Download as pdf or txt
Download as pdf or txt
You are on page 1of 18

Engineering Applications of Computational Fluid

Mechanics

ISSN: 1994-2060 (Print) 1997-003X (Online) Journal homepage: https://www.tandfonline.com/loi/tcfm20

CFD Modelling and Simulation of Jet Mixed Tanks

Kailas L. Wasewar & J. Vijay Sarathi

To cite this article: Kailas L. Wasewar & J. Vijay Sarathi (2008) CFD Modelling and Simulation of
Jet Mixed Tanks, Engineering Applications of Computational Fluid Mechanics, 2:2, 155-171, DOI:
10.1080/19942060.2008.11015218

To link to this article: https://doi.org/10.1080/19942060.2008.11015218

Copyright 2008 Taylor and Francis Group


LLC

Published online: 19 Nov 2014.

Submit your article to this journal

Article views: 3653

View related articles

Citing articles: 5 View citing articles

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=tcfm20
Engineering Applications of Computational Fluid Mechanics Vol. 2, No. 2, pp. 155–171 (2008)

CFD MODELLING AND SIMULATION OF JET MIXED TANKS


Kailas L. Wasewar* and J. Vijay Sarathi**

* Department of Chemical Engineering, Indian Institute of Technology (IIT), Roorkee – 247667 India
E-Mail: klw73fch@iitr.ernet.in (Corresponding Author)
** GE Oil and Gas/EACoE, GE India Technology Centre Pvt. Ltd., John F. Welch Technology Centre, EPIP,
#122, Phase III, Hoodi Village, Whitefield Road, Bangalore – 560066 India

ABSTRACT: Mixing is one of the common unit operations employed in chemical industries. Conventional mixers are
equipped with impellers but are expensive for mixing in large storage tanks and underground tanks. Jet mixers have
become an alternative to impellers for over 50 years in the process industry. For the design of jet mixers, much
experimental work has been done and many correlations have been proposed. However, these correlations are case
specific and are not valid for any generic mixer. In order to establish a generic model for jet mixers, Computational
Fluid Dynamics (CFD) is employed to understand the mixing phenomena and to find out the time required for mixing in
a jet mixed tank. The 3D simulations are carried out using the CFD package FLUENT 6.2 to generate data and validate
them against experimental results. The turbulence model engaged is the standard k-ε model, which is found to show
good agreement between the experimental results and simulated results for Reynolds number of 10000 and above. A
parametric sensitivity analysis is done by changing various parameters to understand how the mixing phenomenon is
affected. The results obtained are explained, and conclusions are drawn. The results obtained give a good understanding
of the mixing process in jet mixed tanks.
Keywords: jet mixing, CFD, flow pattern, mixing time, parametric sensitivity

1. INTRODUCTION also sterilization and maintenance in biochemical


processes and jet mixers are preferable in such
Mixing is one of the common unit operations situations.
employed in chemical industries. It is used for Apart from uses in conventional blending of liquids,
blending of liquids, homogenization of mixtures, homogenization of mixtures, heat transfer
heat transfer operation, mass transfer operation, operations and mass transfer operations such as
prevention of deposition of solid particles, etc. extraction, etc., jet mixers are used for the
Impellers are the conventional devices used for following applications as well: blending the
mixing purpose in industries. But they are very inhibitor into the monomer storage tank to stop
expensive for large storage tanks and underground violent runaway exothermic polymerization
tanks. Jet mixers have become an alternative to reactions (Hoffman, 1996; Mewes and Renz, 1991),
impellers for over 50 years in the process industry. emergency cooling systems of chemical reactors in
In jet mixing, part of the liquid from the tank is case of breakdown in operation (Schmetzek, 1995),
circulated into the tank at high velocities with the biochemical applications (Simon and Fonade,
help of pumps through nozzles. The resulting jet of 1993), and fast competitive consecutive reactions
fluid entrains some of the surrounding fluid and having a mixing sensitive product distribution
creates a circulatory pattern, which leads to mixing (Baldyga, 1994). Jet mixers are more appropriate
in the tank. Jet mixers have several advantages over for mixing processes involving chemically sensitive
conventional impellers. It has no moving parts as in liquids, e.g., preparation of foodstuffs (Lane, 1981)
conventional agitators, thereby reducing and acids mixing (Harnby et al., 1985).
maintenance costs, and it is easy to install when A large number of experimental studies on jet
compared with impellers. Agitators require support mixing in tanks have been carried out. Most of
at the top of the tank, implying a pre-requisite for these studies ended with empirical correlation that
thicker walls of stronger materials. Mechanical relates mixing time to operating conditions and
agitators show disadvantages at an industrial scale geometry. A large number of such correlations are
with regard to investment and energy costs, and available; hence there always exists a dilemma, to

Received: 19 Feb. 2007; Revised: 30 Oct. 2007; Accepted: 31 Oct. 2007

155
Engineering Applications of Computational Fluid Mechanics Vol. 2, No. 2 (2008)

select the proper correlation for calculating the geometric parameters in jet mixing in tanks using
mixing time. CFD can resolve these issues as it can horizontal, inclined and vertical jets. Harnby,
predict the mixing pattern in the tank. Hence, it is Edwards and Nienow (1985) suggested the same.
desirable to study a CFD modelling of jet mixed Yianneskis (1991) investigated the effect of tracer
tanks using commercial package FLUENT and test injection time. He found that the position of the
its validity using experimental results. probe and of the tracer injection point did not have
a significant effect on the final mixing times.
2. PREVIOUS WORK Grenville and Tilton (1996) showed that the mixing
time in jet mixed tanks are governed by the energy
2.1 Experimental studies dissipation rate in a region far from the jet nozzle,
Fossett and Prosser (1949) were the pioneers who where the velocities and turbulent intensities are
first introduced the idea of liquid jet mixers. They much lower. The same study was extended
used an inclined side-entry jet in a flat-base (Grenville and Tilton, 1997). Simon and Fonade
cylindrical tank. This jet was inclined at 45° to the (1993) employed steady and unsteady jets and it
tank base. The liquid depth in the tank was two was concluded that unsteady jets were more energy
third of the tank diameter. Fossett (1951) has found efficient than steady jets. Orfaniotis et al. (1996)
that the jet mixing could occur in a time much studied the effect of viscosity on mixing time in jet
shorter than what was usually taken by mixed tanks. It was found that mixing time
conventional mixing devices. Fox and Gex (1956) increases with liquid viscosity. It was reported that
carried out the investigation for both laminar and alternating jets gives 15% less mixing time at the
turbulent regimes. A comparative study was carried same power input as compared with steady jets.
out using jet mixed tank and propeller mixed tank. Perona et al. (1998) studied the performance of jet
It showed that the most important parameter was mixers in long horizontal tanks.
the momentum flux added to the tank and it also Considerable literature and measurements on jet
correlated the experimental data based on this flux. mixing in tanks are available. Previous efforts
Van de Vusse (1959) and Okita and Oyama (1963) pertaining to the experimental studies of the jet
adopted the correlation of Fossett and Prosser mixing in tanks along with mixing time correlations
(1949). Coldrey (1978) used a bottom-side-entering are summarized in Table 1.
jet inclined at 45° in a flat-base cylindrical tank The available literature on experimental studies and
with the liquid height equal to the tank diameter. correlations has limitation in accounting for jet
Lehrer (1981) formulated a model for a free mixing phenomena. The basic limitation of
turbulent jet of miscible fluids of different densities correlations presented in Table 1 is that they only
in which lateral transfer of momentum was predict well for the range of parameters covered in
considered to be a result of eddy diffusion. It was the study. The correlations of Fossett and Prosser
assumed that the eddy viscosity was the product of (1949), Hiby and Modigell (1978), and Grenville,
the nozzle Reynolds number and molecular Mak and Ruszkowski (1992) did not consider the
viscosity. Lane (1981) studied the jet mixing in liquid height as a parameter. This would imply that
tanks and supported the design of Coldrey (1978). mixing time would be the same in all the tanks that
They suggested that the greater jet length allows the have the same diameter, jet velocity and jet
jet to have a much larger entrainment capacity and diameter irrespective of the liquid height. This is
hence a greater mixing ability. Lane and Rice (1981 obviously not true. There is uncertainty in defining
and 1982) did a comparative assessment of various the jet length. In correlation, normally it is taken as
designs of jet mixed tanks. They concluded that a the length between the jet entry and the point where
hemispherical bottom significantly reduces the the jet axis intersects the liquid surface or opposite
mixing time. Maruyama, Ban and Mizushina (1982) wall. However, in tall tanks, the jet may lose most
found that the mixing time in jet mixed tank of its momentum by the time it reaches the liquid
depended on the liquid depth, nozzle height and surface and the length of the jet may be
nozzle angle. They supported the concept of the overestimated for this case. There is very little
greatest jet length for effective liquid jet agitation in information on the velocity and turbulence fields
a flat-base cylindrical tank. Maruyama, Kamishima and the mixing patterns within the tanks. Effects of
and Mizushina (1986) studied the effect of several the physicochemical properties like density and
viscosity of liquid are also not fully known. Various

156
Engineering Applications of Computational Fluid Mechanics Vol. 2, No. 2 (2008)

correlations are available for a mixing time, but measured and it is not clear if the measurement at a
they are case specific and have only a limited range single point indicates uniform mixing in the entire
of applicability. Many researchers do not tell the vessel.
location of the point where the concentration is

Table 1 Experimental correlations of jet mixing in tanks.

Authors Type of study Geometry Dimensions Correlation Remarks

Fossett and To determine the Inclined D = 1.524 m D2 For large pulse. Did not relate
Prosser minimum velocity of side entry H = 0.9144 m Tm = 9 mixing time to height of liquid.
(1949) the TEL jet required jet and U jd j So not reliable when height
dj = 1.9 mm
to avoid the cylindrical d0 = 2.54 cm changes or when it is different
stratification of the tank θ = 40° from their experiment.
fuel above the Re = 4500–80000
TEL/fuel mixture.

Fossett Modified Inclined D = 1.524 m For large and small pulses. Did
D2
(1951) correlation. side entry H = 0.9144 m Tm = C P not relate mixing time to height
jet and dj = 1.9 mm U jd j of liquid. So not reliable when
cylindrical d0 = 2.54 cm C P = 9 when Tinj > Tm / 2 height changes or when it is
tank θ = 40° C P = 4.5 when Tinj < Tm / 2 different from their experiment.

Fox and
Gex (1956)
For laminar and
turbulent regimes.
Side entry
jet and
D = 0.15–4.27 m
H = 0.15–4.27 m Tm = f
(H 0.5
D ) Did not specify the exact
criteria for degree of mixing as
Correlation based on cylindrical dj = 0.159–3.81 cm (U j d j ) 4
6g 6
1
per Lane and Rice (1982). So
momentum. tank Uj = 0.6–11 m/s accurate results may not be
f = 95.638 Re −0.146 obtained.

Van de Extended the study Inclined -- D2 Tank height was not included in
Vusse of Fox and Gex side entry Tm = 3.68 correlation.
(1959) (1956) for turbulent jet and U jd j
region. cylindrical
tank
Okita and 24% average Inclined -- D1.5 H 0.5 More reliable than the works of
Oyama difference in the side entry Tm = 2.6 Fossett and Prosser (1949),
(1963) turbulent regime for jet and U jd j Okita and Oyama (1963), and
data of Fox and Gex cylindrical Fox and Gex (1956).
(1956), extended tank
same.
Coldrey Proposed a modified Inclined -- D2H A longer jet length produces a
(1978) design for the side entry Tm = 4.507 more effective mixing jet and
inclined side entry jet and Ld jU j therefore reduces mixing time.
jet mixing, utilizing cylindrical Basing this theory on the
the longest possible tank assumption that mixing time is
jet length. inversely dependent on the
amount of liquid entrained by
the jet.
Hiby and Tested mixing in Axial jet -- D2 Mixing time is dependent on a
Modigell flat-base tank. and Tm = T * jet Reynolds number when the
(1978) cylindrical U jd j tank Re is less than 106. Height
tank Re t > 1,000 ,000 T * = 2.3 of tank not included in
Re t < 1,000 ,000 T * ∝ Re correlation.
Lehrer Formulated a model Axial jet Free turbulent jet 0.658 ⎛ ρc ⎞ 5 Developed a model and
Tm = ⎜ ⎟ 8 d 0.25
(1981) for free turbulent jet compared results with that of
U j ⎜⎝ ρd ⎟ j
of miscible fluids of ⎠ Fox and Gex (1956).
3
different densities in ⎛ Uj ⎞ 4

which lateral transfer ×⎜ ⎟


⎜ NjA⎟
(− log(1 − c )) ∗

of axial momentum ⎝ ⎠
occurs by eddy
diffusion.

157
Engineering Applications of Computational Fluid Mechanics Vol. 2, No. 2 (2008)

Authors Type of study Geometry Dimensions Correlation Remarks

Lane and Investigated a Axial jets For axial jets H 0.5 D 0.75 Formulated a correlation for
Rice (1981) vertical jet mixer for and D = 0.31–0.57 m Tm = C1 mixing time showing that the
Re (U j d j )0.5 g 0.25
1.3
a cylindrical vessel cylindrical H/D = 0.5–3.0 m mixing time is strongly
with a hemispherical tank Re = 250–60000 Re < 1800 dependent on the jet Reynolds
base. H 0.5 D 0.75 number in the laminar jet
Tm = C2 0.15 regime and only slightly
Re (U j d j )0.5 g 0.25
dependent in the turbulent jet
Re > 1800 regime.
Recommended design height is
not accurate.
Lane and
Rice (1982)
Extended their
studies and did a
Side entry
jets and
For side entry jets
D = 0.31–0.57 m Tm = f
(H D) 0.5 That their design with the
hemispherical bottom
comparative cylindrical H/D = 0.9–1.10 m (U jd j ) 0.667g0.166 significantly reduced the
assessment of tank Re = 250–60000 mixing time.
designs of Fossett f = 113.133 Re −0.146 Recommended design height is
and Prosser (1949), turbulent not accurate.
Coldrey (1978),
Hiby and Modigell
(1978), and Lane
and Rice (1981).
Maruyama, Investigated the Side entry D = 56,104 cm ⎛ Tm ⎞⎛⎜ L ⎞⎟ The mixing time depended on
Ban and effect of several jets and H = 84, 125 cm ⎜⎜ ⎟⎟ = 2.5 − 8.0 the liquid depth, nozzle height
Mizushina geometric cylindrical hi, ho = 4, 14, 24, ⎝ t r ⎠⎜⎝ d j ⎟⎠ and nozzle elevation angle
(1982) parameters in jet tank 44, 74, 94 cm Re > 30000 Made recommendations for
mixing in tanks (D = 104 cm) optimum nozzle depth in
using horizontal, hi, ho = 4.38, 20.5, circulation flow regime.
inclined and vertical 48.5 cm,
jets. (D = 56 cm)
dj = 0.5, 1, 1.8 cm
θ = 7, 15, 30, 45,
54, 60, 73
Grenville, Series of mixing Side entry -- Height of tank was not
D2
Mak and time experiments in jets and Tm = 5.78 included.
Ruszkowski order to determine as cylindrical U jd j
(1992) to which correlation tank
form of Fossett and
Prosser (1949) or
Fox and Gex (1956)
would be more
reliable for jet mixer
design.
Simon and Steady and unsteady Two jets at D, H = 490 mm Whatever the jet configuration
M = Tm (gH ) DJ s 3 ≈ 1
0.5 2
Fonade jets. H/2 & H/3, dj =10 mm used, the mixing time could be
(1993) horizontally J 2 evaluated through a constant
Js = , J = ρAU j
located ρU j g mixing time factor M, close to
1, which includes jet and tank
characteristics. Alternating jets
appear to be more efficient than
steady jets.
Orfaniotis Unsteady and steady Two jets at D, H = 500 mm ⎡T ⎤ Effect of viscosity on mixing
et al. (1996) jets in order to H/2 & H/3, dj =9, 15 mm M = ⎢ m ⎥ (J s )0.41 = 11.3 time was studied. Variation
⎣ tr ⎦
determine the effects horizontally J with jet momentum becomes
of jet position and located JS = more pronounced at high
ρU j g
liquid viscosity. D viscosities. Alternating jets
J = ρAU j 2 tr = gives more efficient mixing
(gH )0.5 (15% less mixing time) in terms
of mixing time than the steady
jets.

158
Engineering Applications of Computational Fluid Mechanics Vol. 2, No. 2 (2008)

Authors Type of study Geometry Dimensions Correlation Remarks


Grenville Re-analysed the data Cylindrical D = 0.61–36 m For a given volume, an
L2
and Tilton of Grenville et al. tank H/D =0.2–1.0 Tm = 3 optimum geometry exists for a
(1996) (1992), using new dj = 5.8–50 mm U jd j jet mixed tank, which allows a
approach that the Uj =2.2–24.8 m desired mixing time to be
mixing rate and the achieved for minimum power
resulting mixing input when the aspect ratio of
time in a jet mixed the vessel, H/D = 0.7071.
tank are not
governed by the
turbulent energy
dissipation rate (ε) at
the jet or the mean
power input per unit
mass but by ε in a
region far from the
jet nozzle where the
velocities and
turbulence intensity
are much lower.
Grenville Re-analysed their Same as Same as Grenville The attachment to the vessel
D2H
and Tilton previous data. Grenville et al. (1996) Tm = k base by jets with shallow angles
(1997) et al. U jd j L reduces their entrainment
(1996) k = 9.34 , θ > 15° efficiency and requires more
k = 13.8, θ < 15° liquid to be circulating through
the nozzle in order to achieve a
desired degree of mixing. Here
effect of jet angle is considered.
Reliable correlation according
to Patwardhan (2002).

2.2 CFD studies such as chemical reactions by means of computer-


based simulation. This technique is powerful and
From the literature study it is found that different
spans a wide range of industrial and non-industrial
measurement techniques have been used to
applications such as mixing and separation in
determine mixing time in jet mixed tanks. It is one
chemical process engineering, aerodynamics of
of the reasons for obtaining different correlations
aircraft and vehicles, combustion in internal
for mixing time. All the correlations are case
combustion engines and gas turbines, distribution of
specific and cannot be used for the design of a
pollutants and effluents in environmental
generic jet mixed tank system. These types of
engineering, biomedical engineering, meteorology,
difficulties can be resolved by employing CFD
etc. Clearly the investment cost of a CFD capability
because flow variables and concentration anywhere
is not small. However, the total expenses are not as
in the tank would be available for verifying mixing.
great as those of high quality experimental
CFD involves numerical solution of complex
facilities. Some unique advantages of CFD over
equations such as Reynolds transport equations,
experimental CFD approaches to fluid systems
turbulent kinetic energy equation, energy
design are the ability to study systems for which
dissipation equation and continuity equation. In the
controlled experiments are difficult or impossible to
last two decades or so, CFD techniques (Patankar,
perform (very large systems) and the ability to
1980; Anderson et al., 1984; Hirsch, 1988), which
study systems under hazardous conditions at and
are based on numerical solution of the governing
beyond their normal performance limits.
partial differential equations, have become available
So far only few researchers have investigated jet
to the design engineer in the form of computer
mixing using CFD. Brooker (1993) studied the
codes. This allows simulation of a variety of flow
performance of jet mixer using CFD and it was
cases, which can help to assess various alternatives
found that CFD model predicted mixing time with a
and develop designs for better performance and safe
maximum error of 14% when compared with
operation. CFD is the analysis of systems involving
experimental values. For model validation, overall
fluid flow, heat transfer and associated phenomena
mixing time was compared for one nozzle geometry

159
Engineering Applications of Computational Fluid Mechanics Vol. 2, No. 2 (2008)

and one location of the probe. They identified and various locations. Zughbi and Rakib (2002)
examined errors due to discretization, time presented a CFD model of mixing in a fluid jet
integration and round off. They concluded that agitated tank and validated their numerical model
accurate quantitative information could only be against the experimental results. Wasewar and
obtained from numerical simulations if certain Patwardhan (2004) developed an in-house CFD
proper steps such as mesh refinement are taken into code to predict the flow field and mixing
consideration. Hoffman (1996) used FLUENT characteristics in jet mixed tanks. Predicted mixing
software package to find whether jet mixer can be time was compared with experimental
used as an effective device for injection of inhibitor concentration profile and mixing time. Zughbi and
into a tank containing monomer. Simulations were Rakib (2004) presented a CFD model of mixing in a
carried out for only one-half of the tank. Results fluid jet agitated tank. They validated their
were not compared with experimental numerical model against the experimental results of
measurements. Ranade (1996) used CFD code Lane and Rice (1982). Details regarding validation
FLUENT for investigating flow pattern using of numerical model can also be found in Rakib
steady and unsteady jets. The CFD simulations (2000). Some of these studies are summarized in
were carried out using a standard k-ε turbulence Table 2.
model. The CFD model was not validated against In summary, it can be said that, until recently, little
experimental measurements. Jayanti (2001) used a was known about the fluid flow, velocity field and
CFD code, CFX, to investigate the hydrodynamics detailed mixing characteristics of jet mixers. The
of jet mixing in cylindrical tanks. He studied the following are the shortfalls of the previous work on
effects of the flow circulation patterns within the the CFD modelling of jet mixed tanks: small
reactor and their effects on the mixing of a soluble number of grid points and no attempt to study a
salt. The overall mixing time was compared with wide variety of parameters. In the present work, 3D
correlations. Wasewar (2000) developed an in- simulation of jet mixing in tanks with the help of
house CFD code to predict the flow field and commercial CFD software package FLUENT is
mixing characteristics in jet mixed tanks. Predicted carried out for high grid numbers to study the
mixing time was compared with experimental hydrodynamics and mixing in jet mixed tanks so
concentration profile and mixing time. Patwardhan that design of jet mixed tanks can be done for any
(2002) compared experimental results with CFD range of dimensions. The present study investigates
modelling. A poor agreement was observed the effects of jet angle and elevation on mixing in a
between the numerical and experimental results of jet mixed tank using CFD.
concentration profiles as a function of time at
Table 2 Previous CFD studies on jet mixing in tanks.

Author CFD code Type of study Remarks


Brooker (1993) Compared experimental results with CFD Error of 14% between experimental and CFD values.
results.
Hoffman (1996) FLUENT To test injection of inhibitor into tank containing Jet mixer can be used for injection of inhibitor.
monomer.
Ranade (1996) FLUENT Flow pattern prediction with steady & unsteady Alternating jets give better mixing times.
jets.
Wasewar 2000 In-house CFD code was developed and validated using The effect of various parameters was not studied.
experimental data.
Jayanti (2001) CFX To find optimum shape of tank for better mixing. Found that conical bottom is better than hemispherical,
ellipsoidal, flat bottom for mixing.
Patwardhan (2002) FLUENT Compared experimental results with CFD CFD model predicts the overall mixing time well.
results.
Zughbi and Rakib FLUENT Various jets were studied. Validated using experimental results.
(2002)
Wasewar and In-house CFD code was developed and validated using The effect of various parameters was not studied.
Patwardhan (2004) experimental data.
Rakib and Zhugbi FLUENT Investigated effect of jet angle, position of jet Found that blending is a function of jet angle and position.
(2004) and multiple jet in jet mixing. Multiple jets performance depends on the Reynolds number.

160
Engineering Applications of Computational Fluid Mechanics Vol. 2, No. 2 (2008)

3. MODELLING AND SOLUTION predict overall mixing time. The temperature in the
STRATEGY tank was calculated by solving the generalized
scalar transport equation.
3.1 Problem statement
3.3 Flow domain and boundary conditions
A CFD solution of mixing in jet mixed tanks
requires complete specifications of the flow The flow domain consists of a cylindrical tank of a
domain, inlet, outlet, boundary and initial given diameter and height. It has an inlet, a jet of
conditions. A direct, one-to-one comparison small cross-section through which liquid comes in
between CFD results and existing experimental at high velocity, located at the wall above the
results are not possible because all necessary bottom surface as per requirement. The outlet, a
information was not given for any experimental hole of much larger cross-sectional area through
results available in the literature. which the liquid is taken out and is recirculated
The geometry of Lane and Rice (1982), Maruyama without time delay, is located at the desired location
(1982), and Simon and Fonade (1993) were used as per the geometry of study. There are no internal
for validation. A tank dimensions of 0.4 m × 0.4 m parts such as baffles. The flow domain is three-
was used for the case of parametric sensitivity. The dimensional; variations in the flow parameters are
jet diameter was fixed at 0.01 m. The position of the in radial, axial and tangential directions. The flow
jet and jet velocity were changed as per the case domain and grids are shown in Fig. 1. Wall
being investigated. functions were used to calculate the wall parameters
in turbulent flow (Rodi, 1984).
3.2 Governing equations
The governing equations for general mixing
problem solved for a CFD solution are the mass,
momentum, and energy equations. The generalized
equation is given below for a constant-density and
viscosity system after elimination of the appropriate
terms.
∂φ 1 ∂
+ (rUφ ) + 1 ∂ (Wφ ) + ∂ (Vφ )
∂t r ∂r r ∂θ ∂z

1 ∂⎛ ∂φ ⎞ 1 ∂ ⎛ Γ eff ∂φ ⎞ ∂ ⎛ ∂φ ⎞
= ⎜ rΓ eff ⎟+ ⎜ ⎟ + ⎜ Γ eff ⎟ + Sφ (1)
r ∂r ⎝ ⎜ ⎟
∂r ⎠ r ∂θ ⎝ r ∂θ ⎠ ∂z ⎝ ∂z ⎠

Turbulence model supported these equations for


turbulent flows. The standard k-ε model was used.
More sophisticated models such as the RNG k-ε
model and the Reynolds Stress Mode (RSM) were
available. However, these were not used as they
gave almost the same results as the standard k-ε
model did with an error of less than 1% and the Y

computation time being 3 times more (Rakib and


X
Zugbhi, 2004). In the standard k-ε model, two
additional parameters, namely, the turbulent kinetic Fig. 1 Flow domain and grids for jet mixing in tank.
energy and its dissipation rate were used. The
variation of these quantities within the flow domain A no-slip boundary condition is imposed on the
was calculated using additional transport equations. walls. Standard wall function is used to calculate
Details of the turbulence model are well-known and the velocity components near the wall. At the inlet,
are available in various books (Rodi, 1984; Warsi, a constant velocity with axial and radial
1993). The k-ε model was successfully used by components is specified. The outlet is treated as a
many researchers including Patwardhan (2002), mass flow boundary. In the present numerical
Jayanti (2001), and Zugabi and Rakib (2004) to model, temperature rather than concentration is

161
Engineering Applications of Computational Fluid Mechanics Vol. 2, No. 2 (2008)

used to measure mixing time. The initial the corrections for pressure and velocity. PISO
temperature throughout the tank is set to 300 K improves the efficiency of the calculations when
while the initial velocity field corresponds to the compared with SIMPLE by performing additional
steady state velocity field with the given flow corrections, namely, neighbour correction and
domain and boundary conditions. A known volume skewness correction. Neighbour correction can be
of fluid is heated by specifying a patch of the summarized as moving the repeated calculations
desired dimensions. The patch’s temperature was required by SIMPLE inside the solution stage of the
set to 375 K in all the cases. Upon specifying the pressure correction equation. PISO is recommended
patch dimensions, the steady state temperature is for transient calculations such as those carried out
calculated using a heat balance. The heat flux at the in the present study (FLUENT Manuals, 1998).
walls is zero. The properties of the fluid are Power Law Scheme was used to discretize the flow
considered not to vary with temperature and and energy equations with the second order upwind
therefore the flow field is not affected by changes in scheme used in cases of solution divergence to
temperature. Transient state calculations are generate an initial set of data for solution
performed with an initial time step of 0.1 sec, which convergence.
is slowly and gradually increased to 0.5 sec, 1 sec,
2 sec, 5 sec and finally 10 sec. Assigning the value 3.7 Convergence criteria
of the outlet temperature to the inlet temperature As there is no unified method to judge the
value after every iteration simulated the pump convergence of a solution, a common method is to
around. The pump around represents the effect of monitor residuals and the solution is assumed
re-circulation without time delays. converged when the residuals go below a certain
value. Residuals are defined as the imbalance in
3.4 Computer code
each conservation equation following each iteration.
The CFD calculations for the jet mixed tanks were The summation of the imbalance in each
performed using the commercial CFD code conservation variable over all the computational
FLUENT, version 6.2, developed by Fluent Inc. It cells is referred to the unscaled residual. When this
uses a finite volume method-based discretization of residual is scaled using a scaling factor
the governing partial differential equations. representative of the flow rate of the conserved
variable through the domain, it is then referred to a
3.5 Grid scaled residual (Zugabi and Rakib, 2004). A scaled
A non-uniform, staggered grid was used to residual is used in the present study because it
descretize the three-dimensional flow domain. makes it easier to judge convergence. The solution
Typically, 50,000–80,000 cells were used. The is said to have converged if the scaled residuals go
geometry of the tank was modelled in GAMBIT. below values ranging between 10-3 and 10-6. In all
The mesh type in all the cases for validation the cases of FLUENT simulation, the residual value
purposes as well as parametric sensitivity analysis for energy was set to 10-6 and for all other variables
was Tet/Hybrid with the TGrid Algorithm being to 10-3.
used to mesh.
4. RESULTS AND DISCUSSION
3.6 Solver
4.1 Model validation
The segregated solver algorithm was used to solve
Although the commercial code used in the present
the governing equations in all the cases for both
study, namely, FLUENT, has been tested both by
steady state as well as unsteady state calculations.
the vendors and several other users for a number of
SIMPLE algorithm was used for steady state
flow situations, it is always necessary to validate
calculations although PISO (Pressure Implicit with
the calculation methodology by comparing the CFD
Splitting of Operators) could also be used; both
predicted results with experimental/analytical
algorithms gave the same results with SIMPLE
results specific to the phenomena under
algorithm consuming lesser time. PISO is part of
consideration to demonstrate the accuracy of the
the SIMPLE family of algorithms. SIMPLE was
simulations. As the first step of validation, velocity
developed by Patankar (1980). PISO is based on a
profile for free turbulent jet was compared with
higher degree of the approximate relation between

162
Engineering Applications of Computational Fluid Mechanics Vol. 2, No. 2 (2008)

analytical results using the equations of Davies with the analytical results using the equation of
(1972) and Schlichting (1968) since experimental Davies (1972) and Schlichting (1968) are shown in
velocity profile is not available. Then, mixing time Fig. 2 and Fig. 3 respectively. It can be seen that the
was compared with experimental data of Lane and CFD predicted velocity profiles at various axial
Rice (1982), Maruyama, Ban and Mizushina distances collapse on to a single curve and that this
(1982), and Simon and Fonade (1993). curve agrees very well with the theoretical one.
There is systematic deviation from the theoretical
4.1.1 Velocity field curve at large values of radial distance; however,
Davies (1972) gave the equation for the centreline this appears to be the limitation of the theory
velocity for a free turbulent jet as: because similar deviation has been reported
(Davies, 1972; Schlichting, 1968).
d jU j
Vc = 6.4 , for x > 6.4 d j (2)
x 1
where x is the distance from the nozzle axis in the
0.8
direction of jet flow.
For a free turbulent jet, Davies (1972) has given the 0.6
radial distribution of the axial velocity

V/Vc
approximated by a curve of normal error with the 0.4
following equation:
0.2
2
⎡V ⎤ ⎡r ⎤ x
log10 ⎢ c ⎥ = 40 ⎢ ⎥ , for 7 < < 100 (3) 0
⎣V ⎦ ⎣x⎦ dj
0 1 2 3 4
r/(r,V=0.5Vc)
where r is the perpendicular distance to the
direction of jet and V is the axial velocity at any Fig. 2 Axial velocity comparison using the equation of
point in the direction of the jet. Davies (1972) for free turbulent jet.
Schlichting (1968) gave another equation for the ( • Equation of Davies (1972); CFD)
radial distribution of the axial velocity for the free
turbulent jet. 1

3 K 1
V = 0.8
8π v0 x ⎡ 1 2 ⎤ 2 (4)
⎢1 + 4 η ⎥ 0.6
⎣ ⎦
V/VC

0.4
where
1 0.2
1 3 K 2 r 1
η= ; K 2 = 1.59b 1 Vc ;
4 π v0 x 2 0
0 1 2 3 4 5
d jU j r/(r,V=0.5VC)
v0 = 0.256b 1 Vc ; b 1 = 0.0848 x ; Vc = 7.31
2 2 x
Fig. 3 Axial velocity comparison using the equation of
From the above equation we can say that centreline Schlichting (1968) for free turbulent jet.
jet velocity falls to about 5% of the initial value ( • Equation of Schlichting (1968); CFD)
after an axial distance of 100 jet diameters. After
400 jet diameters, the velocities become so low that 4.1.2 Mixing time
the mixing effect of the jet is insignificant at more
For CFD model validation, simulations were
remote positions.
performed to find out the mixing time in jet mixed
Simulations were performed for a tank height of
tank using geometry of Lane and Rice (1982),
1 m and a tank diameter of 1 m. The diameter of the
Maruyama, Ban and Mizushina (1982), and Simon
jet was set to 0.01 m with a jet Reynolds number of
and Fonade (1993). Details of the geometries and
12500. Comparison of the CFD predicted velocity
parameters used are given in Table 3. CFD mixing
profiles of the jet at various distances downstream
times were compared with experimental results.

163
Engineering Applications of Computational Fluid Mechanics Vol. 2, No. 2 (2008)

The mixing time is taken as the time at which the mixing is considered complete when m = 0.05, i.e.,
temperature (T) at the measurement location has mixing time required to achieve 95% mixing.
reached or nearly reached the expected final mean A typical case for the geometry of Lane and Rice
temperature Tmean. Mathematically the mixing time (1982) was simulated for the validation. The
can be defined as the time from temperature tracer geometry of the vessel and the jet were kept the
addition to the time when, same as in the experiments; the fluid was taken to
be water. In the present simulations, the flow field
[T - Tmean] / Tmean = m (5)
in one half of the cylinder was calculated and the
where m is the maximum acceptable value for resulting velocity field is shown in Fig. 4. It can be
deviation from mixing time. When the process of seen that a complicated but strong circulation
mixing just starts, m = 1; when complete mixing is pattern was set up in the vessel. Measurements of
achieved, m = 0. However, in most of the cases, the velocity field in this vessel are not available.

Table 3 Details of the geometries and parameters used for mixing time comparison.

Maruyama, Ban and


Parameters Lane and Rice (1982) Simon and Fonade (1993)
Mizushina (1982)
D 573 mm 560 mm 490 mm
H 573 mm 560 mm 490 mm
dj 10 mm 10 mm 10 mm
Nj 1 1 1
Jet angle 45° 0° 0°
Re j 3000–100000 3000–100000 3000–100000
Degree of mixing 95% 99% 95%
4.38 cm from the bottom
Jet location 1 cm above and away from the wall H/3 at sidewall
at sidewall
at the same wall of jet, near the 48.5 cm from the bottom
Outlet location center of bottom
liquid surface at the opposite wall, radial
at the liquid surface near the wall, at
Tracer location the location of 90° to the jet through nozzle through nozzle
horizontal axis
at the bottom, opposite to the trace the opposite to the jet at 0.1D
Measurement location at outlet
location side to 0.2D away from the wall

A typical predicted variation of the temperature for


the geometry of Lane and Rice (1982) is shown in
Fig. 5 at three locations. All the curves exhibit some
similarities such as the time lag between the time of
injection (heat source) and the response which is
different for each probe and the presence of local
maxima and minima. There is no sharp peak in the
predicted response. This may be a result of
turbulent fluctuations which are “averaged-out” in a
CFD framework; ensemble averaging of a number
of such tracer responses would be a better
experimental measure which can be compared with
Fig. 4 Typical velocity profile for the geometry of
CFD predictions (Jayanti, 2001). Another reason for
Lane and Rice (1982). this could be the injection period as noted above.
Finally, a sharp gradient is likely to be diffused

164
Engineering Applications of Computational Fluid Mechanics Vol. 2, No. 2 (2008)

much faster in a numerical solution as a result of velocity increases, the mixing time value does not
artificial diffusion effects (Jayanti, 2001). decrease appreciably at high Reynolds number,
thereby indicating that mixing time is a weak
function in the turbulent regime.
311

309
Temperature (K)

307

305

303

301

299
0 20 40 60 80 100
Time (sec)

Fig. 5 Typical temperature profiles at various locations


in the geometry of Lane and Rice (1982).
opposite wall 0.285 m above the base
top centre 0.55 m above the base
actual measurement location of Lane and
Rice (1982) Fig. 7 Typical velocity profile for the geometry of
Maruyama, Ban and Mizushina. (1982).

450
400 Lane and Rice
Further the geometry of Maruyama, Ban and
350
(1982)
CFD Mizushina (1982) was used for CFD simulations.
Mixing Time (sec)

300
The geometry of the vessel and the jet were kept the
250
same as in the experiments; the fluid was taken to
200 be water. The CFD simulated velocity field for the
150 geometry of Maruyama, Ban and Mizushina (1982)
100 is shown in Fig. 7. It can be seen that a complicated
50 but strong circulation pattern was set up in the
0 vessel. Measurements of the velocity field in this
0 5000 10000 15000 20000 25000 30000 35000 vessel are not available. A typical predicted
Reynolds Number
variation of the temperature is shown in Fig. 8 at
Fig. 6 Mixing time comparison of CFD results with various locations. All the curves exhibit some
experimental values of Lane and Rice (1982). similarities such as the time lag between the time of
injection (heat source) and the response which is
different for each probe and the presence of local
Mixing time comparisons for the geometry of Lane maxima and minima. There is no sharp peak in the
and Rice are shown in Fig. 6. It can be observed predicted response as in the previous case. Fig. 9
that the simulated mixing time values and the shows the comparison between CFD results and the
experimental results do not correlate well at lower experimental results. There is a good agreement
values of Reynolds number. This could be between the simulated values and the experimental
attributed to the k-ε model being accurate only for values. The standard k-ε model predicts well for
Reynolds numbers above 10000. However, at values above 10000. For Reynolds number of 7000,
higher jet velocities (Re > 10000), there seems to be there is an over prediction.
a very good agreement between experimental and
CFD results. In the case of jet mixing, for Reynolds
numbers less than 10000, the value of local
Reynolds numbers in most of the regions in the tank
is less than 3000. The k-ε model is more accurate
for the turbulent regions and not accurate for lower
Reynolds number region. It can be observed that as

165
Engineering Applications of Computational Fluid Mechanics Vol. 2, No. 2 (2008)

301.5 is concerned. The variation of temperature for the


simulation of Simon and Fonade (1993) at two
different locations is shown in Fig 11. It can be seen
Temperature (K)

301 that both the curves exhibit some similarities, such


as the time lag between the time of injection (heat
source) and the response which is different for each
300.5 probe and the presence of local maxima and
minima. The similar responses were obtained by
Simon and Fonade (1993) on an industrial scale
300 although each response was different. Hence, it is
0 20 40 60 80
Time (sec) more appropriate to compare predicted results with
experimental results. Also, there is no sharp peak in
Fig. 8 Typical temperature profiles at various locations the predicted response. The reason is the same as
in the geometry of Maruyama, Ban and
Mizushina (1982).
that mentioned above (Jayanti, 2001).
bottom 0.15 m above the base
opposite wall 0.25 m above the base
centre 0.28 m above the base

350
Maruyama (1982)
300 CFD
Mixing Time (sec)

250

200

150

100

50

0
0 10000 20000 30000 40000 50000
Fig. 10 Typical velocity profile for the geometry of
Reynolds Number Simon and Fonade (1993).

Fig. 9 Mixing time comparison of CFD results with


experimental values of Maruyama, Ban and
Mizushina (1982). 304

303
Simon and Fonade (1993) studied experimentally
Temperature (K)

the case of cylindrical mixing with two side jets 302

which may be either steady or alternating. One


301
typical case, corresponding to one jet was simulated
in the present study. The geometry of the vessel and 300

the jet were kept the same as in the experiments; the


299
fluid was taken to be water. In the present 0 20 40 60 80
Time (sec)
simulations, the flow field in one-half of the
cylinder is calculated and the resulting velocity Fig. 11 Typical temperature profiles at two locations in
field is shown in Fig 10. It can be seen that a the geometry of Simon and Fonade (1993).
complicated but strong circulation pattern was set centre 0.245 m above the base
up in the vessel. Measurements of the velocity field centre 0.45 m above the base
in this vessel are not available. A typical response
of the conductivity probes to the injection of a One quantitative feature that can be used for
tracer fluid was given by Simon and Fonade (1993). comparison is the overall mixing time. Fig 12
The necessary details for one-to-one comparison shows the comparison of the CFD results and the
were not given in the paper and only qualitative experimental results for the geometry of Simon and
comparison is possible as far as the probe response

166
Engineering Applications of Computational Fluid Mechanics Vol. 2, No. 2 (2008)

Fonade (1993). It was seen that the standard k-ε correlated values of Grenville and Tilton (1997) do
model does not predict well for Reynolds numbers not agree very well. This could possibly be due to
below 10000. There is a good agreement between the wall effects existing at the base of the tank.
the CFD values and experimental values for higher Gaikwad and Patwardhan (2003) also agreed that
jet velocities. Therefore, it can be concluded that horizontal jets required longer mixing time when
the standard k-ε model cannot be used when the compared to inclined jets. The other possibility
Reynolds number values are below approximately could be the presence of local mixing occurring at
10000. the tank base with the upper regions experiencing
relatively less turbulence. This reduces the
700 effectiveness of the jet as a mixer. A similar trend
600 Simon and Fonade (1993)
can be observed for a jet angle of 15° although
CFD there seems to be a fairly good agreement between
Mixing Time (sec)

500
the CFD values and the correlation values. Again,
400
this could possibly be due to the elevation of the jet,
300 which drives away a greater portion of the liquid
200 away from the base, thereby agitating the liquid
100 more. For angles of 30°, 45° and 60°, the simulated
values and the correlated values agree well
0
0 5000 10000 15000 20000 25000 indicating that the mixing phenomena occur to a
Reynolds Number greater extent. This is in accordance with what
Fig. 12 Mixing time comparison of CFD results with
Gaikwad and Patwardhan (2003) had propounded.
experimental values of Simon and Fonade At a jet angle of 45°, a minimum exists giving the
(1993). shortest mixing time, as suggested by Coldrey
(1978). For a jet angle of 45° and an H/D ratio of
4.2 Design parameters 1.0, the jet traverses the maximum possible jet
length, thereby incorporating more turbulence in the
A parametric sensitivity analysis was performed to tank by entraining larger amounts of liquid. For a
analyse the effect of various parameters such as jet jet angle of 60°, the mixing time increases again
position, jet angle, jet Reynolds number, position of since the jet traverses a shorter length when
heat source, and tank height to diameter ratio. In all compared to that traversed with a 45° angle. The
the cases, a tank with dimensions of 0.4 m × 0.4 m results presented here cannot be expected to hold
was used. The jet diameter was fixed at 0.01 m. The under laminar conditions since validation is not
position of the jet and jet velocity were changed as done for such cases. Also for Reynolds numbers
per the case being investigated. The mixing time below 10000, the standard k-ε model does not
corresponded to 95% mixing. The meshing scheme predict well. Hence, the analysis made holds well
used was Tet/Hybrid with TGrid algorithm being only for turbulent conditions with Reynolds
used. The heated fluid’s mass was kept at 1 kg in all numbers above 10000.
the cases. However, the position of the heat fluid
varied in each case since the tank dimensions
changed. In the analysis of the effect of jet angle, a 105
jet velocity of 2.4 m/sec was used. Analysis of 90 Grenville and
different jet positions was done for 3 heights— Tilton (1997)
Mixing Time (sec)

Correlation
75 CFD
0.015 m above the base, 0.18 m above the base and
60
0.25 m above the base. The fluid used was water.
45

4.3 Effect of jet angle at tank base 30

15
The jet angle was varied from 0° to 60° with the
0
intermediate values being 15°, 30° and 45° for an 0 15 30 45 60 75
H/D ratio of one. Fig 13 shows the effect of jet Jet Angle (degrees)

angle on mixing time. The nozzle was placed at Fig. 13 Effect of jet angle at tank base on mixing time.
0.015 m above the tank base. For a jet angle of 0°, it
can be observed that the simulated values and the

167
Engineering Applications of Computational Fluid Mechanics Vol. 2, No. 2 (2008)

4.4 Effect of jet angle with variation in jet 4.5 Effect of H/D ratio
position
To investigate the effect of H/D ratio, H/D ratios of
For an H/D ratio of 1.0, the simulations were 1.0, 1.25 and 1.5 were used. The values of jet
repeated with the jet position and the jet angle being Reynolds number were 14000, 24000 and 32000.
varied simultaneously. The jet angles used were The jet angle was varied simultaneously with jet
-45°, -30°, 0°, 30° and 45°. These angles were tried locations of 0.015 m, 0.18 m and 0.25 m, and also
at 0.18 m and 0.25 m respectively. The jet velocity with the jet angle being varied as 0°, 30°, 45° and
was fixed at 2.4 m/sec. Fig. 14 gives a comparison 60°.
of mixing times for two jet positions. It shows that a Figures 15, 16, 17 give the plots of mixing times
jet angle of 0°, at both locations used, gives the versus H/D ratios for the jet location at 0.015 m
minimum mixing time while a jet angle of 45° above base. For H/D values of 1.0 and 1.5, it is
produces the maximum mixing time. This is due to observed that mixing time is the shortest for a jet
the presence of the jet near the outlet and that most angle of 45°. However, a similar trend does not
of the liquid exits from the tank. Hence, a larger occur to an H/D ratio of 1.25. Therefore, without
portion of the liquid experiences poor shear with further investigations, it is not possible to ascertain
low turbulence existing in most parts of the tank. whether mixing time is the shortest for a jet angle of
Also, in the case where the jet location is at 0.25 m 45°. For H/D ratio, a minimum occurs for the case
above the base, the mixing time is longer as of jet angle of 0°. One result that can be established
compared to the case where jet location is 0.18 m is that mixing time for a given jet angle and jet
above the base. This substantiates the previous velocity increases with increase in H/D ratio.
statement that the closer the inlet is to the outlet, the Grenville and Tilton (1997) also agreed on this. The
greater the bypassing of liquid becomes. Hence, the increase in mixing time is due to an increase in the
cases of 45° for jet locations of 0.18 m and 0.25 m volume of fluid to be processed. For the cases of
above the base are highly undesirable. H/D = 1.5, θ = 0°, Vj = 2.4 m/sec, and H/D = 1.5,
In the cases featuring negative jet positions of -30° θ = 60°, Vj = 2.4 m/sec with jet location at 0.015 m
and -45°, mixing times are slightly longer than that and higher above the base and H/D = 1.5, θ = 60°,
of 0° but shorter than those in the cases involving Vj = 3.2 m/sec, mixing time is constant and their
positive jet angles. This happens since the jet is change, unlike other cases, is negligible. When the
positioned away from the outlet and circulation jet location is changed to 0.18 m above the base, the
occurs to a larger extent. The liquid, before exiting value of mixing time is the highest for a jet angle of
the tank, entrains a greater volume of liquid. -30° for all velocities.
However, it is observed that in all the cases In the case of jet location at 0.18 m above the base,
involving jet locations away from the base, mixing for a given jet angle and a given H/D ratio, the
time is not the shortest since the jet is not traversing mixing time decreases with increase in jet velocity.
the maximum possible jet length. Therefore, for This is expected since the mixing time decreases
tanks with an H/D ratio of 1.0, it is recommendable with increase in Reynolds number although not
to locate the jet close to the base. very appreciably at very high values of about
120
40000. As proposed by Lane and Rice (1982), for
values between 10000 and 40000, re-circulation of
100 fluid in the bulk tank still produces laminar-like
Mixing Time (sec)

80 flow indicating that turbulence exists only in


pockets. For tanks where H/D > 1, it is suggested
60
that multiple jets be used at different locations in
40
CFD, 0.18 m above base
the tank to induce more mixing.
20 CFD, 0.25 m above base The investigations were performed only for H/D
values of 1.25 and 1.5. Therefore, it cannot be
0
concluded as to what the exact nature of the mixing
-50 -40 -30 -20 -10 0 10 20 30 40 50
Jet Angle time dependency is with respect to the H/D ratio.
The mixing time dependency can be better
Fig. 14 Effect of jet angle on mixing time for jet
understood through further investigations of H/D
position of h1 = 0.18 m and h1 = 0.25 m.

168
Engineering Applications of Computational Fluid Mechanics Vol. 2, No. 2 (2008)

ratios between 1 and 1.25 as well as between 1.25 5. CONCLUSIONS


and 1.5.
Mixing is one of the important operations in the
250 chemical industry. Jet mixing is an alternative to
mechanical agitators for large and underground
Mixing Time (sec)

200 storage tanks. This paper explains briefly the jet


mixing process and the various experimental studies
150
conducted in the past. Though many experimental
100
0 degrees
correlations are available, they are case specific and
30 degrees
45 degrees
there is a need for CFD predictions to accurately
50
60 degrees
design jet mixing systems.
0 The commercial CFD code FLUENT was used to
0.75 1 1.25 1.5 1.75 simulate the jet mixing phenomenon in tanks. The
H/D ratio geometry used was constructed using GAMBIT
modeler. Tetrahedral elements and TGrid algorithm
Fig. 15 Effect of H/D ratio on mixing time for jet were used to mesh the geometry. Segregated solver
location at 0.015 m above the base;
was used to solve the governing equations in the
Vj = 1.4 m/sec.
present work. SIMPLE algorithm for steady state
and PISO for transient simulation were used as
pressure velocity coupling schemes. Turbulence
120
was modelled using standard k-ε model. CFD
100 analysis was carried out with the help of various
Mixing Time (sec)

80 post-processing facilities available in FLUENT.


The results were analysed and found to conform to
60
0 degrees
the experimental values although an error of up to
40 30 degrees 15% was noted in a few cases. The simulated
20
45 degrees
results were first validated against experimental
60 degrees
data of Lane and Rice (1982), Maruyama, Ban and
0 Mizushina (1982), and Simon and Fonade (1993). It
0.75 1 1.25 1.5 1.75
H/D ratio was found that the simulated values did not
correlate well with the experimental values at lower
Fig. 16 Effect of H/D ratio on mixing time for jet values of Reynolds number since the standard k-ε
location at 0.015 m above the base; model did not predict well for low Reynolds
Vj = 2.4 m/sec. number. However, there was good agreement
between the simulations and experimental data.
Upon establishing valid simulation results, the
100 study was extended to a parametric sensitivity
analysis wherein the effects of tank parameters
80
Mixing Time (sec)

were studied. The parameters investigated were jet


60 velocity, jet angle, location of jet, effect of heat
source, and tank height to diameter ratio. Results
40 0 degrees
30 degrees
obtained were analysed, and conclusions were
20 45 degrees drawn and verified using literature survey.
60 degrees However, further analysis is to be made before a
0 complete picture of the mixing phenomena can be
0.75 1 1.25 1.5 1.75
H/D ratio
established. It can be concluded that a tremendous
amount of scope exists for CFD in designing of jet
Fig. 17 Effect of H/D ratio on mixing time for jet mixed tanks and this can also be extended to
location at 0.015 m above the base; designing jet mixing of solid suspension systems.
Vj = 3.2 m/sec.

169
Engineering Applications of Computational Fluid Mechanics Vol. 2, No. 2 (2008)

NOMENCLATURE 6. FLUENT (1998). Fluent 6.1 User's Guide.


http://jullio.pe.kr/fluent6.1/help/html/ug/main_p
A Cross-sectional area of jet, m2 re.htm.
c* Degree of mixing 7. Fossett H (1951). The action of free jets in the
Cp Correlation constant (Fossett and Prosser, 1949) mixing of fluids. Transactions of the Institution
D Diameter of tank, m of Chemical Engineers 29:322–332.
dj Diameter of jet, m 8. Fossett H, Prosser LE (1949). The application
do Outlet diameter, m of free jets to the mixing of fluids in the bulk.
f Mixing time factor Proceedings of I Mechanical Engineering
g Acceleration due to gravity, m/s2 160:224–251.
H Height of the tank, m 9. Fox EA, Gex VE (1956). Single phase blending
hi Nozzle height from the bottom of the tank, m of fluids. American Institute of Chemical
ho Outlet height from the bottom of the tank, m Engineers Journal 2:539–544.
J Momentum of jet, kg. m/s2 10. Gaikwad SG, Patwardhan AW (2003). Mixing
Js Specific jet momentum in tanks agitated by jets. Transactions of the
k Correlation constant in Grenville and Tilton Institute of Chemical Engineers 81A:211–220.
(1997) 11. Grenville RK, Mak ATC, Ruszkowski SW
L Maximum jet path length, m (1992). Blending of fluids in mixing vessels by
M Mixing factor in Simon and Fonade (1993) turbulent recirculating jets. The 1992 IchemE
correlation Research Event. University of Manchester
Nj Number of jets Institute of Science and Technology, UK, 128–
r Perpendicular distance from the jet 130.
Re Jet Reynolds number 12. Grenville RK, Tilton JN (1996). A new theory
Ret Tank Reynolds number improves the correlation of blend time data
tr Residence time, sec from turbulent jet mixed vessels. Transaction of
Tm Mixing time, sec the Institution of Chemical Engineers 74A:390–
Tinj Injection time, sec 396.
Uj Jet velocity, m/sec 13. Grenville RK, Tilton JN (1997). Turbulence for
Vc Centreline velocity, m/sec flow as a predictor of blend time in turbulent jet
V Axial velocity, m/s mixed vessels. Recent Progress en Genie des
x Jet path length, m Procedes. Proceedings of the 9th European
z Distance from the jet in the direction of the jet Conference on mixing, France, 11(51):67–74.
flow, m 14. Harnby N, Edwards MF, Nienow AW (1985).
θ Jet angle Mixing in the process industries. London:
Butterworth-Heinemann.
REFERENCES 15. Hiby JW, Modigell M (1978). Experiments on
jet agitation. 6th CHISA Congress, Prague.
1. Anderson DA, Tannehill JC, Pletcher RH 16. Hoffman PD (1996). Mixing in a large storage
(1984). Computational fluid mechanics and tank. AIChE Symposium series no.286 88:77–
heat transfer. Washington DC, USA: 82.
Hemisphere. 17. Jayanti S (2001) Hydrodynamics of jet mixing
2. Baldyga J, Bourne JR, Zimmermann B (1994). in vessels. Chemical Engineering Science
Investigation of mixing in jet reactors using fast 56:193–210.
competitive-consecutive reactions. Chemical 18. Lane AGC, Rice P (1981). An experimental
Engineering Science 49(12):1937–1946. investigation of jet mixing employing a vertical
3. Brooker L (1993). Mixing with the jet set. submerged jet. International Chemical
Chemical Engineering 30:16. Engineering Symposium Series 64:K1.
4. Coldrey PW (1978). Jet mixing. Paper to I. 19. Lane AGC, Rice P (1982). An experimental
Chem. E. Course, Univ. of Bradford, England, investigation of jet mixing employing an
July 1978. inclined side entry jet. Transaction of the
5. Davies JT (1972). Turbulence phenomena. New Institution of Chemical Engineers 60:171–176.
York: Academic Press.

170
Engineering Applications of Computational Fluid Mechanics Vol. 2, No. 2 (2008)

20. Lehrer IH (1981). A new model for free 34. Wasewar KL (2000) Modeling of jet mixed
turbulent jets of miscible fluids of different tanks. Master of Chemical Engineering, Thesis,
density and a jet mixing time criterion. UDCT, Mumbai, INDIA.
Chemical Engineering Research and Design 35. Wasewar KL, Patwardhan AW (2004).
59:247–252. Modeling of jet mixed tanks. Proceedings of
21. Maruyama T, Ban Y, Mizushina T (1982). Jet National Conference on Mathematical
mixing of fluids in tanks. Journal of Chemical Modeling and Analysis. Mathematics Group,
Engineering of Japan 15:342–348. BITS, Pilani-333031, INDIA, Oct.8–9.
22. Maruyama T, Kamishima N, Mizushina T 36. Yianneskis M (1991). The effect of flow rates
(1986). Jet mixing of fluids in tanks. Journal of and tracer insertion time on mixing times in jet
Chemical Engineering of Japan 17:120. agitated vessels. Proceedings of 7th European
23. Mewes D, Renz R (1991). Jet mixing of liquids Conference on mixing, Kiav, Brugge, Sept 18–
in storage tanks. Proceedings of 7th European 20, Belgium, 121–128.
Conference on mixing, Kiev, Brugge, Sept 18– 37. Zughbi HD, Rakib MA (2002). Mixing in a
20, Belgium, 131–137. fluid jet agitated tank. Chemical Engineering
24. Okita N, Oyama Y (1963). Mixing Communication 189:1038–1056.
characteristics of jet mixing. Kagaku Kogaku 38. Zughbi HD, Rakib MA (2004). Mixing in a
27:252–259. fluid jet agitated tank: effect of jet angle and
25. Orfaniotis A, Fonade C, Lalane M, Doubrovine elevation and number of jets. Chemical
N (1996). Experimental study of fluidic mixing Engineering Science 59:829–842.
in a cylindrical reactor. Canadian Journal of
Chemical Engineering 74:203–212.
26. Patankar SV (1980) Numerical heat transfer
and fluid flow. Hemisphere: New York.
27. Patwardhan AW (2002). CFD modelling of jet
mixed tanks. Chemical Engineering Science
57:1307–1318.
28. Perona JJ, Hylton TD, Youngblood EL,
Cummins RL (1998). Jet mixing of liquids in
long horizontal cylindrical tanks. Industrial
Engineering and Chemistry Research 38:1478–
1482.
29. Ranade VV (1996). Towards better mixing
protocols by designing spatially periodic flows:
The case of a jet mixer. Chemical Engineering
Science 51:2637–2642.
30. Schlichting H (1968). Boundary layer theory.
6th ed. McGraw Hill: New York, USA.
31. Schmetzek R, Steiff A,Weinspach PM (1995).
Examination of discontinuous jet mixing for
designing emergency cooling systems of
chemical reactors. Proceedings of 8th European
Conference on Mixing 136:391–398.
32. Simon M, Fonade C (1993). Experimental
study of mixing performances using steady and
unsteady jets. Canadian Journal of Chemical
Engineering 71:507–513.
33. Van de Vusse JG (1951). Vergleichende
ruhrversuche zum mischen losllicher
flussigkeiten en einem 12000 m3 behalter.
Chemie. Ing. Tech. 31:583–587.

171

You might also like