
while validated information about gas phase mixing on
a large scale have been still missing.

S0888-5885(97)00313-8 CCC: $15.00 1998 American Chemical Society


Published on Web 01/01/1998

Ind. Eng. Chem. Res., Vol. 37, No. 3, 1998 735

Figure 2. Scheme of a loop Venturi reactor.

A general question in the whole field of gas-liquid


reactors concerns prediction of the effect of physical
properties of the gas and the liquid phases on the
behavior of gas-liquid dispersions and on the rate of
interfacial mass transfer. The industrial processes are
mostly characterized by undefined components in the
liquid streams which can influence interfacial area to
a large extent. Recently, Martin (1996) emphasized this
problem again in his thesis work and showed the
complexity of the mass-transfer prediction in relation
to the power input.
It can thus be concluded that scale-up of the stirred
gas-liquid contactors is still far from being a routine
task. The engineering people may look for alternatives.
In this respect, the jet loop reactors still have not been
sufficiently evaluated. The jet loop reactors of the
design illustrated in Figure 2 have been claimed to solve
some of the above-raised questions and make the scaleup easier. A characteristic feature of the jet loop
reactors is the requirement of a large pumping capacity
for liquid circulation, and therefore, the pump is a
critical item in the reactor design. It has, however, to
be pointed out that stirrers do pump equal amounts of
liquid volumes around. The relevant design parameters
of the jet loop reactors will be elucidated in more detail
later.
Aspects of Scaling-up the Stirred Gas-Liquid
Reactors
As stated above, the scale-up of a gas-liquid reactor
has to be based on data obtained in a down-scaled
version of the commercial unit. This postulate calls
immediately for a series of boundary conditions, namely
the following:
(1) A minimum scaleable size (T ) 0.15-0.2 m).
(2) A well-defined degree of mixing both in the gas
and the liquid phase.
(3) A well-defined micromixing zone for the fastreacting streams to mix with the bulk phase components.

(4) A mass- and heat transfer volume capacity which


should not vary much with the scale.
(5) A reliable shaft sealing at high pressures.
Beside the questions related to collecting required
information on the small scale, there is still a lot more
to know for reliable designing stirred gas-liquid reactors on the large scale. Crucial questions in this respect
include the following:
(1) The scale-up rules, namely keeping power input
constant, ask for an unambiguous relationship between
the power number (Np ) P/FLN3D5) and the so-called
gas flow number (NA ) g/ND3) (Martin, 1996), which
is not the case in the related literature. In addition, it
is well-known that there exists an extreme difference
in reported values of the local energy dissipation rate,
especially on large scales (up to a factor of 20). Indeed,
this raises the question of how reliable this approach
to the reactor scale-up is. Recently, Wichterle and
Sverak (1996) reviewed the scaling rule UL ) (N - N0)D2/T ) constant and declared this approach as physically well based.
(2) The volumetric mass-transfer coefficient is negatively influenced by the process of bubble coalescence
taking place at large distances from the stirrer. Accordingly, the overall kLa values will decrease with the
reactor scale.
(3) In general, the influence of physical properties on
the mass-transfer rate still cannot be accurately predicted.
(4) The insertion of coils for heating or cooling has a
tremendous impact on the reactor performance. Empirical relations for the mass-transfer rate do not
appropriately account for the influence of these internals
on the interfacial area.
(5) The gas phase is imperfectly mixed. Accordingly,
there is a need for validated axial dispersion coefficients.
In summary, it is thus possible to conclude that the
design of a turbine-stirred gas-liquid reactor is a
complex problem. Therefore, the performance of the jet
loop reactors will be examined in the next paragraph,
with respect to the above-raised questions. An attempt
will be made at demonstrating their favorable scale-up
features.
Ejector Loop Reactors
Working Characteristics. In ejector loop reactors
(ELR), all power input has been supplied for pumping
liquid through a Venturi-type ejector. According to the
ejector position, the reactors can be operated both in
the downflow (Figure 2) and upflow regime. Though,
indeed, these two reactor modifications exhibit some
different operating features; the following analysis of
ELR performance applies generally to both alternative
arrangements. A survey of different ELR modifications
for both semi-batch and continuous flow operations has
been given by Zahradnik and Rylek (1991). A principle
of gas dispersion in ejectors is demonstrated in Figure
3, showing a schematic chart of a typical Venturi-type
ejector. A detailed treatment of underlying theoretical
concepts can be found in the works of Witte (1969) and
Cunningham (1974a,b). The circulating liquid is forced
through an ejector nozzle where the liquid is accelerated
into a jet which due to its momentum entrains reaction
gas into the mixing tube. In the mixing tube, gas and
liquid are intensively mixed in the mixing shock zone
where the gas is finely dispersed as very small bubbles.
As the resulting gas-liquid stream leaves the ejector

736 Ind. Eng. Chem. Res., Vol. 37, No. 3, 1998

Figure 3. Principle of ejector performance.

and enters the reactor vessel, a secondary gas dispersion


of bubbles is obtained in the bulk fluid. Following the
pioneering works of Nagel and co-workers (1970, 1973),
who first proposed the application of ejectors for gas
dispersion in gas-liquid tower contactors, various aspects of ELR performance have been extensively studied
by numerous authors both from academia and from the
industry. Accordingly, the literature, reviewed for
example by Kastanek et al. (1993) or most recently by
Havelka et al. (1997), offers a considerable sum of
information and know-how, insufficiently acknowledged
by the potential users. The most widely known ELR
has been commercialized by Buss AG in Switzerland
and is known as the Buss loop reactor.
Due to their operating principle and construction
arrangement, the ELR exhibits numerous favorable
features regarding their process application as well as
design and scale-up. The mode of dispersion formation,
described above, provides high intensity interfacial
contact and, accordingly, a high rate of mass transfer
in the reactor. Cramers et al. (1992a) reported values
of the interfacial area from 40 000 to 70 000 m2 m-3 in
the ejector and from 500 to 2500 m2 m-3 in the whole
reactor (T ) 0.3 m, H ) 1.5 m), while maximum values
of kLa measured in the ejector reached 6 s-1 for the airwater system (Cramers et al., 1993). For the same
system and reactor size, Havelka (1997) obtained with
an optimized ejector configuration the values for kLa 7.5
and 0.2 s-1 for the ejector and total reactor system. The
application of ELR is, however, particularly advantageous for noncoalescing systems in which fine primary
gas dispersion has been preserved in the whole reactor
vessel. For such a system (0.3 mol L-1 of aqueous
solution of Na2SO4) Havelka (1997) obtained kLa values
of 10 and 1.5 s-1 for the ejector and reactor vessel in
agreement with Nardin (1995) who reported for the
Buss loop reactor overall kLa values up to 1.2 s-1 for
noncoalescing systems as compared with values 0.050.25 and 0.15-0.5 s-1 corresponding for such systems
to bubble columns and stirred tank reactors, respectively. The ELR are thus particularly suitable for fast

reactions in which the liquid phase mass transfer is the


reaction limiting step of the process.
The existence of gas and liquid circulation loops
provides perfect mixing in both phases. In addition, an
external heat exchanger can be suitably inserted into
the liquid circulation loop, eliminating thus the disadvantages of internal coils installation. Gas recirculation
ensures complete gas utilization. Accordingly, the ELR
can be operated at large values of gas throughput,
providing large intensity interfacial contact, without
losses of the active component or requirements for
installation of a circulation compressor. Complete gas
utilization eliminates problems of safety control on the
off-gas streams, and moreover, the gas circulation loop
circumvents the problem of the removal of undesired
volatile components from the gas phase. The liquid
circulation mode and high degree of macroscale turbulence in the reactor vessel provide favorable conditions
for catalyst suspension which may be one of the critical
issues in large scale stirred tank reactors. Further
advantages of ELR, as compared with aerated stirred
tanks, include the absence of moving parts, eliminating
the sealing problems and allowing easier operation at
elevated pressure. Cramers et al. (1992b) reported an
increase of the gas entrainment rate and gas holdup
with increasing pressure due to the favorable effect of
increasing gas density on these ELR characteristics.
Design and Scale-up. The intensity of mass transfer in ELR has been decisively determined by the rate
of energy dissipation in the ejector defined as PL or,
related to a unit of liquid mass in the reactor, PL/
VLFL, where P denotes ejector pressure drop, L liquid
flow rate, VL liquid volume in the reactor, and FL liquid
density. The secondary gas dispersion occurring at the
entrance of the gas-liquid stream from the ejector to
the bulk fluid in the reactor vessel, in connection with
the intensive gas and liquid mixing in the vessel,
ensures uniform radial and axial distribution of gas
bubbles over the entire reactor content (Zahradnik et
al., 1997). As a result, scale-up of ELR has been
considerably safer in comparison with the stirred tank
reactors, and the transfer of data from laboratory or
bench scale units to full-size reactors does not pose
serious problems. In general, ELR scale-up can thus
be based on a constant value of the specific rate of
energy dissipation, PL/VLFL, keeping constant decisive
geometrical characteristics of the ejector distributor,
namely the ratio of the mixing tube and nozzle diameters(d2/d1) and the ratios of the mixing tube and
diffuser lengths to the mixing tube diameter (L1/d2 and
L2/d2, respectively). For practical design purposes, these
geometrical parameters of Venturi-type ejectors should
be kept within the following limits, proved (Henzler,
1983; Zahradnik and Rylek, 1991) to ensure efficient
performance of such gas dispersing devices: d2/d1 )
1.5-4.5; L1/d2 ) 5-8; L2/d2 ) 8-12. Regarding typical
ratios of diffuser outlet to reactor cross-section area,
ejectors can be viewed as typical point (spaceconcentrated) distributors and their distributing efficiency is thus dependent on the reactor height to
diameter ratio (H/T). Obviously, higher H/T values
should be always preferred, at given reactor volume, to
achieve more uniform primary distribution of gas over
the whole reactor cross section. For H/T < 2, a multijet
arrangement has been recommended for full-size reactors and, similarly, such a reactor configuration should
be considered at vessel diameters above 1.5 or 3 m, for

Ind. Eng. Chem. Res., Vol. 37, No. 3, 1998 737

slow bioprocesses (e.g., aerobic fermentations). In this


latter case, the existence of gas and liquid circulation
loops can be advantageously employed for foaming
suppression (Kastanek et al., 1981; Zahradnik and
Rylek, 1991). A favorable comparison of the Buss loop
reactor with stirred reactors with respect to reaction
times, yields, and catalyst concentrations, presented for
selected types of reaction processes by Nardin (1995),
indicates wide potential applicability of ELR, including
notably the high-pressure operations. Obviously these
reactors deserve wider attention of designers and process engineers as a feasible efficient alternative to the
standard types of gas-liquid and gas-liquid-solid
reactors.
Conclusions

Figure 4. Time course of rape-seed oil hydrogenationscomparison


of productionsscale data from the stirred tank and ejector loop
reactors.

upflow or downflow ELR, respectively (Zahradnik and


Rylek, 1991). Alternatively, configurations with a
central draught tube can be advantageously employed
in large diameter units. Due to the principle of ELR
operation, proper selection of a circulating pump is of
crucial importance, regarding namely its flexible working characteristics, sealing system, and erosion resistance for gas-liquid-solid applications.
An illustrative example of a full-scale ejector design
from laboratory data has been presented, for example,
by Zahradnik and Rylek (1991), who reported on the
design of an upflow Venturi-type ejector distributor for
an industrial reactor (T ) 1.6 m, Vr ) 8 m3) for the rapeseed oil hydrogenation catalyzed by Ni on kieselguhr.
The ejector design was based on the results of hydrodynamic and mass-transfer measurements performed
in an ELR 0.3 m in diameter with the liquid volume
0.120 m3 (cold model data) and on the reaction
experiments carried out at real process conditions in a
laboratory scale unit 0.094 m in diameter with an
effective volume of 0.003 m3. The authors reported
superior performance of the industrial ELR over the
equal-size stirred autoclave with propeller agitator (D
) 0.55 m, N ) 3 s-1) used originally for the process,
regarding both the rate of the process (time of a singlebatch hydrogenation) or, alternatively, the catalyst load
corresponding to particular process requirements (reaction time and hydrogenation degree). Typical data from
the industrial units are presented in Figure 4 in the
form of time dependence of the iodine value (I.V.)
commonly employed as the characteristics of oils and
fatty acid hydrogenation degree (Patterson, 1983).
Application. Examples of mass-transfer-limited
processes representing a typical application area for the
ELR include various hydrogenation processes (hydrogenation of double and triple bonds, ring hydrogenation,
hydrogenation of aliphatic or aromatic nitrocompounds,
hydrogenation of aldehydes and ketones, etc.), amination, alkylation, carbonylation, chlorination, oxidation,
and dehydrogenation. Due to their superior gas utilization, the ELR can be, however, suitably used even for

(1) Despite many decades of intensive research, scaleup of a stirred gas-liquid reactor still poses considerable
problems.
(2) Conclusion 1 is supported by the many disagreements between researchers investigating stirred gasliquid reactors. The approach to their scale-up is based
on either power input correlations or proceeds via the
more physically justified route by keeping the effective
stirrer speed constant.
(3) The ejector-type loop reactors allow a simpler
scale-up approach and show better mass-transfer performance than the stirred reactors. The system is,
according to the literature, economically attractive at
pressures above 10 bar.
(4) The ejector loop reactors offer flexible design with
many additional favorable features discussed in this
paper.
(5) Upon conclusions 1-4, assessment of the ELR
potential should become a standard part of the process
of reactor selection for reactions in gas-liquid and gasliquid-solid systems.
Acknowledgment
The Czech co-authors (J.Z. and V.L.) gratefully acknowledge support given to the research of ejector loop
reactors by the Grant Agency of the Czech Republic
through Grant No. 104/97/1170.
Nomenclature
D ) impeller diameter, m
d1 ) nozzle diameter, m
d2 ) mixing tube diameter, m
F ) vessel cross section, m2
H ) reactor height, m
kLa ) volumetric liquid-side mass-transfer coefficient, s-1
L1 ) mixing tube length, m
L2 ) diffuser length, m
N ) impeller rotation speed, s-1
N0 ) minimum stirring speed ensuring gas dispersion, s-1
NA ) aeration number, NA ) g/ND3
Np ) power number, Np ) P/FLN3D5
P ) total power input, W
Pg ) shaft power input under gassing conditions, W
P ) ejector pressure drop, Pa
T ) vessel diameter, m
Ug ) superficial gas velocity, m s-1
UL ) characteristic liquid circulation velocity, m s-1
VL ) liquid volume in vessel, m3
Vr ) reactor volume, m3

738 Ind. Eng. Chem. Res., Vol. 37, No. 3, 1998


Greek Symbols
FL ) liquid density, kg
g ) gas volumetric feed rate, m3 s-1
L ) volumetric liquid flow rate through ejector, m3 s-1
m-3

Literature Cited
Cramers, P. H. M. R.; Beenackers, A. A. C. M.; van Dierendonck,
L. L. Hydrodynamics and mass transfer characteristics of a loopVenturi reactor with a downflow liquid jet ejector. Chem. Eng.
Sci. 1992a, 47, 3557-3564.
Cramers, P. H. M. R.; van Dierendonck, L. L.; Beenackers A. A.
C. M. Influence of the gas density on the gas entrainment rate
and gas holdup in loop-Venturi reactors. Chem. Eng. Sci. 1992b,
47, 2251-2256.
Cramers, P. H. M. R.; Smit, T.; Leuteritz, G. M.; van Dierendonck,
L. L.; Beenackers, A. A. C. M. Hydrodynamics and local mass
transfer characteristics of gas-liquid ejectors. Chem. Eng. J.
1993, 53, 67-73.
Cunningham, R. G. Gas compression with the liquid jet pump.
Trans. ASME: Ser. I 1974a, 96, 203-215.
Cunningham, R. G.; Dopkin, R. J. Jet breakup and mixing throat
lengths for the liquid jet gas pump. Trans. ASME: Ser. I 1974b,
96, 216-226.
Havelka, P. Mass transfer capacity of gas-liquid reactors with
ejector-type gas distributors. Ph.D. Thesis, Prague Institute of
Chemical Technology, 1997.
Havelka, P.; Linek, V.; Sinkule, J.; Zahradnik, J.; Fialova, M.
Effect of ejector configuration on the gas suction rate and gas
holdup in ejector loop reactors. Chem. Eng. Sci. 1997, 52, 17011713.
Henzler, H.-J. Design of ejectors for single-phase material systems.
Ger. Chem. Eng. 1983, 6, 292-300.
Joshi, J. B.; Pandit, A. B.; Sharma, M. M. Mechanically agitated
gas-liquid reactors. Chem. Eng. Sci. 1982, 37, 813-844.
Kastanek, F.; Zahradnik, J.; Rychtera, M.; Kratochvil, J.; Cermak,
J. Application of tower bioreactors with forced circulation in
biomass production. Collect. Czech. Chem. Commun. 1981, 46,
3232-3246.
Kastanek, F.; Zahradnik, J.; Kratochvil, J.; Cermak, J. Chemical
Reactors for Gas-Liquid Systems; Ellis Horwood: New York,
1993.
Mann, R. Gas-Liquid Contacting in Mixing Vessels; IChemE:
Rugby, 1983.
Martin, T. Gas Dispersion with Radial and Hydrofoil Impellers
in Fluids with Different Coalescence Characteristics; H. Utz
Verlag Wissenschaft: Munchen, 1996.
Nagata, S. Mixing: Principles and Applications; Wiley: New York,
1975.

Nagel, O.; Kurten, H.; Sinn, R. Strahldusenreaktoren Teil I: Die


Anwendung des Ejektorprinzips zur Verbesserung der Gasabsorption in Blasensaulen. (Jet Loop Reactors Part I: Application
of the Ejector Principle for Gas Absorption Enhancement in
Bubble Columns.) Chem.-Ing.-Tech. 1970, 42, 474-479.
Nagel, O.; Kurten, H.; Hegner, B. Die Stoffauataschflache in Gas/
Flussigkeits-Kontaktapparaten. Auswahlkriterien und Unterlagen zur Vergrosserung. (Interfacial Area in Gas/Liquid Contactors. Selection Criteria and Scale-up.) Chem.-Ing.-Tech. 1973,
45, 913-920.
Nardin, D. Trends and opportunities with modern Buss loop
reactor technology. Chemspec Europe 95 BACS Symposium,
1995.
Oldshue, J. Y. Fluid Mixing Technology; McGraw-Hill: New York,
1983.
Patterson, H. B. W. Hydrogenation of Fats and Oils; Applied
Science Publ.: London, 1983.
Smith, J. M.; Warmoeskerken, M. M. C. G. The dispersion of gases
in liquids with turbines. Proceedings of the 5th European
Conference on Mixing, Wurzburg, 1985, BHRA: Bedford, U.K.,
1985; Vol. 107, pp 115-126.
Tatterson, G. B. Fluid Mixing and Gas Dispersion in Agitated
Tanks; McGraw-Hill: New York, 1991.
Uhl, V. W.; Gray, J. B. Mixing: Theory and Practice; Academic
Press: New York, 1966, 1967; Vols I and II.
Westerterp, K. R.; van Dierendonck, L. L.; De Kraa, J. A.
Interfacial areas in agitated gas-liquid contactors. Chem. Eng.
Sci. 1963, 18, 157-176.
Wichterle, K.; Sverak, T. Surface aeration threshold in agitated
vessels. Collect. Czech. Chem. Commun. 1996, 61, 681-690.
Witte, J. H. Mixing shocks in two-phase flow. J. Fluid Mech. 1969,
36, 639-655.
Zahradnik, J.; Rylek, M. Design and scale-up of Venturi-tube gas
distributors for bubble column reactors. Collect. Czech. Chem.
Commun. 1991, 56, 619-634.
Zahradnik, J.; Fialova, M.; Linek, V.; Sinkule, J.; Reznickova, J.;
Kastanek, F. Dispersion efficiency of ejector-type gas distributors in different operating modes. Chem. Eng. Sci. 1997, in
press.

Received for review May 2, 1997


Revised manuscript received August 15, 1997
Accepted August 24, 1997X
IE970313B

X Abstract published in Advance ACS Abstracts, Xxxxxxxxx


YY, ZZZZ.

You might also like