Gravity Separation of Particulate Solids

in Turbulent Fluid Flow

JU

RGEN TOMAS
QUERY SHEET
Q1 Au: Is bullet list constructed correctly?
Q2 Au: Should this be Table 4. Not otherwise referred to
Gravity Separation of Particulate Solids
in Turbulent Fluid Flow
JU

RGEN TOMAS
5 Mechanical Process Engineering
Otto-von-Guericke-University Magdeburg
Magdeburg, Germany
The separation characteristics of particle settling velocity, size, density, and shape
are introduced, and the equivalent settling condition for laminar, transition, and
10 turbulent flow are explained. A similarity model of particle transport in a turbulent
flow field is briefly discussed. Typical operation principles in separation apparatuses
for counter current and cross-flow separation are presented. The separation
function, cut point, sharpness, and separation stage utilization coefficient are
determined to assess the efficiency for process sequences for multistage turbulent
15 cross-flow separation. Satisfactory to very good results were achieved in the difficult
separation of a partially liberated aggregate consisting of hardened cement paste
rubble with sharpness from 0.66 to 0.94 at separation stage utilization coefficients of
7 to 87%. Specific mass flow rates of 3 to 16 t=(m
2
h) and mass related energy
consumption of 0.2 to 8 kWh=t were obtained.
20 Keywords: Cross-ow gravity separation model, separation efciency, waste
recycling
Introduction
Typical unit operations of particle separation are introduced to explain the operation
principles like current, counter current, and cross-ow patterns with their physically
25 determined separation characteristics, i.e., settling velocity v
s,i,j
, particle size d
i
,
density r
s,j
, and shape. These processes are assessed by the separation function T
i,j
,
or grading efciency curve, as the ratio of product to feed mass increments of the
separation characteristic fractions i, j. The ratio of secant points of this function near
the cut point, the so-called separation sharpness k, indicates directly the process
30 quality; see Schubert (1989, 1996, 2003) for particle separation processes and
Smigerski (1993) and Leschonski (2003) for air classication.
Wet and dry separations have differing advantages and disadvantages. Dry
separation is more cost-effective in terms of its energy requirement, since with wet
separation problems regarding the treatment of process water and its disposal need to
35 be considered, and dry separation is especially suitable for semi-mobile processing
3b2 Version Number : 7.51c/W (Jun 11 2001)
File path : p:Santype/Journals/Taylor&Francis/Upst/22(2)/22(2)-18467/Upst22(2)-18467.3d
Date and Time : 28/4/04 and 16:07
Presented at the 4th International Conference for Conveying and Handling of Particulate Solids,
Budapest, Hungary, May 2730, 2003.
Address correspondence to Ju rgen Tomas, Mechanical Process Engineering, Otto-von-Guericke-
University Magdeburg, P.O. Box 4120, D-39106 Magdeburg, Germany. E-mail: juergen.tomas@vst.
uni-magdeburg.de
Particulate Science and Technology, 22: 119, 2004
Copyright # Taylor & Francis Inc.
ISSN: 0272-6351 print/1548-0046 online
DOI: 10.1080/02726350490457222
1
plants. Wet separation is also useful for the removal of pollutants from contaminated
building waste (Kalck & Werther, 1990). The liberation of building rubble by com-
minution produces a predominantly mineral mixture with a density distribution of
r
s
1.82.7 g=cm
3
(Tomas et al., 1998; Tomas & Grger, 1999). As a result of the
40 relatively narrow density range, the requirements regarding the sharpness of the
process employed for the separation of partially liberated aggregate and concrete-
brick rubble are very high. The separation processes currently employed in the
recycling of building materials remove mainly the lightweight impurities such as
paper, wood, lms, insulating materials, and pieces of plastic (r
s
0.11.2 g=cm
3
)
45 by wet or dry separation. In such cases, the range of unsharpness beyond the cut
point can be dened in the intermediate range (r
s
1.21.8 g=cm
3
), so that equip-
ment with relatively low separation efciency is still adequate for these applications.
Table 1 provides an overview of the state-of-the-art with respect to separation for
recycling building materials and domestic waste (Bilitewski, 1995; Friedrichs &
50 Tomas, 1996; Hanisch et al., 1991; Melchiorre et al., 1994; G. Schubert, 1999;
Tra nkler, 1992).
For the separation of building rubble, classiers are mainly used in practice. The
upward ow or single-stage cross-ow separators commonly employed for dry
separation in an airow separate the material feed according to the respective settling
55 velocities of the different components. The separation behavior is inuenced deci-
sively by the particle size, particle shape, and particle density of the components to
be separated. An air classier can separate the materials, according to one of these
parameters, providing that the inuence of the other two variables is minimized.
The separation of totally or partially liberated aggregate particles in the size
60 range of d 216 mm, therefore, presents a challenge to the separation sharpness
and efciency of the equipment. Tests were carried out in a zigzag channel to
establish whether the high separation sharpness required can be achieved in a
multistage turbulent cross-ow separation apparatus.
Table 1
Selected separation processes for the recycling of building and domestic waste
Wet separation Dry separation
*
Float and sink cleaning (Tra nkler,
1992; Schu tze, 1987)
*
Classification (Greschner, 1994)
*
Hand sorting (Hanisch et al., 1991)
*
Aquamator (Rohr, 1987; Schu tze, 1987)
*
Automatic sorting (Schubert, 1996)
*
Countercurrent and cross-flow
aeroseparation (Schubert, 1999)
*
Upward flow (Kalck & Werther, 1990)
*
Pneumatic table (Gu ldenpfennig &
Lohr, 1995)
*
Jig (Breuer, 1988; Kellerwessel, 1993)
*
Sloping separating belt
(Schubert, 1996)
*
Slinger (Hanisch et al., 1991)
*
Washing drum (Hentzschel, 1990)
*
Eddy current separator
(Schubert, 1994)
2 J. Tomas
Fundamentals of Aeroseparation of Building Rubble
65 The basis for studying a separation process in a uid cross ow is the balance of the
forces of buoyancy, weight, and uid resistance of a particle. With this balance, it is
possible to obtain a correlation between particle size and the quasi-stationary settling
velocity v
s
in a gravitational eld:
v
2
s

2
c
W

r
s
r
f
r
f

V
p
A
p
g 1
70
where A
p
is the side-fed cross-sectional area, c
W
the drag coefcient of the uid ow
pattern around the particles, V
p
the particle volume, and r
f
the uid density. The
density of the solid particles r
s
depends on the inner porosity (up to 95% for
insulating materials) and pore saturation with a liquid.
75 In the separation of particles of varying density r
s
the principle of what is termed
equivalent falling classes, i.e., classes with equal settling velocity, can be for-
mulated as follows. If the particle shape is constant, large and lightweight
particles settle just as fast as small and heavy particles. With d
i
to d
i 1
as the
particle size of the class i to i 1 as well as r
s,S
and r
s,L
, as the particle densities of
80 the heavy fraction (index S) and the lightweight fraction (index L), the following
applies:
v
s
d
i1
; r
s;L
v
s
d
i
; r
s;S
2
Depending on the particle ow patterns v
s
/ d
a
, (Schubert et al. 1986), and with
85 the relationship
c
W
/ Re
12a
3
3
and the particle Reynolds number Re v
s
d r
f
=Z
f
, then for the size fractions the
following holds:
d
i1
d
i

r
s;S
r
f
r
s;L
r
f

_ _
a1
3a
4
90
and for the settling velocity v
s
related to cut point v
sT
(Table 2):
v
s
v
sT

d
d
T
_ _
a

r
s
r
f
r
s;T
r
f

_ _
a1
3
5
Table 2
Equivalent-falling condition dependent on the particle flow-around pattern
Exponent a
a 1
3 a
Flow pattern Reynolds number Drag coefcient
2 1=2 Laminar (Stokes) Re <1 c
w
/Re
1
2 <a <1=2 1=2 . . .1 Transition 1 <Re <10
3
c
w
/Re
1 . . .0
1=2 1 Turbulent (Newton) 10
3
<Re <(24)10
5
c
w
/Re
0
Gravity Separation of Particulate Solids in Turbulent Fluid Flow 3
First, the counter current particle separation in a eld of gravity has been described
95 by Molerus (1967), Molerus and Hoffmann (1969), Schubert et al. (1986), Schubert
(1989), and Bhme (1989) and centrifugal forces by Husemann (1990) and Husemann
and May (1998). Numerical simulations on basis of computational uid dynamics
(CFD) models were performed by Heiskanen (2002). But the models of counter
current separations formulated by Gorzitzke (1982) and Bhme (1989) to date have
100 proven unsuitable or too complex (Senden, 1979) for the evaluation of the multistage
cross-ow separation in order to describe the process efciency. The separation
tapping model of a turbulent cross-ow hydroclassication derived by Schubert and
Neee (1973), Neee (1969, 1978), and Neee and Schubert (1975, 1976, 1977, 1977)
was used to develop a model for multistage turbulent cross-ow aeroseparation
105 (Tomas & Grger, 1999, 2000). The good separation characteristics of a zigzag
separator, invented by Stebbins (1930), are based on the series arrangement of
several such stages (Kaiser, 1963; Tomas & Grger, 1999, 2000, 2001).
The separation efciency is inuenced by the turbulent ow pattern in the
process chambers (Schubert & Neee, 1973; Neee, 1969, 1978; Neee & Schubert,
110 1975, 1976, 1977). This effective dimensionless turbulence variable can be roughly
described by the averaged apparatus Reynolds number Re ub=n 10
3
10
6
, degree
or intensity of turbulence Tu 0.010.15, and turbulent diffusion coefcient of uid
D
t
12000 cm
2
=s (Table 3). Because of the comparably high degree of turbulence in
the zigzag channel the turbulent particle diffusion or eddy diffusion coefcient of
115 uid D
t,s
%D
t
%(45 cm)
2
=s(63 cm)
2
=s measures high also. Additionally, the
amounts in brackets can be physically correctly interpreted as the local particle
position shift squared, averaged, and related to a time increment, i.e., a second
statistical momentum of particle concentration distribution.
The ratio of convective to diffusive transport expressed by the Bodenstein
120 number Bo ub=D
t
110
3
is large. Hence the particle separation may be described
by a model for random or turbulent mass transport with following prerequisites:
The particle hold up probability distributions (concentration per number c
n,i,j
)
versus height y are independent of each other, i.e., for every particle size fraction i as
well as density fraction j the one-dimensional Fokker-Planck-Equation is assumed to
125 be valid:
Q1
@c
n;i;j
@t
v
s;i:j
_ _

1
1!

@c
n;i;j
@y
D
t;s

1
2!

@
2
c
n;i;j
@y
2
6
For homogeneous eld of turbulence in the process chamber the turbulent dif-
fusion coefcient D
t
%D
t,s
and the particle diffusion coefcient are assumed to be
130 equivalent. The turbulence intensication by the free turbulent particle ow-around
pattern is larger than damping by particle concentration caused by inelastic particle-
particle or particle-wall collisions:
L

u
02
x
_
% const: D
t
7
The macrodimension of turbulence (diameter of largest eddies d
W,max
L=2)
135 depends on the characteristic dimension of a turbulence generating tool, i.e., the
width of process channel L/b. The root-mean-square of turbulent ow rate
uctuations across the principal ow direction is proportional to the eddy
circumferential speed u
f
and to the averaged uid ow rate

u
02
x
_
/ u
j
/ uu.
4 J. Tomas
The particle size is small compared with the channel width d <0.1L<b. The particle
140 size should also be small compared with the microdimension of turbulence (d
W,min
diameter of smallest eddies with circular laminar ow) d<d
W,min
%10l
D
with the
Kolmogorov dimension of microturbulence l
D
n
3
=e
0;25
.
For steady-state condition @ c
n,i,j
= @ t 0 (at bottom y 0, c
n,i,j
c
n,0,i,j
) an
exponential particle number concentration distribution versus height h is obtained
145 from Equation (6), (Molerus, 1967):
c
n;i;j
c
n;0;i;j
exp
v
s;i;j
D
t;s
h
_ _
8
This function is the basis of models that describe the stochastic process
efciency. The so-called separation function or grade efciency curve is used to
150 assess the accuracy of gravity separation of any narrow particle size fraction Dd
i
, as
shown in Figure 1. This gure shows the incremental probability distribution of
particles that are discharged as heavy product. As limits for this S-shaped function,
a step function indicates the ideal or perfect separation and a horizontal line
characterizes the undesired feed mass splitting without any effect related to the
155 separation property, i.e., particle density r
s
(or particle size d in terms of classi-
cation). The difference between the step function and the S-curve characterizes the
amount of misplaced product. Hence, the steeper the S-function, the better is the
separation effectiveness, i.e., sharpness. Generally, this sharpness can be evaluated as
k 0.30.6 being sufcient, k 0.60.8 being good, and k 0.80.9 being very good,
160 which is shown in Figure 1 (Rumpf, 1975).
When one balances the particle ow in one cross-ow element or stage, i.e., a
single channel bend, of the zigzag channel and takes into account the turbulent
particle ow-around pattern in the z
L
overow separation stages or z
S
in the
underow, respectively, the normalized separation function is as follows (
_
VV
L
,
_
VV
S
165 total volume ow rates of lightweight and heavy particles; Tomas & Grger, 2000):
T
z
L
;z
S
r
s;j
r
f
r
s;T
r
f
_ _

dconst:

1
1
_
VV
L
_
VV
S
_ _
1

r
s;j
r
f
r
s;T
r
f
_ 9
This fractional grade function (Equation (9)) corresponds to a probability
distribution of a class j of the measurable density of the porous particles r
s,j
(the pore
170 space in hardened cement paste measures around 2030%) being discharged in the
heavy fraction S. In his case, the cut point (average separation density) is dened
with T
zL,zS
(r
s,T
) 0.5 probability. For equal fractional grade efciencies of the z
L
lightweight fraction separation stages and z
S
heavy fraction separation stages, the
component mass balance of all stages (apparatus elements) returns the total
175 separation probability (feed index A):
T
tot;j

_ mm
S;j
_ mm
A;j
R
m;S

q
S
r
s

q
A
r
s

1
1
1T
z
L;j

z
L
T
z
S;j

z
S
10
Gravity Separation of Particulate Solids in Turbulent Fluid Flow 5
T
a
b
l
e
3
S
e
p
a
r
a
t
o
r
s
w
i
t
h
t
u
r
b
u
l
e
n
t
f
l
u
i
d
f
l
o
w
(
S
c
h
u
b
e
r
t
e
t
a
l
.
,
1
9
8
6
)
,
r
e
-
c
a
l
c
u
l
a
t
e
d
f
r
o
m
s
e
p
a
r
a
t
i
o
n
r
e
s
u
l
t
s
C
r
o
s
s
-

o
w
s
e
p
a
r
a
t
i
o
n
a
p
p
a
r
a
t
u
s
A
p
p
a
r
a
t
u
s
R
e
y
n
o
l
d
s
n
o
.
R
e

u
D
=
n
D
e
g
r
e
e
o
f
t
u
r
b
u
l
e
n
c
e
T
u

u
0
2
_
=
u
T
u
r
b
u
l
e
n
t
d
i
f
f
u
s
i
o
n
c
o
e
f

c
i
e
n
t
(
c
m
)
2
=
s
D
t

u
0
2
_
B
o
d
e
n
s
t
e
i
n
n
u
m
b
e
r
B
o
s

v
L
=
D
t
;
s
S
c
r
e
w
c
l
a
s
s
i
f
i
e
r
n
S
D
2S
n
R
e
c
r
i
t
%
1
0
4
1
0
4

1
0
5
0
.
0
5

0
.
1
5
0
;
0
1
4
n
S
D
2S

0
;
4
8
_
VV
F
B
5

5
0
o
r
%
(
2
)
2

(
7
)
2
n
S
D
2S
D
t
%
1
0
0
R
a
k
e
c
l
a
s
s
i
f
i
e
r
n
R
L
2R
n
1
0
4

1
0
4

0
;
3
1
n
R
L
2R

0
;
4
8
_
VV
F
B
3
0

1
0
0
o
r
%
(
5
.
5
)
2

(
1
0
)
2
n
R
L
2R
D
t
%
1
:
5

3
1
.
3
C
y
c
l
o
n
e
s
u
D
C
n
R
e
c
r
i
t
%
1
0
3
1
0
5

1
0
6
0
.
0
1

0
.
0
5
%
0
.
1
a
t
i
n
p
u
t
H
y
d
r
o
c
y
c
l
o
n
e
:
8

1
0

4
u
D
C
A
e
r
o
c
y
c
l
o
n
e
:
0
:
0
0
3
5
u
D
C
1

2
0
o
r
%
(
1
)
2

(
4
,
5
)
2
u
D
C
D
t
%
1
0
3
6
1
.
4
Z
i
g
z
a
g
s
e
p
a
r
a
t
o
r
a
)
u
b
n
1
0
4

1
0
5
T
u
%
D
t
u
b
%
0
.
1
1

0
.
1
3
(
0
.
1
1

0
.
1
3
)
u
b
2
0
0
0

4
0
0
0
o
r
%
(
4
5
)
2

(
6
3
)
2
u
b
D
t
%
1

1
5
2
.
C
o
u
n
t
e
r
c
u
r
r
e
n
t
c
l
a
s
s
i
f
i
e
r
u
D
n
1
0
3

1
0
6

0
.
0
2
u
D
2
0
0

2
0
0
0
o
r
%
(
1
4
)
2

(
4
5
)
2
0
.
5

5
0
a
S
c
h
u
b
e
t
t
e
t
a
l
.
,
1
9
8
6
.
7
For a symmetrical separation with the same number of stages in the lightweight and
heavy particle ow (z
L
z
S
z), the total separation function can be simplied with
180 Equations (9) and (10) to:
T
tot;j

1
1
_
VV
L
_
VV
S
_ _
1

r
s;j
r
f
r
s;T
r
f
_ _ _
z
11
An increasing number of stages z results in a steeper rising separation function
characterized by a essentially higher sharpness (Figure 2). Hence, the slope of the
185 separation function can be characterized by an elegant analytical formulation of the
overall separation sharpness k
tot
for the ratio of secant points r
s,25
=r
s,75
of 25% and
75% separation probability:
k
tot

r
s;25
r
s;75

z ln
_
VV
L
=
_
VV
S
_ _
ln 3
z ln
_
VV
L
=
_
VV
S
_ _
ln 3
_ _
2
1 12
Separation in the turbulent particle ow pattern can be achieved with appropriate
190 separation sharpness only if the separator has a comparably high number of
separation stages and if a sufciently high ratio between the lightweight and heavy
material volume ow rates
_
VV
L
=
_
VV
S
can be maintained. This is commensurate with
practical experience gained with classication (Kaiser, 1963). Because of this fact, the
multistage separation model is generalized for a wide range of ow patterns
195 concerning the characteristic particle settling velocity (Table 2).
Q2
With the effective total number of separation stages n
e
(the feed stage both in the
overow and the underow is included in this number),
n
e
2 z
e
1 13
Figure 1. Assessment of gravity separation of a particle size fraction Dd
i
.
8 J. Tomas
An additional degree of freedom z z
e
is obtained, which, on the one hand, can
200 be used to t the measured values to a physically valid separation function (Equation
(11)) especially with regard to their S-shape. On the other hand, the so-called
separation stage utilization coefcient represents an additional parameter to assess
separation efciency in the case of small density differences:
Z
T
n
e
=n 14
205 Setup of the Test Rig
For the separation of mineral materials, a test rig, consisting of a zigzag channel
measuring 173 200 mm
2
, a feed unit, a fan, a cyclone, and a cloth lter was set up
as shown in Figure 3. The separation process can be observed through the glass side
walls of the channel. The mass ow rates of the feed materials and the separation
210 products are determined by means of weight cells. In addition, during the tests, the
air volume ow rate, the average channel velocity, the pressure drop versus the
zigzag apparatus, the pressure drop versus the lter, and the temperature and relative
humidity at signicant points can be measured (Friedrichs & Tomas, 1996).
Zigzag classiers are usually classed as counter current classiers (Bhme, 1989).
215 The separation process in the zigzag channel can also be understood, however, as a
series arrangement of cross-ow separation stages. In each stage, so-called vortex
rolls are formed, to which one fractional grade can be assigned in the ideal case
(Senden, 1979).
Figure 2. Inuence of stage number z on total separation function T
tot
of normalized particle
density r
s
=r
s,T
of a multistage cross-ow (zigzag) apparatus.
Gravity Separation of Particulate Solids in Turbulent Fluid Flow 9
Usually, the feed material is added to the separation process at the center
220 relative to the number of stages in the zigzag channel. In each stage of the apparatus,
separation into lightweight and heavy fractions takes place. The heavy particles slide
along the downward sloping channel walls and the lightweight particles are swept up
with the airow at the two upward sloping channel walls. At the bends of the zigzag
channel, these two currents cross the channel so that a cross-ow separation takes
225 place. The good separation characteristics of a zigzag separator are based on the
series arrangement of several such stages (Kaiser, 1963; Senden, 1979; Gorzitzke,
1982; Bhme, 1989; Friedrichs & Tomas, 1996; Tomas & Friedrichs, 1997; Tomas
et al., 1999; Tomas & Grger, 1999, 2000, 2001).
Assessing the Separation Model for Air Classication
230 As part of preliminary studies, the separation model in Equation (11) was applied for
the classication of glass beads, sand=splits, and gravel (Table 5). In Figures 4 and 5,
the measured values for the four separation experiments with the cut particle size
d
T
2.1, 4.6, 5.4, and 6.7 mm are shown. Despite reduction by particle shape impact,
the quasi-stationary settling velocity of spheres v
sT
at this cut point is higher than
235 the averaged channel airow rate u, which is a characteristic of the predominant
cross-ow separation principle.
Figure 3. Setup of the test rig.
10 J. Tomas
T
a
b
l
e
4
A
s
s
e
s
s
m
e
n
t
c
h
a
r
a
c
t
e
r
i
s
t
i
c
s
f
o
r
m
u
l
t
i
s
t
a
g
e
c
r
o
s
s
-
f
l
o
w
s
e
p
a
r
a
t
i
o
n
i
n
a
s
y
m
m
e
t
r
i
c
a
l
a
p
p
a
r
a
t
u
s
w
i
t
h
z
o

z
u

z
n
u
m
b
e
r
o
f
s
e
p
a
r
a
t
i
o
n
s
t
a
g
e
s
a
n
d
t
h
e
e
x
p
o
n
e
n
t
a
a
c
c
o
r
d
i
n
g
t
o
T
a
b
l
e
1
S
e
p
a
r
a
t
i
o
n
f
u
n
c
t
i
o
n
T
t
o
t
(
x
=
x
T
)

C
u
t
c
h
a
r
a
c
t
e
r
i
s
t
i
c
x
T

x
5
0
(
T
t
o
t

0
.
5
)

S
e
p
a
r
a
t
i
o
n
s
h
a
r
p
n
e
s
s
k
t
o
t

x
2
5
=
x
7
5

F
l
u
i
d
f
l
o
w
s
e
p
a
r
a
t
i
o
n
x

v
s
1
1

_
VV
o
_
VV
u
_
_
1

v
z

d
;
r
s

v
s
T

d
T
;
r
s
T

_
_

z
v
s
T

r
s
T

r
f

V
P
;
T

g
c
W
;
T

r
f

A
P
;
T

l
n
_
VV
o
=
_
VV
u
_
_

l
n
3
z

l
n
_
VV
o
=
_
VV
u
_
_

l
n
3
C
l
a
s
s
i
f
i
c
a
t
i
o
n
x

d
r
s

c
o
n
s
t
.
1
1

_
VV
F
_
VV
G
_
_
1

d
d
T
_
_
a
_
_

z
d
T
%
r
f
3

r
s

g
D
t
;
s
h

l
n
_
VV
F
_
VV
G
_
_
_
_
2
f
o
r
s
p
h
e
r
e
s
c
W

0
.
4
4
z

l
n
_
VV
F
=
_
VV
G
_
_

l
n
3
z

l
n
_
VV
F
=
_
VV
G
_
_

l
n
3
_
_
1a
G
r
a
v
i
t
y
s
e
p
a
r
a
t
i
o
n
x

r
s
d

c
o
n
s
t
.
1
1

_
VV
L
_
V
S
V
S
_
_
1

r
s
;
j

r
f
r
s
;
T

r
f
_
_
a

1
3
_
_

z
r
s
T
%
r
f
3

g
D
t
;
s
h

l
n
_
VV
L
_
VV
S
_
_
_
_
2
f
o
r
s
p
h
e
r
e
s
c
W

0
.
4
4
z

l
n
_
VV
L
=
_
VV
S
_
_

l
n
3
z

l
n
_
VV
L
=
_
VV
S
_
_

l
n
3
_
_
3
a

1
_
VV
o
;
_
VV
F
;
_
VV
L
o
v
e
r

o
w
,

n
e
,
o
r
l
i
g
h
t
w
e
i
g
h
t
p
a
r
t
i
c
l
e
s
u
s
p
e
n
s
i
o
n
v
o
l
u
m
e

o
w
r
a
t
e
s
.
_
VV
u
;
_
VV
G
;
_
VV
S
u
n
d
e
r

o
w
,
c
o
a
r
s
e
,
o
r
h
e
a
v
y
p
a
r
t
i
c
l
e
s
u
s
p
e
n
s
i
o
n
v
o
l
u
m
e

o
w
r
a
t
e
s
.
a

2
l
a
m
i
n
a
r
(
S
t
o
k
e
s
)
,
a

0
.
5
t
u
r
b
u
l
e
n
t
(
N
e
w
t
o
n
)

o
w
p
a
t
t
e
r
n
a
c
c
o
r
d
i
n
g
t
o
v
s
/
d
a
(
S
c
h
u
b
e
r
t
e
t
a
l
.
,
1
9
8
6
)
.
11
Table 5
Comparison of classification results with the separation model according to
Equations (11), (12), (13), and (14)
According to Figure 5
According
to Figure 4
m
Sand=Split
*
Split
r
Split=gravel
Particle fraction d
u,i
- d
o,i
, mm 216 0.84 1.616 316
Channel velocity u, m=s 13 7.5 8 10
Air volume flow rate
_
VV
g
, m
3
=s 0.52 0.3 0.32 0.38
Particle settling velocity v
sT
(d
T
), m=s 18.2 11.5 15.6 18.7
Mass flow rate _ mm
s
, t=h 0.12 0.34 0.12 0.16
Specific mass flow rate _ mm
s;A
,
t=(m
2
h)
3.0 8.5 3 4
Particle concentration m
s,g
, g=kg 82 262 82 94
Cut size d
T
, mm 5.4 2.1 4.6 6.6
Separation sharpness k 0.89 0.75 0.7 0.7
n7, effective separation stages n
e
5.8 1.8 1.2 1.2
Utilization of separation
stages Z
T
, %
83 26 17 17
Pressure drop Dp
ZZ
, Pa 440 440 440 700
Specific energy consumption W
m,ZZ
,
kWh=t
1.25 0.39 1.25 1.72
Figure 4. Classication of log normal distributed glass beads and comparison with the
separation model according to Equation (11), Table 5.
12 J. Tomas
The best efciency is obtained for the very sharp separation of ideal-shaped glass
spheres with a steep curve in Figure 4. For the purpose of comparison, the corre-
sponding model curves with tted, i.e., effective, stage numbers z
L
z
S
z
e
1.4,
240 1.1, and 1.1 are also plotted for splits. With this additional degree of freedom z, the
experimental separation processes with the S-shaped curves typical of air classi-
cation (r
s
const.) can be reproduced very well. From these, separation efciencies
of k 0.89 for glass beads and 0.70.75 for split is determined, which can be con-
sidered good (range k 0.60.8). However, for this classication, effective stage
245 numbers in the range n
e
1.21.8 results. This means that only seven separation
stages of the apparatus are utilized satisfactorily, i.e., Z
T
n
e
=n1726%.
Remarkable in technical terms as well are the mass ow rates of 38.5 t=(m
2
h)
related to the apparatus cross-sectional area and the mass-related energy con-
sumption of only 0.41.7 kWh=t, as summarized in Table 5.
250 Results of Gravity Aeroseparation
Tests on the separation of concrete-brick mixtures were carried out. Figure 6 shows
the reproducibility of the separation efciency on the basis of the results of four tests
with a 15-stage unit, all conducted under identical conditions (d 812 mm). It can
be seen that the separation results demonstrate considerable scatter in comparison
255 with the classication according to Figure 5. The plotted model curves limit the
range of separation efciency from k 0.670.91 for n
e
17 (Table 6).
From comparison of the separation experiments with narrowly fractionated
concrete-brick mixtures d810 mm (Figure 7), it can be concluded that the scatter
is caused by the inuence of the particle size. For other tests, Figure 7 shows that the
260 separation model Equation (6) can be tted very well to the measurement results for
narrow particle size ranges. The sharpness of this separation with the seven-stage
Figure 5. Comparison of the classication results with the separation model according to
Equation (11), Table 5.
Gravity Separation of Particulate Solids in Turbulent Fluid Flow 13
Figure 6. Results for the separation of concrete-brick rubble, d 812 mm, n15, Table 6.
Table 6
Assessment of the gravity separation results
Concrete-brick
rubble
Concrete,
rubber
Figure 8
Particle fraction d
u,i
- d
o,i
, mm
812
Figure 6
810
Figure 7
45 56.3
Channel velocity u, m=s 14 12.5 8.5 8.5
Air volume flow rate
_
VV
g
, m
3
=s 0.56 0.51 0.35 0.35
Particle settling velocity v
sT
(d
T
), m=s 20.3 21.7 14.3 14.9
Mass flow rate _ mm
s
, t=h 0.12 0.15 0.15 0.63
Specific mass flow rate _ mm
s;A
, t=(m
2
h) 3.0 3.7 3.7 15.8
Particle concentration m
s,g
, g=kg 50 68 98 417
Cut density r
s,T
, g=cm
3
2.1 2.4 2.1 1.8
Separation sharpness k 0.70.9 0.86 0.80 0.78
Effective separation stages n
e
17 3.8 3 3.4
Utilization of separation stages Z
T
, % 747 54 43 49
Pressure drop Dp
ZZ
, Pa 1600 815 350 350
Specific energy consumption W
m,ZZ
,
kWh=t
8.0 2.75 0.83 0.19
14 J. Tomas
unit can be rated as very good with k 0.86. The utilization of the seven apparatus
stages Z
T
54% at an effective number of stages of n
e
3.8 is considered satisfactory
to good (Table 6).
265 On this basis, the layout for the separation of a light fraction, i.e., rubber
granulate r
s
%1.0 g=cm
3
, is relatively non problematic (Figure 8). Because of high
particle concentration m
s,g
417 g=kg, a comparably small specic energy con-
sumption W
m,ZZ
0.19 kWh=t is generally obtained for good separation efciency
(Table 6). Satisfactory to very good results were also achieved in the difcult
270 separation of a partially liberated aggregate consisting of hardened cement paste
rubble, k 0.660.94 at utilization coefcients of Z
T
787%. Generally, specic
Figure 7. Separation of concrete-brick rubble, d 810 mm, n 7, Table 6.
Figure 8. Separation of concrete-brick rubble with rubber granulate,
*
d 45 mm, md
56.3 mm, Table 6.
Gravity Separation of Particulate Solids in Turbulent Fluid Flow 15
mass ow rates of 3 to 16 t=(m
2
h) and mass-related energy consumption of 0.2 to
8 kWh=t were obtained.
Conclusions
275 To assess the efciency of multistage turbulent cross-ow separation, the separation
functions were derived and compared with test results. With the effective total
number of separation stages an additional degree of freedom is obtained, which can
be used to t measured values to the separation function on a physical basis.
A separation stage utilization coefcient represents an additional parameter to assess
280 the separation efciency in terms of small separation characteristic differences. On
the basis of the well-known separation sharpness, a considerable utilization of
separation stages, and geometrical variability of a zigzag apparatus, it was shown
that this multistage cross-ow separation principle is well suited for the gravity
separation of particulate solids including mineral and waste materials.
285 Nomenclature
A area, m
2
Bo Bodenstein number
c
W
uid drag coefcient
d particle size, mm
290 D diffusion coefcient, m
2
=s
g acceleration due to gravity, m=s
2
h channel height, m
_ mm mass ow rate, t=h
n total number of separation stages
295 q frequency distribution, (kg=m
3
)
1
R recovery
Re Reynolds number
T separation efciency function
Tu degree of turbulence
300 u uid velocity, m=s
v particle velocity, m=s
_
VV volume ow rate, m
3
=h
W work, energy consumption, J
W
m
specic work, kWh=t
305 z fraction number of separation stages
Greek letters
a exponent
Dp pressure drop, Pa
Z utilization coefcient
310 k separation sharpness
m particle concentration, g=kg
x physical separation characteristic
r density, kg=m
3
16 J. Tomas
Subscripts
315 A cross-sectional area related, feed
e effective
f uid
F nes
g gaseous
320 G coarse
i particle size fraction
j particle density fraction
K channel
L lightweight
325 m mass related
p particle
o overow
s solid, settling
S heavy
330 t turbulent
tot total
T cut point, separation
u underow
ZZ zigzag apparatus
335 References
Bilitewski, B. 1995. Vermeidung und Verwertung von Reststoffen in der Bauwirtschaft (Avoid
and reuse of waste of building industry). Beihe. Mull Abfall. 30: 4063.
Bhme, S. 1989. Zur Stromtrennung zerkleinerter metallischer Sekunda rrohstoffe (Flow
separation of grinded metallic scrap). Freib. Forsch. A. 785.
340 Breuer, H. 1988. Operation experiences with Alljig machine to separate organic contaminants
from sand and split. Aufbereit. Tech. 29: 324330.
Friedrichs, J. & J. Tomas. 1996. Aerosortierung von Bauschutt (Aeroseparation of building
rubble). In Arbeits- und Ergebnisbericht des Sonderforschungsbereiches 385 Baustof-
frecycling, ed. by J. Tomas. Magdeburg: Otto-von-Guericke-Universita t. pp. 197250.
345 Gorzitzke, W. 1982. Trockenes Sortieren grober disperser Feststoffe durch Kombination von
Siebung und Windsichtung (Dry separation of coarse particles by combination of sieving
and air separation). Diss. Technische Universita t Clausthal-Zellerfeld.
Greschner, J. 1994. Recycling feasability of building rubble, domestic waste and cohesive solid
mixtures by vibratory screen. Aufbereit. Tech. 35: 1324.
350 Gu ldenpfennig, M. & K. Lhr. 1995. Separation of mixed waste with uniform density by
pneumatic table. Aufbereit. Tech. 36: 314320.
Hanisch, J., H. G., Ja ckel, & M. Eibs. 1991. Processing aspects of building material recycling.
Aufbereit. Tech. 32: 1017.
Heiskanen, K. 2002. Two-phase uid dynamics in mills and classiers. In Proceedings of the
355 10th European Symposium on Comminution, Heidelberg. pp. 123 [On CD].
Hentzschel, W. 1990. Washing and cleaningState of art and application. Aufbereit. Tech. 31:
126130.
Husemann, K. 1990. Modelling of a classication process using a deecting wheel. Aufbereit.
Tech. 31: 359366.
360 Husemann, K. & I. May. 1998. Modelling of dry countercurrent classication. Aufbereit.
Tech. 39: 279286.
Gravity Separation of Particulate Solids in Turbulent Fluid Flow 17
Kaiser, F. 1963. Der Zick-Zack-SichterEin Windsichter nach neuem Prinzip (Zigzag
apparatusan air separator of new principle). Chemie-Ingenieur-Technik. 35: 273282.
Kalck, U. & J. Werther. 1990. Application of countercurrent classiers to separate
365 contaminated excavator sludges. Aufbereit. Tech. 31: 593601.
Kellerwessel, H. 1993. Jigs for recycling purposesFeasability, limits, machines. Aufbereit.
Tech. 34: 521530.
Leschonski, K. 2003. Windsichten (Air classication). In Handbuch der Mechanischen
Verfahrenstechnik, ed. by H. Schubert. Weinheim: Wiley-VCH.
370 Melchiorre, M., M. Gu ldenpfennig, K. Lhr, J. Zu rn. 1994. Teilen und Trennen im Recycling-
Proze (Splitting and separation in recycling). Chemie-Ingenieur-Technik. 66: 661670.
Molerus, O. 1967. Stochastisches Modell der Gleichgewichtssichtung (Stochastic model of
separation equilibrium). Chemie-Ingenieur-Technik. 39: 792796.
Molerus, O. & H. Hoffmann. 1969. Darstellung von Windsichtertrennkurven durch ein
375 stochastisches Modell (Generation of grade efciency curves of air classication by a
stochastic model). Chemie-Ingenieur-Technik. 41: 340344.
Neee, Th. 1969. Der aufbereitungstechnische Klassiervorgang im turbulenten wa ssrigen
Medium (Classication process in turbulent water). Freib. Forsch. H. A. 465: 950.
Neee, Th. 1978. Modellierung der turbulenten Querstromhydroklassierung (Modeling of
380 turbulent cross-ow hydroclasssication). Diss. B. Bergakademie Freiberg.
Neee, Th. & H. Schubert. 1975, 1976, 1977, 1977. Modellierung und verfahrenstechnische
Dimensionierung der turbulenten Querstromklassierung (Modeling and process design
of turbulent cross-ow classsication). Part I: Chemi. Tech. 27: 529533; Part II:
Chemi. Tech. 28: 8083; Part III: Chemi. Tech. 28: 273278; Part IV: Chemi. Tech. 29:
385 1418.
Rohr, W. 1987. Separation and classication developement and results of Aquamator. Auf-
bereit. Tech. 28: 3234.
Rumpf, H. 1975. Mechanische Verfahrenstechnik (Mechanical process engineering). Mu nchen:
Carl-Hanser Verlag.
390 Schubert, G. 1999. Stand und Entwicklungstendenzen bei der Sortierung von Schrotten und
Abfa llen (State of the art, tendencies at separation of scrap and waste). Freib. Forsch. A.
850: 135.
Schubert, H. 1989. Aufbereitung mineralischer Rohstoffe (Mineral processing), Vol. I. Leipzig:
Deutscher Verlag fu r Grundstofndustrie.
395 Schubert, H. 1994. Eddy-current separation, fundamentals, machines, application. Aufbereit.
Tech. 35: 553562.
Schubert, H. 1996. Aufbereitung fester Stoffe (Processing of solids), Vol. II, Sortierprozesse.
Stuttgart: Deutscher Verlag fu r Grundstofndustrie.
Schubert, H. 1999. On the turbulence-controlled microprocesses in otation machines. Int.
400 J. Miner. Process. 56: 257276.
Schubert, H. 2003. Trennprozesse (Separation processes), In Handbuch der Mechanischen
Verfahrenstechnik, ed. by H. Schubert. Weinheim: Wiley-VCH.
Schubert, H. & Th. Neee. 1973. The role of turbulence in wet classication. In Proceedings of
the Tenth International Mineral Processing Congress, London. London: Institution of
405 Mining and Metallurgy. pp. 213239.
Schubert, H., S. Bhme, Th. Neee, & Espig, D. 1986. Classication in turbulent two-phase
ows. Aufbereit. Tech. 27: 295306.
Schu tze, H. J. 1987. Wet processing of building rubble by Aquamator. Aufbereit. Tech. 28:
463469.
410 Senden, M. M. G. 1979. Stochastic model for individual particle behaviour in straight and
zig-zag air classier. Ph. D. diss. Eindhoven University.
Smigerski, H.-J. 1993. Windsichter (Air classiers). Preprints, GVC-Dezembertagung
Feinmahl- und Klassiertechnik, Kln.
Stebbins, A. H. 1930. Air classier. US Patent 1861248.
18 J. Tomas
415 Tomas, J. & J. Friedrichs. 1997. Waste building material separation in a zigzag air apparatus.
Erzmetall. 50: 562571.
Tomas, J. & T. Grger. 1999. Multi-stage turbulent aeroseparation of building rubble. Auf-
bereit. Tech. 40: 379386.
Tomas, J. & T. Grger. 2000. Verfahrenstechnische Bewertung einer mehrstugen Querstrom-
420 Aerosortierung mineralischer Stoffe, (Process assessment of a multi-stage aeroseparation
of minerals). Preprint 1. Otto-von-Guericke-Universita t Magdeburg. pp. 160.
Tomas J. & T. Grger. 2001. Assessment of a multistage gravity separation in turbulent air
ow. In: Handbook of Conveying and Handling of Particulate Solids, ed. by A. Levy and
H. Kalman. Amsterdam: Elsevier. pp. 7385.
425 Tomas, J., S. Ehlers & M. Schreier. 1998. Impact crushing of concrete for liberation
and recycling. In Proceedings of the 9th European Symposium on Comminution, Albi.
pp. 5770.
Tomas, J., M. Schreier, & T. Grger. 1999. Liberation and separation of valuables from
building material waste. In. Global Symposium on Recycling, Waste Treatment and Clean
430 Technology, San Sebastian. pp. 461470.
Tra nkler, J. 1992. Product quality improvement of building material recycling by wet and dry
processing. Aufbereit. Tech. 33: 194202.
Gravity Separation of Particulate Solids in Turbulent Fluid Flow 19