Chemical Engineering Science 63 (2008) 3886 -- 3897

Contents lists available at ScienceDirect

Chemical Engineering Science

journal homepage: w w w . e l s e v i e r . c o m / l o c a t e / c e s

Design aspects of ejectors: Effects of suction chamber geometry

Randheer L. Yadav, Ashwin W. Patwardhan ∗
Department of Chemical Engineering, Institute of Chemical Technology, University of Mumbai, Mumbai 400 019, India

A R T I C L E I N F O A B S T R A C T

Article history: Ejector loop reactors are appealing devices for two fluids mixing. The high interfacial area produced in
Received 14 September 2007 such reactors gives them an advantage over the conventional contactors. Suction chamber where the two
Received in revised form 1 March 2008 streams of the primary and secondary fluid meet is an important part of the overall ejector geometry. The
Accepted 3 April 2008
present work deals with optimization of the geometry of the suction chamber using computational fluid
Available online 9 April 2008
dynamics (CFD). The effect of (i) projection ratio (LTN /DT ), i.e., ratio of the distance between the nozzle
Keywords:
tip and throat (LTN ) to the throat diameter (DT ), (ii) the diameter of the suction chamber (DS ) and (iii)
Ejector the angle of the converging section () on the entrainment rate of the secondary fluid have been studied.
Entrainment It was observed that for low values of projection ratio the entrainment rate was low. It increased with
Fluid mechanics an increase in the projection ratio; however, became constant beyond a particular value of this ratio. The
Hydrodynamics effect of DS on the rate of entrainment was observed to be more complex. The entrainment rate showed
Momentum transfer a maximum value when DS was varied over a wide range. The results obtained for different values of 
Multiphase flow suggest that the optimum lies in the range of 5◦ --15◦ . The results have been explained on the basis of
CFD
flow patterns produced.
© 2008 Elsevier Ltd. All rights reserved.

1. Introduction shock. Here a part of the kinetic energy of the flow is dissipated
in the shock creating the gas--liquid dispersion. The mixing shock
Ejectors are widely used in chemical process industries for results into generation of small bubbles and consequently creation
entraining and pumping corrosive fluids, slurries, fumes and dust- of high interfacial area (∼ 2000 m2 /m3 ). Ejectors thus, give supe-
laden gases, which are otherwise difficult to handle. They may also be rior gas--liquid mass transfer rates and higher rates of reaction as
used for mass transfer viz. liquid--liquid extraction, gas absorption, compared to conventional gas--liquid contact equipments like stirred
stripping, fermentation, hydrogenation, chlorination, etc. (Bhat et al., tanks, bubble columns, packed columns, etc. (Malone, 1980; Mandal
1972; Moresi et al., 1983; Mandal et al., 2005a). Their application et al., 2005b). Depending on its area of application, there could be
in vacuum production and as jet pump (Gamisans et al., 2004) is varying objectives for ejector design. It could be (a) To get large
well known. Fig. 1a shows the typical ejector system in which the entrainment of the secondary fluid, (b) To produce intense mixing
jet of primary fluid issuing out of a nozzle creates a low pressure between the two fluids, or (c) To pump fluids from a region of low
region around it. The pressure differential between the entry point pressure to a region of high pressure. The present work focuses on
of the secondary fluid and the nozzle tip provides the driving force optimizing the ejector geometry with an objective of entrainment of
for entrainment of the secondary fluid. Two principal flow regimes the secondary fluid.
in ejectors are coaxial-flow and froth-flow. The coaxial-flow consti- Key components of an ejector (Fig. 1a) are (a) Inlet for primary
tutes a central core of primary fluid with secondary fluid flowing in fluid, (b) Suction chamber, (c) Inlet for the secondary fluid, (d) Con-
the annular region formed between the jet of primary liquid and verging section of diffuser, (e) Throat of diffuser and (f) Diverging
ejector. Froth-flow regime is a cocurrent flow of fluids with one section of the diffuser. The geometry of ejector can be expressed in
phase completely dispersed in the other. Witte (1969) termed the terms of (a) Throat area ratio (AT /AN ), i.e., area of throat/area of noz-
phenomenon of change from coaxial-flow to froth-flow as mixing zle, (b) Throat aspect ratio (LT /DT ), i.e., length of throat/diameter of
throat, (c) Angle of converging and diverging sections of the diffuser,
(d) Projection ratio (PR) (LTN /DT ), i.e., distance between nozzle tip
and entry to throat and (e) Suction chamber area ratio (AS /AN ) =
∗Corresponding author. Tel.: +91 22 24145616; fax: +91 22 24145614. ((DS2 − DN2 )/D2 ). All these parameters are known to influence the
N
E-mail address: awp@udct.org (A.W. Patwardhan). performance of ejectors.

0009-2509/$ - see front matter © 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ces.2008.04.012
 R.L. Yadav, A.W. Patwardhan / Chemical Engineering Science 63 (2008) 3886 -- 3897 3887

observed from this table that most of the investigators investigated

Inlet for primary fluid (water) the effects of area ratio (AT /AN ), length of the throat (LT /DT ) and
Nozzle angle of divergence of the diffuser. However, very few of them have
Inlet for secondary fluid (air)
reported the effects of the angle of the converging section of the
Dinlet Suction chamber diffuser, diameter of the suction chamber and the PR. Thus, the in-
DS formation about the effects of these geometrical parameters on the
DN
Converging ejector performance needs to be investigated systematically.
section A number of investigators, e.g., Mellanby (1928) and Watson
Throat θ (1933), cf. Kroll (1947), Zahradnik et al. (1982), Henzler (1983) have
DT pointed out that the performance of a liquid jet ejector is a func-
tion of the geometry of the nozzle, size and shape of the chamber
where the gas is introduced (suction chamber) and geometry of the
diffuser (its outlet cross-section area, total length and cone angle).
Axis of symmetry
For a given value of throat length if PR is increased, jet may breakup
Diverging much earlier than at the end of the throat. Thus a large amount of
section energy will be lost. Similarly for small PRs the jet may not break in
the throat due to lower rates of momentum transfer between the
two fluids. Henzler (1983) supported the above observations and
Doutlet suggested that low values of PR result in poor pressure recovery in
Outlet the diffuser. Thus PR is an important design parameter. However it
can be observed that many investigators in Table 1 have not reported
the value of PR for their studies.
Velocity Pressure Among the investigators who have reported the value of PR very
inlet Inlet few have tried to find the optimum value of PR for their studies.
Location for calculating Optimum value of PR obtained by Davies et al. (1967) and Biswas
area available for flow of et al. (1975) was 1.9. Acharjee et al. (1975) studied a vertical liquid
LTN secondary fluid jet ejector with different liquids as primary fluid and air as the sec-
θ ondary fluid. For all the nozzles they observed that a PR 2.0 was
Projection ratio = L TN /D T optimum. Henzler (1983) compiled the experimental data of differ-

(( ) )
Area Ratio = DS2 − D 2N / D2N
ent investigators. The optimum PR suggested by him was in the range
of 0.4--0.9. It was also suggested that the optimum PR depends on
Axis Angle of Converging section = θ the geometry of the entrance to mixing tube (angle and height of
the converging section) and length of the mixing tube.
Kandakure et al. (2005) in a CFD study on jet ejectors predicted
the air entrainment rate of ejectors for different geometries of
diffuser, nozzle velocities, LT /DT ratios, and values of area ratio.
Pressure Their study involves the CFD of ejector geometries of Bhutada and
Outlet Pangarkar (1987). The PR assumed in their simulations was 3.0.
Suction chamber used by them was a cylindrical chamber (0.1 m
Fig. 1. (a) Schematic geometry of an ejector. (b) Geometry used for simulation along
in diameter). The converging section of the diffuser was connected
with boundary conditions.
to the bottom of this chamber. Angle of convergence was 12◦ . The
predicted air entrainment rate was the maximum for the ejector
with LT /DT = 0 and the area ratio of 4. The contour plots of axial
2. Previous work velocity of the secondary fluid showed presence of recirculation
loops in the suction chamber. The presence of such recirculation
The effects of different operating conditions such as nozzle veloc- reduces the net entrainment rate of the secondary fluid. This study
ity, pressure drop and ejector geometry parameters on the perfor- showed that the suction chamber should be designed to minimize
mance of ejectors have been experimentally investigated by several these recirculations.
researchers (Jackson, 1964; Davies et al., 1967; Bhat et al., 1972; Kroll (1947), Engdahl and Holton (1943, cf. Somsak, 2005),
Biswas and Mitra, 1981; Rylek and Zahradnik, 1984; Bhutada and Mellanby (1928), Watson (1933, cf. Somsak, 2005) and Henzler
Pangarkar, 1987; Bhutada, 1989; Havelka et al., 1997; Rusly et al., (1983) suggested that the converging section should be well
2005; Mandal et al., 2005a, b; Das and Biswas, 2006; Sriveerakul rounded, bell-mouthed to have good ejector performance. A conical
et al., 2007; Marynowski et al., 2007). These investigators observed or tapered entry with an angle greater than 20◦ was recommended.
that the optimum performance occurs when mixing of two streams This was to avoid the creation of objectionable shock and eddy
(primary fluid stream and the induced secondary fluid stream) was losses at the convergence inlet (Mellanby, 1928). Watson (1933,
located just upstream of the throat exit, i.e., at the diffuser entrance. cf. Somsak, 2005) did an experiment and stated 25◦ as the best
Witte (1969) proved experimentally that high entrainment ratios convergence angle.
were obtained by means of multi-orifice nozzles and a relatively long Area of the suction chamber is equally important for efficient per-
mixing throat. Whereas Cunningham and Dopkin (1974) studied jet formance of the ejector. However there is no study in the literature
breakup and mixing throat lengths for liquid jet gas pump. Bhutada that has tried to understand the role of suction chamber diameter
and Pangarkar (1987) observed experimentally that the entrainment on entrainment of the secondary fluid.
of the secondary fluid was more when straight portion of the throat The entrainment rate is a function of prevailing fluid dynamics in
is of zero length. the system which in turn is decided by its geometrical and operating
The dimensions of different components of ejectors studied by parameters. Since it is the suction chamber where the two streams
different investigators have been compiled in Table 1. It can be of the primary and secondary fluid meet, its geometry is bound to
3888 R.L. Yadav, A.W. Patwardhan / Chemical Engineering Science 63 (2008) 3886 -- 3897

Table 1
Geometrical parameters of ejectors used by different investigators

References Area ratio Entry to Angle of converging Throat aspect Angle of diverging Projection ratio
(DT /DN )2 the throat section (deg) ratio (LT /DT ) section (deg) (LTN /DT )

Kroll (1947) -- Well rounded 20--25 7 4--10 0.5--5

Davies et al. (1967) 22.6--247 Well rounded -- 7 10 1.9a
Bhat et al. (1972) 3.7--25.1 Conical 28 0 10 8.9
Cunningham and Dopkin (1974) 2.2--6.5 Well rounded -- 12.3--32.4 4.0 3.0a
Acharjee et al. (1975) 5.4--50.4 Well rounded -- 8 10 2.0a
Biswas et al. (1975) 1.4--12.8 Well rounded -- 6 7 1.9a
Zahradnik et al. (1982) 33.8--113.8 Conical -- 0 6.4 --
Henzler (1983) 2.0--25.5 Well rounded -- 7.5 5 0.4--0.9a
Moresi et al. (1983) 1.5--3.5 Conical 17.35 1.8 9.5 --
Ben Ebrahim et al. (1984) 4.0 Conical -- 3.5 2.0 --
Bhutada and Pangarkar (1987) 1.8--10.2 Conical or bell-shaped 12 0--16 5.0 3.0a
Mukherjee et al. (1988) 1--50.6 Well rounded -- 10.8 7 --
Kundu et al. (1994) 29.3--169.8 Conical -- 7 7 --
Kundu et al. (1995) 29.3--169.8 Well rounded -- 9.7 7 --
Havelka et al. (1997) 3.24 Conical -- 5--20 7 --
Fernandez (2001) -- -- -- 6 8 1.0
Cramers and Beenackers (2001) 2.1--9.0 Conical -- 2--10 3.0 --
Elgozali et al. (2002) -- Conical -- 10 7 2.72
Gamisans et al. (2004) -- Conical -- 0.5--1.3 -- --
Rusly et al. (2005) 2.5 Conical 10 5.0 7 5.0
Mandal et al. (2005a) 10.0 Well rounded -- 9.7 9.1 --
Mandal et al. (2005b) 5.6--14.4 Well rounded -- 9.7 9.1 --
Majumder et al. (2005) 7.4--22.5 Well rounded -- 9.6 7 --
Li and Christofides (2005) -- Conical 12 1.1 2 2.17
Das and Biswas (2006) 15.5--59.5 -- -- 7.76 8.6 --
Sriveerakul et al. (2007) 5.6--10.0 Well rounded -- 1--6 10 1.1--6.8
a Indicate the optimum value suggested by the investigator.

have an important effect on the rate of entrainment. However it can 52 354 and 210 042 respectively. Simulations were conducted using
be concluded from the above section that although a large number commercially available software FLUENT 6.2. Standard k-- turbu-
of experiments have been carried out very little has been done to lence model (per phase) was used to calculate the turbulence in
understand the underlying fluid mechanics in the suction chamber. both the phases. Water was taken as the primary fluid and air
Hence the present work was undertaken to understand the role of as the secondary fluid with drop diameter of 1 mm. Air was as-
PR (LTN /DT ), angle of converging section () and diameter of the sumed to obey the ideal gas law. Since the volumetric flowrate
suction chamber ((DS2 − DN 2 )/D2 ) on the rate of entrainment of the
N
of water through the nozzle was known, the velocity inlet was
secondary fluid in a jet of primary fluid. used as the boundary condition for flow of water. Air inlet was
The present work utilizes computational fluid dynamics (CFD) open to the atmosphere; hence the boundary condition for it was
modeling to address the above requirement. The work of a number pressure inlet with 0 Pa g (gauge pressure). The ejector outlet
of researchers (Fernandez, 2001; Szafran and Kmiec, 2004; Somsak, also was considered at 0 Pa g. Pa g is the gauge pressure in Pas-
2005; Kandakure et al., 2005; Rusly et al., 2005; Sriveerakul et al., cal. No-slip boundary condition was enforced at the walls of the
2007; Marynowski et al., 2007) have shown that CFD is a powerful ejector and the nozzle. The second order upwind discretization
tool for predicting the phenomena inside jet ejectors. In the present scheme was used for the momentum, turbulent kinetic energy
work a large number of simulations have been conducted for differ- and turbulent energy dissipation rate and SIMPLE scheme was
ent values of the projection ratio (LTN /DT ), area ratio ((DS2 − DN
2 )/D2 )
N
used for the pressure--velocity coupling. A relaxation factor of 0.2
and angle of the converging section (). This has helped in under- was used for the pressure and momentum, while a factor of 0.1
standing the relation between the different geometrical parameters was used for volume fraction, turbulent kinetic energy and tur-
of the suction chamber and their effect on entrainment of the sec- bulent energy dissipation rate. Solution was initialized by taking
ondary fluid. the ejector geometry completely filled with stagnant air. Initial
simulations were carried out with 1 mm mesh size. It was ob-
3. CFD modeling strategy served that the residues did not tend to zero indicating equations
did not converge. This could be attributed to the inability of this
There are two approaches for the numerical calculation of mul- mesh to capture the large velocity and pressure gradients espe-
tiphase flows: the Euler--Lagrange approach and the Euler--Euler cially near the nozzle tip as well as at the interface of the two
approach. In the latter approach, different phases are treated as fluids. The residues for the equations with mesh sizes of 0.5 and
interpenetrating continua and this approach has been adopted in 0.25 mm, showed low fluctuations and reduced to below 10−3 .
this work. The conservation equations have similar form for all Thus we were able to get converged solutions with 0.5 as well as
phases in this approach. There are two Euler--Euler multiphase 0.25 mm mesh sizes. The rate of entrainment for LTN /DT = 0 and
models: the mixture model and the Eulerian model. As a first step with 0.5 mm mesh size was 2.58 × 10−3 kg/s. Entrainment rate
in this study it was decided to compare the predictive ability of for the same geometrical configuration but with 0.25 mm mesh
Eulerian model with mixture model. Geometry of Bhutada and size was 2.59 × 10−3 kg/s. Thus the percentage difference be-
Pangarkar (1987) and their experimental results have been used tween the rates of entrainment for the two mesh sizes was 0.2%.
for validating the model. A 2D-axisymmetric geometry of ejector However there was large difference between the computational
with LT /DT = 4 and AT /AN = 4 was created. It was meshed with time requirements for the two mesh sizes. With 0.5 mm mesh,
hexahedral elements of uniform size. Three different meshes with about 16 h were required to obtain the converged solution on an
element length 1, 0.5 and 0.25 mm were tested. Total number of AMD opteron machine. The same solution with 0.25 mm mesh size
mesh elements corresponding to these mesh sizes were 13 342, took about 2 days. Thus it was noticed that the results obtained
 R.L. Yadav, A.W. Patwardhan / Chemical Engineering Science 63 (2008) 3886 -- 3897 3889

Table 2 In the mixture model the following momentum equation is solved

Effect of drop size on the rate of entrainment using Eulerian model
for the mixture.
Drop size (mm) Entrainment rate Pressure drop between nozzle
×103 (kg/s) tip and throat (Pa) j
( Vm ) + ∇.(m Vm Vm )
0.1 4.34 −773.8 jt m
1 4.27 −714.3 = −∇P + ∇.[m (∇Vm + ∇VmT )] +  g
5 4.12 −596.2 ⎛ ⎞ m
n

+ F + ∇. ⎝   V V ⎠
k k dr,k dr,k (1)
k=1
Table 3
Effect of different turbulent models on the rate of entrainment of secondary phase where, n is the number of phases, F the body force, m the viscosity

of the mixture, nk=1 k k , Vdr,k is the drift velocity, Vdr,k = Vk − Vm .
Turbulence models Entrainment Percentage
rate from deviation from Slip velocity between phase q (primary phase) and p (secondary
CFD × 103 (kg/s) experimental data phase) is given as,
Eulerian model with standard 3.67 19.9
k-- model Vpq = Vp − Vq (2)
ASMa with standard k-- model 3.09 6.15
ASM with RNG k-- model 3.02 3.97
ASM with RNG k-- model with 2.97 2.36 Mass fraction of any phase (k) is calculated as
differential viscosity approach
ASM with realizable k-- model 2.98 2.68  
Ck = k k (3)
ASM with standard k-- model 2.72 −6.62 m
Experimental entrainment rate = 0.00294 kg/s, Bhutada and Pangarkar, 1987.
a ASM: Hence drift velocity between any two phases can be calculated as
algebraic slip mixture model.
follows:

n

Vdr,p = Vpq − Ck Vqk (4)
for the two mesh sizes are practically the same; however, there k=1
is a large difference in the computational time. Therefore all the
simulations to study the effect of suction chamber geometry on Since in the present study only one secondary phase (air) is present,
the ejector performance were carried out with the mesh size of the above equation is simplified to
0.5 mm.
The effect of bubble size on the rate of entrainment using Eulerian Vdr,p = Vpq − Ck Vpq (5)
model is shown in Table 2. It can be observed from the table that the
rate of entrainment is practically independent of the bubble size. A FLUENT uses its in-built slip velocity formulation given by
change in the bubble size changes the momentum exchange between Manninen et al. (1996). Manninen et al. (1996) have suggested the
the two fluids. However increase in the momentum exchange also following equation for calculating the slip velocity.
leads to changes in the pressure profile. This in turn changes the

driving force for the entrainment of air. Overall, it can be seen that p p − m →
the entrainment rate is practically independent of the bubble size. Vpq = a (6)
fdrag p
In case of algebraic slip mixture model, the slip velocity has been
directly specified through a user-defined subroutine. Bubble size is
where p is the particle relaxation time and it is calculated as
not used in the calculation of the slip velocity and hence there is no
effect of bubble size for algebraic slip mixture model. Bubble size of
p dp2
1 mm has been used for simulations with Eulerian model. p = (7)
The entrainment rate obtained using Eulerian model with stan- 18q
dard k-- model for turbulence is shown in Table 3. The experimen-
tal entrainment rate under similar conditions (bib4Bhutada, 1989) fdrag = 1 + 0.15 Re0.687 for Re  1000
was 2.94 × 10−3 kg/s. Thus the Eulerian model overpredicts the 0.0183 for Re > 1000 (8)
rate of entrainment by 19.9%. Such a large discrepancy between the
experimental and simulation result can be attributed to the nature → jVm
a =g − (Vm .∇) Vm − (9)
of the flow in the ejector. Eulerian model assumes dispersion type jt
flow where the secondary fluid is distributed in the primary fluid.
However, in the present case ejector was arranged such that the wa- For turbulent flow,
ter jet was directed downwards. The entrained air flowed around
the water jet in the annular space between the water jet and the (p − m )dp2 → m
Vpq = a− ∇q (10)
ejector (Bhutada and Pangarkar, 1987). As a result, there was no 18q fdrag p D
bubble formation inside the ejector and both the fluids flowed co-
currently. Since there was no bubble; the virtual mass, which arises where m is the mixture turbulent viscosity and D the Prandtl dis-
from acceleration, and lift force, which arises from unbalanced pres- persion number.
sure forces around the bubble, do not exist. Due to this the Eulerian Thus it can be observed from Eq. (9) that Manninen et al. (1996)
model is not a suitable choice for studying the coaxial type of flows formulation is applicable for flows generated by motion of the dis-
in ejector. persed fluid itself (buoyancy driven flows). In the geometries simu-
It was decided therefore to use the mixture model (algebraic slip lated in the present study the flow was co-current down flow and
model) along with standard k-- turbulence model for CFD of coaxial- the water velocities were many times higher than those that would
flows. This approach required solving a slip velocity equation along be generated by buoyancy effects. Moreover, jetting occurred in a
with the momentum and continuity equations. major portion of the ejector (this was observed experimentally too,
3890 R.L. Yadav, A.W. Patwardhan / Chemical Engineering Science 63 (2008) 3886 -- 3897

Table 4 5
Effect of slip velocity on the air entrainment rate (VN = 21.2 m/s; DT = 0.016 m)

Slip (%) Entrainment rate from Percentage deviation from

CFD × 103 (kg/s) experimental data
4
0 3.82 22.93
10 3.22 8.64

Entrainment x 103 (kg/s)

12 3.04 3.42
14 2.79 −5.49
3
16 3.22 −7.71

Experimental entrainment rate = 0.00294 kg/s, Bhutada and Pangarkar, 1987.

2
Table 5
Geometry for studying the effect of the distance between nozzle tip and throat on
rate of entrainment

((DS2 − DN
2 )/D2 )
N 15 15 15 15 15 15 1
LTN /DT 0 2.5 4 5 10 14.5
 (deg) 10 10 10 10 10 10

0
Table 6 0 5 10 15 20
Geometry for studying the effect of the diameter of the suction chamber on rate of LTN/DT
entrainment

((DS2 − DN
2 )/D2 )
N 4.8 6.6 8 10.8 13.6 16.4 28.8 Fig. 2. Effect of projection ration (LTN /DT ) on the rate of entrainment of air.
LTN /DN 2.5 2.5 2.5 2.5 2.5 2.5 2.5
5 5 5 5 5 5 5
14.5 14.5 14.5 14.5 14.5 14.5 14.5
 (deg) 10 10 10 10 10 10 10 5

Table 7
Geometry for studying the effect of angle of the converging section on rate of
4
entrainment

((DS2 − DN
2 )/D2 ) 8 8 8 8 8 8 8 8 8
Entrainment x 103 (kg/s)

N
LTN /DN 5 5 5 5 5 5 5 5 5
 (deg) 2.5 5 10 15 20 30 45 60 90
3

Bhutada and Pangarkar, 1987). Therefore, in-built slip correlation

was not used for the present work. Instead the slip between the 2
fluids was taken as some percentage of local motive fluid velocity.
The basic idea was to modify Eq. (2) by using the following definition
of slip velocity instead of Manninen et al. (1996).
1
Vpq = 0.13 × Vq (11)

So that the secondary flow velocity will be (1 − 0.13 = 0.87) 0.87

0
times the primary phase velocity. 0 2 4 6 8
For the slip between the fluids, only the local axial velocities of
Area x 104 (m2)
water were considered. The slip between the fluids was implemented
in the form of a user-defined subroutine in the CFD software FLUENT Fig. 3. Effect of area available for flow on the rate of entrainment for different
by using Eq. (11). The solution was iterated until convergence was LTN /DT .
achieved, such that the residue for each equation fell below 10−3 . In
general, it was observed that the residue for the momentum equa-
tions was below 10−5 , that for the turbulent kinetic energy was well of the local water velocity, both, water and air, move at the same ve-
below 10−4 and that for the turbulent energy dissipation rate was locity, i.e., there is substantial momentum transfer from water to air.
well below 10−3 . The continuity equation residue was below 10−3 , This results in higher air velocities leading to the higher values of air
while the residue for the air volume fraction was well below 10−5 . entrainment rate. As the slip between the fluids is increased, the mo-
Table 4 shows the effect of slip between the fluids on the air mentum transfer from water to air decreases resulting in a decrease
entrainment rate in terms of percentage deviation from the experi- in the air entrainment rate. Table 4 shows that when the slip between
mental data for LT /DT = 4 and nozzle velocity of 21.2 m/s. The cor- the fluids is between 12% and 14% of the local water velocity, the
responding experimental data is 2.94 × 10−3 kg/s. The slip between mixture model together with standard k-- model predicts the exper-
the fluids was varied over a wide range of 0--16% of the local water imentally observed air entrainment rates very well. Hence, for the
velocity so as to get a good match between the predictions and the further CFD studies, the slip between the fluids in the axial direction
experimental observations. A two step procedure was followed to get for air was taken as 13% of the local water fluid axial velocity. This
the solution. First the flow, turbulence and volume fraction equations was kept constant for all the other simulations.
were solved without the slip velocity equation. This step is equivalent Kandakure et al. (2005) have presented the model validation for
to no-slip conditions at the water--air interface. In the second step the various liquid flowrates and different geometries of the diffuser sec-
algebraic slip was applied and all the equations including slip velocity tion. They have optimized the area ratio (area of throat to the area of
were solved simultaneously. When the slip between the fluids is 0% nozzle) for three very different types of ejectors with varying liquid
 R.L. Yadav, A.W. Patwardhan / Chemical Engineering Science 63 (2008) 3886 -- 3897 3891

flowrates. They observed that the slip velocity of 13% gives a good As can be observed, the RNG k-- model with differential viscosity
match between the experimental and predicted rate of entrainment approach gives the result which is much closer to the experimen-
for all the ejector geometries and liquid flowrates. Therefore the slip tal data than other models. Hence it was used for carrying out all
velocity of 13% is applicable for different ejector configurations and the further simulations to understand the effect of geometrical pa-
liquid flowrates. rameters of the suction chamber, i.e., PR (LTN /DT ), diameter of the
Table 3 shows the predicted rate of entrainment of the secondary suction chamber (DS ) and angle of the converging section () on en-
fluid and its variation from the experimental entrainment rate of trainment of the secondary fluid. All the studies for optimizing the
(2.94×10−3 kg/s) for different turbulence models. It can be observed suction chamber geometry were carried out for optimum ejector
that the deviation between the rate of entrainment predicted from (LT /DT = 0 and AT /AN = 4) suggested by Kandakure et al. (2005). In
the CFD simulations and that reported by Bhutada and Pangarkar all the simulated cases the nozzle and throat diameters have been
(1987) is less than 10% for all the turbulence models when used kept constant at 8 and 16 mm, respectively. The diverging section
with mixture model approach. Hence any of the models can be used after the throat expands from 0.016 at the throat end to 50 mm on
for predicting the effect of suction chamber geometry on the ejector the other end over a length of 0.2 m giving an angle of divergence as
performance. It can be observed from Table 3 that the rate of en- 4.85◦ . Fig. 1b shows the geometry and boundary conditions used for
trainment predicted with RNG k-- model with differential viscosity simulation. Tables 5--7 show the details of the geometries used for
approach shows the least deviation from the reported value by simulations.
Bhutada and Pangarkar (1987). Therefore this model was used in the The objective of the present work was to understand the effects
present study for predicting the effect of suction chamber geometry of various geometry parameters of the ejector from a fundamental
on the ejector performance. view point. The effects of PR, area ratio and angle of the converging
section have therefore been varied one at a time. Global optimization
Table 8
can be attempted when the effects of various geometry parameters
Area available for flow of secondary fluid calculated between the nozzle outlet and on the performance (entrainment rate) have been understood and
suction chamber for different projection ratios can be modeled from first principles.
LTN /DT Area (m2 ) × 104
4. Results and discussion
0 1.51
2.5 6.61
4 7.54 4.1. Effect of LTN /DT
5 7.54
10 7.54
14.5 7.54
The effect of LTN /DT , i.e., PR on the rate of entrainment is shown
in Fig. 2. It shows that the rate of entrainment of the secondary

L TN /D T Contours of radial velocity (m/s) of air Contours of axial velocity (m/s) of air
-5 0 5 -5 0 25

2.5

10

14.5

1.0 0.5 0.0

Fig. 4. (a) Contours of radial and axial velocities for different projection ratios (LTN /DT ). (b) Volume fraction of air in the ejector for LTN /DT = 5.
3892 R.L. Yadav, A.W. Patwardhan / Chemical Engineering Science 63 (2008) 3886 -- 3897

2000 5

0
4
0 0.1 0.2 0.3 0.4
-2000

Entrainment x103 (kg/s)

Gauge Pressure (Pa)

-4000 3

-6000
2

-8000

1
-10000

-12000
Axial distance (m) 0
0 5 10 15 20 25 30
Fig. 5. Pressure profile along the axis of ejector for different values of projection Area Ratio
ratios ------ PR-0; - - - - PR-2.5; -- · -- · PR-5; ------ PR14.5.
Fig. 7. Effect of area ratio ((DS2 − DN
2 )/D2 ) on the rate of entrainment for different
N
values of projection ratios:  LTN /DT = 2.5;  LTN /DT = 5;  LTN /DT = 14.5.

25

5
20

4
Efficiency (%)

15
Entrainment rate x103(kg/s)

3
10

5 2

0 1
0 5 10 15 20
LTN/DT

Fig. 6. Effect of projection ratio (LTN /DT ) on energy efficiency. 0

0 2 4 6 8 10 12 14 16
2)x104
Area (m
fluid is low for small values of LTN /DT . With increase in LTN /DT the
Fig. 8. Effect of area available on the rate on entrainment of the secondary fluid for
entrainment rate increases and beyond LTN /DT > 5 the entrainment different values of the projection ratios:  LTN /DT =2.5;  LTN /DT =5;  LTN /DT =14.5.
rate is practically the same. Entrainment of the secondary fluid in
the jet of the primary fluid is governed by three factors: (a) Area
available for flow of the secondary fluid, (b) the pressure lost by
the secondary fluid due to generation of radial flow in the suction large amount of radial flow for LTN /DT = 0. Generation of radial flow
chamber, and (c) the pressure driving force between the nozzle tip leads to large pressure loss for the secondary fluid. This leads to a
and the throat point. reduction in the rate of entrainment. It can be observed from Fig. 4
Fig. 3 shows the effect of area available for air flow on the rate that increase in LTN /DT reduces the generation of radial flow and
of entrainment. The areas available for different values of LTN /DT therefore resulting in an increase in the rate of entrainment. However
are tabulated in Table 8. It can be observed that for LTN /DT = 0 the beyond LTN /DT > 5, negligible amount of radial flow is generated
area available for air flow is low which results in a lower rate of en- and hence the rate of entrainment remains constant for different
trainment. With increase in the PR the area available for flow of air LTN /DT > 5.
increased causing a corresponding increase in the rate of entrain- Fig. 4(b) shows the contour plot of the volume fraction of air
ment. Beyond LTN /DT = 5, the area available remains constant and along the ejector length. It can be observed from this figure that
therefore the entrainment rate levels off. two phases flowing through the ejector flow as separated flow.
Fig. 4(a) shows the contour plots of radial and axial velocities Thus there is negligible dispersion of air in water. Similar profiles
of air for different values of the PRs. It shows the presence of a were observed for all the parameters studied in the present work.
 R.L. Yadav, A.W. Patwardhan / Chemical Engineering Science 63 (2008) 3886 -- 3897 3893

Therefore the flow of the phases in the ejector was in the coaxial- This gives the energy efficiency as
flow regime.
(Power)S
Fig. 5 depicts the pressure profile along the axis of the ejector % energy = × 100 (14)
for different values of the PRs. It can be seen that for low values (Power)P
of LTN /DT the pressure driving force between the nozzle tip and Fig. 6 shows the effect of PR on energy efficiency of ejectors. It can
the throat point is low. This results in a lower rate of entrainment. be seen that the energy efficiency of ejectors is affected by the PR. It
The pressure driving force was observed to increase with increase is low for small values of LTN /DT then increases with an increase in
in LTN /DT resulting in corresponding increase in the rate of entrain- LTN /DT ratio and finally attains a constant value. Hence knowledge
ment. It can be observed from Fig. 5 that the pressure profiles for of the optimum PR helps in design of efficient ejectors. Further sim-
LTN /DT > 5 are practically the same. This explains the constancy in ulations to understand the role of DS were carried out with the PRs
the rate of entrainment beyond LTN /DT of 5. of 2.5, 5 and 14.5.
Energy efficiency of the ejector is calculated as the fraction of the
power of the primary fluid imparted to the secondary fluid (Witte,
1969; Henzler, 1983 and Bhutada and Pangarkar, 1987). 4.2. Effect of diameter of suction chamber (DS )
Power of the fluid coming out of the nozzle is given as
The effect of diameter of the suction chamber on the rate of
2 V3 entrainment was studied in terms of the area available for air flow
(Power)P = P DN (12)
8 N to area available for water flow, i.e., ((DS2 − DN
2 )/D2 ). Location for
N
calculating the area available for air flow is shown in Fig. 1(b). This
The power imparted to the secondary fluid was calculated as is defined as a dimensionless parameter ((DS2 − DN 2 )/D2 ). It can be
N
observed from this figure that the area available for air flow varies
(Power)S = (Poutlet − Pthroat ) QS (13)
with changes in the geometrical parameters of the suction chamber.
The above location was chosen because the contact between the two
fluids starts from nozzle tip onwards.
Table 9
For the geometries simulated the DS was varied from 0.0192
Area available for flow of secondary fluid calculated between the nozzle outlet and to 0.0432 m. For these values of DS the parameter ((DS2 − DN 2 )/D2 )
N
suction chamber for different area ratios at varying projection ratios varied from 4.8 to 28.8. Fig. 7 shows the effect of ((DS2 − DN
2 )/D2 ) on
N
((DS2 2 )/D2 )
− DN Area (m2 ) × 104 the rate of entrainment for PRs of 2.5, 5 and 14.5. It can be observed
N
from the figure that low value of area ratio resulted in lower rate of
LTN /DT = 2.5 LTN /DT = 5 LTN /DT = 14.5
entrainment. The entrainment rate increased with an increase in area
4.8 2.39 2.39 2.39
ratio and attained a maximum value. The maximum entrainment
6.6 4.02 4.02 4.02
8 5.31 5.31 5.31 rate for PR 2.5 was obtained for area ratio of 8, whereas for PRs of 5
10.8 6.61 6.76 6.76 and 14.5 it was obtained for area ratio of 6.6. Further increase in the
13.6 6.61 8.36 8.36 area ratio resulted in a decrease in the rate of entrainment. However
16.4 6.61 11.08 11.08
28.8 6.61 14.15 14.15 the entrainment rate remained constant for area ratio beyond 16.4.
It can also be observed from Fig. 7 that the entrainment rates for

AR Contours of radial velocity (m/s) of air Contours of axial velocity (m/s) of air AR Contours of radial velocity (m/s) of air Contours of axial velocity (m/s) of air
-5 0 5 -5 0 25
-5 0 5 -5 0 25

4.8 4.8

6.6 6.6

8 8

10.8 10.8

13.6 13.6

16.4 16.4

28.8
28.8

Fig. 9. (a) Contours of radial and axial velocity profile for different values of area ratios and LTN /DT = 2.5. (b) Contours of radial and axial velocities of air for different
((DS2 − DN
2 )/D2 ) for projection ratio = 5.
N
3894 R.L. Yadav, A.W. Patwardhan / Chemical Engineering Science 63 (2008) 3886 -- 3897

30
10000

25
5000

20
Gauge pressure (Pa)

% Efficiency
0
0.2 0.25 0.3 0.35 0.4 0.45
15

-5000
10

-10000
5

-15000 0
Axial distance (m) 0 10 20 30 40
Area ratio
8000
Fig. 11. Effect of area ratio ((DS2 − DN
2 )/D2 ) on efficiency of ejectors for different
N
6000 values of projection ratios:  LTN /DT = 2.5;  LTN /DT = 5;  LTN /DT = 14.5.

4000
4.6
2000
Gauge pressure (Pa)

0 4.4
0.15 0.25 0.35 0.45
-2000
4.2
Entrainment x103 (kg/s)

-4000
4.0
-6000

-8000 3.8

-10000
3.6
-12000
Axial distance (m) 3.4
Fig. 10. Pressure profile along the axis of ejector for different values of ((DS2 −DN
2 )/D2 )
N
for projection ratio: (a) 2.5 and (b) 5. ------ AR = 4.8; - - - - - - - AR = 6.6; 3.2
-- · -- · -- · --AR = 8; -- · · -- · · -- AR = 10.8. 0 20 40 60 80 100
Angle (degrees)

Fig. 12. Effect of angle of converging section () on rate of entrainment.

LTN /DT = 5 and 14.5 are practically the same. The observed nature
of the rate of entrainment in Fig. 7 can be explained in terms of the
area available for flow of air, the amount of radial flow generated
in the suction chamber and the pressure driving force between the for areas corresponding to area ratios 8 and 6.6 for LTN /DT = 2.5 and
nozzle tip and throat point. LTN /DT = 5 and 14.5, respectively. However, further increase in the
Fig. 8 shows the effect of area available for airflow on the rate of area (due to increase in the area ratio) leads to reduction in the rate
entrainment for PRs of 2.5, 5 and 14.5. The area available for different of entrainment. It can be observed that the reduction in the rate of
values of the PR and area ratios are shown in Table 9. It shows that entrainment is very sharp for LTN /DT =2.5 as compared to LTN /DT =5
the available flow area increased with increase in the area ratio for all and 14.5. This can be explained in terms of generation of radial flow
the PRs except 2.5. For LTN /DT = 2.5 the flow area remained constant in the system.
beyond the area ratio of 10.8. This was because of the geometrical Fig. 9a and b show the contours of the radial and axial velocities
construction of the ejector for LTN /DT = 2.5. For LTN /DT = 2.5 and of air for PR of 2.5 and 5, respectively. The contours for PR 14.5
area ratios  10.8 the beginning of the converging section was found are not shown because they are very similar to the contours for
to be behind the nozzle tip (refer to Fig. 9a). Hence for a given angle the PR of 5. It can be observed from Fig. 9a and b that for low
( = 10◦ in this case), eventhough the area ratio increased the area values of area ratio the air axial velocity is large. This creates a
available between the nozzle tip and the suction chamber remained large pressure loss for air and hence the entrainment is low. As
constant. It can be observed from Fig. 8 that for lower values of the the area ratio is increased from this low value, the axial velocity of
area ratios, the rate of entrainment increased with increase in area the air reduces, causing lower pressure loss and thereby increasing
for all the PRs. The rate of entrainment attained the maximum value the entrainment rate (Fig. 7). However for larger area ratios the
 R.L. Yadav, A.W. Patwardhan / Chemical Engineering Science 63 (2008) 3886 -- 3897 3895

Angle Contours of radial velocity (m/s) of air Contours of axial velocity (m/s) of air
(o)
-5 0 5 -5 0 25

2.5

10

15

20

30

45

60

90

Fig. 13. Effect of different values of angle of converging section () on the rate of entrainments.

8000 30

6000
25
4000

2000
Gauge pressure (Pa)

20
0
% Efficiency

0.2 0.3 0.4

-2000 15
-4000

-6000 10

-8000
5
-10000

-12000 0
Axial distance (m) 0 20 40 60 80 100
Angle (degrees)
Fig. 14. Pressure profile along the axis of the ejector for different values of  for
projection ratio = 5. ------  = 2.5◦ ; - - - - - -  = 10◦ ; ------  = 30◦ ; -- . . -- . .  = 90◦ .
Fig. 15. Effect of angle of converging section () on efficiency of ejectors.

radial flow was observed to be present in a significant portion of the that beyond the area ratio 16.4, the radial as well as axial velocities
suction chamber and this leads to a reduction in the entrainment are lower and hence there is negligible effect of the presence of the
rate for higher area ratios. It can be observed from Fig. 9a and b wall of the suction chamber on the flow profile. Hence the rate of
3896 R.L. Yadav, A.W. Patwardhan / Chemical Engineering Science 63 (2008) 3886 -- 3897

entrainment was observed to remain practically constant beyond results in decrease in the rate of entrainment. The pressure driving
the area ratio of 16.4. The sharp decrease in the rate of entrainment force  = 90◦ is low and hence the corresponding entrainment is also
for PR of 2.5 as observed in Figs. 7 and 8 can be explained with low. Fig. 15 shows the effect of  on energy efficiency of the ejectors.
the help of Fig. 9a. It can be observed from Fig. 9a that for area It can be seen that highest efficiency is obtained at  = 10◦ and large
ratios  10.8 the nozzle tip is located within the converging section. values of  result in poor energy efficiency.
Thus the axial as well as radial velocities are larger. This leads to
larger pressure losses leading to a rapid reduction in the entrainment 5. Conclusions
rate. The pressure profiles along the axis of the ejector for different
values of the area ratio and PR of 2.5 and 5 are shown in Fig. 10a A large number of CFD simulations have been carried to under-
and b, respectively. It can be observed that an area ratio of 4.8 is stand the effects of the geometrical parameters of the suction cham-
not efficient because it gives very low pressure driving force which ber on entrainment of the secondary fluid. The three geometrical
results in a smaller rate of entrainment (Fig. 7). With increase in the parameters investigated in this study are the PR, diameter of the suc-
area ratio the driving force was observed to increase and thereby tion chamber and angle of the converging section. It was observed
resulting in corresponding increase in the rate of entrainment. The that the geometry of the suction chamber has significant effect on
largest driving force was obtained for area ratio of 8 and 6.6 for PR the rate of entrainment of the secondary fluid. Based on the range
of 2.5 and 5, respectively. Hence these area ratios were observed of the three parameters studied following recommendations can be
to give the largest rate of entrainment for corresponding PRs. Thus made for coaxial type of flows in ejectors: A PR of 5 produces the
the entrainment of the secondary fluid is a strong function of the highest rate of entrainment of the secondary fluid. Increasing the PR
pressure drop. For LTN /DT = 2.5 and ((DS2 − DN 2 )/D2 ) = 10.8 a sudden
N beyond 5 does not lead to any significant improvement in the suction
reduction in the driving force was observed. This sudden reduction capacity of the jet. A suction chamber with ((DS2 − DN
2 )/D2 ) = 6.6 was
N
in the driving force could be because of generation of large amount observed to give the largest entrainment. A larger suction chamber
of radial flow for this arrangement (Fig. 9a) as discussed above. This leads to a reduction in suction capacity. It is recommended to have
causes a corresponding decrease in the rate of entrainment. Such the angle of converging section in the range of 5◦ --15◦ .
reduction in the pressure drop was not observed for PR=5. Therefore
this results in a gradual leveling off of the entrainment for LTN /DT =5
Notations
and higher area ratios.
The effect of ((DS2 − DN2 )/D2 ) on energy efficiency of the ejectors
N
for different PRs is shown in Fig. 11. It shows that the maximum AN area of nozzle, m2
ejector efficiency is obtained for ((DS2 − DN2 )/D2 ) = 6.6 and for ((D2 −
N
AT area of throat, m2
S →
2 )/D2 ) > 13.6 it remains constant. Efficiency of the ejector with PR
DN N
a acceleration, m/s2
2.5 was observed to decrease sharply beyond ((DS2 − DN
2 )/D2 ) = 8. Ck mass fraction of phase k
N
This was because of the reduction in the rate of entrainment on Dinlet diameter at water inlet, m,
account of generation of radial flow. Hence an ejector geometry with DN diameter of nozzle, m
((DS2 − DN
2 )/D2 ) = 6.6 is the most suitable since it efficiently uses the Doutlet diameter at mixture outlet, m
N
energy for entrainment of the secondary fluid. Further simulations DS diameter of suction chamber, m
for understanding the effect of angle of converging section () on the DT diameter of throat, m
rate of entrainment of the secondary fluid were done with ejector ((DS2 − DN
2 )/D2 )
N
ratio of area available for air flow to area avail-
geometry of ((DS2 − DN 2 )/D2 ) = 6.6 and L /D = 5.
N TN T
able for water flow
dp bubble diameter, m
4.3. Effect of angle of converging section () F body force, N/m3
fdrag drag force, N
The effect of angle of the converging section of the suction cham- g acceleration due to gravity 9.81 m/s2
ber () on the rate of entrainment is shown in Fig. 12. It can be seen HT height of throat, m
that the rate of entrainment is low for  = 2.5◦ . It increases with in- LTN distance between throat and nozzle tip, m
crease in  and attains a maximum value for =10◦ . Further increase LT length of throat (m)
in  results in a decrease in the rate of entrainment of the secondary n no. of phases
fluid. The effect of  on the rate of entrainment can be explained in (Power)P power of primary fluid, W
terms of the amount of radial flow generated in the suction cham- (Power)S power of secondary fluid, W
ber and the pressure driving force between the nozzle tip and the Poutlet absolute pressure at diffuser outlet, Pa
throat point. Pthroat absolute pressure at throat, Pa
Fig. 13 shows the contour plots of the radial and axial velocities ∇P pressure drop, N/m2
of air for different values of . It can be observed from this figure that p secondary phase (air)
with increase in the angle of converging section there is an increase QS flowrate of secondary fluid, m3 /s
in the amount of radial flow. Largest radial flows were observed for q primary phase (water)
 = 90◦ . This results in a higher form drag which increases the pres- Re Reynolds number
sure drop for the secondary fluid. The higher pressure drop for the Vm velocity of the mixture, m/s
secondary fluid reduces the entrainment rate. Thus the entrainment Vdr,k drift velocity for phase k, m/s
reduces with increase in angle. For maximum entrainment the angle Vpq slip velocity between phases p and q, m/s
of converging section should be between 5◦ --15◦ . Vp velocity of the secondary phase, m/s
Fig. 14 shows the pressure profile along the axis of the ejector for Vq velocity of the primary phase, m/s
different angles studied. To avoid cluttering on the graph pressure
Greek letters
profile for  = 2.5◦ , 10◦ , 30◦ and 90◦ have been shown. It shows that
the largest pressure driving force is generated for  = 10◦ and it k volume fraction of phase k
results in the highest entrainment for this case. With increase in  energy energy efficiency of ejector as defined by
beyond 10◦ the pressure driving force was observed to reduce and it Eq. (13)
 R.L. Yadav, A.W. Patwardhan / Chemical Engineering Science 63 (2008) 3886 -- 3897 3897

 angle of the converging section, degrees Kandakure, M.T., Gaikar, V.G., Patwardhan, A.W., 2005. Hydrodynamic projections
of ejectors. Chemical Engineering Science 60, 6390--6402.
m viscosity of the mixture, kg/ms Kroll, A.E., 1947. The design of jet pumps. Chemical Engineering Progress 1 (2),
m density of the mixture, kg/m3 Kundu, G., Mukherjee, D., Mitra, A.K., 1994. Gas entrainment and depth of penetration
p density of the primary fluid, kg/m3 in a cocurrent gas liquid downflow bubble column. Journal of Chemical
Engineering of Japan 27 (5), 621--626.
D Prandtl dispersion number Kundu, G., Mukherjee, D., Mitra, A.K., 1995. Experimental studies on a co-current
p particle relaxation time, s gas--liquid downflow bubble column. International Journal of Multifluid Flow
m mixture turbulent diffusivity, m2 /s 21 (5), 893--906.
Li, M., Christofides, P., 2005. Multiscale modeling and analysis of an industrial HVOF
thermal spray process. Chemical Engineering Science 60, 3649--3669.
References Majumder, S.K., Kundu, G., Mukherjee, D., 2005. Mixing mechanism in a modified
co-current downflow bubble column. Chemical Engineering Journal 112, 45--55.
Acharjee, D.K., Bhat, P.A., Mitra, A.K., Roy, A.N., 1975. Studies on momentum transfer Malone, R.J., 1980. Loop reactor technology, improves catalytic hydrogenation.
in vertical liquid jet ejectors. Indian Journal of Technology 13, 205--210. Chemical Engineering Progress 78, 53--59.
Ben Ebrahim, A., Prevost, M., Bugarel, R., 1984. Momentum transfer in a vertical down Mandal, A., Kundu, G., Mukherjee, D., 2005a. Comparative study of two fluid
flow liquid jet ejector: case of self aspiration and emulsion flow. International gas--liquid flow in the ejector induced upflow and downflow bubble column.
Journal of Multiphase Flow 10 (1), 79--94. International Journal of Chemical Reactor Engineering 3 Article A13.
Bhat, P.A., Mitra, A.K., Roy, A.N., 1972. Momentum transfer in a horizontal liquid jet Mandal, A., Kundu, G., Mukherjee, D., 2005b. Energy analysis and air entrainment in
ejector. Canadian Journal of Chemical Engineering 50, 313--317. an ejector induced downflow bubble column with non-Newtonian motive fluid.
Bhutada, S.R., 1989. Design of liquid jet loop reactors, Ph.D. (Tech.) Thesis, University Chemical Engineering Technology 28 (2), 210--218.
of Mumbai. Manninen, M., Taivassalo, V., Kallio, S., 1996. On the mixture model for multiphase
Bhutada, S.R., Pangarkar, V.G., 1987. Gas induction and hold-up characteristics of flow. VTT Publications 288, Technical Research Centre of Finland.
liquid jet loop reactors. Chemical Engineering Communications 61, 239--261. Marynowski, T., Desevaux, P., Mercadier, Y., 2007. CFD characterization of geometric
Biswas, M.N., Mitra, A.K., 1981. Momentum transfer in horizontal multi-jet liquid--gas parameters impact on ejector performance. In: Annals of the Assembly for
ejector. Canadian Journal of Chemical Engineering 59, 634--637. International Heat Transfer Conference, vol. 13.
Biswas, M.N., Mitra, A.K., Roy, A.N., 1975. Studies on gas dispersion in a horizontal Mellanby, A.L., 1928. Some phenomenon associated with the flow of steam through
liquid jet ejector. In: Second Symposium on Jet Pumps and Ejectors and Gas Lift nozzles. Transactions of Institution of Chemical Engineers 6, 66--84.
Techniques Cambridge, England, March 24--26, BHRA, E3-27-42. Moresi, M., Gianturco, G.B., Sebastini, E., 1983. The ejector loop fermenter: description
Cramers, P.H.M.R., Beenackers, A.A.C.M., 2001. Influence of the ejector configuration, and performance of the apparatus. Biotechnology and Bioengineering 25,
scale and gas density on the mass transfer characteristics of gas--liquid ejectors. 2889--2904.
Chemical Engineering Journal 82, 131--141. Mukherjee, D., Biswas, M.N., Mitra, A.K., 1988. Hydrodynamics of liquid--liquid
Cunningham, R.G., Dopkin, R.J., 1974. Jet break-up and mixing throat lengths dispersion in ejectors and vertical two fluid flow. Canadian Journal of Chemical
for liquid jet gas pump. Transactions of ASME, Journal of Fluid Engineering, Engineering 66, 896--907.
216--226. Rusly, E., Aye, L., Charters, W.W.S., Ooi, A., 2005. CFD analysis of ejector in a combined
Das, S.K., Biswas, M.N., 2006. Studies on ejector-venturi fume scrubber. Chemical ejector cooling system. International Journal of Refrigeration 28, 1092--1101.
Engineering Journal 119, 153--160. Rylek, M., Zahradnik, J., 1984. Design of venturi---tube gas distributors for
Davies, G.S., Mitra, A.K., Roy, A.N., 1967. Momentum transfer studies in ejectors. bubble type reactors. Collection of Czechoslovak Chemical Communications 49,
Industrial and Engineering Chemistry Process Design and Development 6, 1939--1948.
293--299. Somsak, W., 2005, Optimization of a high-efficiency jet ejector by computational
Elgozali, A., LInek, V., Fialova, M., Wein, O., Zahradnik, J., 2002. Influence of viscosity fluid dynamics software. Thesis for Master of Science, Texas A&M University.
and surface tension on performance of gas--liquid contactors with ejector type Sriveerakul, T., Aphornratana, S., Chunnanond, K., 2007. Performance prediction of
gas distributor. Chemical Engineering Science 57, 2987--2994. steam ejector using computational fluid dynamics: part 1. Validation of the CFD
Engdahl, R.B., Holton, W.C., 1943. Overfire air jets. In: Transactions of the ASME, results. International Journal of Thermal Sciences 46, 812--822.
vol. 65, No. 10, pp. 741--754. Szafran, R.G., Kmiec, A., 2004. Application of CFD modelling technique in engineering
Fernandez, R., 2001. Calculation of injector characteristics using a commercial calculations of three-fluid flow hydrodynamics in a jet-loop reactor. International
CFD code. Thesis for Diploma. The Institute of Thermal Turbomachines and Journal of Chemical Reactor Engineering 2, A30.
Powerplants, University of Technology, Vienna. Watson, F.R.B., 1933. The production of a vacuum in an air tank by means of a
Gamisans, X., Sarra, M., Lafuente, J., 2004. Fluid flow and pumping efficiency in an steam jet. Proceedings of the Institution of Mechanical Engineers 124, 231--238.
ejector venture scrubber. Chemical Engineering and Processing 43, 127--136. Witte, J.H., 1969. Mixing shock in two fluid flow. Journal of Fluid Mechanics 36,
Havelka, P., Linek, V., Sinkule, J., Zahradnik, J., Fialova, M., 1997. Effect of the ejector 639--655.
configuration on the gas suction rate and gas hold-up in ejector loop reactors. Zahradnik, J., Kratochvil, J., Kastanek, F., Rylek, M., 1982. Hydrodynamic
Chemical Engineering Science 52, 1701--1713. characteristics of gas--liquid beds in contactors with ejector type gas distributors.
Henzler, H.J., 1983. Design of ejectors for single-fluid material systems. German Collection of Czechoslovak Chemical Communications 47, 1939--1949.
Chemical Engineering 6, 292--300.
Jackson, M.L., 1964. Aeration in Bernoulli types of devices. American Institution of
Chemical Engineers Journal 10, 836--842.