Separator Operation PDF

Download as pdf or txt
Download as pdf or txt
You are on page 1of 21

GRINDING MANUAL

Operation of Separators

Walter Zeller
VA 93/4052/E

Holcim Group Support Ltd / Cement Manufacturing Services

1/21

TABLE OF CONTENTS
1.
2.

3.

4.

INTRODUCTION
GENERAL CACULATING METHODS

3
3

2.1

Definition of a Closed Grinding Circuit

2.2

Basic Equations (Mass Balance)

2.3

Circulating Load

2.3.1

Calculation of Circulating Load

2.4

Separator Efficiency

2.5

Classification (Tromp) Curve

2.6

Features of the Classification Curve

2.6.1

Cut Point

2.6.2

Sharpness of Separation

2.6.3 By-Pass Effect


TEST PROCEDURE

3
3

3.1

Target of Test and Conditions

3.2

Sampling and Duration of Test (Fig. 11, 12, 13)

3.3

Sieve Analysis

3.4
Evaluation of Test Results
PRACTICAL CALCULATION AND EXAMPLE

3
3

4.1

Data of Separator and Mill

4.2

Test

4.3

Sampling and Sieve Analysis

4.4

Evaluation

The copyright for this document and all appendices are reserved by Holcim Group Support Ltd

2/21

1.

INTRODUCTION
The ideal situation in grinding would stipulate that a particle of feed would be discharged
from the grinding circuit as soon as it has been reduced to the required size. Thus, the
grinding forces would be applied only to the oversize particles. However, due to the
Action of tumbling mills, any attempt to reduce all the feed to a finished product in one
step results in costly overgrinding.

To control top size of discharge from the grinding circuit, sizing apparatus such as an air
separator is employed, and the oversize constitutes the circulating load to the mill. In this
manner, particles greater than a specified maximum size are prevented from leaving the
circuit, and particles below the desired size are not recirculated through the mill.

The separation itself has an essential influence on the grinding performance in the mill,
and therefore it is necessary to evaluate characteristic performance figures.

Proper functioning is mainly influenced by:

Separator adjustment and technical condition, e.g. distributor plate speed, number
and position of spin rotor blades, wear on fan and spin blades, air in-leaks etc.
Separator feed, e.g. feed rate, particle size distribution, moisture content, density,
etc.
Separation air, e.g. temperature, density, viscosity, moisture content, etc.

Four levels of representation of separator performance are recognized, in which:


a) Only the total mass of the coarse (or fine) stream is considered relative to the mass
of feed, expressed by the term "yield". This is the simplest form of assessment.
b) Cumulative size distributions of feed, fine and coarse streams are included, leading
to the term "recovery" or "efficiency".
c) The extent to which a feed size class appears in the fine and coarse streams is
calculated for all size classes and leads to the term "classification function" or "Tromp
curve" with its characteristic parameters cut point, sharpness of cut and by-pass.
d) Stochastic errors are taken into account (not dealt with in this paper).

3/21

Fig. 1, 2 & 3

FIG. 1

Separators

Mechanical air separator


Fan

Separator

R
F
FIG. 2

Cyclone air separator (Rotor type)


A

Dedusting
Filter

Fan

Cyclones
Separator

R
FIG. 3

Single pass separator (Rotor type)


A

Filter
Fan

Separator

F
CC-0040E.dsf / Kma 20.11.02

4/21

2.

GENERAL CACULATING METHODS

2.1

Definition of a Closed Grinding Circuit


If the oversize in the mill discharge is recycled to the mill after a classification stage, the
mill is considered to be operating in a closed circuit.
Symbols used for mass balances (Figure 4):
Description

2.2

Mass flow
[t/h]

Fineness
[%] passing

Spec. surface
[cm2/g], Blaine

Mill feed

Separator feed
(=mill discharge)

BA

Fine fraction
(Product)

BF

Coarse fraction
(Tailings)

BR

Basic Equations (Mass Balance)


In steady-state operation, the mass flow of feed is equal to the sum of mass flows of fine
and coarse streams.
A=F+R

eq. (1)

A mass balance can also be formulated for a portion of the streams finer than a certain
particle size x [m] (see Fig. 5).
Let a, f and r denote the percentage finer than size x in the feed, fine and coarse stream.
As long as no comminution takes place in the separator, the amounts of material less
than size x entering and leaving the separator are equal, hence:
A.a = F.f + R.r

eq. (2)

The values of a, f and r are found by a particle size analysis (e.g. sieve analysis, laser
diffraction analysis etc.).
Yet another formulation of the mass balance can be made for a portion which is limited
by a lower and upper particle size x1 and x2 (see Fig. 5).
Let a, f, and r denote the percentage of material finer than size x2 and larger than size x1
in the feed, fine and coarse stream. As long as no comminution takes place in the
separator, the amounts of material between x1 and x2 that enter and leave the separator
are equal, hence:
A. a = F. f + R. r

eq. (3)

The values of a, f and r are found by a particle size analysis.

5/21

FIG. 4

Symbols

2.3

A, a, BA

Separator

R, r, BR

F, f, BF

M
Mill
Basic equations
= F+ R

A
A
A

FIG. 5

= F

a = F

+ R

f + R

Definitions fineness

passing (%)
f
a
f
a

r
0
1

10

100
particle size

1000
( um )

6/21

Circulating Load
The circulating load in closed-circuit grinding is defined as the mass of the separator feed
A divided by the mass of the separator feed A divided by the mass of the fine fraction F,
or:

A
F

u=

eq. (4)

The actual value of the circulating load depends on various factors, such as mill design,
grinding efficiency, product fineness, etc., but as a guide the following values can be
given:

2.3.1

Cement mill

Raw mill

low fineness
high fineness

u = 1.5 to 2
u>2
u = 2 to 2.5

Calculation of Circulating Load


Formula (4) can be used directly if the mass flows of fines and feed (or rejects) are given.
Most mills are equipped with weigh feeders, so M is known, which is equal to F in
steady-state operation.
If no weighing equipment for separator feed or rejects is installed, u must be determined
using particle size analysis data and formula (1), (2).
A = F + R (1)

A.a = F.f + R.r (2)


u=

( f r)
(a r )

u = A/F (4)

eq. (5)

It is recommended to use the u-values calculated by this formula with caution, because
inevitable errors in the determination of the particle size distributions affect the result
considerably.

2.4

Separator Efficiency
Efficiency takes some account of size distribution, to the extent that it is defined as the
recovery of a feed size class (0 to x m)
(x) =
(x) =

F. f
A.a

100 [%]

f .
100 [%]
ua

eq. (6)
eq. (7)

If x is chosen as the maximum particle size present in the feed, (x) is equal to 1/u.
Fig. 6 shows efficiency curves for two different values of u.

7/21

Fig. 6: Circulating Load, Efficiency

2.5

Circulating load

Separator efficiency h

u=

A
F

u=

f-r
a-r

f
a

efficiency (%)
100
90
80
u = 1,5

70
60
50

u = 2,5
40
30
20
10
0
0

10

20

30

40

50

60

70

particle size

( um )

8/21

Classification (Tromp) Curve


The information given by the efficiency curve is not detailed enough because the defined
fractions 0 to x [m] are wide.
If we ask for the recovery of a feed size class with lower limit x1 and upper limit x2 (x1
greater zero) into the coarse stream, we arrive at the Tromp value for the coarse stream.
tr =

R r
100 [%]
A a

eq. (8)

1
r
(1 ) 100 [%]
a
u

eq. (9)

or
tr =

If this ration is obtained for a number of size classes and plotted against x, there results
the classification curve (Fig. 7). It is also called the Tromp curve after the name of the
man to whom it is assigned.

2.6

Features of the Classification Curve

2.6.1

Cut Point
The cut point d50 corresponds to 50 % of the feed passing to thw coarse stream as seen
in Fig. 8. It is therefore that size which has equal probability of passing to either coarse or
fine streams.

2.6.2

Sharpness of Separation
The sharpness of separation is defined as follows
k=

d 75
d 25

eq. (10)

where d75 and d25 denote the sizes with Tromp values of 75 % and 25 % (Fig. 8).
For an ideal separation k would be 1.
2.6.3

By-Pass Effect
Ideally, the Tromp curve is asymptotic to the abscissae at ordinate values of zero and
unity.
In practice, it is often the case that the lower asymptote occurs at ordinate values a'
greater than zero (Fig. 9), i.e. a portion of each size fraction bypasses the classifying
action. Expressed in an other way, part of the feed reports to the coarse stream
independently of its particle size.
Experience has shown that the bypass parameter a' varies with classifier feed rate, and
hence it is difficult to describe a single Tromp curve which is representative of the
classifier.

9/21

Fig. 7, 8 & 9

Tromp Curves

Basic equations
1
r
)
tr=
(1 u
a

100

FIG. 7
% into fines

% into coarses

0 (%)

100 (%)
fines

coarses

63 (%)

37 (%)

100 (%)

0 (%)
0

FIG. 8

d
sharpness of cut k =

particle size

( um )

d
50 75
particle size

( um )

particle size

( um )

75
25

1,00
0,75

cut point

0,50
0,25
0

25

FIG. 9
bypass a'

a'
0

CC-0042E.dsf / Kma 20.11.02

10/21

with mechanical air separators of Heyd or Sturtevant type it frequently happens that the
bypass value exceeds 50 %, so that the definitions given above for d50 and k are no
more applicable.
It is then common practice to introduce a separator model which explicitely includes the
bypass effect, shown in Fig. 10.
The "reduced" Tromp curve with
tr* =

(t r a ' )
100 [%]
(100 a ' )

allows for determination of d50 and k*.

11/21

Fig. 10

Separator Model

3.

a'

t r =

F
*
tr

* =

* =

*
t r

*
t r

R
A

r - a'

*
*

r
a

*
*

( 100 - a' )

r - a'
( 100 -

a' )

( t r - a' )
( 100 - a' )
CC-0041E.dsf / Kma 20.11.02

12/21

TEST PROCEDURE

3.1

Target of Test and Conditions


The separation result is characterized by weight and fineness of coarse and fine
streams. These values and the operating variables which affect separator performance
(e.g. separator adjustment, feed, separating air etc.) must be recorded during the test.
Since separator performance is strongly dependent on feed rate and feed size
distribution, it is of utmost importance that stable operation of the mill is maintained
during the test, which in this context means constant classifier feed rate.
Variables correlated with this figure are e.g. power consumption of the bucket elevator or
separator rejects mass flow (provided a scale is installed). If no automatic control system
is available, the fresh feed rate to the mill must be regulated such that the power
consumption of the bucket elevator is constant.

3.2

Sampling and Duration of Test (Fig. 11, 12, 13)


The most accurate determination of A, F and R can be done if two of the values can be
measured by scales and the third value is calculated by means of equation (1). In this
case, the duration of the test is about 10 minutes, and samples should be taken in
intervals of 1 to 2 minutes.
Then the individual samples of each sampling point can be homogenized to one
composite sample for A, F and R. Before doing this, it is recommended to check for
outliers due to e.g. instable mill operation by checking the residue on a sieve of medium
size, e.g. 30 m. Samples with extreme values should be eliminated.
Two separators operating in parallel usually do not have equal feed rates, therefore it is
recommended to carry out a separate test for each unit.

3.3

Sieve Analysis
As mentioned earlier, it is necessary to determine the particle size distribution (PSD) of
the three samples A, F and R.
If many plants it is only possible to carry out dry sieving tests down to 30 m. For an
accurate separator judgement the PSD must be known also in the finer range. Below
30 m, wet sieving is applied.
A more advanced method of PSD determination would be the application of a laser
diffraction analyzer.

13/21

Fig. 11, 12 & 13

Sampling Points

FIG. 11 Single pass separator

3.4

FIG. 12 Cyclone air separator

R
F
M

FIG. 13 Mechanical air separator

Sampling points

CC-0038E.dsf / Kma 20.11.02

14/21

Evaluation of Test Results


The further procedure comprises calculation of:

Mass balance
Circulating load
Efficiency, including graph
Tromp values, including graph
Cut point, bypass and sharpness of separation

The report at hand explains the basic methods how to carry out a separator performance
test. These methods enable to define and evaluate the main operational characteristics
of a closed grinding circuit, to compare them with other cases and with ideal values.
The optimum solution for a particular case must be found in several steps, the evaluation
of each single step indicating a better choice of operating variables for the next one. No
universal method is available to find the ideal solution at the writing desk without
experiments.

4.

PRACTICAL CALCULATION AND EXAMPLE

4.1

Data of Separator and Mill


Plant

Date of test

12.06.1991

Separator:
Make, type
Rotor diameter
Rotor height
Rotor speed
Installed motor power
Separation air amount
Fineness regulation
Year of start-up

O&K QS 225 (HES)


2.25 [m]
2.25 [m]
120 [rpm]
220 [kW]
170'000 [m3/h]
rotor speed variation
1991

Mill:
Type
Diameter, length
Installed motor power
Year of start-up

Cement Mill, No. 4


Two compartment, end discharge
4.2 / 15.4 [m]
3' 600 [kW]
1980

Mill feed

Weigh feeder

Control

automatic mill control system

Flowsheet

Fig. 14

15/21

Kiln
Dust

Clinker

Gypsum

Cyclone air separator

Product
CC-0045E.dsf / Kma 21.11.02

Cooler

Fig. 14
Flowsheet

16/21

4.2

Test
During the test a cement has been produced with 93.1 % clinker, 4.6 % gypsum and
2.3 % kiln dust. Since the system was equipped with a scale for the returns, test duration
was only 30 minutes.
Average mill feed rate was 120 t/h, average returns rate 126 t/h.

4.3

Sampling and Sieve Analysis


Sampling points:

shown in Fig. 14, marked

Number of samples:

16 per sampling point

Sieve analysis:

Particles > 200 m dry screening


Particles < 200 m CILAS laser analyzer

The particle size distributions of feed, coarses and fines are listed in Fig. 15, the
corresponding graphs are shown in Fig. 16.

4.4

Evaluation

Circulating Load u
1) Calculation by means of measured mill feed rate and rejects rate
u=
2)

A ( R + F ) (126 + 120)
=
=
= 2.05
F
F
120

Calculation by means of particle size distributions


u=

( f r)
(equation 5)
(a r)

The circulating load for each particle size (e.g. 1,2, 4 m etc.) can now be
calculated and then the mean value of the circulating load can be determined.
A simpler form is to calculate the sum of a, f, and r and then to calculate u by
means of the formula:
u=

( f r )
( a r )

u=

(607.7 285.1)
( 463.1 285.1)

u = 1.81

17/21

Fig. 15 & 16

Size Distribution

FIG. 15
size
( um)

feed

1
2
4
8
16
24
32
48
64
96
200

3,6
6,1
11,0
18,5
31,2
39,7
47,8
60,6
67,7
81,8
95,1

1,9
3,5
6,3
10,1
14,4
16,1
19,1
28,1
37,1
59,5
89,0

4,9
8,3
14,7
25,4
44,9
58,8
71,1
87,0
92,6
100,0
100,0

463,1

285,1

607,7

Summe

coarses
(% passing)

fines

FIG. 16 Particle size distribution


100
90
80
70
60
50
40
30
20
10
0

f
a
r

10
100
particle size ( um )

1000

18/21

Separator Feed A
The mill feed M is known and is equal to F. The separator feed A can now be
calculated by means of equation 4:
u=

A
A = u.F = u.M
F

A = 1.81 * 120 = 217 [t/h]

Coarse Fraction R
R = A - F = 97 t/h

Efficiency
Equation 7 is used for calculating the efficiency.
=

f
100 [%]
au

The calculated values are listed in Fig. 17, and the corresponding graph is shown
in Fig. 18. It is to state that for the calculation the value u = 1.81 has been used.

Tromp Values
Equation 9 is used for calculating the Tromp values.
tr =

1
r
(1 ) 100 [%]
u
a

The calculated values are listed in Fig. 17, and the corresponding graph is shown
in Fig. 18.

Cut point
The cut point was determined graphically, and its value is
d50 = 52 [m]

Sharpness of Separation
Equation 10 was used for the calculation
k=

d 75
= 3.4
d 25

Bypass
The minimum of the Tromp curve is located at 20 m, the bypass value is 9 %.

19/21

Fig. 17 & 18

Efficiency, Tromp Curve

FIG. 17 Calculation
size
( um)

efficiency
(%)

Tromp
(%)

1
2
4
8
16
24
32
48
64
96
200

75,2
75,2
73,9
75,9
79,5
81,9
82,2
79,4
75,9
67,6
58,1

23,6
28,6
25,6
22,7
15,1
8,9
16,6
31,4
56,7
71,1
99,2

FIG. 18 Efficiency, tromp curve


100
90
80
70
60
50
40
30
20
10

a'

d25 d50

d75

0
1

10
particle size( um )

100

20/21

Specific Air Loads


Equation 10 was used for the calculation

feed
217'000[kg / h ]
=
= 1.28 [kg/m3]
air amount 170'000 m 3 / h

guide value: <=2.5 [kg/m3]

fines
120'000[kg / h ]
=
= 0.71 [kg/m3]
air amount 170'000 m 3 / h

guide value: <=0.8 [kg/m3]

Specific Rotor Load


D [m]
rotor diameter
H [m]
rotor height

120
fines[t / h ]
=
= 6.1 t / m 2 / h
D 3.14 H 2.5 x 3.14 2.5

guide value: <= 10 [t/h m2]

Conclusion
The evaluation of the separator performance test shows that separator efficiency is
good, bypass is low and sharpness of separation is sufficient.

21/21

You might also like