You are on page 1of 9

Granular Matter 6, 159–165 

c Springer-Verlag 2004
DOI 10.1007/s10035-004-0166-x

Measurements and comparison of saltation and pickup


velocities in wind tunnel
Mario Hubert, Haim Kalman

d U ρ
Abstract This work presents experimental results of the Rep particle Reynolds number Rep = p µgg g
pickup velocity measurements for variety of particulate Ug mean gas velocity [m/s]
solids. The experiments were carried out in a horizontal Ugpu is pickup velocity (average gas velocity when
wind tunnel. The influence of three initial arrangements pickup starts) [m/s]
of particle(s) on pickup velocity was investigated. Parti- Up particle velocity [m/s]
cle(s) were placed either on the bottom surface of the tun- µg gas dynamic viscosity [kg/ms]
nel or on the horizontal layer of particles or the heaps of ρg gas density [kg/m3 ]
particles with various heights were created in the wind ρp particle density [kg/m3 ]
tunnel. A high-speed video camera was used to visual-
ize the particle entrainment. However, the pickup velocity
was defined in a qualitative manner by extrapolating the 1
carried weight of particles as a function of air velocity. Introduction
Comparison between the saltation velocity in dilute phase
Operation of the pneumatic conveying systems is mainly
flow and the pickup velocity, both measured in the hori-
influenced by several important parameters [1,2] such as
zontal wind tunnel, provide an insight and guideline how
the minimum operating air velocity in a pneumatic con- • Shape and size of particles
veying line could be defined and determined. • Cross-section dimension of duct
• Type of flow in the pipe
Keywords Pickup, Saltation, Pneumatic conveying, • Porosity of the layer of particles
Particle arrangements • Drag coefficient
• Density of gas
• Pressure in pipe
Nomenclature • Particle-particle friction
• Way of particle feeding
D pipe inside diameter [m] • Operating gas velocity
dp particle diameter [m] • Particle size distribution
Froude number F rp = √ p
U
Frp
gdp
• Thickness of laminar sublayer
g acceleration due to gravity [m/s2 ] • Velocity profile
• Cohesion
• Density of solids
Received: 20 July 2003
• Pipe roughness
M. Hubert (&) • Adhesion
205 Particle Science and Technology, • Particle-wall friction
University of Florida, • Shape of the layer of particles
PO Box 116135, Gainesville, • Initial arrangement of particles
FL 32611-6135, USA
e-mail: mhubert@erc.ufl.edu
The operating air velocity influences the system perfor-
Tel.: (352)846-1194 mance and its economic efficiency to the largest extent.
Fax: (352)846-1196 Prediction of the minimum conveying velocity is closely
related to the understanding of the saltation and pickup
H. Kalman mechanisms. Pickup differs from saltation in relation to
Department of Mechanical Engineering, the initial position of particles. For pickup mechanism
Ben-Gurion University of the Negev, particles initially lie at rest on the bottom of the pipe
P.O.Box 653, Beer-Sheva
whereas for saltation particles are initially suspended.
84105, Israel
Pickup velocity has been defined as the fluid velocity
e-mail: hkalman@bgumail.bgu.ac.il
Tel.: +972-8-6477113,
required to initiate sliding, rolling and suspension of parti-
Fax: +972-8-6472811 cles [3]. However, the most general definition is as follows:
The velocity required to resuspend (blowing away) a par-
Authors would like to thank Mr. Rhye Hamey and Mr. Milorad ticle initially at rest on the bottom of pipe is the pickup
Djomlija for their valuable comments and proofreading. velocity.
160

Fig. 1. Scheme of the experimental setup

Several researchers investigated pickup velocity. For (a) particle on the bottom of the pipe; (b) particle on the
transport of single particle in a gas stream it is possible layer of particles; and (c) the heap of particles that blocks
to generalize that pickup velocity is up to 2 to 2.5 higher various ratios of the tunnel cross sectional area.
than saltation velocity [1, 4]. Cabrejos and Klinzing [2]
found that similar conclusion is also valid for very dilute
gas-solid flow. They also concluded that for higher solids 2
loading ratios the saltation velocity approaches the pick Experimental Setup
up velocity [2]. Noteworthy is the fact that pickup velocity
Experimental setup is depicted in Fig.1. The main part
is assumed to be independent of the solids loading ratio [1].
of the setup is a plexiglass wind tunnel, which consists
Although the pickup and saltation represent two com-
of 10 square ducts. Each duct is 0.6 m long and its inter-
pletely different mechanisms they create a controversy on
nal dimensions are 0.1×0.1 m. Square cross section was
how to define the minimum conveying velocity in pneu-
preferred to circle one because it seems to be more appro-
matic conveying pipelines. Which of the two velocities
priate for studying the fundamental phenomenon of the
should be considered to be the minimum conveying veloc-
pickup and saltation. For this cross-section the width of
ity is very much dependent on material to be conveyed
the initial layer of particles is the same as the wind tunnel
and conveying regime required. If, for instance, partic-
width and does not change with height of the layer.
ulate matter is to be transported in homogenous flow
A compressor was used as a source of air, i.e., positive
then pickup velocity seems to be the appropriate veloc-
conveying system was employed. So-called laminator was
ity. Sticky and very cohesive materials may need to be
embedded between duct 1 and 2 to improve uniformity of
conveyed at pickup velocity too.
velocity profile across the wind tunnel. Average gas veloc-
Cabrejos and Klinzing [2] studied pickup mechanism
ity was measured by a mass flow meters for flow rates bel-
and proposed the following equation to calculate the
low 2500 l/min (accuracy of ±0.1%) and by a thin-plate
pickup velocity.
orifice (estimated accuracy of ±3%) for higher flow rates.
 0.25  0.75
U D ρp The use of the thin-plate orifice was engendered by limi-
gpu = 0.0428Re0.175
p (1) tation of the mass flow meter. Gas velocity in the tunnel
gdp dp ρg
was regulated by globe and needle valves. Particulate sol-
This relationship is valid for 25 < Rep < 5000, 8 < ids were separated from gas and collected in a cyclone. For
(D/dp ) < 1340, and 700 < (ρp /ρg ) < 4240. One can see fine particles the cyclone could be equipped with a sleeve
that the equation includes a few of the earlier mentioned filter on the top.
parameters influencing the pickup velocity. However, the Particles were placed in section 4. Two types of pickup
effect of initial particle condition is not included in the velocities are defined at this stage: pickup velocity from a
equation. Particles were picked up from the layer of par- smooth wall and pickup velocity from a layer (or heap) of
ticles where the surface of the layer was not at the same particles. In case of pickup of particles from the bottom of
height as the bottom of the wind tunnel. That is one of the tunnel (smooth wall) they were placed directly on the
the points in which our experiments differed from Cabrejos bottom plate (Fig. 2a). Average gas velocity at which the
and Klinzing’s experiments. Authors of this article believe particle(s) started rolling is considered to be the pickup
it is important to investigate the influence of initial condi- velocity for this case. In case of pickup of particles from
tion of particles in order to better understand the pickup particle layer the particle layer had to be created. The
mechanism. This study could also provide some insight layer was created by filling the hole in the bottom of the
into the saltation mechanism. wind tunnel (Fig. 2b). Top surface of the layer matched
Influence of initial particle condition on pickup veloc- the bottom surface of the tunnel. In case of pickup from
ity was investigated. Following three initial conditions for a heap of particles (Fig. 2c), an initial heap was built on
the pickup experiments will be considered and examined the bottom of the tunnel.
161

Fig. 2. Schematic representation of the initial particle con-


ditions. a, particles scattered on the bottom of the tunnel,
b, particles arranged in a layer, c, heap of particles

the surface of the layer is rough the particles then started


to bounce due to collisions with other particles in the layer.
Thirdly, the newly entrained particles collided with par-
ticles already carried with the air. All these three steps
represent pickup mechanism as it was observed in the
experimental setup. However, defining the pickup veloc-
ity by visual observations could lead to significant errors.
Therefore, we preferred to measure the pickup velocity
by a qualitative manner, although it requires some de-
gree of extrapolation. By plotting the amount of entrained
particles (weight reduction of the layer) as a function of
operating gas velocity it is possible to determine pickup
velocity. The pickup velocity is determined as intersec-
tion of extrapolated curve passing through the measured
points and abscissa (Fig. 3). The amount of entrained par-
ticles can be easily calculated since the initial and final
amounts of particles in the bottom hole are known. They
are weighed before and after the experiment.
In this way, the problem of identifying the first particle
that was picked up could be overcome. The identification
is based on qualitative measurement rather then on vi-
sual observation, although it is based on extrapolation of
an experimental curve. In order to get the curve, a few
measurements of the layer or heap weight loss as a func-
tion of the air velocity had to be carried out. Obviously,
as more measurements are taken the accuracy improves.
The accuracy of the extrapolation increases if measure-
ments with very small weight losses can be achieved. It
also should be noted that as the weight loss is higher,
the weight measurement accuracy is higher. Therefore,
the duration of the pickup experiment should be selected
appropriately and has to be kept the same for each line
subject to extrapolation (as shown in Fig. 3). Obviously, as
the duration of a pickup experiment is longer, the weight
Fig. 3. Determination of the pickup velocity for layer of of the picked up particles is higher. Using two different
alumina spheres times of pickup measurements result with two different
curves of loss of weight versus the air velocity. However,
the extrapolation of these two lines results with the same
By using high-speed video camera we were able to pickup velocity, as is shown in Fig. 3, since the starting
observe the pickup, i.e., movement of particles once the velocity for pickup is not dependent on the test duration.
pickup was initiated. When particles were picked up from Therefore, any test duration can be selected to pick up
the layer of particles it was possible to observe and distin- significant weight of particles but prevent picking up the
guish three stages particles went through. Initially, par- whole layer. The test duration should also be long enough
ticles started rolling along the layer of particles. Since to enable to neglect the start-up and shut-down periods
162

Table 1. Particle and gas characteristics

Particle Particle density [kg/m3 ] Shape of particles Particle diameter [mm]


1 alumina spheres 1700 spherical 1.6–1.7
2 Glass beads 2500 spherical 0.75–1
3 Sand 2700 irregular 0.71–1
4 Potassium sulfate 2670 irregular 1.8–2.0
5 Limestone 2756 irregular 0.0008–0.0132
6 TiO2 4052 irregular 0.0006–0.0047
7 Talk 3459 irregular 0.0012–0.0081
Gas Gas density [kg/m3 ] Kin. viscosity [m2 /s] Temperature [o K]
air 1.2 1.5 E-5 293.15

Table 2. Experimental and theoretical results for saltation and pickup velocities
Type of particles (according to Table 1) 1 2 3 4 5 6 7
Pickup velocity for the smooth surface [m/s] 1.7 2.26 4.13 3.335 3.53 3.8 2.4
Pickup velocity for the layer of particles [m/s] 8.55 8.78 8.6 9.52 6 – 5.98
Boundary saltation velocity (when particles can be
conveyed over long distances) [m/s] 2.54 2.5 4.28 3.65 – – –

during which particles will not be picked-up since the


velocities are lower. Figure 3 also presents the method by
which the pickup velocity is determined for the flat layer of
particles or heap of particles. It is important to note that
the accuracy of the pick up curve (shape and slope) is not
important, as long as the extrapolation to zero picked up
weight is accurate.
Experiments were carried out with several materials
that are listed in Table 1. The density was measured by a
Helium picnometer. All tests were conducted with a nar-
row range of particle sizes prepared by sieving.

3
Results and Discussion
3.1
Pickup Velocity
Figure 3 shows experimental results of measuring pickup Fig. 4. Comparison between experimental and calculated Fro-
ude numbers
velocity for alumina spheres when the particles were en-
trained from the layer of particles. As it was explained
above the pickup velocity is obtained from extrapolating conditions (smooth surface and particle layer) are pre-
curve, which passes through measured points. Thus, the sented, although Eq. (1) was suggested for the second case
point where the curve intersects the abscissa determines only. As the particles are relatively big (within the limits
magnitude of the pickup velocity for the layer of particles of Eq. 1) the experimental Froude numbers are slightly
initial condition. smaller then those predicted by Cabrejos and Klinzing [2]
Hubert and Kalman [5] measured saltation velocity in (that is d = 0.075 mm and Re = 25). However, for much
dilute system for the same material and in the same exper- smaller particles the experimental Froude number is much
imental setup. The pickup and saltation velocities are larger then the predicted one since eq. (1) is out of validity
summarized in Table 2. Comparing the boundary salta- range.
tion and pickup velocity for initial condition of the layer of
particles one can conclude there is a significant difference.
The ratio between the two velocities varies up to about 3,
which corresponds well to other researcher’s findings. 3.2
Experimentally obtained pickup velocities for the Comparison Between Saltation and Pickup Velocities
smooth surface initial condition are also listed in Table 2. The saltation velocity was defined as such that particles
These are the velocities when loosely scattered particles stop rolling or sliding on the bottom wall of the wind
on the bottom of the wind tunnel started moving or roll- tunnel. Such a definition should be related to the pickup
ing. Figure 4 shows the comparison between the measured velocity measured on the smooth surface. Figure 5 shows
Froude numbers and the calculated ones based on Eq. (1). the comparison between the pickup and saltation veloc-
The measured Froude numbers for both initial particle ity (in fact boundary saltation velocity). The measured
163

and Klinzing’s conceptual model of pickup and saltation


mechanisms in horizontal pneumatic conveying systems.
Figure 6 shows this conceptual model for the case when
particles were picked up from the layer of particles. The
pickup velocity is compared with the boundary salta-
tion velocity. Once the particles are entrained during the
pickup they will keep moving. Particles will keep mov-
ing until gas velocity drops under the boundary saltation
velocity. Pickup curve is shifted rightward from saltation
curve. The ratio of pickup to saltation velocity varies
about value of 3.
For case of smooth surface initial condition the authors
believe there is no significant difference between F rp −Rep
curves for saltation and pickup.

3.3
Pickup Velocity from the Heap of Particles
The shape of the heap depends on the angle of repose
Fig. 5. Comparison between pickup velocity (the smooth sur- of the material. The pickup velocities of several materi-
face condition) and saltation velocity for particles 1–4 (see als were measured in the same way as described earlier
Table 1) (extrapolating the weight loss function) and they are pre-
sented in Fig. 7 as a function of the wind tunnel cross sec-
pickup velocity is slightly smaller than the saltation veloc- tion area that is occupied by the particulate heap. Pickup
ity probably because of the difference between the static velocity for zero cross section area ratio represents the
and the dynamic friction forces. Moreover, if pickup veloc- measured pickup velocity for initial condition of the layer
ity was compared to saltation velocity for short conveying of particles. The results that are shown in Fig. 7 pres-
distances instead of boundary saltation velocity (long con- ent the average gas velocity in free cross section of the
veying distances) [6] the velocities would be virtualy the wind tunnel. This velocity is also called superficial veloc-
same. ity. As expected, the pickup velocity decreases as the heap
Knowing that pickup velocity depends on initial par- of particles occupies a bigger area of the wind tunnel cross
ticle conditions we can incorporate this fact into Cabrejos section. In other words, a smaller superficial velocity is

Fig. 6. Conceptual representation of the saltation and pickup


from the layer of particles (experimental data for alumina
spheres were used in this illustration)
164

needed to pickup particles from the heap as the height of


the heap increases.
Figure 8 reproduce the results shown in Fig. 7, but the
average gas velocity at the clearance between the heap
top and the top wall of the wind tunnel is plotted instead
of the superficial gas velocity. The velocity in Fig. 8 is
called the corrected pickup velocity. It is noticable that
there is difference between the behavior of coarse particles
(solid curves) and fine powders (dashed curves). Figure 9
presents difference between the superficial and corrected
pickup velocity in graphical way. It can also be seen how
the heap of particles eroded as gas velocity increased.
Difference between coarse and fine materials is even
clearer in Fig. 10 in which the corrected pickup velocity
is normalized by the pickup velocity measured for initial
condition of the layer of particles. The pickup velocity
Fig. 7. Superficial pickup velocity for the heaps of particles as increases as the cross sectional areas occupied by parti-
a function of cross section area ratio cles increases, for coarse particles, while it is decreasing
for fine powders. The main reason for that is due to the

Fig. 8. The corrected (at the top of the heap) pickup velocity
as a function of cross section area ratio

Fig. 9. Erosion of the heap of glass spheres (0.15–0.25 mm) as


a function of gas velocity
165

Fig. 10. The normalized corrected pickup velocity as a func-


tion of cross section area ratio

permeability of the materials. For coarse particles, a part particles (particles loosely scattered) at the bottom of the
of the air is flowing between the particles and reduces wind tunnel is almost the same as the boundary saltation
the real velocity at the heap top, and also changing the velocity. In general, it can be said that pickup velocity
pressure distribution in the heap. is about 2.5–3 times higher than saltation velocity. The
pickup velocity for heap of particles depends on the heap
size and behaves differently for coarse particles and fine
4 powders.
Conclusions
Influence of three initial arrangements of particles on
pickup velocity was investigated. A few conclusions can References
be stated. A new method of measuring the pickup veloc- 1. F. Rizk, G.E. Klinzing, R.D. Marcus and L.S. Leung, Pneu-
ity is presented and used successfully. Initial arrangements matic Conveying of Solids, Chapman and Hall (1997)
of particles may be varied and controlled if the proposed 2. F.J. Cabrejos and G.E. Klinzing, Powder Technology 79,
method is used. These are the main advantages offered by 173–186 (1994)
this method. 3. J. Hallow. Chemical Engineering Science 28, (1978)
The pickup velocity of coarse particles for initial con- 4. F. Zenz. I&EC Fundamentals 3(1), 65–75 (1964)
dition of the layer of particles is slightly smaller then pre- 5. M. Hubert and H. Kalman, Determination of Length
dicted by existing empirical correlation, while it is much Dependent Saltation Velocity in Dilute Flows, to be pub-
larger for fine powders. The pickup velocity for individual lished, 2003

You might also like