Powder Technology 197 (2010) 1724

Contents lists available at ScienceDirect

Powder Technology
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / p ow t e c

Experimental investigation of solid bed depth at the discharge end of rotary kilns
Eckehard Specht a,, Yi-Chun Shi a, Herrmann Woche a, Joern Knabbe b, Uwe Sprinz b
a
b

Institute of Fluid Dynamics and Thermodynamics, Otto von Guericke University of Magdeburg, Universitaetsplatz 2, 39106 Magdeburg, Germany
Claudius Peters Technologies GmbH, Schanzenstrae 40, D-21614 Buxtehude, Germany

a r t i c l e

i n f o

Article history:
Received 7 March 2009
Received in revised form 15 June 2009
Accepted 14 August 2009
Available online 6 September 2009
Keywords:
Rotary kiln
Solid depth
Solid end depth
Initial condition
Out-owing
Particle movement

a b s t r a c t
The solid bed depth at the discharge end of rotary kilns was experimentally investigated for different mass ow
rates, rotational speeds, inclination angles and materials using two lab kilns with sizes of 0.4 m (ID) 5 m (L) and
0.25 m (ID) 6.7 m (L), respectively. The solid depth at the discharge was found to be several more times higher
than the particle diameter. All parameters according to Saeman's model were combined in a newly developed
dimensionless Bed depth number designated as Bd. The lling degree of a solid bed at the discharge can be
correlated with F0 = 1.75Bd0.5 (for an inclination angle between 1 and 4). The range of the researched Bed
depth number (Bd) is suitable for all industrial kilns. These values should be used as the initial condition, which
was still unknown before, to solve the differential equation for the prole of the solid bed depth through the
cylinder.
2009 Published by Elsevier B.V.

1. Introduction
Rotary kilns are widely used in chemical industries, cement
industry, metallurgical industry, waste treatment, drying process,
incineration process etc. For process modeling and simulation the
solid bed depth in the axial direction of the kiln must be known. The
solid bed depth inuences the heat penetration into the bed and the
contact area to the internal wall. These areas are necessary to calculate
the heat transfer among gas, solid and internal wall in rotary kilns. The
solid motion in the transverse view has been deeply investigated by
Mellmann [1]. Different forms of solid motions, such as sliding,
slumping, rolling, and the transition behavior between these motions
are well described. The solid transverse motions are found to
inuence the solid axial motion in the kiln. Furthermore, it is also
found that the change of the bed depth in axial direction inversely
inuences the solid transverse motion [2], as well as the residence
time of the particles in the active layer [3] and the hold-up of solids in
kilns [4]. For the axial bed depth some models exist. These models
require the bed depth at the discharge end as initial condition to solve
the differential equations. However, this actual depth is still unknown
for different operational conditions.
Sullivan et al. [5] rstly carried out experimental work on solid axial
transport and hold-up in kilns in the manufacture of Portland cement
clinker. A model for the steady state transport of a granular solid through

Corresponding author. Tel.: +49 391 6718765; fax: +49 391 6712762.
E-mail address: eckehard.specht@ovgu.de (E. Specht).
0032-5910/$ see front matter 2009 Published by Elsevier B.V.
doi:10.1016/j.powtec.2009.08.024

a kiln drum which was slowly rotated was rst developed by Saeman
[6]. His model is often used to calculate the local depth change h shown
in Fig. 1. This gure also depicts the solid depth variation from the
feeding end till the discharge end of the cylinder. The differential
equation to solve the solid depth variation along the cylinder is given in
Eq. (1)

dh
0:75 tan M
tan
2
2 3 = 2
=

R  h  R 

dx
n
cos ;
s

where M is the mass ow; n is the rotation speed; R is the internal

radius of cylinder; is the inclination angle; s is the solid bulk
density and is the dynamic angle of repose, see Fig. 1. To solve this
differential equation an initial condition or boundary condition is
needed. Because of a lack of experimental data, Saeman set h = 0 at
discharge end x = 0 of the kiln as the initial condition to numerically
solve the equation.
Hogg et al. [7] continued to study axial transport of dry particles
through horizontal rotary drums of both the constricted-end and the
open-end based on the physical model of Saeman [6]. He rstly
considered particles owing out of the kiln. In this case the differential
equation was built up with the change of the lling angle in Eq. (2)
d
3 M tan
=
;
4
dx
4 s n R4 sin

where is the half of the lling angle and B is the fraction of time
spent in bed to total time. For a dynamic angle of repose of 38 and a

18

E. Specht et al. / Powder Technology 197 (2010) 1724

Fig. 1. (a) Solid bed depth in the transverse view and relevant parameters in the kiln; (b) the bed depth change along the cylinder (x = 0 to x = L).

bulk density of 1.46 g/cm3, the value B = 0.51 is recommended. The

simple initial condition for open discharge end = 0 at x = 0 was
used, which is similar to h = 0 at x = 0. For this case he got the
analytical solution
sin 4  8 sin 2 + 12 =

24M x tan
:
s n R4

For practical application the lling degree is more suitable which

can be calculated from the lling angle
F=

2  sin2
:
2

Taking this advantage, the lling degree was approximated using

the term sin4 8sin 2 + 12 from Eq. (3) with
0:64

F0:0735 sin 4  8 sin 2 + 12

which is valid in the range of 0.04 F 0.40. The lling degree changes
with the axial position then results in
M x tan
F = 0:56
s n R4

!0:64
:

where d is the lling angle at the overow given as

d = arccosRd = R

with Rd as the radius of the overow discharge hole.

Afacan and Masliyah [10] studied the ow of granular solid in a
horizontal drum with and without lifters and end constrictions. Their
experimental results were compared with equations from different
authors [6,8,9]. Based on numerical solutions of Saeman's equation,
they estimated the solid end depth to be twice the particle diameter
h0 = 2dp in the drum without end constriction.
Austin et al. [11] developed a comprehensive model to derive the
solid depth prole along the kiln. They recommended an initial
condition, which depends on all variables, such as geometrical and
operational parameters as well as material properties. However, this
condition is still assumed that the bed depth at the discharge end
approaches a minimal value. The actual values of the bed depth at kiln
end are still unknown.
Recently Spurling [12] carried out experimental and theoretical
studies on hold-up of solids based on the theory of Saeman's model in
cylinders with different types of dams and without dam. He solved the
equation with h0 = dp for kilns without a discharge dam. For kilns
with discharge dam h(x = 0) = dp + hdam was used.
On the other hand, for a cylinder without dam Spurling [12] also
explained another alternative boundary condition to solve the Eq. (1),
which was recommended by Saeman [6]. Physically the highest slope
of a solid bed at a boundary axial position xbc near the discharge end

For a rotary kiln with a central overow discharge end, the

equation
"
F = F0 1 +

0:406 tan M x
s n R4 F01:56

#0:64
7

was recommended which is based on the Eq. (6). Here F0 is the lling
degree at the discharge end. Based on experiments this lling degree
was correlated with the mass ow rate
2
F0 = 41 +

!1 = 4 3
M
5Fmin :
3:65

Here Fmin is the minimum lling degree at the discharge given by

the geometry
Fmin =

d  0:5 sin2d
;

10

Fig. 2. The boundary condition solving the solid bed depth in kilns.

E. Specht et al. / Powder Technology 197 (2010) 1724

19

Fig. 3. Schematic structure of lab rotary kiln1 (in [mm]).

cannot be greater than the angle of repose of the granular solids

considering the inclination of the cylinder,

2. Experiments
2.1. Rotary kilns

dh
dx

x = xbc

tan 

11

Combining Eq. (1) and (11), we get

tan  =

h
i
0:75 tan M
tan
2
2 3 = 2

R  hbc  R

n
cos:
s

12

From Eq. (12) the depth hbc is shown in Fig. 2. Assuming a constant
gradient between xbc and the discharge end, we get the axial position
from Eq. (11)

xbc =

hbc
:
tan 

13

Thus, a value for the boundary condition (xbc,hbc) was derived.

However, with this boundary condition the solid depth at x = 0 is still
regarded as zero.
These authors calculated the axial prole of the solid bed depth
with a xed or assumed values for the bed depth the discharge end.
However, this end depth varies with the operational conditions of
rotary kilns. Therefore, in this study the bed depth at the discharge
end was experimentally investigated in details.

Two laboratory kilns were used to carry out the experiments. Kiln1
had an internal diameter of 400 mm and the length of 5 m. The discharge of the kiln1 was designed with a conical dam as shown in Fig. 3.
Although the length of the discharge dam of 90 mm was relatively short,
the diameter of 350 mm was applied for the analysis. The inclination
angle of the kiln was adjusted at 1, 2, 3, 4 and 5 for measurements.
Kiln2 had a uniform internal diameter of 0.25 m without installation
of a dam at discharge end, see Fig. 4. This kiln was inclined at 1 and 2 to
perform experiments.
2.2. Experimental method
All experiments and measurements were carried out under steady
state of solid ow, i.e. M out = M in. No gas ow passed through the
cylinder and a heating process did not exist. The solid depth at
discharge end is so small that accurate and direct measurements were
too difcult. An easy and relatively accurate way was to measure the
bed surface width which could be done with normal measuring tools.
A photo of the measuring area at kiln discharge end is shown in Fig. 5.
Then the solid end depth was calculated with Eq. (14)

h0 = R 

s
 2
L
R2  0 :
2

14

In order to reect the practical situation the measurements were

carried out always under dynamic conditions. The bed surface width

Fig. 4. Schematic structure of lab rotary kiln2 (in [mm]).

20

E. Specht et al. / Powder Technology 197 (2010) 1724

Table 2
Physical property of experimental materials.
Materials

Quartz sand

Clinker

Glass bead

Diameter distribution [mm]

Mean diameter dp [mm]
Bulk density s [kg/m3]
Dynamic angle of repose []
(Measured in kilns)

0.10.4
0.25
1570
32 (at kiln1)

112
4.5
1410
31 (at kiln1)

0.40.84
0.72
1560
21 (at kiln1)
28 (at kiln2)

3. Results and discussion

3.1. Dimensionless description

Fig. 5. Dynamic measuring the bed surface width at discharge end of kiln.

Table 1
Overview of experimental parameters.
Parameters

Kiln1

Kiln2

Mass ow [kg/h]
Rotational speed [rpm]
Kiln inclination angle []

45440
18
1, 2, 3, 4, 5

2590
3.5
1, 2

was measured without stopping the rotation of the kiln and solids
kept owing out. At discharge end the kiln wall was scaled by grids in
order to clearly recognise the width of the out-owing solid bed. Due
to the sensitive uctuation of the bed surface width at the discharge
end during rotation, measurements were performed 45 times. The
mean value was recorded nally.

In the practical operation of rotary kilns, the end depth of the solid
bed must be greater than zero as long as there is a solid ow out of the
rotary kiln. Furthermore, the end depth is also not a xed value
because it must depend on the different conditions. We assume that
the parameters inuencing the end depth are the same as those
inuencing the solid depth prole inside the kilns. Saeman's model [6]
has been proved applicable for a different load in kilns by a lot of
researchers [5,7,8,10,11]. From Eq. (1), it can be seen that a) the
operational parameters: mass ow, rotational speed and cylinder
inclination angle; b) the kiln geometry: radius of kiln; c) the particle
properties: particle size, bulk density and dynamic angle of repose,
determine the solid bed depth h(x) in axial direction
n; ; R; d ; ; :
hx = f M;
p
s

15

Considering the dimension of these parameters Eq. (1) can be

rewritten in the form of Eq. (16)

 3 = 2

dh
h 2
= Bd 1  1 
B;
dx
R

16

2.3. Experimental parameters and materials

with
In order to nd out the end depth of the solid bed, a wide range of
mass ow, rotational speed and kiln inclination angle have been
chosen according to the capability of both rotary kilns and other
equipments. Rotational speed and inclination angle chosen in
experiments were identical with practical industrial situations. An
overview of experimental parameters is listed in Table 1.
More than 100 measurements have been performed in both rotary
kilns with different kinds of materials. The images of particles are shown
in Fig. 6. The physical properties concerned here were practically
measured by us, see Table 2. A glass bead was employed in both kilns.
Remarkably, it was found that it had different dynamic angles in both
kilns. This was caused by different friction factors of the internal wall
materials in both kilns as well as different value of dp/D [13]. Quartz sand
and clinker have no measurement in the kiln2.

B=

tan
:
cos

17

and
Bd =

0:75 M tan
:
n s R 3

18

All expressions dh/dx, Bd, h/R and B are dimensionless. Each unit
has a great advantage to transfer the deduced results into any other
application. B contains both angles of kiln inclination and solid repose
so that it is named as the Angle constant. Bd includes mass ow,
rotational speed, kiln diameters, and particle properties. It dominates

Fig. 6. Real images of experimental materials (unit of scale in [mm]).

E. Specht et al. / Powder Technology 197 (2010) 1724

21

the bed depth change in kilns so that it is designated here as Bed depth
number. Meanwhile, Bd can be physically interpreted as the ratio of
two velocities
Bd = 1:5 tan

va
;
vk

19

where

va =

M
;
s R2

20

is the axial velocity of the solid in a fully lled kiln and

vk = 2Rn

21

is the tangential velocity of the cylinder, which is also the maximum

tangential velocity of the solid near the internal wall. Therefore, the
formulation in Eq. (19) can be also interpreted as the ratio of inertial
force to the centrifugal force for a certain material.
3.2. Correlation with the end depth of the solid bed

Fig. 8. Variation of the ratio of the experimental end depth to the particle diameter for
glass beads with the bed depth number in kiln1 and kiln2.

inuence of the inclination cannot be seen in the range between 1

and 4 in kiln1. Only for 5 the values are again a little bit smaller. In
kiln2 measurements were performed at inclinations 1 and 2, mass
ows between 25 kg/h and 185 kg/h and a rotational speed of
3.5 rpm. The values depend on Bd with the same tendency. However,
the values are signicantly lower than those measured in kiln1. For
the glass beads the end depth is 38 times larger than the particle
diameter. This is caused by its smaller particle diameter.
The sand has the smallest size (see Table 2) among the three
materials. The end depth of the sand was measured at mass ows
between 100 kg/h and 250 kg/h and rotational speeds between 1 rpm
and 8 rpm at 3 inclination angle in kiln1. The values are shown in Fig. 9.
In this case the end depth was found to be 1025 particles.
The initial condition in all existing models was assumed to be one or
two particle diameters, as explained in the beginning. This assumption is
not suitable for very ne materials and especially for industrial big kilns,
where the ratio of the kiln diameter to the particle diameter is much
larger than that in the laboratory kilns. As a consequence, also for clinker
the end depth must be much higher than the particle diameter in
industrial kilns.

Some authors assumed h0 = 2dp[10] or h0 = dp[12] as the initial

conditions, as explained in the beginning. For the purpose of
comparison, the solid end depth from our measurements is scaled
by the particle diameter as h0/dp. This also represents the number of
particles heaped at mid-chord of the kiln discharge end (at the
position of the line EF in Fig. 1). Fig. 7 shows the variation of the
number of clinker particles with respect to Bd in kiln1. The data were
measured in the range of mass ows between 45 and 440 kg/h and
rotational speeds between 1.5 rpm and 7 rpm, and were sorted by the
inclination angle. It can be seen that all data lie closely together. The
inclination angles at 14 didn't strongly inuence the correlation
between h0/dp and Bd, however at 5 h0/dp was a little bit smaller. The
bed depth of clinker varies in the range about 0.62 particle diameter
depending on Bd. For low Bd the end depth was found to be smaller
than one particle diameter. In this case an out-ow of a single particle
occurs. This results in a very low bulk density of the particle ow.
However, the bulk density of the material in the kiln is used in the Bed
depth number. Therefore, the average value of the bed height can be
smaller than one particle diameter. This problem can always occur for
the particles with big size.
The end depth of glass beads was measured in both kilns, see Fig. 8.
In kiln1 the mass ow and rotational speed were adjusted between
60 kg/h and 400 kg/h and 1.5 rpm and 2 rpm, respectively. The ratio
h0/dp sorted by inclination angles increases with Bd as before. An

The end depth is not a proper parameter for the initial condition
because it depends on the particle size and on the kiln diameter.
Therefore, the lling degree will be used as a parameter. The lling
degree is a commonly used parameter for rotary kilns. It can be easily

Fig. 7. Variation of the ratio of the experimental end depth to the particle diameter for
clinker with the bed depth number in kiln1.

Fig. 9. Variation of the ratio of the experimental end depth to the particle diameter for
sand with the bed depth number in kiln1.

3.3. Correlation with the end lling degree

22

E. Specht et al. / Powder Technology 197 (2010) 1724

Fig. 10. Variation of the experimental end lling degree with the bed depth number for
clinker sorted by the inclination angle.

calculated directly from the measured surface bed width L0 and the
kiln radius R
 
L
0 = arcsin 0 ;
R

F0 =

20  sin20
:
2

22

23

The lling degree was correlated with the Bed depth number (Bd)
using the inclination angle as the parameter as before. The results for
clinker are shown in Fig. 10. It is obvious that the values can be tted
with only one curve for = 14. At 5 inclination the values are a
little bit lower designated with the dashed line.
Fig. 11 shows the results for glass beads. All values can be tted
with the same curve as before. For 5 inclination the values are again a
little bit smaller. The measured values in both kilns can be tted with
this curve. As a consequence, the inuence of the kiln radius can be
described with this Bed depth number (Bd).
3.4. Inuence of the materials

Fig. 12. Inuence of the particle property on the BdF0 correlation based on the
experimental results.

at = 1 for clinker in kiln1 and glass beads in both kilns where they
had a small deviation of the dynamic angle of repose because of a
different wall friction. Fig. 13 presents the values at = 3 for all three
materials in kiln1. Fig. 14 shows the values at = 4 for clinker and
glass beads in kiln1. It is obvious that for all cases the values can be
tted with the same curve in each gure. Consequently, the inuence
of the materials can be described also with the Bed depth number (Bd).
According to Eq. (16) the bed depth in the kiln is not only inuenced
by Bd but also by B which was named as the Angle constant. To check if
B inuences the end lling degree, the values of B are given in the
Figs. 1214, These values vary between 0.019 and 0.082. They are mainly
inuenced by tan . Because the dynamic angle of repose lies in the
range between 21 (glass beads) and 32 (sand), the cosines of this angle
changes only between 0.93 and 0.85. As could be seen before in the
gures, the inclination angle has no inuence for values lower than 4.
Therefore, also B has no inuence on the end lling degree.
3.5. Equations for BdF0 correlation
All values of the end lling degree for the three materials are
nally presented together in Fig. 15. For the inclination 14, the
values can be correlated with the power function

The inuence of the material on the end lling degree has not yet
been investigated in details. Therefore, the values of the lling degree
associated to three materials are compared. Fig. 12 presents the values

F0 = 1:75Bd

Fig. 11. Variation of the experimental end lling degree with the bed depth number for
glass beads sorted by the inclination angle.

Fig. 13. Inuence of the particle property on the BdF0 correlation based on the
experimental results.

0:5

= 1-  4-: in %

24

E. Specht et al. / Powder Technology 197 (2010) 1724

23

diameter of 34 m (up to 6 m). Rotational speed varies between

2 rpm and 4 rpm. The values of Bd for clinker lie therewith in the
range of about 0.0080.1. For rotary kilns in ceramic industry typical
value of mass ow is 26 t/d. The kiln diameter is between 0.5 and
1.5 m. Rotational speed varies between 2 and 3 rpm. The values of Bd
lie in the range of 0.0040.08. In metallurgical industry for reduction
treatment or recovery process, the mass ow of the metal oxide is
about 30 t/d in the rotary kilns. The typical kiln diameter and the
rotational speed are 0.44 m and 23 rpm, respectively. The corresponding value of Bd is about 0.080.1. The range of Bd investigated in
our laboratory conditions coincide with the industrial applications.
4. Conclusions

Fig. 14. Inuence of the particle property on the BdF0 correlation based on the
experimental results.

The deviation of the tting curve is within 10%. At 5 inclination

angle another tting function
0:58

F0 = 1:86Bd

= 5- in %

25

is recommended.
The fact, why the values for 5 inclination are signicantly lower
than those for 14 cannot be explained. Tests with inclination at 4.5
or large than 5 with our kilns were not possible. However, kiln
inclinations higher than 4 have no high industrial relevance.
3.6. Transfer to industrial kilns
Both equations are derived from our experimental conditions in
the laboratory. However, some parameters are not comparable with
industrial situations, such as mass ow and kiln size. One purpose of
the dimensional description was to transfer the results into industrial
applications. Therefore we have to prove the availability of the range
of Bd which has been used in the laboratory conditions. For example,
in the cement industry the typical range of the mass ow for rotary
kilns is generally 1000 t/d4000 t/d (up to 12,000 t/d) with a kiln

The end depth of the out-owing material in rotary kilns depends

on the mass ow, kiln diameter, rotational speed, angle of repose of
the material and the inclination angle lower than 5. It changes from
one to several particle diameters. The inuence of all these
parameters can be described using only a one dimensionless number,
the so called Bed depth number (Bd). This number was derived from
Saeman's model for the axial prole of the bed depth. The end bed
depth for different kiln diameters can be described using the end
lling degree by the correlation F0 = 1.75Bd 0.5. The accuracy of this
equation to the experimental values are within 10%. The range of
the investigated Bed depth number covers the range of all processes in
industrial rotary kilns.
Nomenclature
B
Angle factor in Eq. (17). []
Bd
Bed depth number in Eq. (18) []
dp
Particle diameter [L]
F
Filling degree [] or [%] in Eqs. (24) and (25)
F0
Filling degree at discharge end [%]
h
Solid depth in a cylinder [L]
Lbw
Surface width of a solid bed [L]
L0
Surface width of a solid bed at discharge end [L]
M
Mass ow rate [M/T]
n
Rotational speed [1/T]
R
Radius of cylinder [L]
Rd
Radius of discharge orice [L]
va
Nominal axial transport velocity of particles through cylinder
[L/T]
vk
circumferential velocity of cylinder [L/T]
x
Axial distance from discharge end [L]
Greek symbols

Half of the lling angle [rad.]

d
Half of lling angle at the kiln end with the central overow
discharge [rad.]

Dynamic angle of repose [rad.]

Inclination angle of the kiln [rad.]

s
Bulk density of solids [M/L3]

Fraction of time spent locked in bed to total time []

References

Fig. 15. Derivation of the tting equation for the end lling degree F0 of all materials
based on Bd.

[1] J. Mellmann, The transverse motion of solids in rotating cylindersforms of

motion and transition behavior, Powder Technology 118 (2001) 251270.
[2] X.Y. Liu, E. Specht, J. Mellmann, Slumpingrolling transition of granular solids in
rotary kilns, Chemical Engineering Science 60 (2005) 36293636.
[3] X.Y. Liu, E. Specht, O.G. Gonzalez, P. Walzel, Analytical solution for the rolling-mode
granular motion in rotary kilns, Chemical Engineering and Process 45 (2006) 515521.
[4] X.Y. Liu, E. Specht, Mean residence time and holdup of solids in rotary kiln,
Chemical Engineering Science 61 (2006) 51765181.
[5] J.D. Sullivan, C.G. Maier, O.C. Ralson, Passage of solid particles through rotary
cylindrical kilns, U.S. Bureau of Mines Technical Paper, vol. 384, 1927.

24

E. Specht et al. / Powder Technology 197 (2010) 1724

[6] W.C. Saeman, Passage of solids through rotary kilns: factors affecting time of
passage, Chemical Engineering Progress 47 (1951) 508514.
[7] R. Hogg, K. Shoji, L.G. Austin, Axial transport of dry powders in horizontal rotating
cylinders, Powder Technology 9 (1974) 99106.
[8] M. Hehl, H. Kroeger, H. Helmrich, K. Schuegerl, Longitudinal mixing in horizontal
rotary drum reactors, Powder Technology 20 (1) (1978) 2937.
[9] A.Z.M. Abhouzeid, D.W. Fuerstenau, A study of the hold-up in rotary drums with
discharge end constrictions, Powder Technology 25 (1980) 2129.
[10] A. Afacan, J.H. Masliyah, Solids hold-up in rotary drums, Powder Technology 61 (1990)
179184.

[11] L.G. Austin, K. Shoji, R. Hogg, J. Carlson, R.L.C. Flemmer, Flow rates of dry powders
in inclinded rotating cylinders under open-ended discharge conditions, Powder
Technology 20 (1978) 219225.
[12] Spurling, R.J. Granular ow in an inclined rotating cylinder: steady state and transients.
Ph.D. Thesis, Department of Chemical Engineering, University of Cambridge (2000).
[13] X.Y. Liu, E. Specht, J. Mellmann, Experimental study of the lower and upper angles
of repose of granular materials in rotating drums, Powder Technology 154 (2005)
125131.