Impeller Perfomance in Stirred Tanks
Impeller Perfomance in Stirred Tanks
Impeller Perfomance in Stirred Tanks
Mixing:
Impeller Performance
in Stirred Tanks
Characterizing mixer impellers on the basis of power, flow, shear and efficiency
M
Richard K. ixing has been defined as “the
Grenville NOMENCLATURE
application of mechanical mo- A Constant in Equation (30)
and Jason J. tion in order to create fluid dy-
Giacomelli ADIS Discharge area for primary flow from impeller
namic effects that achieve a de- D Impeller diameter
Philadelphia Mixing
sired process result” [1]. The process result d32 Sauter mean droplet size
Solutions Ltd.
is the objective of the vessel operator and Fl Flow or pumping number (= Q/(ND3))
Gustavo Padron
and David A. R.
will be a transformation of the ingredients K Ratio (= MAX/)
Brown fed to the vessel into a product. The goal kMAX Maximum kinetic energy in trailing vortex
BHR Group of the equipment supplier will be to under- l0 Diameter of trailing vortex
stand the role of mixing in promoting the N Impeller rotational speed
transformation and choosing an impeller P Power
IN BRIEF that will create the appropriate fluid-dy-
namic effects to do this.
Po Power number [= P/(N3D5)]
Q Flow rate generated by impeller
TURBULENCE Processes carried out in stirred tanks can be R Impeller radius
IMPELLER GEOMETRIES generally divided into the following two classes: r Radial position in impeller discharge for estimating
• Those relying on flow generated by the velocity gradient
HYDRAULIC EFFICIENCY Re Impeller Reynolds number (= ND2/µ)
impeller creating motion throughout the
SHEAR T Vessel diameter
fluid, such as blending of pigments into
U Mean velocity in impeller discharge (= Q/ADIS)
PROCESS RESULTS a resin or emulsion in paint manufacture
VTIP Impeller tip speed
where homogeneity of the vessel contents
TRAILING VORTEX vH High velocity in impeller discharge for estimating
is critical to product quality velocity gradient
APPLICATIONS • Those relying on “shear” to reduce vL Low velocity in impeller discharge for estimating
CONCLUSIONS
the size of a second dispersed phase, velocity gradient
whether gas bubbles, liquid droplets or w Projected blade height
particles, such as a hydrogenation reactor x Ratio of impeller to trailing vortex diameters (= D/l0)
where smaller bubbles provide more sur- y Distance (in definition of shear rate)
face area for mass transfer from the gas Constant (= vH/VTIP)
into the liquid phase Constant (= vL/VTIP)
Impellers are often described qualitatively . Time-averaged velocity gradient
Power input per mass of fluid in vessel
as, among others, high flow, high shear or
MAX Local energy dissipation rate in trialing vortex
high efficiency, and the choice of equip- Efficiency defined as mass of fluid pumped per unit
ment required to achieve the process result energy input by impeller
most efficiently is made on this vague basis. Shear rate constant
This article describes how the performance Liquid density
characteristics of impellers commonly used Efficiency defined as kinetic energy of fluid divided
in stirred tanks can be quantified, thereby by mechanical energy input by impeller
enabling engineers to make educated deci- Constant (= rH/R)
sions about which ones to use in order to Constant (= rL/R)
achieve their desired process results. Subscripts
AX Axial PBT Pitched-blade tur-
Turbulence HYDFL Hydrofoil bine
Turbulent flow is characterized by the pres- IMP Impeller RD Radial
ence of random fluctuations in velocity, so- HYDR Hydraulic RUSH Rushton
called eddies, that are superimposed on the MECH Mechanical
mean, time-averaged flow. There will be a
46 CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM AUGUST 2017
Credit for a–g: PMSL
FIGURE 1. Four general classes of impellers are used in stirred tanks operating at low to medium viscosities in the turbulent regime. These impellers primarily
generate: axial flow (a, b, c); mixed flow (d), radial flow (e, f, g) or dispersion or de-agglomeration (h)
range of eddy time and length scales associated with Pitched-blade turbines have flat blades that are usu-
a particular flow field. The size of the largest eddies will ally angled at 45 deg, although shallower and steeper
be on the order of the size of the equipment generat- angles are sometimes used.
ing the flow (for example, the blade width of an impel- 3. Radial flow. These impellers generate a strong ra-
ler). The size of the smallest eddies is the Kolmogorov dial component of velocity directed at the vessel wall. A
length scale. The eddies also have a lifetime, with the pitched blade turbine with 90-deg blade angle generates
larger eddies existing for a longer period than the small radial flow and is commonly called a flat-blade turbine
ones. Understanding the role turbulence plays in mixing (Figure 1e).
processes is critical to successful design and scaleup [2]. Impellers used for processes requiring dispersion of
gas bubbles also generate a primarily radial flow, but
Impeller geometries have blades attached to a disk. The Rushton (Figure 1f)
There are four general classes of impellers used in stirred and Smith (Figure 1g) turbines are commonly used for
tanks operating in low to medium viscosity fluids in the these processes. The disk ensures that bubbles fed into
turbulent regime (Re > 104): the vessel beneath the impeller must flow through the
1. Axial flow. The primary flow generated by an axial-flow blades where the local “shear” breaks them up, creating
impeller is directed down toward the base of the vessel. high interfacial area for mass transfer. The Rushton tur-
Hydrofoils with narrow or wide blades are in this category. bine is generally considered to be “high-shear” [4].
Hydrofoils have profiled blades that may be narrow 4. High-speed dispersers. These impellers look like
like an airplane wing (Figure 1a) or wide like a marine circular-saw blades with alternating teeth angled up and
propeller (Figure 1b). These impellers were developed down (Figure 1h). They operate at high rotational and tip
to generate the same velocity profile as a propeller, but speeds and are used almost exclusively for processes
to be fabricated rather than cast to reduce the impeller’s that require significant size reduction, such as dispersion
weight and cost. They are also easier to install since and de-agglomeration of dry powder when preparing a
they can be supplied as a hub and blades that are as- slurry from liquid and a dry powder.
sembled inside the vessel [3]. These impellers are gen-
erally considered to be “low-shear” [4]. Hydraulic efficiency
An anti-ragging hydrofoil (Figure 1c) is used in waste- Impellers in stirred tanks are machines that move fluid;
water applications. It has blades that are swept-back essentially they are pumps. Like pumps, their efficiency
preventing build-up of fibrous matter, which is commonly can be defined and calculated. The hydraulic efficiency
present in municipal wastewater, on the leading edge of of a pump is the ratio of the kinetic energy of the flowing
the blades. fluid to the mechanical energy input by the impeller.
2. Mixed flow. These impellers generate both axial and The mechanical power input by an impeller in a stirred
radial components of velocity and the distribution be- vessel is calculated from the following equation (Note: all
tween the two can be controlled by adjusting the impeller nomenclature are defined in the box on p. 46):
diameter to vessel diameter ratio. Pitched-blade turbines
(Figure 1d) are in this category. PMECH = Po N 3 D 5 (1)
CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM AUGUST 2017 47
60%
Fl N D 3 Fl D
U= N D (5)
Pitched-blade turbine
=
w D w
Narrow hydrofoil
50%
Wide hydrofoil
Flat-blade turbine
Rushton
The energy dissipation rate, or power, of the flowing
40%
High-shear disperser sawtooth fluid is the product of the flowrate and the head that the
30%
Wide hydrofoil pump develops:
Narrow hydrofoil
HYDR, %
Pitched-blade turbine
20% Anti-ragging turbine PHYDR = Q H (6)
10% Where:
2
U
H= (7)
2
0%
0 1 2 3 4 5 6
Po, dimensionless
FIGURE 2. This graph plots the hydraulic efficiency, HYDR, versus the power Combining Equations (2), (4), (6) and (7), for axial flow
number for various impellers impellers:
2
4Fl 8Fl 3
Po is the impeller’s power number and it is a drag coef- PHYDR = Fl N D 3
N D = N 3 D5 (8)
ficient that is determined by the geometry of the impeller 2 2
HYDR, kg/J
This quantity has units of kilogram of fluid pumped per Narrow hydrofoil
Joule of energy input by the impeller. Pitched-blade turbine
Anti-ragging turbine
The power input per unit mass of fluid, for a vessel 0.2
where depth is equal to vessel diameter, can be calcu-
lated from: 0.1
Po N 3 D5 4 Po N 3 D 5
= = (13) 0
( 4) T 3
T3 0.2 0.3 0.4 0.5 0.6
D/T, dimensionless
FIGURE 3. Shown here is a plot of the hydraulic efficiency, HYDR, versus
Re-arranging for impeller speed: impeller-to-vessel-diameter ratio
where vH and vL are the high and low velocities in the
13
T 3 gradient and rH and rL are the radial positions corre-
N= (14) sponding to the locations where these velocities were
4Po D 5 measured. Since the velocities are normalized by the
Substituting Equation (14) into Equation (12) gives the impeller tip speed and the radial positions by the impeller
following: radius, Equation (16) can be re-written as follows:
23 43
Fl 4 4Po D 5 Fl D 23
VTIP
= 1.175 2 3 ( T) VTIP
=(
HYDR = )
Po D 2 T3 Po T = (17)
23
(15) ( ) R R
43
o D5 Fl D 23
= 1.175 2 3 ( T) Values of , , , and are given in Table 2. Also
T3 Po T the ratio of / HYDFL is shown and, at equal tip speed
The hydraulic efficiency data plotted in Figure 2 are and impeller diameter the Rushton generates the high-
replotted in Figure 3 using the new definition from Equa- est shear rate followed by the pitched-blade turbine and
tion (15) with a power per mass of 1 W/kg and vessel then the hydrofoil.
diameter of 1 m. The effect of impeller diameter is now Engineers are concerned with the power drawn by
taken into account and large diameter impellers (D/T the impeller since this determines the size of the agitator
≈ 0.5) are more efficient than smaller ones (D/T ≈ 0.3), needed to achieve the desired process result. Equation
pumping approximately twice the mass of fluid per unit (13) can be rearranged to express the power input by
of energy input. the impeller per unit mass of fluid in terms of tip speed:
Shear 2
4 VTIP
3
D (18)
In any flowing system, the shear rate is the time-aver- = Po
aged velocity gradient [7].
4
T T
Oldshue [3] has compared the time-averaged veloc-
ity gradients in the discharge of a hydrofoil and pitched- The π3 term must be introduced because VTIP = πND.
blade and Rushton turbines to show that the Rushton Comparing different impellers of equal diameter at the
generates higher shear than the pitched-blade, which same scale:
generates higher shear than the hydrofoil. This has be-
come the conventional wisdom in the mixing field. VTIP Po 1 3 (19)
Figure 4 shows the mean velocity profiles for a hydro-
foil (in green) and pitched-blade turbine (in red), which Comparing any impeller with the hydrofoil:
were measured using particle-image velocimetry in the
PMSL laboratory. The dashed lines show the average IMP IMP VIMP
velocity gradient in the discharge. Figure 5 shows the = (20)
mean velocity profile for the Rushton turbine and, again, HYDFL HYDFL VHYDFL
the dashed lines show the average velocity gradient in
the discharge. The shear rate is described by the follow- Substituting Equation (19) into Equation (20):
ing equation: 13
v v IMP IMP PoHYDFL
= H L (16) = (21)
rH rL HYDFL HYDFL PoIMP
facial tension between the two liquids and the viscosity of Where VTIP = πN.D.
the dispersed phase liquid. The standard deviation for this correlation is ±10%.
Figure 6 shows the Sauter mean droplet size plotted The maximum energy dissipation rate within the vortex
versus the average power input per unit mass for Rush- is given by the following equation [13]:
ton, two pitched-blade turbines, with blades angled at kMAX
32
45 and 60 deg, a hydrofoil and a high-shear disperser MAX = A (23)
impeller. The experiments were carried out with low vis- l0
cosity silicone oil as the dispersed phase and at a very l0 is a length scale related to the flow near the impeller
low concentration so that the effect of coalescence on and it is a fraction of the impeller diameter. Substituting
the droplet size can be ignored. Equation (22) into Equation (23) and setting l0 = D/x and
If the hydraulic efficiency and shear rate comparison A = 1:
quantify the performance characteristics of the impellers,
when compared at the same power input per mass, the = 1.04 x Po3 4 N 3 D 2 (24)
MAX
15%
wp/R, %
0%
The ratio is weakly dependent on the type of impeller -5%
(Po), dependent on the scale of the vortex (x) and strongly -10%
dependent on the size of the impeller (D/T). The reason -15%
for this is that a small-diameter impeller must operate at -20%
a higher tip speed than a larger one to input the same -25%
power and the maximum kinetic energy is proportional to 0% 10% 20% 30% 40% 50% 60% 70%
v/VTIP, %
the tip speed squared.
FIGURE 5. This graph shows a plot of the mean velocity profiles for the Rush-
The droplet size data plotted versus the average power ton turbine
input per mass in Figure 6 are replotted in Figure 7 versus
the maximum energy dissipation rate in the trailing vor- as defined in Equation (26), of 15, 25 and 50 s–1. The
tex. The variations in the trailing vortex energy dissipation corresponding values of vessel-averaged power input per
rate generated by the impellers and the effects on the mass are 2.25 10–4, 6.25 10–4 and 2.50 10–3 W/
droplet size are now correctly accounted for, including kg. Grenville and Spicer [16] have re-analyzed these data
the high-shear disperser. and the floc length versus the maximum kinetic energy
The conventional wisdom in the mixing industry has dissipation rate, calculated using Equation (23), is plotted
been that hydrofoil impellers generate “low shear” and in Figure 8. This approach to the analysis correlates the
Rushton turbines generate “high shear” [3, 4] and this data and suggests that the concept of a G-value should
is true if only the time-averaged velocity gradients are work for agitator design provided that it is based on the
compared. The maximum kinetic energy dissipation rate maximum energy-dissipation rate in the trailing vortex —
within the trailing vortex, MAX, generates the stresses not the vessel averaged power per volume.
that break-up droplets, or any other second phase, in an The selectivity of competitive reactions carried out
agitated vessel. Rather than describing these impellers in semi-batch mode is determined by the local mixing
as “high shear,” it is more rigorous to call them “high dis- rate [17], the micro-mixing rate, in the region where
sipation” or “high stress.” the added reactant is introduced to the vessel [18].
Bourne and Dell’ava [19] have shown that the selec-
Applications tivity of an azo-coupling reaction can be maximized
There are many processes in which the fluid dynamic ef- by feeding the added reactant at the impeller where
fect that achieves the process result is commonly con- the trailing-vortex energy dissipation rate determines
sidered to be “shear” although, strictly, the process result the rate of micro-mixing. They, and Nienow and others
is determined by the maximum energy dissipation rate [20], have also shown that, provided the feed location
within the trailing vortex. One example of a “shear” driven is geometrically similar, the selectivity of the reaction
process is flocculation of fine particles. Agitators are de- can be maintained on scaleup if the trailing-vortex en-
signed to provide a desired shear rate, or G-value. G is ergy dissipation rate is the same at the two scales.
defined as: This has also been shown to apply to precipitation re-
actions where the particle size and morphology need
12 12 to be controlled [21, 22].
P Po N 3 D5
G= = (26) 1000
μ V μ V Narrow hydrofoil
Pitched-blade turbine, 45 deg
Pitched-blade turbine, 60 deg
This shear rate is based on the vessel-averaged
Rushton
power input per volume and the fluid’s dynamic vis-
High-sheear disperser sawtooth
cosity. Equation (26) suggests that, provided that the
average power per volume is kept constant, the same
d32, µm
100
G-value will be generated and the flocculation perfor-
mance will be the same. Benz [14] has written a review
of the problems that will be encountered taking this
approach to agitator design, especially the fact that it
takes no account of impeller type or diameter. He con-
cludes that “G-value has no legitimate use in designing
or specifying agitators.” 10
0.01 0.10 1.00 10.00
Spicer and others [15] have measured the size and , W/kg
structure of flocculated polystyrene particles using a hy- FIGURE 6. This graph shows the behavior of the Sauter mean diameter, d32,
drofoil, pitched blade and Rushton turbines at G-values, versus vessel-averaged power input per unit mass
Circle xx on p. 66 or go to adlinks.chemengonline.com/66431-xx