Fluidization of Bulk Solids

Ing. (grad.)
Manfred Heyde
Flow and exchange
behaviour
in theory and experiment
_



3
The published measuring results for the different
phenomena, that occur when dealing with bulk ma-
terials, were combined by using uniform and in-
termeshing physical basics. The developed algo-
rithms allow for example estimations for the mate-
rial discharge from silos, for the design of pneu-
matic conveying installations, and for the heat
transfer and the catalytic gas phase reaction in flu-
idized beds. These algorithms apply mostly up to
industrial dimensions. In addition in many cases
the boundaries have become clear which in labora-
tory scale experiments should not be exceeded. For
the direct use of the calculation methods in the pro-
cedural practice is the Windows software FLUIDI
with integrated material database intended. It is
written in VISUAL BASIC and runs even under
Windows 7 (32 bit). An English version of the pro-
gram can be downloaded as a ZIP file from the
Internet:
https://skydrive.live.com/#cid=0239E78E2F136DDE&id=23
9E78E2F136DDE%21936
For the discharge from silos becomes clear that
the amount of material is influenced by the ratio of
outlet diameter and particle diameter. This effect is
also noticeable on other occasions by restricted
mobility of bulk solid beds. The calculation ap-
proach using measured values from small-scale ex-
periments is completely confirmed by the results
from large-scale plants.
For the pneumatic conveying is from the opera-
tional data of different industrial conveyor systems
a characteristic field been developed, which in-
cludes in particular the dense phase conveying. As
dimensionless parameters serve the well-known
Reynolds number together with a factor that comes
from a physical model for the turbulent pipe flow.
Short-term contact and the height-dependent seg-
regation determine the size of heat transfer at ex-
changer installations in fluidized beds. In catalytic
gas-phase reactions the segregation effect is re-
sponsible for the decrease of the reaction turnover.

1. Properties of pure substances and mixtures
1.1 Phase boundaries .............................................................. 5
1.2 Thermal equations of state................................................ 7
1.3 Viscosity........................................................................... 8
1.4 Thermal conductivity and heat capacity............................ 9
1.5 Moisture.......................................................................... 10
1.6 Flammable mixtures ....................................................... 11
2. Characteristics and properties of bulk materials
2.1 Particle size .................................................................... 13
2.2 Particle size distribution ................................................. 13
2.3 Void fraction, particle shape and density ....................... 14
2.4 Adhesive forces .............................................................. 15
2.5 Moisture ......................................................................... 15
2.6 Thermal conductivity and heat capacity......................... 16
3. Mechanisms of heat transfer
3.1 Conduction..................................................................... 17
3.2 Convection and short-term contact ................................. 17
3.3 Radiation........................................................................ 19
4. Dimensionless numbers for flow and
Transport processes........................................................... 20
5. Flow of bulk solids
5.1 Material flow in silos .................................................... 21
5.1.1 Silo design............................................................... 21
5.1.2 Discharge rate ......................................................... 22
5.1.3 Measures against outflow restrictions ..................... 23
5.2 Mechanical moving of bulk materials ............................ 23
5.2.1 Heat transfer in thin layer contact devices............... 24
6. Single phase flow through pipes
6.1 Continuity and energy conservation ............................... 26
6.2 Flow pressure loss .......................................................... 26
6.3 Heat transfer................................................................... 28
6.4 Turbulence structure and exchange behaviour................ 28
6.4.1 Pressure loss............................................................ 28
6.4.2 Speed profile ........................................................... 29
6.4.3 Heat transfer............................................................ 29
7. Pneumatic conveying of bulk materials
7.1 Conveying conditions ..................................................... 31
7.2 Pressure loss................................................................... 32
7.2.1 Fine-grain solid, horizontal ..................................... 32
7.2.2 Material acceleration............................................... 33
7.3 Heat transfer at the tube wall.......................................... 34
7.3.1 Conveying vertical upward...................................... 34
7.3.2 Cyclone separator.................................................... 35
8. Flow through packed beds of bulk materials
8.1 Pressure loss................................................................... 37
8.1.1 Flow around a single particle .................................. 37
8.1.2 Flow through bulk materials.................................... 38
8.2 Heat transfer................................................................... 39
8.2.1 Exchange at the particle surface.............................. 39
8.2.2 Exchange at the tube wall........................................ 40
9. Fluidized bed
9.1 Minimum fluidization velocity....................................... 41
9.2 Appearance..................................................................... 42
9.3 Expansion behaviour ...................................................... 42
9.3.1 Homogeneous expansion and change to
inhomogeneous state ............................................... 43
9.3.2 Inhomogeneous expansion....................................... 44
9.3.3 Bubbles rise............................................................. 45
9.4 Material entrainment ...................................................... 46
9.5 Penetration by gas jets.................................................... 47
9.6 Heat transfer on internals ............................................... 47
9.6.1 Influence of bulk solid movement and expansion ... 48
9.6.2 Comparison of measurements and calculations....... 49
9.6.3 Influence of fluidized bed dimensions..................... 52
9.7 Heat and mass transfer between gas and particles.......... 52
9.8 Catalytic gas phase reactions.......................................... 52
9.8.1 Reactor model for reactions of 1. order ................... 53
9.8.2 Application of the reactor model ............................. 53
9.8.2.1 Laboratory scale............................................. 53
9.8.2.2 Semi-technical scale....................................... 55
9.8.2.3 Measurements on large-scale reactor ............. 55
9.9 Operation as thermal dryer ............................................. 56
9.9.1 Heat requirement ..................................................... 56
9.9.2 Influence of the gas distribution plate construction. 56
9.9.3 Multi-stage fluidized beds ....................................... 58
9.9.4 Heating surfaces internals ....................................... 60
9.9.5 Spray fluidized bed.................................................. 60
10. Solids removal by cyclone separators............................65

4
Symbols

A Surface
Ar Archimedes number
a thermal conductivity
c specific heat capacity
c unreacted proportion
c' unreacted proportion in fluidized bed reactor
D Apparatus and pipe diameter
d Diameter
F Force
f Cross sectional area
Fr Froude number
Ga Galilei number
g Acceleration due to gravity
h Height
h Enthalpy
h Enthalpy difference
k Reaction rate constant
L Pipe length
l Length
M Mass
%
M Molar mass
N
r
Number of reaction units
Nu Nusselt number
n,Z Number
Pr Prandtl number
p
D
Vapour pressure
p Pressure
p Pressure loss
Q Heat quantity
q related heat quantity
R Residue
R Radius of curvature
%
R molar gas constant
Re Reynolds number
T absolute temperature
t Time
u Velocity
V Volume
v specific volume
x Moisture content
x Coordinate in the direction of flow
y Coordinate transverse to the flow direction



Heat transfer coefficient
Material transition coefficient
Accommodation coefficient
Diffusion coefficient
Dynamic Viscosity
Angle
Temperature
Temperature difference
mean free path length of gas molecules
thermal conductivity
solid/gas rate
kinematic viscosity
Resistance coefficient
Density
Shear stress
Sphericity
relative humidity
bubbles-flowed proportion
Porosity, void fraction



5
1 Properties of pure substances
and mixtures
1.1 Phase boundaries
In apparatuses and equipment of process engineering
exist the materials to be processed depending on the
prevailing conditions, in solid, liquid or gasous state. If
several substances with the same aggregate state are
situated together in a space, one speaks of single-phase
systems. With two aggregate states side by side, we
have a two-phase systems, and a multiphase systems, if
all three exist side by side.
In chemical engineering, multiphase systems are used,
to bring gaseous and liquid media with each other or
with solids in contact. On the other hand, substances
can be transformed from one phase to another, in order
to separate them from each other. During the thermal
drying, for example, the material moisture is being
vaporized by supply of heat, and thus removed.
Whether a solid at a certain pressure during heating
passes over into the liquid state or directly into the
gaseous state, is dependent on the locations of the
various equilibrium curves. These vapor pressure
curves represent the pressure exerted by a steam in
thermal equilibrium with its liquid or solid phase. The
curve of the vapor pressure p
D
over the liquid phase is
the boiling line, over the
solid phase the
sublimation line. In
addition there is the
fusion line, which
represents the transition
from the solid to the
liquid state.
Fig. 1.1 is a general p
D
-
T-diagram that shows
the courses of boiling
line, fusion line and
sublimation line. The
three curves meet up in a
point that is known as
the triple point. In this
point are solid, liquid
and gaseous phase in
thermodynamical
equilibrium. The boiling
line ends in the so-called
critical point, where the boiling line and the
condensation line meet up each other. At temperatures
higher than the critical temperature, there is no sharply
defined boundary between gas phase and liquid phase.
The sublimation pressure curve that ends at the triple
point, separates the regions of vaporous and solid
phase. Except helium has every substance, that itself
not already decomposes at lower temperatures, a triple
point. Its position is decisive for the sublimation ability
of a substance. Is the vapor pressure at the triple point
larger than 1 bar, sublimates the substance at
atmospheric pressure (simple sublimation). However,
the vapour pressure is usually significantly lower, so
that the simple sublimation is relative rare. For its
technically sublimability an organic substance must
have at not too high temperatures (to avoid
decomposition) below the fusion point a sufficiently
high vapour pressure (over 1.5 mbar).
In the processes of thermal drying and heating, the
sublimation ability of solids is sometimes unpleasantly
noticeable. As long as the partial pressure of the solid
in the surrounding drying gas is lower than the
saturation vapour pressure at the temperature of the
system, the solid passes directly over into the vaporous
state. The vapor de-sublimates subsequently on colder
parts of the system and forms relative solid wall layers,
which can influence the function and safety of the
respective facility. Pipes and apparatus walls must in
such cases be heated in addition.
Mathematical representations for the vapor pressure of
pure substances, are derived from the Clausius-
Clapeyron vapor pressure equation and deliver simple
vapor pressure relationships for small temperature

Fig. 1.1:
In the general state diagram
of a substance meet boiling
pressure, melt pressure and
sublimation printing curve at
a point

Fig. 1.2:
Vapor pressure curves of some solvents in the plot log p
D
against 1/T: a water, b methyl alcohol, c ethyl alcohol, d
n-propyl alcohol, e carbon tetrachloride, f benzene, g
toluene, h o-xylene

6
ranges, in which the evaporation
enthalpies themselves not change
significantly:
log p
T
D
= +
A
B (1-1)
In the plot of log p
D
against 1/T the
vapour pressure curve should form a
straight line. Fig. 1.2 shows the vapor
pressure curves of some solvents in the
temperature range from 0 to 100 C.
The Antoine equation provides better
results for larger temperature ranges:
log p
T
D
=
+
+
A
C
B (1-2)
The following correlation for the saturation vapor
pressure has been found on the basis of an extended
investigation of existing measurement values for a large
number of substances by Patel, Schorr, Sha, and Yaws
[1.1] to be the best:
log log p
T
T T T
D
= + + + + A
B
C D E
2
(1-3)
In many cases, this correlation is valid from the triple
point to the critical point. Table 1.1 indicates the
correlation coefficients A, B, C, D and E, with whose
help the saturation vapour pressure of various solvents
can be calculated in Torr using the temperature T in K.
The validity ranges are also given.
The heat that is required for the vaporization of a
certain quantity of liquid can be estimated after the
Trouton rule. Experience teaches namely, that the
molar entropy change for the evaporation at boiling
point for many substances is 80 to 110 kJ /(kmol K).
With the approach to the critical point,
the heat of vaporization becomes smaller
and disappears at the critical point. The
heat of vaporization h
v
can be
calculated for any temperature T with
good accuracy by the correlation of
Watson [1.2]:
h h
T T
T T
n
v v
kr
kr
=

|
\

|
.
|
1
1
(1-4)
The correlation coefficients h
v1
, T
1

and n were specified by Thakore, Miller
and Yaws [1.3] for many substances.
They are listet in Table 1.2 that also
indicates the critical temperature of the
substances.
The phase transformation heat changes itself
at the triple point abruptly, because
vaporization heat and fusion heat add up
themselves to the sublimation heat. The fusion
enthalpy h
s
can be calculated according to the
equation of Clausius-Clapeyron using the
change of the specific volume from v
s
to v
l
:

( ) h
s
v
l
v
s
T
dp
dT
D
= (1-5)
Liquid A B C D 10
3
E 10
6
Range in C
Water 16.373 -2818.6 -1.6908 -5.7546 4.0073 0 to 347.2
Methyl alcohol -42.629 -1186.2 23.279 -35.082 17.578 -67.4 to 240
Ethyl alcohol -10.967 -2212.6 10.298 -21.061 10.748 -45 to 243
n-Propyl alcohol -338.31 5127.5 148.80 -175.79 74.666 0 to 263.7
Carbon tetrachlorid 50.612 -3135.7 -16.313 7.8036 - -69.7 to 283.2
Benzene 51.204 -3245.7 -16.403 7.54 - 7.6 to 289.4
Toluene 115.21 -4918.1 -43.467 38.548 -13.496 -60 to 320.6
Table 1.1
Correlation coefficients to calculate the vapour pressure curve of some
solvents according to Eq. (1-3)
Liquid h
v1
T
1
T
kr
n Range
kcal/kg C C - C
Water 538.7 100 374.2 0.38 0 to 347.2
Methyl alcohol 260.1 64.7 239.4 0.4 -97.6 to 239.4
Ethyl alcohol 202.6 78.3 243.1 0.4 -114.1 to 243.1
n-Propyl alcohol 162.3 97.2 263.6 0.4 -126.2 to 263.6
Carbon tetrachlorid 46.55 76.7 283.2 0.38 -22.9 to 283.2
Benzene 94.1 80.1 288.94 0.38 5.53 to 288.94
Toluene 86.1 110.6 318.8 0.38 -95.0 to 318.8
Table 1.2
Correlation coefficients of Eq. (1-4) for calculating the heat of
vaporization of some solvents
Liquid A B T
kr
Range
C at 25 C C
Water 0.3471 0.2740 374.2 1.0 0 to 374.2
Methyl alcohol 0.2928 0.2760 239.4 0.79 -97.6 to 239.4
Ethyl alcohol 0.2903 0.2765 243.1 0.79 -114.1 to 243.1
n-Propyl alcohol 0.2915 0.2758 263.6 0.80 -126.2 to 263.6
Carbon tetrachlorid 0.5591 0.2736 283.2 1.58 -22.9 to 283.2
Benzene 0.3051 0.2714 288.94 0.87 5.53 to 288.94
Toluene 0.2883 0.2624 318.8 0.86 -95.0 to 318.8
Table 1.3
Correlation coefficients of Eq. (1-6) to calculate the density of
liquids in the saturation state

7
The values of the molar entropy of fusion h M T
s

%
/ are
according to [1.4] approximately:
9.2 kJ/(kmolK) for metals,
22 bis 29 kJ/(kmolK) for organic compounds,
38 bis 58 kJ/(kmolK) for anorganic compounds.

For the density
l
of liquids in the saturation state as a
function of the temperature T, have Sha und Yaws [1.5]
the following correlation introduced.

( )

l
T
r
=

A B
1
2 7 /
(1-6)

Calculated values for
l
in g/cm
3
using the respective
constants A and B are listet in Table 1.3. T
r
= T/T
kr
is
the reduced temperature.

1.2 Thermal eqation of state
For each of the three phases of gas, liquid and solid,
the thermal equations of state

p = p (v,T)

or

v = v (p,T)

specify the relationship between the thermal state
variables. These functions are very complex and not yet
known exactly for the three phases.
Gases behave at low pressures and enough distance
from the boiling state almost exactly after the thermal
equation of state for so-called ideal gases:

p v T = R (1-7)

R is the gas constant that can be calculated by using
the universal gas constant
%
R and the molar mass
%
M.


K kmol
kJ
8.317 = R
~
mit M
~
/ R
~
R

= (1-8)

In addition, one should know that the volumes of a
kilomole of any ideal gases at the same temperature and
pressure are equal. At 0 C und 1013.25 mbar, the
molar volume is v
~
= 22.414 m
3
/kmol.
Looking at the behavior of gases at higher pressures or
in the vicinity of the boiling state, deviations from the
ideal gas are observed. Fig. 1.3 shows some isotherms
of CO
2
; they extend to the region, in which
condensation occurs. To recognize is the increasing
deviation from the expected hyperbolic curve shape for
ideal gases at low temperatures above the critical point.
The ideal gas law is therefore a limit law. Gases,
whose behavior differ, are designated as non-ideal or
real gases. However, in first approximation many gases
can, at least in certain ranges of pressure and
temperature, be treated as ideal gases.
For the description of the temperature- and pressure-
dependent behavior of liquids and solids is often being
used the following in T und p linear approach:

( ) ( )
| |
v T p v T T p p ( , ) = +
0 0 0 0 0
1 (1-9)

In Eq. (1-9) means the volume change that is caused
due to a temperature change at constant pressure
(volume coefficient of expansion):

(
1
v
v
T
p


(1-10)

and is the change in volume that is caused by the
change of pressure at constant temperature (isothermal
compressibility coefficient):

Fig. 1.3
Deviation from the expected hyperbolic curve shape of
the isotherms for CO
2
at low temperatures above the
critical point (according to [1.6])

8



1
T
p
v
v
(

(1-11)
at the by index 0 marked reference state.
For solids, the coefficient of volume expansion is being
calculated by using the linear expansion coefficient
that is experimentally easy to determine:

(
1
l
l
T
p


(1-12)
l in Eq. (1-12) is a characteristic length of the
investigated solid. Because the volume V is
proportional to l
3
, follows that = 3. Values for ,
and can be found in many table works.
Liquids and solids are changing their volume under
the influence of pressure and temperature very little; v
is thus constant. Gases, however, show elastic
behaviour already at little pressure and temperature
change. The compressibility of gases plays for example
a role in pneumatic conveying. Due to the flow pressure
loss decreases the gas pressure and thus the density of
the gas in conveying direction. Therefore, the gas
velocity and thus the pressure loss in non-widened
pipes rise significantly.
The incompressible behavior of liquids plays, for
example, in the testing of pressure vessels with water
an important role. If once a weld really tears, reduces a
very small volume increase the pressure to zero. Would
air be used for the pressure build-up, could the work
capacity that is contained
in the large density change
of the air, the vessel tear
explosively apart.
For liquids is the
influence of temperature
on the pressure change to
note, if the specific
volume is kept constant.
In such cases rises the
pressure steeply. Therefore, may vessels for the storage
and transport of liquids, which are not equipped with a
pressure relief valve, never be completely filled, to
avoid the dangerous condition of v=const.

1.3 Viscosity
If a real fluid is subjected a change in shape, the
internal friction of the fluid causes resistance forces. In
case of moving a plate on a fluid layer (Fig. 1.4),
applies for the resistance force F related to the plate
surface A:

F
A
du
dy
= = (1-13)

is the shear stress, du/dy the velocity gradient und
the dynamic viskosity, a measure of the size of the
internal friction. With higher temperatures, the dynamic
viscosity of gases increases, while that of liquids
decreases. The ratio of dynamic viscosity and fluid
density is called kinematic viskositt : = /.
The relationship between dynamic viscosity
g
of
gases and the temperature T can be described by a
polynomial approach:

g
T T = + + A B C
2
(1-14)

The correlation constants A, B, and C, determined by
Miller, Schorr and Yaws [1.1], are listed for some
technically important gases in Table 1.4. Inserting the
temperature T in K in Eq. (1-14), results the dynamic
viskosity
g
in P.
The kinetic theory of gases describes the relationship
between the viscosity and the molecular properties. On
this basis, the mean free path length of gas molecules
can be expressed with the gelp of the viskosity
g
.

=



16
5
~
~
R
2 M

T
p
g

(1-15)

For liquids, the viscosity as a function of the
temperature must be correlated with a slightly more
complicated equation after Miller, Gordon, Schorr and
Yaws [1.1]:

log
l
T
T T = + + + A
B
C D
2
(1-16)

The correlations constants for different liquids are
contained in Table 1.5. Inserting of A, B, C und D as
well as T in K into Eq. (1-16) results the viscosity
l
of
the liquid in the state of saturation in cP.
The fact that according to Eq. (1-13) the shear force
per unit of surface area is proportional to the negative
local velocity gradient, calls one the Newtonian flow
law. Liquids, which behave in this way, are so-called
Newtonian fluids.
Within the procedural practice one often has to work
with materials - pastes or liquids - which behave
differently as to be expected after the Newton's law of
flow. So it can happen that a supposedly solid material,

Fig. 1.4
A moving plate, which slides
over a layer of real fluid,
induces a shear stress

9
that is grinded in a mill, becomes suddenly
liquid. The flow characteristics of such
substances is in contrast to the Newtonian
behavior dependent of the size of the shear
rate. There are structural viscosity (pseudo
plasticity), Dilatancy, Bingham behaviour
and elasto-viscous behaviour.
In case of pseudoplastic behaviour the
viscosity decreases during increasing shear
rate, however, it remains constant at constant
shear rate. With changing shear rate, the
viscosity changes also.
Dilatant materials behave exactly opposed.
The viscosity is being increased with
increasing shear rate. This property is in
practice relatively rare.
The characteristic of a Bingham liquid is,
that the flowing starts only at a certain
minimal shear stress. From then on, the
course of the flow curve can be both,
pseudoplastic and dilatant.
A fourth group of non-Newtonian fluids is
called elastoviskos. The behaviour of these
substances is a combination from elastic
solids and liquids. During the stirring of such
substances there is for example no trombe
(funnel). Instead of this, the fluid climbs on the stirring
element upwards. In the vicinity of a radially pumping
agitator can also a reversal of flow occur.
The relationship between shear stress
xy
and velocity
gradient -du
x
/dy for Newtonian, dilatant und Bingham
liquids shows Fig. 1.5.
In addition to the dependence of the viscosity from the
shear rate, exists in non-Newtonian behavior often still
a time-dependence. Thereby a distinction is between
thixotropic and rheopex behavior. Thixotropy is being
observed during the stirring of a substance in a
container over a longer period of time. Despite constant
shear rate, the viscosity decreases more and more. After
switching off of the agitator the viscosity increases
back to its original value. The opposite behavior,
namely the increase of the viscosity with increasing
duration of mixing, is known as rheopex.
In practice show the majority of the non-Newtonian
substances structural viscosity and thixotropic
behavior.

1.4 Thermal conductivity and heat capacity
Properties with significant influence on the heat
transfer during flow processes are the thermal
conductivity and the heat absorption ability of the
substances. Mathematically, the temperature-dependent
material behaviour is likewise be described with the
help of polynomial-approaches. The following equation
describes the thermal conductivity g of gas at
pressures around 1 bar after Miller, Sha and Yaws [1.1]
as function of the temperature T.

3
D
2
C B A T T T
g
+ + + = (1-17)
For the thermal conductivity
l
of liquids is one term
less necessary [1.5]:

l
T T = + + A B C
2
(1-18)
In Table 1.6 are the correlation constants A, B, C und
D for some important technical gases and solvents
listet. In the Eq. (1-17) and (1-18) must the temperature
Gas A B 10
2
C 10
6

g
at 25 C Range
P C
Hydrogen 21.87 22.20 -37.51 84.7 -160 to 1200
Nitrogen 30.43 49.89 -109.3 169.5 -160 to 1200
Carbon dioxide 25.45 45.49 -86.49 153.4 -100 to 1400
Sulphur dioxide -3.793 46.45 -72.76 128.2 -100 to 1400
Ammonia -9.372 38.99 -44.05 103 -200 to 1200
Table 1.4
Correlation coefficients of Eq. (1-14) for the calculation of the
dynamic viscosity of some technically important gases at low
pressures
Liquid A B C 10
2
D 10
6
Range in C
Water -10.73 1828 1.966 -14.66 0 to 374.2
Methyl alcohol -17.09 2096 4.738 -48.93 -40 to 239.4
Ethyl alcohol -2.697 700.9 0.2682 -4.917 -105 to 243.1
n-Propyl alcohol -5.333 1158 0.8722 -9.699 -72 to 263.6
Tetrachlorkohlenstoff -5.658 994.5 1.016 -8.733 -20 to 283.2
Benzene 2.003 64.66 -1.105 9.648 5.53 to 288.94
Toluene -2.553 559.1 0.1987 -1.954 -40 to 318.8
Table 1.5
Correlation coefficients of Eq. (1-16) for the calculation of the
dynamic viscosity of some liquids in the saturation state
Fig. 1.5
Relationship between shear stress
yx
and speed gradient
-du/dy for Newtonian, Bingham, pseudoplastic and
dilatant liquids

10
T be used in K. The thermal conductivity is then
calculated in cal/(scmK).
Measurements of the specific heat capacity cp of
gases and liquids were correlated with the help of a
polynomial-approach [1.3 und 1.5]:
c T T T
p
g l ,
= + + + A B C D
2 3
(1-19)

In Table 1.7 are the correlation constants A, B, C and
D listed. In Eq. (1-19) must the temperature T be used
in K. The molare heat capacity cpg for gases at low
pressures gets is then calculated in in
kcal/(kmol K); For liquids in the state of
saturation gets one the specific heat capacity
in kcal/(kg K).
The amount of heat
&
Q, that is necessary to
heat a substance to a certain temperature, is
being expressed in thermodynamics as
follows:
&
Q h c T
p
= =

d (1-20)
Insertion of Eq. (1-19) into Eq. (1-20) and
the integration lead to following expression
for the amount of heat
&
Q:
&
Q T T T T
T
T
= + + + A
B C D
2 3 4
2 3 4
1
2

(1-21)
T
1
and T
2
are the starting and final tem-
peratures of the substance.
1.5 Moisture
Gases can be mixed up to certain limits
with steams, i.e., with substances, which are
in the examined temperature range con-
densable. The most important example of
such gas vapour mixtures is the humid air,
which for air conditioning and in the drying
technology as well as in meteorology plays
a major role.
Expediently are the state variables of gas-
vapour mixtures not related on the mixture,
but on the dry gas, because in facilities its
amount is constant, in complete contrast to
the total amount of moist air. The moisture
content x is being specified as the mass ratio
of steam to dry gas:

x
M
M
D
g
= (1-22)
Because for ideal gases the mole ratio
equals the ratio of steam partial pressure to
gas partial pressure is, can for the specific moisture
content x be written:

x
M
M
p
p p
D
g
D
D
=

~
~
(1-23)

How far moist gases are away from the saturation, is
expressed by the relative humidity . is the ratio be-
tween the actual steam partial pressure p
D
and the satu-
ration pressure p
S
at the mixture temperature, or the ra-
Liquid / Gas A B 10
2
C 10
4
D 10
8
Range in C
Water -916.62 1254.73 -152.12
Methyl alcohol 770.13 -114.28 2.79 -97.6 to 210
Ethyl alcohol 628.0 -91.88 5.28 -114.1 to 190
n-Propyl alcohol 1442.74 -8.04 -5.29 -126.2 to 220
Carbon tetrachlorid 383.95 -45.45 -0.24 -22.9 to 224
Benzene 424.26 1.14 -9.03 5.53 to 260
Toluene 485.1 -53.84 -0.59 -95 to 308
Hydrogen 19.34 159.74 -9.93 37.29 -160 to 1200
Nitrogen 0.9359 23.44 -1.21 3.591 -160 to 1200
Carbon dioxide -17.23 19.14 0.1308 -2.514 -90 to 1400
Sulphur dioxide -19.31 15.15 -0.33 0.55 0.0 to 1400
Ammonia 0.91 12.87 2.93 -8.68 0.0 to 1400
Table 1.6
Correlation coefficients of the Eq. (1-17) and (1-18) for calculating
the thermal conductivity of gases at low pressures and liquids in the
saturation state
Liquid / Gas A B 10
3
C 10
6
D 10
9
Range in C
Water 0.6741 2.825 -8.371 8.601 0 to 350
Methyl alcohol 0.8382 -3.231 8.296 -0.1689 -97.6 to 220
Ethyl alcohol -0.3499 9.559 -37.86 54.59 -114.1 to 180
n-Propyl alcohol -0.2761 8.573 -34.2 49.85 -126.2 to 200
Carbon tetrachlorid 0.01228 2.058 -7.04 8.610 -22.9 to 260
Benzene -1.481 15.46 -43.70 44.09 5.53 to 250
Toluene -0.1461 4.584 -13.46 14.25 -95 to 310
Hydrogen 6.88 -0.022 0.21 0.13 25 to 1227
Nitrogen 7.07 -1.32 3.31 -1.26 25 to 1227
Carbon dioxide 5.14 15.4 -9.94 2.42 25 to 1227
Sulphur dioxide 5.85 15.4 -11.1 2.91 25 to 1227
Ammonia 6.07 8.23 -0.16 -0.66 25 to 1227
Table 1.7
Correlation coefficients of the Eq. (1-19) for calculating the specific
heat capacity of gases at low pressures and liquids in the saturation
state

11
tio between the amount of steam in the volume unit of
moist air
D
and the largest possible value of
S
for
equal total pressure and equal temperature:

= =
p
p
D
S
D
S
(1-24)
For the density of moist air below the saturation
state applies for a certain temperature the following
equation:
=

+
+
p
R T
x
M
M
x
D D
g
1
~
~
(1-25)
State changes of humid gases can be pursued in sim-
ple and clear way in the Mollier-h-x diagram. The en-
thalpy of the amount of (1 + x) kg of humid gas con-
sists of the enthalpies of dry gas and steam:

( )
h c h c x
x p
g
v p
D
1+
= + + (1-26)

By this equation can for certain temperatures and
various steam contents x the enthalpy-values be deter-
mined. The isotherms are straight lines in the h-x-chart.
By inserting the saturation value of the vapour partial
pressure p
D
at a certain temperature into Eq. (1-23) and
the use of the calculated value for x can the enthalpy of
the saturated steam-gas mixture be determined. With
the help of for various temperatures calculated values,
is it possible to plot the saturation curve = 1.
Due to the poor legibility of the diagram, has Mollier,
as is known, an oblique-angled presentation proposed,
in which the x-axis is tilted, until the iso-
therme for = 0 C is horizontal. This is
the case, when at the coordinate x = 1 the
evaporation enthalpy h
v
is displayed
downwards. Fig. 1.6 shows the Mollier-
h-x diagram for moist air for a tempera-
ture range of 0 to 90 C, at a pressure of
1 bar.
For the determination of the curve of
equal relative humidity, Eq. (1-23) can
be written also:
x
M
M
p
p p
D
g
S
S
=


~
~

(1-27)


1.6 Flammable mixtures
Just like technical gases and heating oil
in the mixture with air can be burned, many other
substances can form flammable mixtures with air. For
the operating of facilities are mixtures dangerous, in
which the flame front itself after an ignition explosively
spreads. One speaks of an explosion, if the combustion
is connected with a clear pressure increase. Such a
situation can occur, if in a closed space a large mass of
fuel is burned and the pressure noticably goes up, or if
the fuel is combusting so quickly, that the surrounding
gas due to its mass inertia, causes a large pressure
increase.
The explosion becomes a detonation, if the progress -
for example, through turbulence influences - itself as
far as speeded up, that the mixture will no longer being
Gas / Vapour Explosion limits in air Ignition Ignition group
(1013 mbar, 20 C) Temp. in dependence
lower upper of the ignition
Vol.-% C temperature
Hydrogen 4.0 75.6 560 G1
Carbon monoxid 12.5 74 605 G1
Hydrogen sulfide 4.3 45.5 270 G3
Ammonia 15.0 28.0 630 G1
Methyl alcohol 5.5 31/44 455 G1
Ethyl alcohol 3.5 15.0 425 G2
n-Propyl alcohol 2.1 13.5
Benzene 1.2 8.0 555 G1
Chlorbenzene 1.3/1.5 7/11 (590) G1
Toluene 1.2 7.0 535 G1
o-Xylene 1.0 6.0/7.6 465 G1
Table 1.8
Key figures of some combustible gases and vapours

Fig. 1.6
Mollier-h-x diagram for moist air at temperatures from 0 to 90 C and
atmospheric pressure [1.9]

12
ignited by the heat of the flame front, but by the
resulting blast. The pressure wave moves ahead with
sound velocity, in case technical fuels in air thus with
approximately 1000 m/s.
The boundary between deflagration and explosion is
not precisely defined. In the parlance, deflagration is a
weak explosion, which causes only minor damages.
Also dusts can together with air form explosive
mixtures. The lower explosion limit lies between 15
and 45 g/m
3
.
The first requirement for the presence of an explosive
mixtures is the concentration of the combustible
portion in the air, which must lie in the range between
lower and upper explosion limit. Moreover needs a
surface, that touches the ignitable mixture, at least a
temperature as large as the ignition temperature of the
mixture. In addition, is a certain minimum energy
necessary, to ignite the mixture. The minimum energies
for dust/air mixtures are two or three powers of ten
larger than those of gas(steam)/air mixtures. The size of
the ignition energy influences the explosion limits
likewise.
The explosion intensity is also a function of the vessel
dimensions. While the explosion pressure in vessels
with larger dimensions remains the same, changes the
maximum temporal pressure rise (dp/dt)
max
with the
volume V of the vessel ("cubic law")

dp
dt
V konst
G
|
\

|
.
| = =
max
/ 1 3
K (1-28)

K
G
in barms
-1
is under otherwise equal conditions a
material-specific constant. The cubic law is valid also
for dust-air mixtures. Dusts are characterized by using
Kst-values. Dust explosions are at least as fierce - if
not more fierce - as gas explosions.
In Table 1.8 are listed the safety technical key figures
for some inflammable gases and vapours. Tables with
comprehensive information were published by Nabert
and Schn [1.7]. Regarding the intensity of dust
explosions can be key figures found in the VDI-
Richtlinie 2263 [1.8].

Literature of chapter 1.
[1.1] Yaws, C.L.; J.W. Miller; P.N. Sha; R.R. Schorr und P.M. Patel: Corre-
lation constants for chemical compounds. Chemical Engineering No-
vember 22(1976), S.153-162.
[1.2] Watson: Ind. Eng. Chem. 35(1943)398, zitiert in [1.4].
[1.3] Yaws, C.L.; R.W. Borreson; C.E. Gorin II; L.D. Hood, J.W.Miller;
G.R. Schorr und S.B. Thakore: Correlation constants for chemical
compounds. Chemical Engineering August (1976)16, S.79-87.[1.4]
Perry, R.H.; und C.H. Chilton: Chemical engineers handbook, 5th ed.
New York, Tokyo: McGraw Hill/Kogakusha, 1973.
[1.5] Yaws, C.L.; J.J. McGinley; P.N. Sha; J.W. Miller und G.R. Schorr:
Correlation constants for liquids. Chemical Engineering October
(1976)25, S.127-135.
[1.6] Eastman, E.D.; und G.K. Rollefson: Physical Chemistry. New York:
McGraw Hill Book Co., 1947.
[1.7] Nabert, K.; und G. Schn: Sicherheitstechnische Kennzahlen brenn-
barer Gase und Dmpfe. 2. Aufl., Braunschweig: Deutscher Eich-
verlag.
[1.8] VDI-Richtlinie 2263: Verhtung von Staubbrnden und Staub-
explosionen.
[1.9] Buchholz, E.: Das i-x-Diagramm von Mollier und seine Anwendung
bei der Bedienung von Luftbehandlungsanlagen. Energie 6(1964),
S.316-323.

13
2 Characteristics and Properties of
bulk materials
The behaviour of bulk materials during the discharge
from silos, during the pneumatic conveyance, or in flu-
idized beds, is decisively influenced by certain recur-
ring material characteristics. The individual particles
are of different shape and size, often show a porous
structure, and are often randomly arranged within the
bulk solid. Under certain circumstances friction and
adhesion forces are acting between the particles.
These properties affect the movement and the ex-
change processes in technical devices and facilities. For
a mathematical description of the applicable physical
laws, must bulk materials therefore physically meaning-
ful and clearly be marked. Process selection and quality
evaluation requiring likewise clear specific data.
2.1 Particle size
The particle diameter dp in this context means not the
geometric dimension of the particles, but describes the
dimensions of particles of defined form with the same
properties. It is named equivalent diameter. For exam-
ple, the terminal velocity equivalent diameter is the di-
ameter of a sphere with the same terminal velocity as it
the particle has. In practice, one often uses the sieve
mesh width of a sieving as a not very meaningful, but
quickly and easily determinable equivalent diameter.
Other equivalent diameters are:
geometric diameter: diameter of a sphere of equal
volume or equal surface,
Equivalent diameter of the particle projection: di-
ameter of the circle with equal area or equal circum-
ference,
flow-mechanical diameter: diameter of the sphere
with equal flow resistance or equal sinking speed,
Diameter of the sphere with equal scattered light in-
tensity,
Diameter of the sphere with equal electrical resis-
tance change.
It is not possible to determine particle diameter of
very different sizes by the same way. The analysis
method is being chosen depending on particle size and
other properties of the bulk material. In addition, per-
fect results are anyway just to expect if the investigated
sample is representative of the bulk material.
During the sieve analysis, the bulk material by test
sieves of different mesh size is decomposed into vari-
ous classes of particles. Without any additional means,
particle sizes above 40 to 60 m can in this way be
seized. The screening is improving, by sucking and elu-
triating the particles through the sieve. In this way par-
ticle sizes smaller than 5 m can be measured.
The sedimentation analysis works, using a suspension.
From the determined sinking velocities during the
sedimentation of the particles, the particle diameters of
individual classes are calculated. In the gravitational
field with this method particle diameters between 2 and
50 microns can be measured, in the centrifugal field
such with 0.01 to 3 microns in diameter.
Wind sifting within gravity or centrifugal fields is
suitable for dry bulk materials. During the wind sifting,
the particles with a smaller sinking velocity than that of
the upward airflow are discharged by the fluid. In the
gravity field, particle sizes between 5 and 60 m, in the
centrifugal field particle sizes down to 1 m can be
measured by this method.
In the light microscope can particles with sizes be-
tween 1 and 150 m be analysed.
In the so-called coulter counter, particles, which are
suspended in an electrically conductive fluid, are
sucked through a current-carrying capillary bore. This
changes the electrical resistance of the liquid approxi-
mately proportional to the volume of a single particle.
In the scattered light analyser, dust loaded gas is
blown into an illuminated volume. There, each particle
scatters a to its size corresponding light proportion,
which is photometrically registered.
2.2 Particle size distribution
Usually, the particles in a bulk material have no uni-
form size, but there is a more or less broad range in
which the dimensions vary.
Most of the usual analysis methods provide informa-
tion about the quantity of particles of a specific size in
a bulk solid. These quantities can be plotted in a chart
as proportion of the number, mass, volume, or surface
dependent on the investigated particle size intervals
(called classes). It is common, to depict the proportion
of the total amount as relative value, as so-called fre-
quency, based on the investigated class width. This ap-
proach results in a step diagram as shown in Fig. 2.1, in

Fig 2.1
Relative occurrence of particle-classes in a bulk solid

14
which in case of a sufficient number of particle classes
a steady course can be plotted.
This characteristic line can also as cumulative fre-
quency curve be plotted, that indicates, which propor-
tion of the particles greater or less than a certain parti-
cle diameter dp is. Following the sieve analysis, these
graphics are designated as backlog and passage cumula-
tive curves. The particles, which are larger than the
mesh size, are forming the backlog, and those, which
are smaller, the passage. Fig. 2.2 shows an example.
In order to describe the measured particle size distri-
butions of a bulk material mathematically, different
equations were proposed. The equations are, however,
all of empirical nature, and describe the actual grain
sizes only in limited ranges accurately.
In addition to the arithmetic and logarithmic normal
distribution, as well as the power-law distribution,
should the RRSB distribution according to Rosin,
Rammler and Sperling [2.1] be mentioned, after which
the backlog cumulative curves of many bulk solids fol-
low the Eq. (2-1).

R
d
d
p
p
n
=
|
\

|
.
|
|

(
(
(
exp
'
(2-1)
In Eq. (2-1) means n an evenness coefficient, and d
p

is the particle size at the value R = 0.3679. In a coordi-
nate system, consisting out of a double logarithmic or-
dinate and a logarithmic abscissa, the RRSB backlog
cumulative curve is a straight line. In practical cases
are, however, those lines usually slightly curved. With
a polynomial function is it possible to reproduce their
course. Transformation of analysis results into the
RRSB coordinate system and subsequent determination
of the polynomial function constitutes the mathematical
basis for the numerical evaluation of separating and
grinding processes.
The various nets for the representation of the particle
size distribution are standardized according to DIN
66141 (basics), DIN 66143 (power-law distribution),
DIN 66144 (logarithmic normal distribution), and DIN
66145 (RRSB distribution).
For calculations, must the particle size distribution of
a bulk material to be expressed by a measure for the
average particle size. This so-called mean particle di-
ameter d
p
m
is defined as follows:
1
d
x
d
p
m
i
p
i
= (2-2)

x
i
is the weight proportion of a particle class, and d
p
i

is the arithmetic mean value of the equivalent diameter
of a particle class. With the particle diameter dp is in
the following always a medium value meant.
The published results of studies on bulk solids were
mostly won with the help of sieved particle fractions.
For most of the discovered laws one must assume
therefore, that their validity only for particle size distri-
butions is guaranteed, in which the ratio between the
largest and the smallest particle size is not larger than
4. Despite this restriction, belong the majority of prob-
lems in practice to this scope. A greater proportion of
fine particles or discontinuous particle size distribu-
tions have, however, a sustainable influence on the mo-
tion and exchange behaviour of bulk materials.
2.3 Void fraction, particle shape and density
Like the dimensions of the individual particles of a
bulk solid, so also differ the shapes from each other.
Particle and bulk solid properties are dependent of this
shape. To take this into account, were shape criteria be-
ing established, which be mostly used in connection to
the particle size. So makes for example the sphericity
Particle type
p
Average particle size in m
20 50 70 100 200 300
Angular sand 0.67 - 0.60 0.59 0.58 0.54 0.50
Rounded sand 0.86 - 0.56 0.52 0.48 0.44 0.42
Rounded sand
mixture
- - - 0.42 0.42 0.41 -
Carbon and glass - 0.72 0.67 0.64 0.62 0.57 0.56
Anthracite 0.63 - 0.62 0.61 0.60 0.56 0.53
Activated carbon - 0.74 0.72 0.71 0.69 - -
Catalyst 0.58 - - - 0.58 0.56 0.55
Abrasive - - 0.61 0.59 0.56 0.48 -
Table 2.1
Experimental void fractions of some bulk solids [2.2]

Fig. 2.2
Cumulative curves of particles of a bulk material: a pas-
sage, b backlog

15
p the comparison between the surface of a particle
and a sphere of equal volume:

particle single a of surface
volume equal with sphere a of surface
=
p
(2-3)
The ratio of the free volume between the particles of a
packed bed and its total volume is referred to as poros-
ity or void fraction
0
:

0
0
p 0
V
V V
=
+
(2-4)
V
0
is the volume of gas, and Vp is the volume of the
particle mass.

0
depends on both the sphericity and the particle size
distribution, and it can be calculated only under very
restricted conditions. For practical use, the void frac-
tion is therefore experimentally determined or on the
basis of known examples being estimated. It rises ac-
cording to Table 2.1 (after [2.2]) with decreasing parti-
cle size and sphericity.
According to the definition, the void fraction
o
is an
average value, in order to characterise the overall bulk
solid. The local void fraction
o
is however very dif-
ferent. Independent of the shape and size of the parti-
cles reaches the void fraction due to the point contacts
of the bulk solid at the apparatus wall the value of 1. It
becomes with increasing distance from the wall smaller
and reaches at a distance of half particle diameter a
minimum value. Then it rises again, but not to the value
1, because the second row of the particles lays itself in
the gussets of the first one. The arrangement of each of
the following rows of particles is more and more left to
chance. From a certain distance of the wall is the distri-
bution of the particles only pure random.
Fig. 2.3 shows experimentally determined porosities
for spheres of equal size [2.3]. It is to recognise that the
size of the local pore volume according to the transition
from the well-ordered bulk material to the randomly
distributed bulk material is swinging around a mean
with steadily decreasing amplitude. Upon reaching the
pure random distribution this oscillation is completely
subsided.
The change of the void fraction
0
of the total bulk,
which is caused by the wall influence, decreases with
increasing ratio between vessel diameter and particle
diameter.
An also important influencing factor is the density p
of the particles. It must however be taken into account,
that the particle density p and the actual solids density

s
in case of a porous structure are possibly not equal.
In practice, the particle density is therefore also desig-
nated as apparent density.
The bulk solid density
Sch
is the measure for the den-
sity of the whole bulk solid volume. Because of the gas
proportion, which is considered as weightless, its value
is reduced:


Sch p 0
= ( ) 1 (2-5)

If the void fraction
o
of the bulk solid is known, the
particle density p can be calculated with the help of
the bulk solid density.
2.4 Adhesive forces
The different mechanisms of adhesion can cause in
moving bulk materials disabilities. During drying for
example, meets the average particle size often not the
actual particle diameter, because agglomerates may be
formed. So it comes to deviations in the flow and mov-
ing behaviour of bulk goods.
The main adhesion mechanisms occur due to liquid
bridges, van-der-Waals and electrostatic interactions, as
well as solid-state bridges. Liquid bridges between two
solids surfaces cause always an attraction due to the
surface tension of the liquid. Van-der-Waals forces
make themselves only noticeable in case of very small
particles. Electrostatic forces arise due to surplus
charges and can act both attractive and repulsive. Ex-
cess charges form themselves for example because of
frequent collisions of the particles against walls or with
each other, if one of the contact partners has the charac-
teristics of an insulator. Solid-state bridges can for ex-
ample arise during sintering, melting and crystallizing.

Fig. 2.3
Position dependency of the void fraction

o
in a bulk solid with sphere-shaped par-
ticles [2.3]

16
2.5 Moisture
Liquid-solid mixtures can exist as a real solution, col-
loidal solution, fixed colloids (gels), suspensions or as
crystalline solids with dispersed fluid inclusions. De-
pending on the type of mixture, are the liquids in dif-
ferent ways bound to the solids.
During the crystallization of solutions are many of the
separated substances not fluid-free, but they incorpo-
rate liquid molecules into their crystal lattice. So, we
have it to do with a molar binding.
One speaks of adsorption, if the liquid molecules are
bound due to van-der-Waals forces. The idea in this re-
gard is, that several layers of liquid molecules them-
selves settle with decreasing binding energy from layer
to layer on the free surface. Significant quantities of
liquids can be adsorbed in this way only of gels with
their very large specific inner and outer surface. There-
fore, this type of binding plays a role particularly in
case of small liquid contents.
With increasing number of molecular layers of liq-
uids, the binding energy becomes smaller. When reach-
ing the value zero, no more moisture can adsorptively
be absorbed. However, there are colloidal materials,
which consist of a broad mixture of various molecular
weights. The low-molecular materials are still partially
soluble in the liquid. In contrast, the high-molecular
materials form an insoluble skeleton. This skeleton
houses the soluble portions, which cannot pass through
the cell walls. If, however, the low-molecular liquids
can pass the cell walls, an osmotic process arises. The
amount of liquid that can be taken by such a Material in
this way is under circumstances a multiple of the ad-
sorbed liquid. One speaks of osmotically bound or
structural moisture. On free solids surfaces and in cap-
illaries the moisture is mechanically bound. As is
known, the surface tension of the liquid is the cause of
this binding.
All known kinds of the binding, which cause a lower-
ing of the vapour pressure above the surface of the
moisture, for mechanically bound moisture this is only
the case in the capillaries of the fine-pored materials,
are summarized under the term sorption. By thermal
drying of these so-called hygroscopic solids, sorbed
liquid can be removed only up to the point, when the
vapour pressure inside the substance equals the partial
pressure in the surroundings. It is almost impossible to
distinguish the binding types strictly, so that sorption
isotherms, thus the curves of the liquid content X in the
material in dependence of the relative air humidity
are experimentally determined. Fig. 2.4 shows the sorp-
tion isotherms of some plastics; the sorbed substance is
water vapour.
2.6 Thermal conductivity and heat capacity
The thermal conductivity of solids can only be deter-
mined with the help of experiments. Thereby was ob-
serve, that electrical conductors much a higher thermal
conductivity have, than electrical insulators.
Material c
C kg/m
3
kJ/(kg K) W/(m K)
Aluminium 99 75 20 2700 0.896 229
Lead, pure 0 11340 0.128 35.1
Bronze 20 8800 0.377 61.7
V2A-Steel 20 8000 0.477 15
Copper 20 8300 0.419 372
Concrete 20 2200 0.879 1.28
Ice 0 917 1.93 2.2
Soil, coarse 20 2040 1.84 0.52
Glass 20 2480 0.7/0.93 1.16
Granite 20 2750 0.75 2.9
Coal 20 1300 1.26 0.26
Table 2.2
Density , thermal conductivity and specific heat ca-
pacity c for some solids
Fig 2.4
Sorption isotherms of plastics (sorbed
substance is water vapour): a polysty-
rene granules, b polyethylene powder,
c polyethylene granules

17
The thermal conductivity of metals is thereby the
higher, the greater the value of the electrical
conductivity is. A connection to other properties of the
substances is not recognize. The temperature
dependence of the thermal conductivity of non-metals
is various. An influence of the density is however in
such a way noticeable, that the thermal conductivity
with increasing density becomes greater. Table 2.2
shows the published values of density, thermal
conductivity and specific heat capacity of some solids.
The effective thermal conductivity of non flowed
through bulk solids is not a simple material constant
any more, but is influenced by the molecular thermal
conductivity of the gas in the overall connected gas
volumes. In addition play the following influences a
role: the conductivity in the gas-filled gussets between
the particles, the thermal conductivity of the particles,
the contact between the particles, and the radiation
exchange between the particle surfaces. These
dependencies can no longer be represented by means of
simple calculation approaches.
How much the gas proportion influences the thermal
conductivity of a bulk solid, can one recognize by the
characteristics of porous insulating materials. The
insulating effect results almost completely from the
poor thermal conductivity of the air in the pores. The
solid material sceleton - the solid material conducts
heat much better - is only used to prevent the
convective heat transfer due to free moving air.

Schrifttum zum Abschnitt 2.
[2.1] Rammler, E.: Zur Auswertung von Krnungsanalysen in Krnungsnet-
zen. Freiberger Forschungsheft, Reihe A4, 1952.
[2.2] Leva, M.: Fluidization. New York: McGraw-Hill Book Company,
1959.
[2.3] Ridgeway, K; und K.J. Tarbuck: Voidage fluctuations in random
packed beds of spheres adjacent to a containing wall. Chem. Eng. Sci.
23(1969), S.1147-1155.
3 Mechanisms of heat transfer
3.1 Heat conduction
The temperature fields during the heating and cooling
of materials are usually temporally variable. Calcula-
tion basis for the thermal conduction is the Fourier
equation that describes the relationship between the
spatial and temporal change of temperature.

=
2
1

a t
(3-1)

The thermal diffusivity a = /(c) of a substance is a
property just as , c und , for example with the dimen-
sion m
2
/s.
For the amount of heat, which passes through a sur-
face area of 1 m
2
, applies the following equation [3.1]:

& q c t =
2
0

(3-2)

The heat flow grows thus with t . Die physical term
c is a pure material property that is known as heat
penetration coefficient, but could illustratively be
named heat storage capacity.
The comparison between the thermal diffusivity and
the heat penetration coefficient shows the different ef-
fects of the thermal capacity per volume unit in regard
on the temperature field and on the stored amount of
heat. Accordingly the material for a heated wall must
be chosen under consideration of the respective neces-
sities. In some cases should be the propagation of initial
temperatures as slowly as possibly, and in other cases
should be the heat flow into the wall in equal time in-
tervals as small as possible. For example in case of a
fire-retardant wall the value would be kept small, and in
case of the isolation of a discontinuous-powered fur-
nace the value of c would be kept small.
3.2 Convection and short-term contact
The size of convective heat transport is usually being
described with the heat transfer coefficient . The de-
fining equation for is:

=

&
Q
A
(3-3)

is the temperature difference between the surface
and the along flowing fluid, A is the heat exchange sur-
face, and
&
Q is the per unit time exchanged quantity of

18
heat. The size of the heat transfer coefficient depends
on the flow conditions, which are being influenced by
the substance properties and the geometrical condi-
tions.
If the wall is touched by bulk solid, applies for the
heat transfer coefficient in case of not too short contact
times t and constant wall temperature the following re-
lationship [3.2]:

=
2
( ) c
t
Sch
(3-4)
According to Eq. (3-4) is the heat transfer coefficient
proportional to the square root of the heat penetration
coefficient (c)
Sch
of the bulk solid and inversely
proportional to the square root of the contact time t.
Forever shorter contact, the heat transfer reaches fi-
nally a maximum value and remains constant despite
further reduced contact time. For the limit of vanishing
short contact applies after Schlnder [3.2, 3.3] for the
heat transfer coefficient p

between a spherical indi-
vidual particle and a wall:
lim ln
max,
t
p p
g
p p
p
rd
d
d

= =


+
|
\

|
.
|

+
|
\

|
.
|

(
(
+
0
4
2
1
2
1 1

( ) 3 5
g signifies the thermal conductivity of the gas be-
tween the particles, and dp is the particle diameter.
= 2 2 ( ) / can be determined, using the mean
free path of the gas molecules after Eq. (1-15).
is the so called accommodation coefficient. This co-
efficient takes into account the imperfection of the en-
ergy exchange during the collisions of the gas mole-
cules against the wall and against the particle surface.
Fig. 3.1 depicts for different temperatures values for
in dependency from the molecular mass
~
M
g
of the
gases. In addition, the influence of the gas-solid pair-
ings is to recognize.
Fig. 3.2 shows by Eq. (3-5) for air as gas without tak-
ing into account the proportion of radiation rd com-
puted values for the maximum heat transfer coefficients
between a single particle and a wall during short-term
contact as function of the particle diameter dp. Parame-
ter is the average temperature
m w p
= + ( ) / 2 be-
tween wall and particle. To recognize is the significant
increase of the heat transfer coefficient max,p with
decreasing particle size and increasing temperature.
Thereby outweighs the influence of the particle diame-
ter by far.
Eq. (3-5) applies to the heat transfer between a spheri-
cal single particles and a wall. During the heat transfer
between a bulk solid and a wall the void fraction of the
packed bed must be considered. Because the empty

Fig. 3.3
Heat transfer coefficient between bulk solid and wall as
function of the contact time t

Fig. 3.2
Maximum heat transfer coefficient max,p

between wall
and particles
Fig. 3.1
The accommodation coefficient as a function of the mo-
lecular mass
~
M
g
of the gas; parameter is the mean tem-
perature between wall and particles

19
space provides a very small amount of heat transfers,
the effective transfer surface is smaller than the actual
one. The missing proportion of the heating or cooling
area corresponds with the average void fraction of the
bulk material, so that for the maximum heat transfer be-
tween the bed and wall applies:
( )
max max,
= +
p rd
1
0
(3-6)
For a through a pipe sliding bulk solid, called Moving
Bed, has Ernst [3.8 and 3.9] the heat transfer coeffi-
cients measured between the bulk solid and a heated
ring surface at very short contact times. Table 3.1 gives
a comparison between the measured maximum values
for three solids with different mean particle diameters
and the according to Eq. (3-6) calculated values. The
accordance is remarkably good, so that the void frac-
tion also in other cases as a measure of the missing
proportion of the actual heat transfer surface can be
used.
d
p

0

max
in W/m
2
K
m - calculated measured
150 0.48 1774 1740
400 0.42 894 904
600 0.42 639 696
Table 3.1
Comparison of measured and with Eq. (3-6) calculated
heat transfer coefficients
max
for bulk solids, which
slides through a pipe

The general relationship between the heat transfer coef-
ficient for a bulk material (packed bed) and its contact
time on a wall is once more in Fig. 3.3 depicted,
whereby the radiation component was neglected.
While for the heat transfer in the validity range of the
t -law according to Eq. (3-4), the material properties
of the bulk material and the contact time are relevant, is
the heat transfer in the validity range of the
max
-law
according to Eq. (3-5) and (3-6) only dependent of the
particle diameter, the bulk solid void fraction and the
properties of the interspace gas.
3.3 Radiation
During the exchange of heat between two bodies (for
example, wall and bulk solid), the amount of heat that
is exchanged as a result of the radiation effect can often
not be neglected. The frequently occurring diatomic
gases N
2
and O
2
do not hamper this exchange because
they are diatherm.
Because radiation values in practice be rarely known
or are determinable, one uses as basis for the mathe-
matical description the model of the complete radiation
exchange between themselves enclosing grey bodies or
flat surfaces. The exchanged amount of heat then is
written:
&
Q A C
T T
rd
=
|
\

|
.
|
|
\

|
.
|

(
(
1 12
1
4
2
4
100 100
(3-7)
with
( )
C
C A A C C
rd
12
1 1 2 2
1
1 1 1
=
+ / / / /
(3-8)
For technical surfaces one can use for the radiation
coefficient C sufficiently accurate a value of 4.6 W/(m
2

K
4
). For the equivalent heat transfer coefficient
rd
one
uses for the radiation in case of not too great tempera-
ture differences the well-known relationship:

rd
m
C
T
=
|
\

|
.
| 0 04
100
12
3
, (3-9)
T
m
is the mean temperature between T
1
and T
2
. Fig.
3.4 shows the radiant heat transfer rates, calculated ac-
cording to Eq. (3-9) for temperature differences below
200

C; For C
12
has been used a value of 4.6 W/(m
2

K
4
).
It is noteworthy, and it is often misjudged, that the
heat transfer by radiation at room temperature lies in
the order of magnitude of the heat transfer for free con-
vection. The propor-
tion of radiation
must thus in certain
cases to be consid-
ered in heat transfer
calculations.
Another character-
istic of the radiant
heat exchange that is
particularly noticea-
bly at high tempera-
tures, as they prevail
in reactors concerns
the increase of the
equivalent heat trans-
fer coefficient with
decreasing tempera-
ture differences be-
tween the heat ex-
changing surfaces. Recognizable become the functional
interrelations in the complete equation for
rd
:

( ) ( )

rd
c T T
T T
=

12 1
4
2
4
1 2
100 100 / /
(3-10)
Fig. 3.4
Equivalent heat transfer coeffi-
cient
rd
for the radiation ex-
change between completely en-
closed surfaces of approximately
equal size

20
Literature of chapter 3.
[3.1] Grber, Erk und Grigull: Grundgesetze der Wrmebertragung. Berlin,
Heidelberg, New York: Springer-Verlag, 1981.
[3.2] Schlnder, E.U:: Wrmebergang an bewegten Kugelschttungen bei
kurzfristigem Kontakt. Chem.-Ing.-Tech. 43(1971)11, S.651-654.
[3.3] Wunschmann, J.; und E. U. Schlnder: vt "verfahrenstechnik"
9(1975)10, S.501-505.
[3.4] Zehner, P.: VDI-Forschungsheft 558. Dsseldorf: VDI-Verlag, 1973.
[3.5] Reiter, F.W.; J. Camposilvan und R. Nehren: Akkomodationskoef-
fizienten von Edelgasen an Pt im Temperaturbereich von 80 bis 450
K. Wrme- und Stoffbertragung 5(1972)2, S.116-120.
[3.6] Eckert, E.R.G.: McGraw-Hill Inc., 1959.
[3.7] Ebert, H.: Physikalisches Taschenbuch. Wiesbaden: Vieweg Verlags-
gesellschaft, 1976.
[3.8] Ernst, R.: Der Mechanismus des Wrmebergangs an Wr-
meaustauschern in Fliebetten (Wirbelschichten). Chem.-Ing.-Tech.
31(1959)3, S.166-173.
[3.9] Ernst, R.: Wrmebergang an Wrmeaustauschern im Moving Bed.
Chem.-Ing.-Tech. 32(1960)1, S.17-22.

4 Dimensionless numbers for flow
and transport processes
For the scaling of devices and summarising relation-
ships between physical conditions there is the problem
that a variety of possible geometries and conditions
have to be set in relation to each other. Because the dif-
ferential equations for flow and transport processes
generally cannot be integrated, model theories are used.
When using such relationships, the question arises, un-
der which circumstances two conditions are equivalent
to each other.
For the comparability of flow and transport processes
must first the geometric similarity be met. For this pur-
pose often is used the ratio of two lengths, and the
equality of these ratios in two different versions fulfils
then the similarity conditions. For the pipe flow for ex-
ample one uses the relationship between the roughness
k of the pipe wall and the pipe diameter D.
Flowing fluids are being influenced by forces, which
attack at the fluid elements. Similarity of processes is
given, if certain relationships of forces are equal. The
ratio between inertia force and resistance force is the
so-called Reynolds number Re:

Re =
u l

(4-1)

u is the flow velocity, l is a characteristic length, and
is the kinematic viscosity.
The ratio of inertia force and gravity force is called
Froude number Fr:

Fr =

u
l g
2
(4-2)

u and l have the same meaning as in the Reynolds
number, g is the acceleration due to gravity.
A further dimensionless number, which can be derived
for the here interesting incompressible flow without
free surface, is the Euler number Eu, which sets the
pressure differential p p
1 2
and the product of fluid
density
f
and the square of velocity in relation:

Eu =

p p
u
f
1 2
2

(4-3)

The combinations of dimensionless numbers provide
new ones, for example, the Galilei number Ga:

Ga
Fr
= =
Re
2 3
2
l g

(4-4)

To take into account the buoyancy, one further
dimensionless number was formed, the Archimedes
number:
Ar Ga =

=

s f
f f
l g
3
2

(4-5)
For the description of mass and heat transfer
processes, the use of dimensionless transfer coefficients
has become customary, especially the Nusselt number
Nu and the Sherwood number Sh:

Nu
l
=

(heat) Sh
l
=

(mass) (4-6)

These dimensionless numbers depend on the Reynolds
number, and are also influenced by some others for the
capture of the ratio between momentum transfer and
heat transfer, for example the Prandtl number Pr and
the Schmidt number Sc:
Pr =

a
(heat) Sc
a
=

(mass) (4-7)
The Prandtl number, and the Schmidt number are
based on physical characteristics, just as the ratio of the
two, the Lewis number Le:

Le
Sc
= =
Pr
a

(4-8)

21
5 Flow behavior of bulk solids
For moved bulk solids has one to distinguish between
cohesive and non-cohesive behavior. A non-cohesive
material like dry sand takes the form of a circular cone
when it is being heaped on a horizontal level.
Depending on the particle density and the internal
friction arises
always the same
rise angle
between the
surface line of
the cone and the
horizontal level.
The angle of
repose is thus
reproducible.
Cohesive
material sets its
deformation an
additional
resistance
against. The size
of this resistance
is dependent of
the compaction state of the material. For example
expires moist compacted sand not from the form, even
if it is standing on its head. Experience has shown, that
the angle of repose of such materials is little
reproducible.
The internal friction of non-cohesion bulk materials -
only the flow behavior of such materials is the subject
of all further considerations - is affecting all processes,
in which relative movements in the bulk material occur.
Quantitatively can this influence, for example be
considered in the relationship for the discharge of bulk
solids from silos.
5.1 Material flow in silos
Due to the gravity flows bulk solid by itself out of
bunkers and silos and can be passed through pipes to
their destination. The designing engineer has to do it
with two issues. A silo must be constructed in a way,
that the whole contained material runs evenly out. The
outlet cross section must be large enough to prevent,
that stable material bridges, which hinder the flow of
the bulk material, occur.
For the given dimensions of the silo then arises also yet
the question, how the per time unit outflowing material
mass be influenced by the material properties. If the
necessary geometrical conditions, for example, from
lack of space can not be met, or if bulk materials with
very different properties be stored in a bunker, a
smooth and trouble-free operation is often only with the
help of additional discharge aids guaranteed.
5.1.1 Silo design
Two different modes of flow can be observed if a
bulk solid is discharged from a silo: mass flow and
funnel flow (Fig. 5a). In case of mass flow, the whole
contents of the silo are in motion at discharge. Mass
flow is only possible, if the hopper walls are
sufficiently steep and/or smooth, and the bulk solid is
discharged across the whole outlet opening.
If a hopper wall is too flat or too rough, funnel flow
will appear. In case of funnel flow (Fig. 5b), only that
bulk solid is in motion first, which is placed in the area
more or less above the outlet. The bulk solid adjacent
to the hopper walls remains at rest and is called dead"
or stagnant" zone.
Two steps are necessary for the design of mass flow
silos for bulk material: The calculation of the required
hopper slope which ensures mass flow, and the
determination of the minimum outlet size to prevent
arching. These parameters are measured in dependency
on the consolidation stress with shear testers, e.g. with
the Jenike shear tester or a ring shear tester.
The shear cell of the shear tester introduced by Jenike
[5.1] consists of a closed ring at the bottom, a ring of
the same diameter (so-
called upper ring) lying
above the bottom ring
(Fig. 5.2), and a lid. The
sample of bulk solid is
poured into the shear
cell. The lid is loaded
centrally with a normal
force N. In addition, a
bracket is fixed to the
lid. The upper part of
the shear cell is
displaced horizontally
against the fixed bottom
ring.
The hopper slope
required for mass flow
and the minimum outlet
size to prevent arching can be calculated with the
measured values using Jenikes theory [5.5].
5.1.2 Discharge rate
Equations for the discharge rate from a silo were
established so far only empirically and almost
exclusively for non-cohesive bulk materials. Mostly
describe these equations only the respective trials with
good accuracy.
In silos with symmetrical funnels is the mass flow of
non-cohesive bulk material, whose effective friction

Fig. 5.1
During discharge of bulk materials
from silos can occur funnel flow (a)
and mass flow (b)

Fig. 5.2
Jenike shear tester for
determining the friction
inside the bulk solid (a) and
between bulk solid and wall
(b)

22
angle corresponds to its angle of repose, being
influenced by numerous parameters and dependencies
[5.6].
8 . 2 5 . 2
to
A A s
d d M
&


0 5 . 0
to h h M
s

&


&
M
s p


1 18 . 0
to

p p s
d d M
&


5 . 0
g M
s

&


d
A
is the outlet diameter, h is the height of the filling,

p
is the particle density, d
p
is the particle diameter, g
is the acceleration due to gravity. In addition there is
the influence of
the opening
angle in radian
measure, that
can be assumed
as follows [5.7]:

36 . 0

s
M
&


In addition there
is the effects of
the inner friction
coefficient of
the bulk
material. is
being equated to the tangens of the angle of repose :
= tan (5-1)
The influence of the bed height h on the outflow can
be neglected. The silo bottom pressure is changing
itself only at very small filling heights, and the outflow
becomes then non-stationary [5.6]. In an analysis of the
measurements from three studies [8.5 to 10.5] could the
very useful Eq. (5-2) for the calculation of the mass
flow be developed [5.11].

36 . 0
5 . 2 5 . 0

|
|
.
|

\
|
=
A p
p
A
s
d g
d
d
k M
&
(5-2)
As the in Fig. 5.4 depicted evaluation resulted, exists
between the proportionality factor k and the ratio of
outlet diameter to particle diameter approximately an
exponential relationship. The graph indicates also the
limits: at values of d
A
/d
p
less than 5 to 10 is the outflow
of bulk material no longer possible.
( ) 1 3 . 0
/ 1 . 0
p A
d d
p
A
e
d
d
k

=
|
|
.
|

\
|
(5-3)
The for the evaluation of the investigation results used
data, were determined in experiments with glass, sand,
lead, clay, resin and coal. The ranges of the individual
parameters are stated in Table 5.1. The particle sizes of
the investigated bulk solids range from 150 to 4000
microns.
With the help of the measurements by Taubman [5.12]
could be proven, that Eq. (5-2) together with Eq. (5-3)
also for much larger particle sizes can be applied. The
results from experiments using gravel with particle
sizes from 3 to 65 mm are reproduced in Fig. 5.5. The
measurments for three outlet diameters show good
agreement with the calculated values.
literature sign bulk sol. d
p
d
A

mm mm grd grd

[5.8]
sand
glass
sand
550
1100
2540
10
to
50
180 34
25
36

[5.9]
sand 160
to
910
6,6
to
11,9
60
to
180
35
to
42

5.10]
lead,
clay,
sand,
glas,
fertilizer,
coal
790
to
4000
10
to
58
180 24
to
39
Table 5.1
Range of variation of the individual parameters
in the investigations, evaluated in Fig. 5.4

Fig 5.4
Evaluation of eq. (5-2) with the help of
measured values for cohesionless
solids [5.8 bis 5.10]
Fig. 5.5
Comparison of the calculation according to eq. (5-2) with
the measured values on gravel [5.12]

23
In Eq. (5-2) is recognizable, that for the outflow of
bulk materials from vessels (silos) especially the
Froude number, formed using the outlet diameter d
A
as
characteristic length, plays a role:

Fr
u
g d
M
g d
A
s
p A
=

=

2 2
2 5
&

(5-4)

5.1.3 Measures against outflow restrictions
During the storing of bulk materials in silos are being
mainly discharge devices and discharge aids used as
additional facilities. Discharge devices serve the dosing
of outflowing bulk material and are thus at the same
time discharge and feeding device. In contrast to that,
discharge aids have the task to support the material
flow from the silo against the occurring disabilities.
Some discharge devices can also be used as discharge
aids..
The flow of the bulk material is always then hindered,
if the dimensions of the silo are not being attuned to the
bulk material properties. Various reasons may be
responsible. Either must the silo be installed without
any knowledge of the properties of the material, or the
silo could due to the spatial conditions not be designed
according to the known material properties. Very
different bulk materials are often being stored in one
and the same silo. Then could in case of bad flow
behavior of the material, discharge devices be needed,
in order to protect the bin against bridge formation and
blocking, and in order to change from a current funnel
flow to the mass flow, if necessary.
The flow behavior of bulk material in the bunker can
be changed fundamentally in two ways. One can
influence the material properties, or uses constructive
measures, to create a favourable flow profile of the
material in the bunker.
A way to affect the properties of the bulk materials, is
adding dispersants, also known as lubricants. Such
lubricants are magnesium oxide, aerosil, urea, or
diatomaceous earth with particle sizes down to 10-5
mm. The small particles distribute themselves among
the actual particles of the bulk material and prevent in
this way the cohesion effect.
Another way is the fluidization of the bulk solid by
blowing air, in order to loosen the packed bed up. This
measure is tantamount to the reducing the effective
friction angle of the bulk material. To execute this,
appropriate air distribution elements in the outlet
section or at the outlet cone must be installed.
The measures described above, however, are often not
suitable. On the one hand change the added dispersants
the quality and composition of the bulk material, on the
other hand can the loosened bulk material, for example,
only poorly be bagged, because the bulk density be
negatively affected through the additional air in the
material.
Tapping with hammers and poking around with lances
are for example measures, which the properties of the
bulk materials not influencing, and at the same time the
sliding of material on the wall of the outlet funnels
ensure. Much more convenient and without great
personnel effort can Vibrators be operated, which be
installed outside the silo wall and be moved by a
vibrator. Also can in silos inserted cushions be
pulsatingly inflated by compressed air.
With the material and the surface quality of the outlet
funnel can the wall friction angle between bulk solid
and hopper wall, and thus the critical dimensions of the
silo be influenced. Also with the help of plastic
coatings of the walls can one achieve something.
While measures, taking effect on the wall, are only be
useful, if mass flow occurs, can built-in discharge
devices in the silo also be used in case of funnel flow.
Also with bunker vibrators of different shapes and
arrangement, which are being moved from outside, and
with the help of agitators, can the flow of the bulk
material be stimulated. By stationary devices and with a
suitable shape of the hopper can the flow profile in the
bunker be likewise favourably influenced. Discharge
devices, which work at the same time as discharge aids,
are for example flat floors with rotating arm or
bottoms, which are elastically connected with an
vibrating outlet.
5.2 Mechanical movement of bulk materials
In case of mechanically moved bulk materials, for
example in screw conveyors, mixers and dryers, there
are apart from the flow properties of the bulk materials
additional parameters like the geometric conditions of
devices and mechanical equipments as well as the
manner of energy input. Therefore, it is in regard to the
required drive powers and attainable mixing times or
heat transfer coefficients, accordingly difficult to get
information by calculations. Basic researches, in
particular taking into account the aspects of scale-up,
are only in insufficient extent available, so that for the
planning appropriate documentation is missing. In
practice is the plant engineer therefore dependent on
the experiences of the respective device manufacturer.
The respective movement behavior of mechanically
moving bulk materials becomes for example indirectly
recognizable through the interpretation of heat transfer
measurements during the heating of bulk materials in

24
horizontal thin film dryers, which were published by
Klocke [5.13].
5.2.1 Heat transfer in horizontal thin film dryers
Thin layer contact apparatuses with vertical heating
surfaces have long been known for the vaporisation of
solutions with low viscosity and for the rectification or
distillation of fluid mixtures. Its rotating internals
create mechanically a thin liquid film along the inner
surface of a heated cylinder, and regenerate this film
constantly. In this way high heat and material transfer
can be achieved.
The thin film principle can also be used, for the
cooling, heating and drying of pasty materials, powders
and granules, as well as in order to perform reactions,
where solids and liquids are involved. Fig. 5.6 shows
the scheme of such
apparatus, that is
equipped with four
blade rows, looking
like paddles. The
effective surfaces of
the blade rows overlap
themselves, and the
residence time of the
material can by the
change of the angle of
attack be adjusted. To
increase the so-called
hold-up, it has proven
expedient, to arrange
the heating surfaces
horizontally. In addition, the design of the rotor was
modified accordingly to the granulate form of the
products, so that the material is distributed as evenly as
possible along the wall in order to improve the heat
exchange.
In Experiments in two facilities with diameters of 210
and 250 mm as well as heating surfaces of 1.2 and 1.5
m
2
, five bulk materials were heated. In all experiments
was the orientation of the blades identical: "transport in
flow direction" in the entry zone, "transport against the
flow direction" in the discharge zone and "neutral
position" in the remaining area. Would be more
"transport against the flow" chosen, so that the dwell
time would increase, the wall coverage could be
certainly improved, on the other hand would the drive
power likewise grow considerably. Under consideration
of the heat transfer on the one hand and the necessary
drive power on the other hand, the largely neutral
position of the blades has itself proven, as far as in case
the use of drying gas not an additional conveying effect
occurs.
Fig. 5.7 depicts the impact of the short-term contact
during the heat transfer between the bulk material and
the surface, as well as the variation of rotor speed.
Represented is the size of the heat transfer coefficient
as function of the contact time t. The contact time is in
thin film contact devices the time between two
rearrangements at the wall. Its size depends on the rotor
speed n and the number Z of blade rows:

t
Z n
=

60
(5-5)

The relationship between the Froude number Fr, for
which the diameter D of the device is used, and the
contact time t looks like follow:

Fr
g
D
t Z
=

2 2 /
(5-6)

Due to the circulation of the product ring at the wall
of the thin film device, is the actual contact time of the
product greater than the value of t, which is being
calculated by Eq. (5-5). This fact is here ignored.
The measurement results for the heat transfer
coefficient, related on the whole heated area, which
were determined at constant throughput and variable
rotor speed, can be interpreted according to Fig 5.7 as
follows. In the region of large contact times (low
rotation speeds), grows the material coverage
significantly with growing rotation speed, and the heat
transfer rises steeply. After the operating point P
K1
is
reached, the increase of heat transfer corresponds to the

Fig. 5.7
Heat transfer coefficient in dependency of the contact
time t in a thin film dryer during heating of various bulk
materials

Fig 5.6
Horizontal thin film dryer
with four blade rows (picture
credits: BSH)

25
t - law for short-term contact; the material coverage
remains seemingly constant.
The optimum operating point is being reached at P
K2

because by the further increase of the rotation speed
(shorter contact times) the heat transfer is no longer
improved. Increasing frictional resistance of the bulk
material on the wall or changed conditions for the force
transmission between blades and bulk solid seem to
prevent a further reduction of the contact time between
the bulk solid and the wall.
Additional conclusions are possible, if one tries to
summarize the measurement values with a single curve.
Fig. 5.8 shows such a presentation, in which in addition
to the heat transfer coefficient and the modified
contact time, the influence of the heat penetration
coefficient (c)
Sch
, the mass flow rate
&
M
s
and the
diameter d
p
of the particles is becoming recognisable.
As Fig. 5.8 shows, grows the heat transfer coefficient
with increasing bulk material throughput steeply -
namely with
&
M . Due to the larger quantity of bulk
material is obviously also a larger material volume for
the coverage of the apparatus wall present.
As expected, leads a greater heat penetration
coefficient of the bulk material likewise to an increased
heat transfer. The numeric value of 0.33 for the
exponent, is already in other apparative constellations
been determined.
The effect of the particle diameter that influences the
contact time inversely proportional with the third root ,
corresponds to the behavior of bulk solids during the
outflow from silos. There, smaller particle sizes causing
larger material discharges.
Obviously, eases the larger number of particles per
volume unit its relocation, and less resistance is being
opposed to the shape change. In thin film contact
apparatuses has this mechanism an opposite effect. In
case of smaller particles, the frictional connection
between the blades and the bulk solid is being
deteriorated, so that higher rotor speeds are required, to
achieve similar heat transfer rates.



Literature of chapter 5.
[5.1] Jenike, A.W.: Gravity flow of bulk solids. Engineering Experiment
Station Bulletin 108. University of Utah, 1961.
[5.2] Molerus, O.: Fluid-Feststoffstrmungen. Berlin, Heidelberg, New
York: Springer-Verlag, 1982.
[5.3] Schwedes, J.: Entwicklung der Schttguttechnik seit 1974. Aufberei-
tungstechnik 23(1982)8, S.403-410.
[5.4] Jenike, A.W.: Das Flieen und Lagern schwerflieender Schttgter -
Ein berblick. Aufbereitungstechnik 23(1982)8, S.411-421.
[5.5] Rumpf, H.: Mechanische Verfahrenstechnik. Mnchen, Wien: Carl
Hanser Verlag, 1975.
[5.6] Brauer, H.: Grundlagen der Einphasen- und Mehrphasenstrmung.
Aarau, Frankfurt a. M.: Verlag Sauerlnder, 1971.
[5.7] Riedel, K.: Der Ausflu von Schttgtern aus Bunkern. Studienarbeit
am Lehrstuhl fr Thermodynamik und Verfahrenstechnik der TU
Berlin, 1965.
[5.8] Brown, R.L.; und J.C. Richards: Exploratory study of the flow of
granules through apertures. Transactions of Institution of Chemical
Engineers 37(1959)2, S.108-119.
[5.9] Rose, H.E.; und T. Tanaka: Rate of discharge of granular materials
from bins and hopper. Engineer 208(1959)5413, S.465-469.
[5.10] Franklin, F.C.; und L.N. Johanson: Flow of granular materials
through a circular orifice. Chemical Engineering Science 4(1955)3,
S.465-469.
[5.11] Heyde, M.: Merkmale und Flieverhalten von Schttgutmassen.
Maschinenmarkt 89(1983)46, S.1047-1050.
[5.12] Taubmann, H.: Technologie der Schttgter. Aufbereitungstechnik
23(1982)2, S.77-83.
[5.13] Klocke, H.-J.: Wrmebergang im Dnnschichtkontakttrocknern - ein
Beitrag zur Vorausberechnung und bertragung von Versuchswerten
auf Betriebsverhltnisse. Vortrag im GVC-Fachausschu Trocknung-
stechnik, 10./11.4.1975

Fig. 5.8
Summarizing presentation of the heat transfer in thin
film contact dryers

26
6 Single-phase flow through pipes
One speaks of flow, if movements of liquids and gases
are being caused by the total pressure difference. In
contrast, is by partial pressure differences caused mo-
tion known as diffusion.
The one-dimensional representation of a flow process
is based on the assumption of the mass flow through a
given cross-section. The relationship for the flow rate
of a liquid with its constant density is accordingly:

u
M
f
=

&

(6-1)

If the pressure differences are small, the rules of the
hydraulic transport apply also to gases. In case of larger
pressure differences, the gas density changes itself in
dependence of pressure and temperature. Processes of
this kind are dealt with in the gas dynamics.
6.1 Continuity and energy conservation
In the case of the one-dimensional steady state pipe
flow, applies for the mass flow
&
M in the axial direction
at the points 1 and 2:

&
M u f u f = =
1 1 1 2 2 2
(6-2)

The mass flow
&
M thus remains constant; the Eq. (6-2)
is therefore referred to as continuity equation. If the
pipe cross section remains equal (f
1
= f
2
), the current
density changes neither:

&
M
f
u u u = =
1 1 2 2
oder = const. (6-3)

In case of an incompressible fluid, the flow velocity u
remains thus equal.
The integration of the general Euler Equation for the
one-dimensional, unsteady case at constant density,
results in the Bernulli Equation:

1
2
1
2
1
2
1 1 2
2
2 2
+ + = + + +

u p g z u p g z p
(6-4)

The Bernoulli equation connects the dynamic
pressure, the initial gage pressure, and the static
pressure with each other. During the frictionless pipe
flow remains the sum of the three pressures along the
pipe constant. In the case of the viscous pipe flow, this
sum is being reduced by the amount of the pressure
loss.
Dividing Eq. (6-4) by the fluid density leads to a
form, in which the sum of energies per unit mass is
represented, namely the kinetic energy, the pressure
energy and the potential energy.

1
2
1
2
1
2
1
1 2
2
2
2
+ + = + + + u
p
g z u
p
g z
p

(6-5)

Along the flow path, a part of the energy is
transformed into heat and must be registered as a loss.
If one relates each term of the the Eq. (6-5) to the
acceleration due to gravity g, results a equation with
altitude values, of which h

the friction head loss is.

u
g
p
g
z
u
g
p
g
z h h
ges
1
2
1
1
2
2
2
2
2 2
+

+ = +

+ + =


(6-6)

This form is especially suitable for the graphical
representation of the flow process; at every point of the
flow, the sum of velocity head, pressure altitude,
geodetic height and friction head loss is equal to the
constant total height h
ges
.
That additional energy is needed for the acceleration of
the fluid on its flow velocity, is often forgotten in
practice. In case of small frictional pressure losses, the
dynamic component of the total pressure loss gets more
and more weight. If this proportion would neglected in
specifying the blower performance, this would have
unpleasant consequences for the operation. Also,
pressure loss measurements on facility parts should be
performed at places with equal flow cross section, so
that the measurement results are directly comparable.
6.2 Flow pressure loss
Basically, one must distinguish between the laminar
and turbulent flow. In laminar state, the fluid layers
move side by side without any exchange of fluid
elements . In this form of flow, the shear stress at the
boundary between two fluid layers alone by the
viscosity of the fluid is determined. In contrast to this,
the wall friction generates during the turbulent flow
"vortex bales", which wander to the pipe axis. The
pressure loss is thereby being caused by the
acceleration and deceleration of the vortex bales.
The relation between inertia and resistance force in a
pipe flow decides whether the flow is turbulent or
laminar. The significant Reynolds Number is being
formed with the pipe diameter D and the mean velocity
u:

27
Re =
u D

(6-7)
In turbulent pipe flows are velocity fluctuations and
highly disordered motion typical, which be not
suppressed by the viscous forces. Flows in pipes at
Reynolds numbers larger than 5000 are typically (but
not necessarily) turbulent.
The pressure loss p of a fluid, which flows along a
pipe section with diameter D and length L, can be
calculated with the help the general resistance law:

p
u L
D
=



2
2
(6-8)

The resistance factor for the theoretically
describable laminar flow can for Reynolds numbers
smaller than ca 5000 be calculated after following
equation:
=
64
Re
(6-9)
For the turbulent flow, the resistance factor is only
with the help of measurements to determine. Because
the conditions at the pipe wall have the greatest
importance, is in particular the influence of the wall
roughness noticeable.
Fig. 6.1 shows the resistance coefficients for the
laminar and turbulent flow in pipes. For different
roughnesses of the pipe wall there are separate graph
curves, depending on the ratio of the diameter D to the
mean pipe roughness factor k.
The turbulent flow in technically smooth pipes
depends only on the Reynolds number. Between Re =
2300 and Re = 10
5
applies for the resistance coefficient
the empirical equation of Blasius [6.1]:


Re
3164 , 0
25 . 0
=
(6-10)
Up to arbitrarily high Reynolds numbers can, based on
the theory of Prandtl and Karman, be an implicit form
of spedified:

1
4
1
2
51 . 2
Re
log
(
(
(
(

|
|
.
|

\
|
=

(6-11)

This function represents in Fig. 6.1 the left boundary
of the turbulent range, which is referred to as
hydraulically smooth.
Additional pressure losses are being caused by pipe
bends and by internals. Especially the shedding of the
flow leads to large pressure losses, because the cinetic
energy, which is needed for the acceleration of the
fluid, can no longer be recovered.
Special cases are shown in Fig. 6.2: the sudden
constriction of the flow cross section from f to f
0
and
the sudden extention from f
0
auf f.
For the pressure loss applies for suddenly constricted
pipes
p
f
f
u
e
=
|
\

|
.
|
0
2
0
2
1
2

(6-12)
and for suddenly extended pipes
p
f
f
u =
|
\

|
.
|
1
2
0
2
0
2

(6-13)
The ratio f f u u
e e
/ /
0 0
= is named contraction
coefficient . f
e
is the narrowest flow cross section.
For sharp edged openings apply the following
contraction coefficient [6.2]:
61 . 0
2
=
+
=

(6-14)

Fig. 6.2
Suddenly constricted tube (left) and
suddenly exstended tube (right)
Fig. 6.1
Resistance coefficient as a function of the Reynolds
number for various ratios of D/k

28
6.3 Heat transfer
For laminar pipe flow, the heat transfer conditions can be
theoretically described. For a long pipe, in which velocity and
temperature profiles are fully develloped, one obtains for the
Nusselt number, by using the pipe diameter D, constant
values:
66 . 3 =

D
Nu
(6-15)
In case of a very short pipe, it results:

Pr Re 664 . 0
3 / 1 2 / 1
=

D
Nu (6-16)

In case of turbulent pipe flow, approaches must be used, in
which the correlations between the dimensionless parameters
have empirically been determined. According to Hausen
[6.3] applies for Re greater than 2300 the following equation:

(
(

|
.
|

\
|
+ =

=
3 / 2
42 . 0 4 / 3
1 Pr ) 180 (Re 037 . 0
L
D D
Nu

(6-17)
One should realize, that upon heating a substance in a
heated tube, the temperature difference between the
wall and the materials, which are flowing on the inside
and the outside, is inversely proportional to the size of
the heat transfer coefficient. During heating of crude oil
before the distillation is, for example, water vapor
being blown into the pipes in order to generate
additional velocity and turbulence in the pipe. In this
way, the wall temperature on the oil side and the
temperature of the oil flow are being adjusted to each
other, to avoid deposits of cracked oil on the tube wall.
6.4 Turbulence and exchange behavior
Generally valid equations for the heat transfer in the
pipe or broader insights into the turbulence dependent
pneumatic conveying process are only to obtain with
the help of detailed considerations. Prandtl made the
following picture: during turbulent flow arise "fluid
bales" which have an own mobility, and which
themselves over a certain distance in the longitudinal
and transverse directions as cohesive entities move.
Such a fluid bale covers a distance relative to the
surrounding fluid, proportional to its diameter, which is
referred to as mixing length, before it itself with the
surrounding fluid mixes and its individuality looses. If
such a fluid bale is due to a transverse movement being
moved from its original location to a neighboring
volumen, then its velocity is compared to its new
surroundings higher or lower. The displacement effect
of the fluid bales causes fluctuations, which are of the
same magnitude as the speed differences. These
considerations are the basis for Prandtl's Mixing Length
Hypothesis and the approach for the turbulent shear
stress
turb
[6.4] :

turb
l
u
y
u
y
=
2




(6-18)

In the usual approaches, the analogy to Newton's law
of viscosity is searched by summing up the proportions
of the molecular and the turbulent shear stresses. In
contrast, here is assumed, that the turbulent shear stress
itself proportionally to the velocity gradient and the
viscosity of the flowing Fluid alters:

turb
k
u
y
=


(6-19)

Thus, one can insert the Eq. (6-19) into the Eq. (6-18),
and one gets an expression for the turbulent shear
stress, in which the gradient of the flow velocity is no
longer included:

turb
k
l
=

2 2
2
(6-20)

6.4.1 Pressure loss
The relationship between the wall shear stress in the
pipe and the pressure loss is as follows:
p
L D
w
=
4
(6-21)
The dimension
of the mixing
length can be
introduced into
Eq. (6-21) with
the help of the
shear stress in
accordance to Eq.
(6-20). The
mixing length
relates to the flow
state along the
pipe wall and ha
the symbol l
w
.
p k
L
l D
k k
w
=


=

2
2
2
4 mit (6-22)

Fig. 6.3
Ratio between pipe diameter and
wall-related mixing length l
w
, plottet
against the Reynolds number Re

29
Is the proportionality factor k chosen such, that the
wall-related mixing length at a Reynolds number of
5000 is equal to the tube diameter, then give the
pressure loss measurements by Nusselt [6.5] the
function in Fig. 6.3. Plotted is the ratio between the
pipe diameter D and the wall-related mixing length l
w

against the Reynolds number Re / = u D . The
proportionality factor has the value k=4.5 10
5
.
The functional relationship in Fig. 6.3 can be
mathematically expressed as follows:

Re 10 64 . 4
9 . 0 4
=

w
l
D
(6-23)
Insertion of Eq. (6-23) into Eq. (6-22), gives the
following equation for the pressure loss p in a
hydraulically smooth pipe:

Re 097 , 0
3
2
8 . 1
D
L
p

(6-24)

The validity of Eq. (6-24) can easy be verified by
comparison with the usual formulation:

3
2
8 . 1 2
Re 097 , 0
2 D
L
u
D
L

This comparison gives for the friction coefficient :

Re 194 . 0
2 . 0
= (6-25)
The graph for the friction according to Eq. (6-25)
has a slightly flatter course than that after Blasius and
averages virtually the boundary curve for the
hydraulically smooth pipe in the Range of Re = 2300 an
Re = 10
7
.
6.4.2 Velocity profile
Reflections regarding the velocity profile, which itself
in the pipe developes, confine themselves usually to
the area, in which the turbulent exchange authoritative
is. The area of the pipe wall with its laminar rules is left
disregarded.
For the zone of turbulent mixing, the Eq. (6-18) and
(6-19) can be equated. The result is an expression for
the velocity gradient u y / :

u
y
k
l
=

2
(6-26)
Because the mixing length l inside the pipe is
depending on the wall-related mixing length l
w
, the
searched law for the velocity distribution must be
precisely built up like Prandtl's "law-of-the-wall" [6.6].
This requirement can be fulfilled by the choice of an
appropriate relationship between the mixing length l in
the distance y from the wall and the wall-related mixing
length l
w
:
l
l
y
l
w w
2
2
=

=

proportionality factor (6-27)

Insertion into Eq. (6-26) gives for the velocity
gradient:

u
y
k
l y
w
=


1
(6-28)

Designates one the expression k l
w
/ ( ) , which has
the dimension of a velocity, with u

, provides the
integration of the Eq. (6-28) the well-known expression
[6.6]:

u u y C = +

( ln )
1
(6-29)

With the ratios u/u

and y/l
w
Eq. (6-29) becomes
u
u
f
y
l
y
l
C
w w

=
|
\

|
.
| = +
1
ln (6-30)

By twice inserting the maximum speed in the middle
of the tube (calculated using the velocity ratios, which
are to be found in the Dubbel [6.7]), was for a value
of 0.44 and for C a value of 20.5 determined. Eq. (6-30)
is now:
5 . 20 ln
44 . 0
1
|
|
.
|

\
|
+ =
w
l
y
u u

(6-31)
6.4.3 Heat transfer
The physical ideas about the heat transfer in the pipe
flow consider the conditions in the boundary zone as
crucial for the amount of heat, that is between pipe
wall and fluid exchanged. The heat transport through
turbulent exchange surpasses the molecular Exchange
till close to the wall by far, so that the temperature
gradient in this zone is small. In contrast to that, are in
the adhering wall layer due to the missing exchange
very steep temperature gradients present. On the
boundary of these two zones exist a speed u and a
temperature T.

30
This model concept can be used for deriving a
relationship for the dimensionless Nusselt number Nu
[6.6]:
( ) | |
Nu
D D
u u
w
=

=

+

1
(6-32)
=

1
m Pr

Eq. (6-32) contains the unknown velocity u on the
boundary of the two zones, and also the unknown ratio
m A A
q
= /

between the apparent thermal conductivity
A
q
and the apparent viscosity A

. Eq. (6-30) now offers

the opportunity, to connect u with the conditions on
the pipe wall. For this, the wall-related mixing length
l
w
is used in place of y.

5 . 20
337
1 ln
1

= + =
|
.
|

\
|
w
l
C u u

(6-33)

After inserting the Eq. (6-23) for l
w
in the above
equation, the expression for u is:

Re 2 . 3
9 . 0
D
u

(6-34)

Using the expression u D = Re / for the mean flow
velocity in the pipe, the relationship between u and u is
dependent on the Reynolds number:
Re 2 . 3
1 . 0
=

u
u
(6-35)
Taking into account that Eq. (6-20) with l = l
w

describes the wall shear stress, can Eq. (6-32) be
rephrased:

Re 0076 . 0
1 . 0
Re
Pr 2 . 3
1 . 0
Re 1
8 . 0

=
m
D
Nu

(6-36)
The calculation results from above equation, are for
the Prandtl number 1 equal to the published
measurement values, if m = 6.25 is used. In case of
other Prandtl numbers there are deviations, which can
be corrected by the factor Pr
0.4
For the accordance in
case of small Nusselt numbers, must still the laminar
proportion (Nu5) be taken into account, so that after a
small rephrasing generally applies:

05 . 0 Pr Re 05 . 0
Pr Re 0076 . 0
5
1 . 0
4 . 1 9 . 0
+

+ =

D
Nu (6-37)

Fig. 6.4 shows the accordance between the Eq. (6-37)
and the measurement results from a large number of
researches, which have been compiled by Churchill
[6.8]. It is at all cases fully developed turbulence in
smooth pipes at moderate temperature differences. The
equation captures not only the area, where the
measured values in the double logarithmic system
straight run, but also the curved curves at Prandtl
numbers below Pr = 0.72.



Literature of chapter 6.
[6.1] Blasius, H.: Das hnlichkeitsgesetz bei Reibungsvorgngen in Fls-
sigkeiten. VDI-Z. 56(1912).
[6.2] Mersmann, A.: Thermische Verfahrenstechnik. Berlin, Heidelberg,
New York: Springer-Verlag, 1980.
[6.3] Hausen, H.: Neue Gleichung fr die Wrmebertragung bei freier oder
erzwungener Strmung. Allg. Wrmetech. 9(1959), S.75-79.
[6.4] Prandtl, L.: Z. angew. Math. Mech. 5(1925), S.136-139.
[6.5] Nusselt, W.: VDI-Forschungsheft Nr. 89, 1910.
[6.6] Prandtl, L.; K. Oswatitsch und K. Wieghardt: Strmungslehre, 7. Au-
flage. Braunschweig: Vieweg-Verlag, 1969.
[6.7] Dubbel: Taschenbuch fr den Maschinenbau. Bd. 1, 13. Auflage. Ber-
lin, Heidelberg, New York: Springer-Verlag, 1970.
[6.8] Churchill, W.S.: Ind. Eng. Chem., Fundam. 16(1977)1, S.109-116.
Fig. 6.4
Comparison of measurement results with the
calculation after Eq. (6-37)

31
7 Pneumatic conveying of
bulk materials
Through pipelines are most commonly liquids and
gases sent, but also powders and bulk materials,
whereby in case of pneumatic conveying mostly by
blowers or fans supplied air or compressed air is used.
The pneumatic conveyance utilizes the capability of
flowing gases to transport under certain preconditions
solids with their higher specific weight. This technique
is however only usable for relatively short distances of
up to some hundred meters.
Despite some disadvantages, which are associated
with pneumatic conveying of bulk materials, there are
some reasons, which make the difference of this way of
bulk material transport especially in small and medium-
sized internal conveying systems. It is often the only
viable alternative, for example in case of tight spaces in
mining or in case of subsequent automation of transport
processes, if not enough space for a mechanical solu-
tion is available. Weber [7.1] listed the pros and cons.
The operation of pneumatic conveying systems de-
mands a considerable amount of devices, which func-
tionality is owed the several decades of experience of
the manufacturers.
7.1 Conveying states
During the pneumatic conveying of bulk material
through a tube, the material is, depending on the re-
spective flow conditions, in very different ways across
the cross section distributed. This includes: uniform
distribution over the pipe cross-section, subsets as
strands on the bottom of the pipe, as well as bales,
dunes and plugs. The different flow states Cambric
[7.2] has summarized in a statechart diagram: therein is
the pressure loss at constant material throughputs plot-
ted against the gas velocity in the empty pipe cross sec-
tion (Fig. 7.1).
At large velocities and respective large pressure
losses, the state of dilute phase conveying is present, in
which the solid particles pass the conveying line float-
ing and jumping. With decreasing gas velocities, the
pressure loss graph passes a minimum, at which parts
of the loose material be deposited on the bottom of the
pipe and be conveyed as strands. At even smaller flow
rates, the strands cumulate to dunes or fill as bales and
plugs partially the entire cross section of the pipe. In
case of the further decrease of the gas velocity the pres-
sure loss rises again.
To note is that in the region of the so called dense
phase on the left, the required blower performance de-
creases despite the increase in the pressure loss, be-
cause the effect of the decreasing air flow is much lar-
ger, at least up to certain gas velocities. In addition, the
considerably more gentle transport due to the lower
mechanical stress on the pipe walls and the particles
plays a role. For this reason one is today strived, indus-
Advantages Disadvantages
Simplicity relatively high energy demand
Customizability Wear
low space need Abrasion
easy routing Danger of clogging
Possibility of branching relatively low flexibility
Controllability
can be automated Restriction as to the suitability of
conveyed materials
Availability
low environmental impact
Low inflation rate possibly complicated particle
processing
Serviceability
Integrability
linkable to processes possibly costly dust separation
Table 7.1
Advantages and disadvantages of pneumatic conveying
installations
Fig. 7.1
Diagram of pneumatic conveyance of pourable bulk ma-
terials with particle sizes larger than 500 m

32
trial conveying installations under dense phase condi-
tions to operate.
After Krambrock [7.2] can most bulk materials with
particle sizes over 50 m be in simple, smooth pipe-
lines reliable conveyed under dense conditions. In re-
cent years, it has been found, that particularly bulk ma-
terials with narrow particle size distributions and aver-
age particle sizes between 0.5 and 5 mm can be con-
veyed without major problems even if the gas velocity,
related on the free cross-section, is smaller than the
sinking velocity of the individual particles.
In the diagram are two characteristic conveying condi-
tions recognizable. In case of dilute phase conveying, is
after passing through the minimum of the pressure loss
the point reached, where the bulk material possibly
clogs the pipeline. Two expressions are being com-
monly used for this boundary.
"Choking velocity" means the minimum velocity that
is required to maintain solids in a vertical conveying
line in the dilute-phase mode (its value is influenced by
the particle's terminal velocity).
"Saltation velocity" means the
minimum velocity that is re-
quired to maintain solids in a
horizontal conveying line in the
dilute-phase mode. The value of
this is 3 to 6 times the "Choking
velocity".
Mostly, however, this condition
is not the end of the pneumatic
conveying, but the start of unsta-
ble conditions, which with the
help of a high blower power can
be bridged. Further lowering of
the air velocity, however, en-
ables often again a stable opera-
tion at constant pressure loss. Fi-
nally however the conveying
comes to a standstill. This condi-
tion is represented by the left
boundary of the statechart.
Not all bulk materials can be
conveyed in dense phase without
any problems, so that especially
for fine, adherent and plug-
forming materials additional
measures be needed - for exam-
ple the use of secondary air.
7.2 Pressure loss
In pneumatic conveying lines occurring pressure loss
are caused by the opposite to the flow acting resis-
tances. Wall friction and pipe bends cause the pressure
loss of the pure airflow. The additional losses are
caused mainly by the acceleration of the bulk material,
the wall friction and the material weight.
Usually, one tries to capture the individual proportions
of the pressure loss with the help of resistance coeffi-
cients, as they are used for the calculation of the pure
gas flow. Really satisfying results, which take in par-
ticular the dense phase state into account, however, are
up until now not available.
7.2.1 Fine-grain material in horizontal pipes
Muschelknautz and Krambrock [7.3] as well as
Muschelknautz and Wojahn [7.4] published the condi-
tions for the conveyance of fine-grained bulk materials
through horizontal pipelines. These data sets are sum-
marized in table 7.2. More measurement results were
published by: Bohnet [7.5], Krambrock [7.2], Matsu-
No. Material dp50
m
L
m
h
m
D
mm
&
M
s

t/h
&
M
g

Nm
3
/h
p
bar
u
m/s

-
1 Cement raw flour 60 146 - 36 1.4 11 2 3.15 102
2 PAN powder 75 150 - 36 1.2 19 1.4 3.95 53
3 Soda 100 200 30 100 35 1050 2.5 18 28
4 Filler (SiO
2
) 10 30 - 65 4 100 0.3 7.3 33
5 Moist sand 200 120 - 100 19.5 300 3.5 3.75 54
6 Filler (SiO
2
) 15 58 - 70 1.4 30 0.14 2.1 39
7 Filler (SiO
2
) 15 10 - 70 1 29 0.023 2.25 29
8 Organic material 275 37 1.5 66 4.1 63 0.7 3.6 54
9 Soda 75 25 - 65 20 200 0.3 13 83
10 Fly ash 15 1200 15 200 50 4000 3.5 13 10
11 PVC powder 60 115 20 70 7 240 0.65 13.5 24
12 Rock salt (hose) 700 26 2 50 5.8 110 0.8 11 44
13 Organic material 150 6,5 - 20 0.4 8 0.17 6.5 41
14 Organic material 80 25 10 70 17 300 1.2 13 47
15 Clay 45 130 15 100 8 300 2.5 5 22
16 Plastic powder 40 300 - 100 10 310 1.5 6.8 27
17 Rubber chips 25000 45 6 125 1 1100 0.04 27 0,8
18 Plastic granules 2000 30 12 65 1 200 0.14 18 4
19 Kontakt (catalyst) 1200 - 16 100 9 460 0.1 17.6 16
20 Abbrand (burnup) 120 92 20 95 4 300 0.8 12.5 11
Table 7.2
Data of industrial pneumatic conveying installations

33
moto [7.6], Molerus [7.7] and Siegel [7.8].
Pneumatic transport of solids mainly takes place by
transversal movements in the gas flow, and is therefore
closely linked to the laws of turbulent pipe flow. This
fact can be accounted with the help of parameters,
which characterize the turbulent flow condition. In ad-
dition to the Reynolds number, which is being formed
using the pipe diameter, comes yet the ratio between
the diameter D and the wall-related mixing length l
w
in
question. This ratio can be specified according to the
Eq. (6-23) and (6-24) as a function of the pressure loss
p, the pipe length L, the gas density
g
, the pipe di-
ameter D and the dynamic viscosity :

D
l
p
L
D
w
g
=

0 0015
3
2
,

(7-1)

In Fig. 7.2, the values of D/l
w
, calculated using the
data from the above-mentioned publications, are plotted
against the Reynolds number. With the help of the sin-
gle points and graphs has been constructed a general
statechart similar to that of Kram-
brock [7.2]. This diagram shows
that industrial conveying installa-
tions are mostly be operated in
dense phase mode. The importance
of the turbulent flow condition is
underlined by the fact, that the dia-
gram is located in the area above
the critical Reynolds number of
2300. Regarding the operating
conditions should be noted, that
two conveying installations with
relatively large throughputs be-
cause of not recognizable reasons
be operated not in the dense phase
region, but on the boundary of the
dilute phase region.
With the help of the plotted val-
ues and graphs, it is possible, to
estimate the lines of equal material
throughput. However, the pressure
losses, which were published by
Molerus, are too low, because the
particles of the used bulk material
are very large in size. The rolling
particles offer only low resistance
against the gas flow. Moreover,
one has to be aware that in most
researches only the pressure losses
for the non-accelerated flow in a
test assembly were measured. Un-
der such conditions occurring pres-
sure gradients are not directly comparable with those of
industrial convey-
ing installations.
The left bound-
ary of the general
statechart in Fig.
7.2, which repre-
sents the maxi-
mum material
throughputs, in-
cludes the well-
known, by Fig. 7.3
illustrated fact,
that in case of
constant pressure
loss, the maximum
amount of mate-
rial with increasing pipe length is becoming smaller
[7.2].

Fig. 7.3
Maximum material throughput
&
M
s
of PE granules dependent on
the pipe diameter D and the pipe
length L

Fig. 7.2
General statechart for the pneumatic conveying of pourable bulk materials, con-
structed on the basis of characteristic parameters for the turbulent pipe flow,
using measured values of various researchers and data of industrial facilities.

34
7.2.2 Material acceleration at the beginning
and after pipe bends
After the product was fed into the pipeline, and after
pipe bends, the bulk solid is accelerated until reaching
its final velocity. The required energy is taken from the
airflow, so that an additional pressure loss occurs. In
short conveying lines, this pressure loss can be the
main proportion.
Reference values for the size of these additional pres-
sure losses can be found in the literature. After the
feeding the material in horizontal pipelines requires the
acceleration of 1 kg of material per kg of air, depending
on the particle diameter, the following pressure losses:
250 N/m
2
(d
p
= 8000 m) or 550 N/m
2
(d
p
= 1000
m) [7.8]. These values apply to a gas velocity of 20
m/s and a particle density of 1000 kg/m
3
.
In case of the straight up directed conveying of bulk
solids are about 100 N/m
2
for the acceleration of 1 kg
of solids per kg of air necessary [6]. This value was de-
termined at a gas velocity of 23 m/s and for a particle
density of 2600 kg/m
3
.
The required pressure losses for the acceleration of
the bulk solids in-
crease itself in gen-
eral with decreasing
settling velocities of
the particles. The
reason lies in the
fact that smaller or
lighter particles at
constant mass flow
have higher end
speeds.
During the flow in
pipe bends, the bulk
solid is by the cen-
trifugal forces being
pressed outward and
moves along the pipe wall. This results in increased
friction between the particles and the tube wall, so that
the velocity of the material decreases. In the subsequent
straight pipeline, the solid is then again accelerated to
the speed of the steady state. In this way, the pressure
loss p
Kr
is generated.
After Schuchart [7.10] the additional pressure loss
p
Kr
for horizontal deflections is dependent on the ra-
tio 2R/D between pipe bend radius R and tube diameter
D. The diagram in Fig. 7.4 depicts the related pressure
loss for the pipe bend in relation to the related pressure
loss in a straight pipe for air with and without solids
loading.
For the frequently used ratio of 2R/D=10 is the pres-
sure drop for the pure air flow by a factor of 2, and for
the bulk solid conveying by a factor of 15 greater than
in a straight pipe. However, it remains unclear, on what
basis the specified ratios be applied in the estimation of
pressure losses.
7.3 Heat transfer at the pipe wall
Also regarding the heat transfer at the pipe wall, the
presence of the bulk solid is during the pneumatic con-
veyance noticeable. Depending on the flow conditions,
the heat transfer coefficients are higher or lower, com-
pared to the pure airflow.
7.3.1 Conveying vertically upward
During the vertical steady-state pneumatic convey-
ance, in addition to the pipe diameter, especially the
particles size of the bulk material and the gas velocity
have influence on the surface-related heat transfer coef-
ficient. Fig. 7.5 shows the differently orientated meas-
urement graphs of the heat transfer coefficients plotted
over the solid/gas-ratio [7.11]. The gas velocity in
these investigations has been kept constant.
For sand with densities of 2600 (left chart), low mate-
rial loads cause obviously an impulse loss, and the re-
duced turbulence leads to reduced heat transfer. The

Fig. 7.4
Dependence of the pressure loss
ratio on the diameter ratio 2R / D
in horizontal pipe bends [7.10]

Fig. 7.6
By the solids fraction caused additional heat transfer
-
g
during the stationary vertical pneumatic convey-
ance

35
particles-wall-contact is not able to compensate this
deficit. Only at higher material loadings, the value for
the pure gas flow is reached again and with increasing
loading also outbid. With decreasing particle size
rises faster and the rise takes place sooner.
The measurement values by Stockburger [7.12] in the
right diagram, for which particles of smaller density
were used, show a slightly different characteristic.
Because of the lighter particles, a reduction of the heat
transfer under the initial value is not noticable. The
gradient of the heat transfer coefficient in dependance
on the solid/gas rate, reduces itself with higher gas
velocities. The particle size d
p
was constant during
these investigations, and the particle density was 1000
kg/m
3
.
As already shown [7.13 and 7.14], must be assumed,
that the contact times between particles and wall be
located in the amax region, so that the contact time
itself plays no role for the size of the heat transfer
between wall and particles. Thus, can the influence of
the particle size be considered by the use of the
maximum value for a single particle according to Eq.
(3-5). This tool can be used for the summarizing
presentation of the by the flow conditions influenced
changes of the heat transfer coefficient.
In Fig. 7.6 is the heat transfer coefficient -
g
, that is
caused only by the material proportion, plotted over an
expression, that apart from the maximum heat transfer
coefficient for a single particle
max,p
, takes the
influences of the solid/gas ratio , the Froude number
Fr as well as the density ratio
p
/
g
takes into account.
The Froude number is being formed using the
difference between the gas velocity u and the sinking
velocity u
t
of a single particle as well as the pipe
diameter D ; g means the acceleration due to gravity. In
the evaluated measurements from five researches [7.11,
7.12, 7.15 bis 7.17] vary the particle diameters between
30 and 570 m and the tube diameters between 17 and
50 mm.
The influence of the particle size on the increase of
the heat transfer becomes in the diagram as parameter
visible. Really significant improvements of the heat
transfer are after that only be expected in case of
particle sizes below 80 m. The reason for this is
mostly the low average volume concentration of the
solid in case of the dilute phase conveying. The solids
concentration on the pipe wall is therefore also small.
Particle diameters of 50 m show according to the
diagram the most favourable results. For a material load
of 10 kg/kg, that at a particle density of 2600 kg/m
3

corresponds to a mean volume concentration of 0.4
percent, the proportion on the maximum value of heat
transfer by a packed bed lies according to Eq. (3-6) at
about 4 percent. Are the particle diameters greater than
200 m, this percentage drops below 1 percent.
7.3.2 Cyclone separator
Increased solid concentrations therefore improve the
surface-related heat exchange, for example in cyclone
separators, in which the material, due to the acting
centrifugal forces, preferably is moved along the wall.
Fig. 7.7 shows measurements in a cyclone separator
with a diameter of 100 mm and an overall height of 330
mm, which were published by Szkely and Carr [7.18].
The particle sizes range between 150 and 1200 m, and
the densities of the solids between 2600 and 8800
kg/m
3
. Plotted is the ratio and the measured heat
transfer coefficients to the maximum value
max,p
(1-

0
) for the packed bed (moving bed), according to Eq.
(3-6), against a non-dimensionless expression, using the

Fig. 7.5
Influence of the solids loading , the particle diameter d
p
as well as the gas velocity u on the heat transfer coefficien
at the tube wall

Fig. 7.7
The ratio between the measured value
exp
and the
maximum heat transfer coefficient ( )
max, p
1
0
for a
moving bed in a cyclone separator, plotted against a non-
dimensionless expression with the particle diameter d
p
,
the particle density
p
and the solid flow
&
M
s


36
material flow
&
M
s
, the particle density
p
and the
particle size d
p
.
In the tested cyclone separator, the surface occupancy
reaches values up to a maximum of 40 percent,
although the average volume concentration is only
about 0.2 percent. These are much higher values than
during pneumatic conveyance. In larger cyclone
separators, may the occupancies, however, be lower. In
horizontal thin film contact dryers, in which the
materials be also forcibly guided along the wall, are
values known of up to 10 percent.

Literature of chapter 7.
[7.1] Weber, M.: Grundlagen der hyddraulischen und pneumatischen
Rohrfrderung. VDI-Berichte, Nr. 371(1980), S.23-29.
[7.2] Krambrock, W.: Dichtstromfrderung. Chem.-Ing.-Tech. 54(1982)9,
S.793-803
[7.3] Muschelknautz, E.; und W. Krambrock: Vereinfachte Berechnung
horizontaler pneumatischer Frderleitungen bei hoher Gutbeladung
mit feinkrnigen Partikeln. Chem.-Ing.-Tech. 41(1969)21, S.1164-
1172.
[7.4] Muschelknautz, E.; und H. Wojahn: VDI-Wrmeatlas 1973, Kap Lh
[7.5] Bohnet, M.: Fortschritte bei der Auslegung pneumatischer Frder-
anlagen. Chem.-Ing.-Tech. 55(1983)7, S.524-539.
[7.6] Matsumoto, S.; M. Hara, S. Saito und S. Maeda: Minimum transport
velocity for horizontal pneumatic conveying. J. Chem. Engng. Jpn.
7(1974)6, S.425-430.
[7.7] Molerus, O.; und K.-E. Wirth: Die Stopfgrenze der horizontalen
pneumatischen Frderung. vt "verfahrenstechnik" 15(1981)9, S.641-
645.
[7.8] Siegel, W.: VDI-Forschungsheft 538: Experimentelle Untersuchung
zur pneumatischen Frderung krniger Stoffe in waagerechten
Rohren und berprfung der hnlichkeitsgesetze. Dsseldorf: VDI-
Verlag, 1970.
[7.9] Kerker, L.: Druckverlust und Partikelgeschwindigkeit bei der verti-
kalen Gas-Feststoff-Strmung. vt "verfahrenstechnik" 11(1977)9,
S.549-559.
[7.10] Schuchart, P.: Widerstandsgesetz beim Transport in Rohrkrmmern.
Chem.-Ing.-Tech. 40(1968)21/22, S.1060-1067.
[7.11] Brtz, W.; J.W. Hiby und K.G. Mller: Wrmebergang auf eine
Flugstaubstrmung im senkrechten Rohr. Chem.-Ing.-Tech.
30(1958)3, S.138-143.
[7.12] Stockburger, D.: Der Wrmeaustausch zwischen einer Rohrwand und
einem turbulent strmenden Gas-Feststoff-Gemisch (Flugstaub).
VDI-Forschungsheft 518, Dsseldorf: VDI-Verlag, 1966.
[7.13] Heyde, M.: Der Wrmebergang an der Rohrwand und das Druckver-
lustminimum bei der pneumatischen Frderung. Chem.-Ing.-Tech.
51(1979)11, S.1138-1139, MS 746/79.
[7.14] Heyde, M.: Heat transfer and minimum pressure drop in pneumatic
conveying. Ger. Chem. Engng. 3(1980)3, S.203-209.
[7.15] Jepson, G.; A. Poll und W. Smith: Heat transfer from gas to wall in a
gas/solids trasport line. Trans. Instn. Chem. Engrs. 41(1963), S.207-
211.
[7.16] Farbar, L.; und C.A. Depew: Heat transfer effects to gas-solids mix-
tures using solid spherical particles of uniform size. I&EC Funda-
mentals 2(1963)2, S.130-135.
[7.17] Farbar, L.; und M.J. Morley: Heat transfer of flowing gas-solids mix-
tures in a circular tube. Ind. and Engng. Chem. 49(1957)7, S.1143-
1150.
[7.18] Szekely, J.; und R. Carr: Heat transfer in a cyclone. Chem. Engng.
Sci. 21(1966)12, S.1119-1132.



37
8 Flow through packed beds
of bulk materials
For considerations regarding the pressure loss in
flowed-through packed beds, information on the spe-
cific flow velocity are necessary. The difficulty thereby
is, that the precise determination of the local velocities
in the bed is not possi-
ble. Measurable how-
ever is the velocity dis-
tribution behind a
packed bed. Fig. 8.1
shows the qualitative
results of such meas-
urement.
Thereby is to recognise
that the originally pis-
ton-shaped velocity pro-
file looks behind the bed
different, showing a
maximum near to the
wall. Direct conclusions on the flow velocity inside the
packed bed, can from such measurements not be de-
rived. But it stands to reason, that inside the bed the
flow velocity close to the wall likewise is larger, be-
cause the void fraction near the wall increases, so that
in case of the constant static pressure at each level, the
flow velocity also rises. The preferred flow of the fluid
in the vicinity of the wall is known as "Randgaengig-
keit" (wall effect).
For quantitative considerations on the single-phase
flow in packed beds, is used the mean velocity, related
on the empty cross-section.
8.1 Pressure loss
8.1.1 Flow around a single particle
A part of the flow behaviour in packed beds is closely
with the conditions connected, which occur at the flow
around a single particle.
On a particle, that itself moves along a straight-line
path at constant velocity, acts a resistance force F

.
Normally, F

is related on the largest cross-sectional

area f
p
of the particle, and is expressed as the propor-
tion of the dynamic pressure (
f
/2) u
rel
2
, using the
relative velocity u
rel
. The definition equation is there-
fore:
p
F
f
u
p
f
rel
= =

2
2
(8-1)
is also known as resistance coefficient, and its size
depends on the Reynolds number Re
rel
that is formed
with the relative velocity. In Fig. 8.2 is the resistance
coefficient for a single sphere plotted as function of the
Reynolds number [8.1].
For very small Reynolds numbers applies the Stokes
law of resistance:
=
24
Re
rel
(8-2)
In the range of mean Reynolds numbers (Re 10
3
to
10
5
) is the resistance coefficient constant at 0.4. In case
of Reynolds numbers greater than 510
5
drops the value
to 0.1.
The balance of forces for a sphere, that moves itself
vertically, is expressed as follows:
( )
F F F
g A
= (8-3)
F

means the resistance force according to Eq. (8-1), and Fg-

F
A
is the by the buoyancy force F
A
reduced inertia force Fg:
( )
F F g V
g A f p
p f
f
=
|
\

|
.
|
|

(8-4)
By substituting Eq. (8-4) and (8-2) in the Eq. (8-3) re-
sults in:

( )
( )

Re = =
f
t p p f p
u d g d
2 4 6
2 2 3
(8-5)

The sinking velocity u
t
is tantamount to u
rel
. Eq. (8-5)
can also be written in a dimensionless form:

Ga Ar

= =

p f
f
t
3
4
(Re) Re (8-6)
The approximation equations for (Re) necessitate for
the determination of the sinking velocity often an itera-
tive calculation [8.2]. For the practice can the method
be recommended, that was proposed by Martin [8.3].
Accordingly can in the whole interesting Re
t
-range

Fig. 8.1
Velocity distribution in a gas
stream after the flow through
a packed bed
Fig. 8.2:
Resistance coefficient for a flow-around single sphere
as a function of Re
rel


38
from 0 to 10
5
be correlated by using the following ap-
proximate equation:
= +
|
\

|
.
|
|
1
3
72
1
2
Re
t
(8-7)
In this way can Eq. (8-6) be resolved explicitly for the
sinking velocity:
Re /
t
t p
u d
=

= +
|
\

|
.
|

18 1 9 1
2
Ar (8-8)

8.1.2 Flow through bulk materials
The flow through bulk materials can be laminar or
turbulent, just like in the pipe flow. In case of the lami-
nar flow, it seems to be appropriate, to deduce the ap-
plicable laws on the basis of the conditions at the flow
around spherical particles. Because of the many
branches and curvature of the flow channels in a bulk
solid, the various individual streams overlap so that no
solid flow profile is formed. Because of this fact, the
laws of the laminar pipe flow, which are in particular
connected with the pronounced flow profile, are not
suitable as a basis.
The pressure loss, related on the largest cross-
sectional area of a single particle is according to Eq. (8-
1):

p
F
f
u u
p
p
f
f
= = =

24
2
12
2 2
Re Re
(8-9)

For the adding up of individual pressure losses in a
bulk solid, can Eq. (8-9) only be used, while taking into
account the fact, that in disordered bulk materials, in
which the particles each other overlap, the effective
cross section per particle length (diameter) in the flow
direction f
p
is larger than the maximum cross sectional
area of a single particle. As a simple correction, the
void fraction comes in question, so that the following
formula applies:
= f f
p p
1
0

(8-10)

For a packed bed the Eq. (8-9) therefore can be writ-
ten:

=

= p
F
f
u
p
p
f

12
2
Re
(8-11)

For the force F

results thus:

F u f
f p
=
12 1
2
0
Re
(8-12)

The number of particles in the entire packed bed is
calculated from the ratio between the volume of the en-
tire material V(1-
0
) and the volume of the single par-
ticle V
p
:

( ) ( ) ( )
n
V
V
d
V
d
p
p
p p
=

=

1 6 1 6 1
0 0
3
0
3


f h
0
(8-13)
The entire cross-sectional area f can be also expressed
with the help of the free cross-sectional area f

:
f f =

/
0

Eq. (8-13) can therefore be written:

n
d
f h
p
p
=

6 1
3
0
0
0

(8-14)

The entire pressure drop in a laminar flowed-through
packed bed is now the sum of all individual forces F

according to Eq. (8-12), multiplied with the number of
particles np according to Eq. (8-14), related on the
empty cross section f

of the packed bed, which is also

contained in Eq. (8-14).

p
n F
f
u
d
d
h
lam
p
f
p
p
=

=

12
4
1 6 1
2
2
0
3
0
0
0
Re

(8-15)
In Eq. (8-15) is used the mean velocity u in the free
cross section f

. With the help of the velocity u in the

empty cross section becomes u=u/
o
. After a respec-
tive extension, the equation for the laminar pressure
loss p
lam
looks then like this:


1
Ga
Re
18
81 . 9
3
0
0
0

g h
p
f
lam
(8-16)

The division of p by 9.81 becomes necessary, be-
cause the dimension of the right side due to the exten-
sions is altered. The dimensionless form of Eq. (8-16)
looks as follows:

Ga
Re
6 . 176
1
0
3
0
0
=

g h
p
K
f
lam
lam
p

(8-17)

39
Except the factor, Eq. (8-17) equals the theoretical re-
sistance law after Brauer [8.1]. Brauer has empirically a
factor of 160 determined for the pressure loss in the
laminar flow through disordered bulk solids with
spherical particles.
For the description of the turbulent flow condition in
packed beds, one can orientate themselves at the laws
of the turbulent pipe flow. In case of the flow through
channels with non-circular cross sections, is being used
a so-called hydraulic diameter D
h
as characteristic di-
mension. D
h
is for bulk solids defined as ratio between
the empty volume of the packed bed and the entire par-
ticle surface. For the use of the for circular cross sec-
tions valid laws, the diameter D must be replaced by
4D
h
. Because of the - compared to tubes - not only in
one flow direction orientated bulk solid surfaces, the
factor 4 is however meaningless and is being not con-
sidered here. The defining equation for the hydraulic
diameter D
h
of a packed bed is therefore as follows:
D
V V
A
V
A
d
h
s
Sch
p
p
p
=

=

0
0
0
0
1
1
6 1
(8-18)
The pressure loss Eq. (6-8) with the resistance law ac-
cording to Eq. (6-10) results together with the hydraulic
diameter for the turbulent pressure p
turb
in the turbu-
lent flow:

2
Re 194 . 0
0
0
1 6
1
0 2 2 . 0
p
d
f
turb
h
u p

=

(8-19)
The extension of Eq. (8-19) as in case of the laminar
pressure loss leads to the following form:


1
Ga
Re
e R 58 . 0
81 . 9
3
0
0
2
2 . 0
0

g h
p
f
turb

(8-20)

For Re must instead of d
p
the hydraulic diameter D
h

and instead of u the actual velocity u = u/
0
be used,
so that for the pressure loss in the turbulent flow
through bulk solids the following relationship is valid:

( )

Ga
Re
1 . 8
1
8 . 1
2 . 1
0
3
0
0
=

g h
p
K
f
turb
turb
p

(8-21)

The pressure loss equations apply strictly speaking only
for an average void fraction. At small ratios between
the diameter of the apparatus and the average particle
diameters, wall effects are noticeable. Fig. 8.3 clarifies
the functional relationships in Eq. (8-17) and (8-21).
The plotted measurement values originate from four in-
vestigations with air at pressures up to 5 bar and water
[8.4 to 8.7]. The theoretical relationships and in par-
ticular the transition from the laminar to the turbulent
flow regime are being confirmed by the measurement
results.
8.2 Heat transfer
8.2.1 Exchange at the particle surface
For the summary presentation of the measured values
of mass transfer coefficients and heat transfer coeffi-
cient in flowed-through bulk solids, the same dimen-
sionless numbers Ga and Re/Ga must be considered,
which are important for the calculation of the pressure
loss [8.8]. Fig. 8.4 shows the correlation between the
Nusselt number combined with the void fraction and
the quotient of the Reynolds number and the Galilei
number. In the double logarithmic presentation, Galilei
number and Prandtl number appear linked with each
other as parameter.
There are three flow regions with different gradients
for the curves of equal parameter values. In addition
varies the exponent n of the Prandtl number from re-
gion to region. n takes values of n = 1, n = 2/3 and n =
1/3.
The presentation was developed with the help of
measurement results from eleven studies [8.5, 8.9 to
8.18], which are plotted in Fig. 8.4 in addition. The
bulk solids were arranged in form of packed beds. The
tests were always carried out only in one of the three
Fig. 8.3
Pressure loss in the flow through bulk solids with spheri-
cal particles: comparison of theory with measurement re-
sults

40
flow regions, so that the transition from one region to
the other remains unclear. The experimental conditions
for the mostly disordered bulk solids are varying
widely. The used fluids were: H
2
O, air, CO
2
, H
2
und
He. The values for Pr and Sc varied between 0.68 und
1650, for Ga between 10
-3
and 310
9
, for Re between
10
-3
and 2.510
4,
and for
0
between 0.26 and 0.48.
8.2.2 Exchange at the pipe wall
Measurements of the heat transfer coefficient at the tube
wall have been performed, for example, by Leva [8.19] and
Kling [8.20]. The flow conditions corresponded to the
uppermost portion of the diagram in Fig. 8.4. The measured
heat transfer coefficients have thereby approximately the
magnitude, which can be determined for the particle surface
with the help of the dimensionless representation. There is
also no reason why this should not be so. Representations,
which are at diameter ratios orientated [8.21], take into
account, if it is no real wall effect, indirectly the
influence of the pressure, that cannot be captured by
means of the Reynolds number alone.
Literature of chapter 8.
[8.1] Brauer, H.: Grundlagen der Einphasen- und Mehrphasenstrmung.
Aarau und Frankfurt a.M.: Verlag Sauerlnder, 1971.
[8.2] Krten, H.; J. Raasch und H. Rumpf: Beschleunigung eines kugelfr-
migen Feststoffteilchens im Strmungsfeld konstanter Geschwindig-
keit. Chem.-Ing.-Tech. 38(1966)7, S.941-948.
[8.3] Subramanian, D.; H. Martin und E.U. Schlnder: Stoffbertragung
zwischen Gas und Feststoff in Wirbelschichten. vt "verfahrenstech-
nik" 11(1977)12, S.748-750.
[8.4] Kling, G.: Das Wrmeleitvermgen eines Kugelhaufwerkes. Diss. TH
Mnchen, 1937.
[8.5] Glaser, M.B.; und G. Thodos: Heat and momentum transfer in the flow
of gases through packes beds. Amer. Chem. Engng. J. 4(1958)1, S.63-
68.
[8.6] Krischer, O.: Vorgnge der Stoffbewegung durch Haufwerke und
porige Gter bei Diffusion, Molekularbewegung sowie laminarer und
turbulenzartiger Strmung. Chem.-Ing.-Technik 34(1962)3, S.154-
162.
[8.7] Wilhelm, R.H.; und M. Kwauk: Fluidization of solid particles. Chem.
Engng. Prozess 44(1948)3, S.201-218.
[8.8] Heyde, M.: Empirische Darstellung des Wrme- und Stoffaustausches
in durchstrmten ruhenden Kugelschttungen. Chem.-Ing.-Technik
52(1980)8, S.654-655.
[8.9] Lf, C.O.G.; und R.W. Hawley: Unsteady-state heat transfer between
air and loose solids. Ind. Engng. Chem. 40(1948), S.1061-1070.
[8.10] Grootenhuis, R.C.A.; und A.O. Saunders: Heat transfer to air passing
through heatet porous metals. Proc. Instn. Mech. Engrs. (1951),
S.363-366.
[8.11] Eichhorn,J.; und R.R. White: Particle-to-fluid heat transf. in fixed and
fluidi. beds. Chem.Engng.Prog. Symp. Ser. 48(1952), S.11-18.
[8.12] Kunii, D.; und J.M. Smith: Heat-transfer characteristics of porous
rocks. AIChEJ 7(1961)1, S.29-34.
[8.13] Donnadieu, G.: Transmission de la chaleur dans les milieux granulai-
res. Revue Inst. Fr. Petrole 16(1961), S.1330.
[8.14] Mimura, R.: Studies on heat transfer in packed beds. Graduate thesis,
University of Tokyo, 1963.
[8.15] Rowe, P.N.; und K.T. Claxton: Heat and mass transfer from a single
sphere to fluid flowing through an array. Trans. Instn. Chem. Engrs.
43(1965), S.T321-T331.
[8.16] Bhattacharyya, D.; und D.C.T. Pei: Heat transfer in fixed gas-solid
systems. Chem. Engng. Sci. 30(1975), S.293-300
[8.17] Thoenes, D.; und H. Kramer: Mass transfer from spheres in various
regular packings to a flowing fluid. Chem. Engng. Sci. 8(1958)3/4,
S.271-283.
[8.18] Littman, H.; R.G. Barile und A.H. Pulsifer: Gas-particle heat transfer
coefficients in packed beds at low Reynolds numbers. I/EC Funda-
mentals 7(1968)4, S.554-561.
[8.19] Leva, M.: Heat transfer to gases through packed tubes. Ind. Eng.
Chem.
[8.20] Kling, G.: Versuche ber den Wrmeaustausch in Rohren mit kugli-
gen und zylindrischen Fllungen. Chem.-Ing.-Tech. 31(1959)11,
S.705-710.
[8.21] Schumacher, R.: Wrmebergang an Gasen in Fllkrper- und Kon-
taktrohren. Chem.-Ing.-Tech. 32(1960)9, S.594-597.
Fig 8.4
With the help of
measurements from
various researchers
developed dimension-
less representation of
the relationship be-
tween Nu, Re, Pr and
Ga for the heat trans-
fer at the particle sur-
face in flowed-through
packed beds


41
9 Fluidized bed
Flowed-through packed beds remain with increasing
gas velocity so long in rest until the so-called minimum
fluidization velocity is reached. Fig. 9.1 depicts the
pressure loss p that occurs in the bed, as function of
gas velocity u in the empty cross-section of the bed. In
the fixed bed rises the pressure loss with increasing gas
velocity, and remains after having reached the mini-
mum fluidization, despite further increase of the gas ve-
locity constant. Over the whole state of fluidization, the
pressure loss corresponds to the weight of the bulk ma-
terial.
Determining the minimum fluidization velocity u
mf
is
complicated by the fact, that the function profile of the
pressure loss in case of increasing gas velocity often
does not have a pronounced kink, but a transition re-
gion. Therefore, the point of the minimum fluidization
velocity is defined as the intersection of the extended
function curves of the fixed bed and the fluidized bed.
The fluidized bulk material has virtually no inner
friction angle, and its undulating movements are
reminiscent of a liquid. In bulk solids with very broad
particle size distributions, smaller particles are being
earlier fluidized than large ones, and the full
fluidization of the bulk solids is being only at fluid
velocities achieved, which are higher than those,
defined by the intersection of the characteristic curves
9.1 Minimum fluidization velocity
At a flowed-through bulk solid arises at the fluidiza-
tion point a balance between the resistance force
pf
Sch
that acts in the bulk solid, and the oppositely
directed mass force V
s
(
p
-
f
)g. With the void fraction
at the fluidization point according to Eq. (2-4) one ob-
tains from the balance of forces the following relation-
ship for the pressure loss p
mf
:

( ) ( )
p g h
mf p f
=
0 0
1 (9-1)

One can also write:

( )

p
g h
mf
f
p f
f

=

0
0
1

(9-2)

Replacing the left side of Eq. (9-2) by the pressure
loss Eq. (8-17) for the laminar flowed-through bulk
solid, leads to the following form:

( )
0
1
1
Ga
Re
5 . 160
3
0
0
f
f p mf

= (9-3)

Out of this arises for the Reynolds number of the
minimum fluidization velocity for the laminar flow:

3
0
3
0
5 . 160
1
5 . 160
1
Re =

= Ar Ga
f
f p
lam
mf

(9-4)

Analogous to this, delivers the equilibrium of forces
together with the pressure loss Eq. (8-21) for a
turbulent flowed-through bulk solid, the following
relationship for the minimum fluidisation velocity:
( )

1
Ga 12 . 0 Re
2 . 0
0
3
0

=
f
f p
turb
mf

(9-5)
or in other notation:

Fig. 9.1
Pressure loss p and behaviour of flowed-through bulk
solids as function of the fluid velocity u

Fig. 9.2
Minimum fluidization velocity for bulk materials: com-
parison between theory and measurements

42

( )

1
Ar 31 . 0 Re
11 . 0
0
67 . 1
0 0.56

=
turb
mf
(9-6)

Fig. 9.2 depicts the functions according to the Eq.s (9-
4) and (9-6),whereby the term Re
mf
(1-
0
)/
0
3
was
plotted against the Archimedes number Ar.
Measurement results from experiments on bulk solids
[1.9 to 3.9], flowed through by water and air, confirm
the theoretically deduced relationships.
The width of the transition region between laminar
and turbulent flow as well as the spread of the
measurement results are not insignificant. This fact and
the circumstance that in industrial practice often rough
estimations without detailed knowledge of the
characteristic data of a bulk material must be carried
out, justifies the use of empirical calculation
approaches. For a void fraction at the fluidization point
of
0
= 0.45 reproduces the following equation the
measurement results with good accuracy [9.4]:


Ar 5.22 + 1400
Ar
Re =
mf
(9-7)

9.2 Appearance
As known from many experiments, there are two
forms of fluidized bed [9.5]. If the bulk solid is
fluidized by a liquid, the concentration of particles is
generally in the entire fluidized bed volume temporally
and spatially constant. This state is called as
homogeneous. In contrast, most gas-solid fluidized beds
show gas bubbles, which rise in the surrounding
suspension virtually solids-free up - the fluidized bed is
inhomogeneous. Experiments show, however, that also
gas-solid fluidized beds in case of conditions just above
the minimum fluidization velocity can be
homogeneous, to turn at a higher gas velocity into the
inhomogeneous state.
The growth of the bubbles with increasing height above
the gas distribution goes hand in hand with an increase
in the bubbles ascent rate. Due to the "continuity of
flow", the proportion of the cross sectional area, that is
flowed-through of bubbles, reduces itself with increas-
ing height above the gas distribution. Because of that,
fluidized beds have in case of equal gas throughput
with increasing filling height smaller volume-related
turnovers in catalytic gas-phase reactions, as well as
smaller volume-related heat transfer coefficients in de-
pendence on the installation height of heat transfer sur-
faces.
9.3 Expansion behaviour
One distinguishes between void fraction
b
that oc-
curs during the expansion of a fluidized bed, and the
void fraction
0
in the suspension phase. With Vp as
volume of the material within the fluidized bed, with V
0

as the gas volume in the state of minimum fluidization,
and with V
b
as expansion volume, apply using the usual
definitions for the void fraction, the following equa-
tions:


0
0
0 0
=
+
=
+ +
V
V V
V
V V V
p
b
p b
und
b


For the entire void fraction applies:

=
+
+ +
V V
V V V
b
p b
0
0


The connection between these three variables pro-
duces the following relationship:

( ) ( ) ( ) 1 1 1
0
=
b
(9-8)

The void fraction
0
is relatively easy to measure.
The expansion void fraction
b
, in contrast, is being
influenced by the bulk solid properties, the flow condi-
tions and the fluidized bed geometry.

Fig 9.3
In inhomogeneous fluidized are rising
gas bubbles up, which grow due to
coalescence

43
9.3.1 Change from homogeneous to inhomogene-
ous expansion
In Fig. 9.4 plotted measurement values from studies
indicate for the homogeneous expansion of water-
fluidized beds, that for the minimum fluidization veloc-
ity in a turbulent flow region approximately Eq. (9-6)
applies, if one replaces the void fraction of the packed
bed
0
through the void fraction of the expanded flu-
idized bed:
( )

1
Ar 31 . 0 Re
11 . 0
67 . 1
56 . 0

= (9-9)
The few measurements from the analysed
study [9.1] that lie in the laminar region,
reveal unfortunately no uniform overall
picture.
Several researches provide values of gas
velocities for the change from homogene-
ous to inhomogeneous fluidization [9.1,
6.9 to 8.9]. Especially in one study [9.6]
have been the gas pressure and thus the
ratio between the particle density and the
gas density systematically varied. An in-
fluence of the gas distributor plate on the
behaviour of the fluidised bed has been
largely excluded by use of porous plates.
Fig. 9.5 shows very clearly the influence
of the density ratio
p
/
f
on the position of
the transition point. In case of density ra-
tios above 1000, the fluidization state be-
comes already in the vicinity of the mini-
mum fluidization velocity inhomogeneous.
In contrast, remains a water-fluidized bed
with a density ratio of 1.6 up until fluidi-
zation numbers of u/u
mf
14 homogenous.
The increasing bed expansion with de-
creasing density ratio makes the behaviour
of liquid-fluidized beds plausible, which
remain throughout the whole range of flu-
idization homogeneous.
Fig. 9.4
Homogeneous expansion of liquid-fluidized beds: Com-
parison of calculation with measured values

Fig. 9.5
Boundaries between homogeneous and inhomogeneous
fluidization state according to measurement results from
four studies
Lit. d
p

p
Fluid p Ar (
p
-
f
)/
f
Re
mb
Re
mb

-
m

kg/m
3 - bar - - - Gl.(9-10)
[9.6] 125 1185 Air 1 86 650 0.16 0.19
2 162 475 0.33 0.32
4 322 237 0.66 0.63
8 650 118 1.4 1.27
14 1135 68 2.5 2.20
186 1185 Air 1 266 650 0.3 0.39
2 532 475 0.65 0.68
4 1066 237 1.33 1.37
8 2124 118 2.67 2.73
14 3740 68 4.68 4.79
121 240 Air 1 15 192 0.09 0.09
2 40 96 0.18 0.22
4 65 48 0.38 0.39
[9.1] 282 2600 Water 1 371 1,6 3.50 3.96
[9.7] 39 3186 Air 1 6 2600 0.05 0.02
58 21 2600 0.093 0.045
Table 9.1
Boundaries between homogeneous and inhomogeneous fluidization
state: Comparison of measured and by Eq. (9-10) calculated values
for Re
mb


44
The connection in Fig. 9.5 can empirically be captured
with the following equation:
Ar Ar 1 . 0 Re
35 . 0
(
(

=
f
f p
mb

(9-10)
Table 9.1 contains the comparison between calcula-
tion results and measured values. One investigation
[9.7] shows thereby greater deviations that under cir-
cumstances can be explained with the different diame-
ter of the test apparatus. However, this problem should
be here not pursued.
Because of the high fluidization numbers u/u
mf
com-
mon in practice, technical gas-solid fluidized beds at
atmospheric pressure are virtually always inhomogene-
ous. In addition, the used gas-distributions are usually
equipped with discrete holes, so that the gas jets hinder
the formation of a homogeneous state in principle.
9.3.2 Inhomogeneous expansion
As already mentioned, rises the gas in gas-solid fluid-
ized beds, which are operated at large ratios
p
/
g
be-
tween particle density and gas density, as bubbles up.
The proportion of the cross sectional area, that is
flowed-through of bubbles, reduces itself thereby with
increasing height above the gas distribution. This proc-
ess can be described as follows [9.9 and 9.10]:

( ) h
h
h h
=

+
=
+
+
2
20 h 20 cm, h cm
+
(9-11)

h
+
represents the fluidized bed level, in which no in-
fluence of the height is noticeable, while h is the height
of the expanded fluidized bed. In Fig. 9.6 is shown as
a function of the fluidized bed height h.
Between the height h of a fluidized bed and the height
h
0
of the resting bulk solid exists the following connec-
tion:

h
h
0
0
1
1
=

(9-12)

It follows from Eq. (9-8):

h
h
h h
b
=

+ 0
1
fr (9-13)
Regarding greater heights, it should be noted that the
void fraction decreases with increasing altitude. The
ratio between the average void fraction in the volume
between the bed levels h and h
+
and the void fraction in
the volume up to the level h
+
, is indicated by the factor
(h). On this basis one can formulate for the average
void fraction
b
in the entire bed volume:

( )
( ) ( ) h h h h h
b b b
= +
+ +
0 0 ( )

This results in:

( )
| |

b
b
h
h h h h = +
+ +
( ) (9-14)

Thus applies to larger fluidized bed heights:
h
h
h h
b
=

0
1

+
(9-15)
Values for
b
were measured in a large number of
published studies. For evaluating these measurement
values, they must be related on the bed height h
+
.
There, the void fraction is independent of the bed

Fig. 9.6
Of bubbles flowed-through fluidized bed proportion as
a function h of the expanded fluidized bed height h

Fig. 9.7
Expansion void fraction
b
, related on the height h
+
and
therefore independent of the height, connected with flu-
idization number and Galileo number

45
height. In this way can the corresponding void fractions

b
be determined for all measured values. The rela-
tionship for the conversion results from the Eq. (9-14)
and (9-15):

( ) ( )

b
b
h
h h h h
h h
h h h h
=

+
=

+
+ + + +
( ) ( )
0
(9-16)

Measurement values from eight studies [9.1, 9.7, 9.8,
9.11 to 9.15] have been evaluated, in which various
fluidized bed systems were investigated, but only at
ambient temperature and only with air as fluidizing gas.
The evaluation of these data showed, that a relationship
between the calculated void fraction
b
of the
expanded fluidized bed, the fluidization number u/u
mf

and the Galileo number Ga exists. For constant Galileo
numbers arise themselves namely within a series of
measurements for the expression
( )
( ) ( )
75 . 0
1 /
1 ln

mf
b
u u

constant numeric numbers. The complete analysis is
depicted in Fig. 9.7.
The straight line, that in the double-logarithmic
coordinate system can be approximately laid through
the measurement points, delivers the expansion law for
the by the height unaffected region of the fluidized bed:
1 Ga 12 . 0 exp 1
75 . 0
29 . 0
(
(

|
|
.
|

\
|
=
mf
b
u
u
(9-17)
The use of porous gas distributor or perforated plates
does not seem to produce any significant differences
[9.15].
The evaluation of the Eq. (9-17) with the help of a
model for the heat exchange between fluidized bed and
internals showed, that the factor -0.12 in front of the
Galileo number applies only for air as the fluidizing
gas. For other gases with other temperatures and
pressures and other materials, the ratio between
material density and gas density must be considered, so
that for
b
in general applies:
(
(

|
|
.
|

\
|

|
|
.
|

\
|
=
75 . 0
29 . 0
54 . 0
1 Ga 002 . 0 exp 1
mf g
p
b
u
u

(9-18)
With the help of Eq.s (9-11), (9-14), (9-15) and (9-18)
can the expanded fluidized bed height h iteratively be
calculated. After the determination of
b
, using an
estimated value for h, is being calculated the bed
proportion (h), that is flowed-through by bubbles.
From this results the value for
b
. The new value for h,
that is determined with the help of Eq. (9-15), serves,
where appropriate, as the start value for the new loop.
The iteration converges very rapidly, so that only two
or three iterations are necessary.
9.3.3 Bubble rise
The growth of the bubble diameter d d
b b

0
with
increasing height above the gas distribution is,
according to measurement results [9.16 to 9.21] for
constant particle sizes dependent on the fluidization
number u/u
mf
. In addition, there is an influence of the
flow conditions at the fluidization point, that can be
expressed by the Froude number
( ) g d u
p mf mf
= / Fr
2

Within a measurement series one obtains for the
expression
d d
h
b b
mf

0
Fr

constant numeric values. Fig. 9.8 shows the converted
measurement results, plotted against the
Fluidisationszahl. The initial bubble diameter d
bo
was
used according to the course of the measurements.
In the log-log plot can be two different regression
lines specified, without that for that a special reason
being recognizable. The slope is for both straights
nearly the same . Thus, applies for the bubble growth
the following law:
1 Fr a
7 . 0
0
|
|
.
|

\
|
=

mf
mf
b b
u
u
h
d d
(9-19)
For air as the fluidizing gas lie the numeric values for
the factor a between 0.08 and 0.2.

Fig. 9.8
Influence of the fluidization number u/u
mf
on the change
d
b
-d
bo
of the bubble diameter with increasing height h
above the gas distribution

46
The bubble rise velocity is often described being
proportional to the root of the bubble diameter [9.22]:
u g d
b b
= k (9-20)
With the bubble diameter d
b
according to Eq. (9-19)
and without taking into account an initial bubble
diameter, this relationship is written:
1 Fr k
7 . 0
|
|
.
|

\
|
=
mf
mf b
u
u
h g u (9-21)
9.4 Material entrainment
When the gas flows through the bulk material in a
fluidized bed, a certain material amount is being
entrained. This share should be recycled to the
fluidized bed, or be replaced in some way. Especially
for the downstream separator may be the knowledge of
the discharged amount of bulk material of interest.
In inhomogeneous gas-solid fluidized beds, particles
are hurled into the free space above the fluidized bed.
This is mainly being caused
due to the rising gas
bubbles, which burst when
leaving the fluidized bed
(Fig. 9.9). The initial
velocity of the particles is
dependent on the height h
of the expanded fluidized
bed. With increasing
clearance height H above
the fluidized bed, an
increasing part of the
discharged solid falls back
into the bed, so that the
actual amount of discharged
material is reduced.
There are only few
published studies on the
material entrainment, which
can be evaluated [9.23 to
9.25]. Two of these
investigations range
however in technically
interesting areas with
equipment diameters of
0.45 m and 0.6 m. Further experimental data are
contained in the Table 9.2.
If one the expansion
b
of the fluidized bed considers
as a relevant state variable, one can the test results
accordingly to Fig. 9.10 summarize in one single
diagram. It can be seen that the influence of the
expanded height h on the material entrainment
&
M
s
is
greater than that of the clearance height H. As indicated
at lower expansion values, the material entrainment
above certain clearance heights is no longer influenced
by the fluidized bed conditions and is then only still
dependent on the flow conditions and the material
properties.

Literature Material
p
d
p
H h
- kg/m
3
m m m
[9.23] Catalyst 1000 67 0,87 1,0
[9.24] Catalyst 1000 58 0.3-3.0 1.05
[9.25] Catalyst 2600 120/260 0.2-0.55 0.3/0.2
Table 9.2
Experimental data of the material entrainment in fluidized
bed facilities

Unwanted material entrainment from the fluidized bed
can be prevented not only by appropriately
dimensioned clearance heights. Much more effective
are for example flow obstructions (baffle plates) or
flow diversions, which ensure, that the particles lose
their upward directed proportion of kinetic energy and
fall back into the bed.

Fig. 9.9
Material entrainment in
inhomogenous gas-solid
fluidized beds

Fig. 9.10
Influence of the void fraction
b
of the expanded bed
and the height dependence of in gas-solid fluidized
beds in conjunction with the bed height h and the height
H of the free space above the bed on the amount of
entrained material

47
9.5 Penetration of gas jets
When gases be injected by nozzles into a fluidized bed,
it forms in front of the nozzle usually a jet, which the
surrounding material entrains and accelerates. The
erosive effect in this
fluidized bed region,
makes itself extremely
negative noticable, for
example, if heat
exchanger tubes are not
placed at a sufficient
distance. But also
apparatus walls and gas
distribution plates can be
affected. During the
spray drying in fluidized
beds, where the moist
material is distributed
through nozzles into the
bed, must the mutual
interference of the jets
be prevented, and the
jets must have a
corresponding distance
from the apparatus wall.
Empirical relationships
for the depth of penetration
of jets were publicated by
Merry [9.26 and 9.27]. For
the description of the
vertical and the horizontal penetration, two different
equations were needed. These dimensionless equations mix
the characteristics of the jet and the fluidized bed with each
other and appear therefore confusing.
The jet effects and
the conditions in the
fluidized bed can be
separated from each
other using the Froude
number, which is
formed with the jet
velocity u
0
at the
nozzle exit and the
nozzle diameter d
0
.
For the description of
the fluidized bed
remain as influencing
factors the
fluidisation number
u/u
mf
, the Galilei
number Ga and the
density ratio (
p
-
g
)/
g
. The evaluation of
measurement values from investigations on horizontal
gas jets leads to the following relationship:
( )

Ga
/
8 . 3
5 . 0
08 . 0
5 . 0
45 . 0
0
2
0
0
|
|
.
|

\
|

|
|
.
|

\
|

=
g
g p
mf
u u
g d
u
d
L

(9-22)
Fig. 9-11 shows the comparison of calculated and
measured values.
Some values for the propagation angle of jets were
likewise published by Merry [9.26]. Fig. 9.12 shows
that especially the ratio of nozzle diameter to particle
diameter d
0
/dp in certain areas plays a role. In addition,
the density ratio, that goes with the fourth root, is
seemingly again important.
9.6 Heat transfer between fluidized bed and heat
exchange surfaces
Gas-solid fluidized beds with immersed heating and
cooling elements are being used in a variety of
technical processes, such as cooling, heating and drying
of bulk solids, as well as endothermic or exothermic
reactions. Advantageous are the ease of use of the bulk
materials in the fluidized state, the good mixing of the
material and, not least, the high heat transfer
coefficients between fluidized bed and heat exchange
internals.
Because of the great mobility and intensive mixing of
the particles in the fluidized bed, and because of the
large specific surface of the material mass, is the
temperature of the fluidised bed in the entire volume
practically equal. Therefore, the heat transfer
coefficient , related on the heat transfer surface
w
,
can be defined, using the bed temperature
Sch
:

( )

=

&
Q
A
w w Sch
(9-23)
The heat transfer to the heating or cooling surfaces
takes place in three ways, which can be approximately
regarded as being independent of each other, namely by
Partikelkonvektion (
pc
), by gas convection (
gc
) and
by radiation (
rd
). The resulting heat transfer
coefficient is then the sum of these three
components:
= + +
pc gc rd
(9-24)
Fig. 9.13 shows the course of the heat transfer
coefficient in dependance of the gas velocity u. In the
stationary packed bed (a), the heat is being transferred
only by conduction in the bulk solid and by convection
as well as radiation. Therefore, the heat transfer is bad
and increases only gradually with rising gas velocity.
After reaching the point u
mf
(a/b), at which the
fluidization starts, increases the heat transfer

Fig. 9.11
Comparison of penetration
depths of jets, calculated
according to Eq. (9-22), with
measured values

Fig. 9.12
Influence of the diameter ratio
do/dp and the density ratio p/ g
on the propagation angle of gas
jets

48
coefficient itself rapidly, because due to the material
movement, the bulk solid layer at the transfer surfaces
is frequently
renewed. With
increasing gas
velocity (b) grow
ascent speed and
frequency of the
bubbles and thus
the mixing of the
fluidized bed. At
the same time
disappear the
temperature
profiles, so that
the heat exchange
within the bed is
no longer
inhibited. Therefore increases the heat transfer,
although due to the bed expansion the mean number
of particles per unit volume of the fluidized bed
becomes smaller. Outweighs the effect of the bed
expansion, the heat transfer can despite increasing gas
velocity no longer be increased (b/c). The heat transfer
reduces itself again.
9.6.1 Influence of material movement and bed
expansion
It can be assumed that also the particle-convective
heat transfer between fluidized bed and internals has to
do with the short-term contact between the individual
particles and the wall [9.28]. For the bulk material in
the fluidized bed, however, must except the void
fraction
0
of the resting bed, in addition the expansion
and movement of the fluidized bed be considered. By
summarizing these influences in an alternative heat
transfer void fraction

, can the Eq. (3-6) for the

stationary fixed bed be rewritten on the particle-
convective heat transfer coefficient:
( ) ( )
pc
=
max, p
1 1
0
(9-25)
The heat transfer void fraction

is firstly
determined through the material occupancy of the heat
exchange surface, that is directly connected with the
expansion of the fluidized bed. In addition, however,
the proportion of the heat exchange surface must be
given, which is in contact with a material amount, that
fullfills the conditions of the maximum heat transfer,
namely sufficiently short contact time and vanishing
temperature profile. As a measure for this, the ascent
velocity of the bubbles can be used, so that with the
help of Eq.s (9-17) and (9-21) for 1-

can be written:

(
(

|
|
.
|

\
|

|
|
.
|

\
|
=
75 . 0
29 . 0
7 . 0
1 Ga exp 1 Fr 1
mf mf
mf
u
u
n
u
u
k

(9-26)
Insertion into Eq. (9-25) gives for the particle-
convective heat transfer coefficient
pc
:

( )
1 exp
1 1
75 . 0
29 . 0
7 . 0
0 p max,
(
(

|
|
.
|

\
|

|
|
.
|

\
|
=
mf
mf
mf pc
u
u
Ga n
u
u
Fr k
(9-27)

By evaluating data from many studies with several
gases at various temperatures could be determined, that
for the coefficients k and n in Eq. (9-27) an additional
influence through the density ratio
p
/
g
has to be
considered [9.28]
3 . 2 k
27 . 0
|
|
.
|

\
|
=
g
p

(9-28)
und
54 . 0
289
1
= n
|
|
.
|

\
|

g
p

(9-29)
For the heat transfer by radiation, which is mainly at
high temperatures significant, applies according to Eq.
(3-9):

K 100
04 . 0
3
12
|
.
|

\
|
=
m
rd
T
C (9-30)
For the evaluation of the measurement results from
the literature, a mean value of 4.5 W/(m
2
K) was used
for C
12
. T
m
is the mean absolute temperature between
wall and fluidized bed. For the minimum fluidization
velocity u
mf
was used the value calculated according to
Eq. (9-7).
The proportion of the direct heat transfer between the
gas and the heating or cooling surface rises with
increasing gas velocity. For very large particles (dp
3mm), which require correspondingly high gas
velocities for its fluidization, this proportion can play a
dominant role. After Baskakov [9.29] can gc be
estimated for gas velocities above
max
with the help of
the following simple relationship:

Fig. 9.13
Heat transfer coefficient at heat
exchanger internals in fluidized
beds as function of the gas velocity u

49
Ar Pr 009 . 0
3 / 1
max
gc
=
p
g
d

(9-31)
For gas velocities between u
mf
and u
max
, this
maximum value must still be multiplied by
(u/u
max
)
0.3
.
9.6.2 Comparison with measured values
Investigations in laboratory fluidized beds have been
carried out with submerged bodies of very different
shape. Regarding the comparability of the different
experimental setups, there are no knowledges, so that
the determination of the coefficients k and n in Eq. (9-
27), was possible only under consideration of certain
effects, which were caused by the experimental setups.
If one compares only the measured maximum values of
the heat transfer coefficients, which are listed in Table
9.4, becomes apparent, that between the investigated
bulk solids, and from author to author at otherwise the
same size and density of the particles significant
differences occur.
Fig. 9.14 shows, for example, measurements in
systems with very different density ratios p/g

[9.30].
In the experiments, a rod with fairly small diameter was
hung coaxially in a laboratory fluidized bed. The
tendency of calculated and measured values correlates
quite well. That in case of the lead particles the
measured values do not quite match with the calculated
ones, could lie at the specific material properties of
lead.
In Fig. 9.15, trend and magnitude of the calculated
values [9.31] correspond to the measurement results,
although in one case, due to the large particle size, the
convective proportion of the heat transfer was relatively
great. In these experiments, cooling coils were installed
in a laboratory fluidized bed.
Different again are the conditions in the experiments,
which were described by Karchenko [9.32]. These
investigations have been carried out at partly very high
gas temperatures. In a laboratory fluidized bed was
above the gas distributor plate a sphere installed, which
has been selected with respect to the fluidized bed
dimensions extremely large. In such a case one must
probably assume, that the mobility of the bulk material,
that rises in center of the bed with the bubbles, is
limited by the large volume of the sphere. Therefore,
the material occupancy of the surface is bigger, than
due to the average void fraction of the fluidized bed
would be expected. After Fig. 9.16, the heat transfer
coefficients are about 30 percent higher, as may be
expected according to the invoice or similar
measurements of other researchers. Only in case of the
largest investigated particles, which already at low

Fig. 9.15
Heat transfer at fluidized bed internals as function of the
fluidization number: comparison with measurement
results

Fig. 9.16
Heat transfer at fluidized bed internals as function of the
fluidization number: comparison of measurement results
with widely varying temperatures

Fig. 9.14
Heat transfer at fluidized bed internals as function of the
fluidization number: comparison of measurement results
with widely varying density ratios

50
fluidization numbers show a relative high expansion,
this effect was no longer present.
A larger material occupancy of the heat exchange
surface is also to be expected, if the heat transfer
surfaces are situated at the casing wall of the fluidized
bed. In laboratory apparatuses, the particles move down
along the wall in the direction of the gas distribution
plate. They are however no longer involved in the
actual material movement, so that the conditions equal
a moving bed. Fig. 9.17 shows the results of such
experiments [9.33]. The tendency of the calculated and
measured values are being completely equal, however,
the measured heat transfer coefficients are
considerably higher, and their values equal almost
exactly the maximum values of a fast-moving bed of
densely packed particles.
In some investigations (eg [9.34]), the behavior of
bulk materials with particle sizes below of about 120
microns differs from the described laws. Partly there
are no maxima of the heat transfer coefficients, and the
achievable heat transfer coefficients are higher than
usual.
Table 9.4 contains a number of maximum values for
the heat transfer and the associated fluidization
numbers which are being compared with the from the
Eq. (9-25) and (9-27) derived relationships for the
optimal fluidization number
Ga 685 1
72 . 0
39 . 0
max

|
|
.
|

\
|
|
|
.
|

\
|
+ =
g
p
mf
u
u

(9-32)
and the maximum heat transfer coefficient
( )

-

Ga Re 1 14
52 . 0
38 . 0 5 . 0
0 max, max
rd gc
g
g p
mf p

+ +
|
|
.
|

\
|
=

(9-33)


































Fig. 9.17
Heat transfer coefficients at heating surfaces, which are
situated on the casing wall: comparison of measured
values with those, which were calculated for immersed
heating elements



Table 9.4
Comparison of measured and calculated fluidization numbers u/u
mf
und as well as heat transfer coefficients
for the state of maximum heat transfer coefficient at heat transfer installations
Experimental conditions Measurement Calculation
Lit. Gas p
Material d
p

p

0
u/u
mf

max
u/u
mf

max,p

max

- - bar C - m kg/m
3
- - W/m
2
K - W/m
2
K W/m
2
K
[9.35] Air 1 20 redurit 52 4000 0,6 21 435 30 7500 550
[9.36] 95 0,6 13 370 15 4760 430
[9.37] 150 0,6 7 336 9,5 3300 355
260 0,57 3,2 284 5,5 2115 299
460 0,55 2 232 3,3 1320 240
mullite 350 1000 0,5 272 3,2 1660 267
700 203 2 930 223
[9.30] Air 1 20 abrasive 65 1600 0,45 25 707 22 6300 682
dust 150 6,3 570 8,9 3300 492
river 315 2600 0,45 3,5 360 3,3 1810 356
sand 450 2,7 290 2,5 1350 300
750 1,9 255 1,8 875 238
alumina 310 2700 0,45 2,2 284 3,3 1830 356
grit 450 2 250 2,5 1345 300
750 1,7 220 1,8 875 238
lead powd. 125 11000 0,45 2,7 395 3,4 3815 528
glas 900 2600 0,40 1,7 226 1,7 750 234
H
2
1 20 river sand 315 2600 0,45 3,9 1080 2,5 7880 990
Al grit 310 2700 0,45 2,8 900 2,5 7970 990
CO
2
1 20 river sand 315 2600 0,45 2,7 278 3 1244 272
Al grit 310 2700 0,45 1,8 185 3 1260 272
[9.31] Air 1 40 polyamide 2240 1135 0,47 1,7 150 1,4 346 143
Al 450 2700 0,42 2,8 330 2,5 1360 306
[9.38] Air 1 50 carbon 245 1230 0,56 6,3 360 6,5 2390 329
346 0,61 5,7 310 4,7 1803 254
447 0,59 4,8 270 3,8 1460 237
560 0,61 3,7 255 3,1 1210 206
690 0,61 2,5 230 2,7 1020 189
775 0,63 2,5 220 2,5 924 172
980 0,63 2 210 2,1 760 158
280 0,50 350 5,8 2140 348
sand 224 2600 0,52 5 450 4,6 2563 369
296 3,6 406 3,6 2047 328
447 0,50 2,5 360 2,6 1460 283
561 1,8 300 2,2 1210 256
[9.32] Air 1 900 quartz 340 2600 0,43 8 730 5,9 2540 567
700 sand 7,5 660 5,5 2346 451
500 7,5 590 4,9 2220 384
300 7 510 4,2 2026 348
500 clay 420 2300 0,53 4 390 4,3 1880 315
710 3 300 2,8 1235 273
1660 1,7 210 1,7 615 223



9.6.3 Influence of fluidized bed dimensions
Measurements in apparatuses with diameters from 70
to 4500 mm and mounting heights of the heat
exchanger elements to about 200 mm, do not give any
evidence to a dependence between the average heat
transfer and the diameter of the fluidized bed [9.9 and
9.10]. Important is seemingly not the diameter ratio of
gas bubbles to fluidized beds [9:39], but a sufficient
distribution of the bubbles over the entire cross
sectional area. In
low laboratory flu-
idized beds the lat-
ter is ensured.
Like the bed
expansion, is also
the heat transfer
affected by the
with increasing
height decreasing
proportion of the
bubbles flowed-
through cross sectional area. With increasing height,
increases the proportion of the fluidized bed, in which
the heat transfer only correspond to that of a slowly
moving bed, so that the average heat transfer is
significantly reduced.
This relationship is being confirmed by the
measurement values in Fig. 9.18. The ratio between the
maximum heat transfer coefficient at the bed height h
and the bed height h
+
gets with increasing height
above the gas distribution plate constantly smaller. h
+

means the height, up to which a sufficient distribution
of bubbles across the cross sectional area is present.
The evaluation of the measurement gave for h
+
a
value of about 20 cm.
For the particle convective proportion of the
maximum heat transfer, the influence of the fluidized
bed height can be considered accordingly [9.9 and
9.10]:
( ) ( )

pc
h
u
h
ob
pc
h
u
h
ob
max
;
max
; = (9-34)
describes the mean bubble flowed-through
proportion of the fluidized bed between the lower and
upper installation height h
u
and h
ob
of the heat
exchanger surfaces. From Eq. (9-11) results the
following relationship:
( ) h h
h h
h h cm
u ob
u ob
ob u
; =

+

4 5
20 (9-35)
9.7 Heat and mass transfer between the fluid and
particles
Investigations of heat and mass transfer between fluid
and particles were mainly carried out in extremely low
fluidized beds, in which the initial bed heights were
only in the centimeter range. By plotting the
measurement results similar to those of fixed beds,
results the diagram in Fig. 9.19.
Eight studies have been evaluated [9.43 to 9.50].
Plotting the Nusselt number Nu against the quotient
Re/Ga, the expression GaPr
1/3
appears as a parameter.
Measurements in systems with Galileo numbers less or
equal to 4.3, can however only be classified, if the
parameter value is corrected by the factor of 6. Because
these facts apply to the investigations of various
researchers, it might be assumed, that for the single
particles at Galileo numbers of about 5, the flow
conditions themselves change fundamentally . Another
peculiarity in fluidized beds is the fact that, dependent
on the parameter values, the Nusselt values stagnate
above certain limits of Re/Ga.
The evaluated measurement results originate from
experiments with air (20 und 1000C) and H
2
O. The
range of the experimental conditions is not very broad,
so that the depicted connections apply only roughly. In
particular, the impact of larger fluidized bed heights
remains unclear.
9.8 Catalytic gas-phase reactions
Heterogeneous catalytic gas-phase reactions in
fluidized are like the heat transfer being influenced by
the expansion behaviour of the fluidized bed [9.51].

Fig. 9.18
The heat transfer between fluidized
bed and internals as function of the
fluidized bed height h

Fig. 9.19
With the help of measurement values, which have been
published by various researchers, detected connection
between the dimensionless numbers Nu, Pr, Re and Ga
for the heat and mass transfer in low fluidized beds

53
Thereby, the following idea can serve as discussion
basis:
A portion of the total amount of gas flows with the
minimum fluidizing velocity through the suspension
phase of the fluidized bed, while the excess gas in form
of bubbles ascends. Close to the gas distribution, gas
bubbles and material are being mixed ideally. The same
holds for larger hights in the fluidized bed, but for a
decreasing amount of material. The size of the ideally
mixed proportion of the fluidized bed is described by
the function (h) in accordance to Eq. (9-11).
Additionally must still be taken into account, that in the
bubbles flowed-through and therefore ideal stirred
reactor volume the expansion
b
according to Eq. (9-
18) prevails. Accordingly applies in this area for the
suspension fraction 1-
b
.

9.8.1 Reaction model for first order reactions
For the homogeneous catalytic reaction of first order
in an ideal stirred reactor, applies for the unreacted
proportion c [9.52]:
c
N
r
=
+
1
1
(9-36)
N
r
is being designated as number of the reaction units.
In accordance to the prevailing conditions in the
fluidized bed it must N
r
be given separately for the
quantity of gas, which flows through the void fraction
of the suspension phase, and for the quantity of gas,
which ascends in the fluidized bed as bubbles. The
expression for the first-mentioned gas proportion is the
same as for a with minimum fluidization velocity u
mf

flowed-through fixed bed:
N
h
u
r
mf
0
0
=
k
(9-37)
k is the reaction rate constant, related on the
suspension or fixed bed volume. For the as bubbles
ascending gas is being written under consideration of
(h) and 1-
b
:
( ) ( )
N
h h
u
r
b
b
=
k 1
(9-38)
The unreacted portion c in the fluidized bed then is
the sum of the proportions, which are calculated under
use of N
ro
und N
rb
:

=
+
+

+
c
u
u N
u u
u N
mf
r
mf
r
b
1
1
1
1
0
(9-39)
9.8.2 Application of the reaction model
By comparing published measurement results of
studies on first order reactions can the useability of the
reaction model be demonstrated.
9.8.2.1 Laboratory scale up to fluidized bed
diameters of 23 cm
The catalytic low temperature oxidation was
investigated by Massimilla and Johnstone [9:53]. The
reaction proceeds at 250 C in the presence of
manganese and bismuth oxide, that has been applied on
a Al
2
O
3
catalyst
support. The reaction
gas consisted of 90
percent oxygen and
10 percent ammonia.
The experiments
were conducted in a
cylindrical fluidized
bed of 11.4 cm in
diameter with various
initial heights. The
reaction rate constant
k is a mean value,
that has been
determined from the
measurement results
[9:54]. Fig. 9.2
shows the
comparison of
measured and
calculated values.
Investigations for
the hydrogenation of
ethylene on nickel contact were carried out by Lewis,
Gilliland and Glass [9:55]. The reaction proceeded at
temperatures around 110 C at ethylene surplus with 10
percent hydrogen and 90 percent ethylene, so that with
respect to H
2
, a first-order reaction can be taken as a
basis. The diameter of the fluidized bed was 5.2 cm,

Fig. 9.21
Ethylene hydrogenation at ethylene surplus: comparison
of measurement and calculation

Fig. 9.20
Catalytic NH
3
oxidation:
comparison of the by Werther
evaluated measurements with the
calculation

54
was thus rather small. The fixed bed heights were
around 0.4 m.
The measurement results in Fig. 9.21 show no full
accordance with the
calculation, but
trend and magnitude
are correct in any
case. The
differences have
certainly something
to do, with the very
small diameter of
the experimental
apparatus.
The same reaction,
the hydrogenation
of ethylene, was
also investigated by
Heidel, Fetting,
Schgerl and
Schiemann [9.56].
However, the
reaction was
verified at
hydrogen surplus
with 70 percent
hydrogen and 30
percent ethylene at
temperatures of 400
K on a copper-
bearing contact. The
reactor had a
diameter of 7.5 cm,
was thus again
relatively small. In
separate series of
measurements, the
reaction kinetics
have been studied in a fixed bed reactor of 2 cm in
diameter. For the chosen operating conditions was
found out, that the conversion of ethylene can be
described by a first-order reaction. The value of the
reaction rate constant k comes also from the fixed bed
experiments.
Fig. 9.22 shows that close to the minimum fluidization
velocity like in the previously described 5 cm reactor a
tendency to better reaction conversions prevails, than
according to the calculation would be expected.
Results for the catalytic decomposition of nitrous
oxide N
2
O at manganese oxide-bismuth-contact were
published by Shen and Johnstone [9.57]. The reaction
was carried out at temperatures around 400 C in a
reactor with 11.4 cm in diameter. The reaction gas
consisted of 99 percent air and 1 percent of nitrous
oxide. The values for the reaction rate constant k were
taken from the publikation of Werther [9.54].
As the comparison of the measurement results in Fig.
9.23 shows, there is
for a fluidized bed
diameter of 11.5
cm an exact
accordance with the
results of the first
investigation. This
also applies to the
area near the
minimum
fluidization velocity
and for fixed bed
heights of up to 1.1
m.
In literature is the
catalytic ozone
decay a widely held
model for first-order
reaction, which proceeds at temperatures slightly above
the ambient conditions. For comparison purposes, are
being used here two investigations in devices with
diameters of 10, 15 and 23 cm [9.52 and 9.58]. The fill
heights were 0.6 m and 1 m.
In Fig. 9.24, the unreacted proportion c is plottet
against the reaction rate constant k
m
, that is related on
the catalysator mass. In Fig. 9.25 is c plotted against
the expression kh
0
/u, formed by using the reaction rate
constant k. The connection between k
m
and k, that is
related on the volume of the suspension phase or the
fixed bed, is as follows:

Fig. 9.22
Ethylene hydrogenation at
hydrogen excess: comparison of
measurements and calculation

Fig. 9.23
Catalytic decomposition of N
2
O:
comparison of measurements and
calculation

Fig. 9.25
Catalytic decomposition of ozone:
comparison of measurements and
calculation

Fig. 9.24
Catalytic decomposition of ozone: comparison of
measurements and calculation

55
( ) k
p m
= 1
0
k (9-40)
As on the basis of the gained findings is to be
expected, there is even in these cases an extensive
accordance between the measured results and the
calculation.
9.8.2.2 Semi-technical scale
Greater reaction volumes than in the previously
evaluated studies were investigated by Hovmand,
Freedman and Davidson [9.59]. Here again, the
catalytic ozone decomposition was chosen as a model
reaction of first order. The fluidized bed had a diameter
of 46 cm, and it was being operated with fixed bed
heights of 1.3 m and 2.6 m. Apart from the commonly
used porous gas distributors, were in these studies in
addition two perforated plates used, which rather
fullfill the industrial requirements. One of them had 14
holes with 6.4 mm diameter, the other had 230 holes
with 2.7 mm diameter.
The course of the measurement values in the
semi-industrial fluidized bed installations indicate
a behaviour, that in laboratory apparatuses could
not be observed. The achievable reaction
conversion strives namely with increased reaction
rate not against the 100% value. This implies, that
a certain proportion of the gas from the holes of
the gas distribution plate directly into the forming
bubbles flows, and thereby no contact with the
material has, and there also takes place no further
exchanges with the surrounding suspension phase.
This gas proportion leaves the fluidized bed
unreacted.
Mathematically the described situation can be
taken into consideration, by modifying the Eq. (9-
39) with an unreacted proportion c'
byp
that
represents the proportion of the bypass gas.
( ) =
+
+

(
(
+ c c
u
u N
u u
u N
c
byp
mf
r
mf
r
b
byp
1
1
1
1
1
0

(9-41)
For the size of c
byp
shall here no general law be
given. The comparisons of measurements and
calculation in Fig. 9.26 show, however, that in the
investigated area measured and calculated values
can be matched to each other - here was c
byp
used
just as an adjusting factor. The values of c
byp
are
the greater, the smaller the fluidized bed height is.
Moreover, it is striking that even the porous gas
distributer plate, especially at low bed heights and
higher gas velocities, causes a bypass flow.
9.8.2.3 Large-scale reactor
One of the few studies at commercial reactors
were published by de Vries and his co-workers
[9.14]. The large-scale reactor for the Shell process for
catalytic oxidation
of HCl had a
diameter of 3 m and
a hight of 10 m. In
the experiments was
the effect of a
targetedly set fine
grain content
determined. As fine-
grained proportion
has been defined the
bed material with
particle sizes
smaller than 44
microns.
Fine grain
proportion in bulk

Fig. 9.26
Catalytic decomposition of ozone in semi-technical scale:
comparison of measurements and calculation according to Eq. (9-
41)
Fine-grain
content
u Convers
ion
m/s %
20 0.2 95.7
17 95.0
12 93.5
7 91.0
Calculation 88
Table 9.5
Reaction conversion in a large-
scale reactor for the Shell process
for the catalytic oxidation of HCL
depending on the fine grain
proportion (k = 0.68 1/s)

56
solids reduces the friction between the bulk of the
larger particles, so that the gas bubbles coalesce slower
with the height. Therefore increases the proportion of
the ideal with gas commingled bulk material, and the
catalytic reaction conversion increases as well. The
measurement results in Table 9.5 confirm the with
increasing fine grain content improved catalytic
reaction. The calculation by Eq. (9-39) delivers in this
case for the reaction conversion a value of 88 percent,
thus exactly the magnitude, that according to the
experiments with close particle size distributions is to
be expected.
9.9 Operation as thermal dryer
With the notion of the fluidized bed drying, is
connected for a long time the use of a single-stage
convection dryer for powder, crystalline, granular, or
short fibrous goods. For some time there are also multi-
stage facilities and facilities with internals for heating
and cooling in drying technology. A relatively recent
development is the spray fluidized bed for drying and
granulating solutions and suspensions.
The advantages of the fluidized bed are decisive for
their wide dissemination in drying technology: high
heat and mass transfer, no product residues on hot
surfaces, convektive drying with longer dwell times of
good, no mechanically moving parts and therefore low
repair requirements, constant pressure loss even with
fluctuating gas throughput, low maintenance and
investment costs and good space/time yield.
The construction of a fluidized bed drying facility is
shown in Fig. 9.27. The hot drying gas flows through a
gas distribution, for which plates of various kind are
being used. Subsequently, the material on the plate is
being fluidized by the gas. The gas transfers a part of
its heat to the product, so that liquid evaporates. The
steam-gas mixture contains dry particles from the
fluidized bed, which will be separated in a downstream
device, such as cyclone, textile filter, or wet scrubber.
Fluidized bed dryers are continuously and
discontinuously operated, with fresh air, recirculated
air or inert gas.
9.9.1 Heat requirement
Fluidized bed dryers are mainly suitable, if on the one
hand the drying rate is depending on the heat and mass
transfer between gas and moist material, and if on the
other hand the moisture does not as pure surface
humidity exist, so that longer dwell times are required.
Due to fact, that the physical parameters, which
determine the diffusion processes, usually are not
known, are for the design of dryers model experiments
necessary.
As lower limit can be determined the smallest gas-
solids ratio
&
/
&
M M
g s
for continuously operating dryers.
For a simplification, the cross flow is being neglected,
what because of the considerable cross-mixing and the
thus nearly isothermal conditions seems justified.
Under the additional condition of non-hygroscopic
material, and the assumption that the surface
temperature is equal to the cooling limit temperature
(first drying section), applies the following simple
equation.

( )
&
&
M
M
q X X
c
g
s
s s
g g g
=


|
\
|
.



(9-42)

During the drying of water-wetted drying-goods, the
heat consumption q for 1 kg of evaporated water with
surcharges for the heat loss of the apparatus lies
between 3000 and 3300 kJ/kg. The gas outlet
temperature is being equated in a first approximation
with the cooling limit temperature. The gas inlet
temperature depends on the product. To avoid sticking
at the gas distribution plate, the gas temperature should
preferably lie below the softening point of the dry
material. Results of the Eq. (9-42) lie frequently quite
near the actual gas solids ratios in dryers, because due
to the neglection of the cross flow, a too high gas
throughput is being calculated (fluidized beds work
between the extremes of an ideal mixer and an ideal
cross-flow device), and because of the fact that for
reasons of economic efficiency in single-stage fluidized
bed devices, only goods with a not too large "second
drying section" are treated. In case of a very large
"second drying section", is for the post-drying a second
facility with low gas flow advisable, for example a
multistage fluidized bed.

Fig. 9.27
Basic construction of a fluidized bed drying facility

57
9.9.2 Influence of the gas distribution plate
construction
The size of the cross-sectional area of a fluidized bed
dryer is being calculated on the basis of the thermally
necessary quantity of gas according to Eq. (9-42) and
the fluidization velocity. For the required gas velocity
can be assumed, that fluidized beds with expansion
values of
b
0,5 in terms of bed mobility and solid
material discharge be advantageously operated. The
fluidization number, that belongs to this bed expansion
value, can be determined by using Eq. (9-18). For large
particles, the gas velocities are by a factor of 1.5 to 3
higher, than the minimum fluidization velocity. For
very small particle sizes lie the values of this factor
between 15 and 30. Because of the particle size
dependence of the minimum fluidization velocity,
remains the actual gas velocity despite different particle
sizes however in the same order of magnitude.
The products that are to be dried often have wide par-
ticle size distributions and consist of agglomerates,
which were formed due to moisture bridges, and which
during the drying process partially disintegrate. There-
fore, the fluidization velocity must so be chosen, that
the agglomerated material is likewise being moved, and
that the gas as little as possible fine particles carries
with it. Because during the drying process the average
particle size, the moisture content, as well as the parti-
cle shape constantly changing, the optimal fluidisation
velocity can only experimentally be determined.
A constructive way, to reduce the proportion of parti-
cles that is entrained by the gas, is the enlarging of the
cross section of the fluidized bed dryer above the fluid-
ized bed up to twice the size. The amount of the en-
trained material of fine particles can also be influenced
through the position of the gas outlet (Fig. 9.28).
For the even distribution of the gas over the cross sec-
tion of the fluidized bed, is a gas distributor plate nec-
essary that must meet the following requirements:
Uniform gas distribution over the cross-section of
the drying facility can be realised best with a large
number of smaller gas inlets. Prerequisite is a
minimum pressure loss at the gas distributor plate.
No zones with resting material in immediate vicin-
ity of the gas distributor plate, especially in cases,
in which the gas inlet temperature has to be higher,
than the softening temperature of the material.
No trickling through of solid particles during the
drying process, often also during interruptions,
when the material as packed bed in the facility re-
mains.
Holes should not plug, and the material to be dried
should form no deposits at the gas distribution
plate.
Minimum particle crushing by gas jets from the gas
distribution plate.
Crushing and abrasion of particles is closely con-
nected with the construction of the gas distribution de-
vice. Particles, which are accelerated by the gas jets
from the holes of the plate, cause increased friction and
collision between the particles in the jet region. In addi-
tion to the jet velocity, plays especially the jet diameter
a role, because its influence on the jet pulse is stronger
than that of the gas velocity. As is to be expected, in-
creases the crushing of particles with larger nozzle di-
ameters, even in case of equal fluidizing velocities.
The described relationships are being confirmed by
abrasion-measurements on a fibre precursor (Fig 9.29),
which were published by Stockburger [9.62]. A sin-
tered metal plate caused the smallest abrasion. This
kind of gas distributors, with a average pore size of just
0.035 mm, produce the most even gas distribution and
the thinnest gas jets.
The hole plate with holes of 1.6 mm diameter pro-
duces a greater abrasion than a fine-hole-sheet with
openings of 0.35 mm diameter. And that, even though
its opening ratio is twice as large, and therefore the ve-
locity of the leaking gas only half as large.

Fig. 9.28
Influence of the gas outlet location on the amount of
product, which is entrained by the fluidizing gas

58
The gas distribution plate types, which are commonly
used in the drying technology, are shown in Fig. 9.30.
Only in laboratory apparatus one uses still frits plates.
This kind of plate ensures due to the many small pores
an even distribution of gas, but tends - especially if
dust-laden drying-air is used - easily to the clogging.
Generally, this type of gas distribution plate causes a
very high pressure loss.
For facilities on a technical scale one often uses plates
with cylindrical holes as gas distributor. The holes are
usually larger than the particles. Thus, the plates tend
less to the clogging, but have the disadvantage that the
gas is unevenly distributed and in thick jets into the
fluidized bed flows. Moreover, trickles the dry material
through the gas distribution plate.
Fine hole sheets have themselves very successful
proven and are most commonly used in fluidized layer
dryers. The holes are triangular to half-eliptical and
strong conical. The passage of the drying gas is
inclined. The holes are kept very small (up to 60 m),
so that an even distribution of the gas and low abrasion
of the product can be achieved. The trickling through of
product during operating interruption is no problem at
this type of gas distribution plate.
Trickling through of product and clogging of holes can
be avoided best by using gas distributor heads or bell-
shaped distributing elements. Such plate constructions
are however expensive to produce, and the gas
distribution is very uneven, so that the bulk solid
movement between the gas distribution elements very
small is, and partly also dormant product accumulations
occur. In case of gas distributor heads, the product
abrasion is still far greater than with perforated sheets
with round holes, because the thick gas jets are in
addition directed against each other. If the nozzles are
directed down to the plate, erosion is to be expected.
Such gas distribution plates are being rarely used in
drying technology.
A special feature is the so-called spouted bed, which
has no gas distribution plate, but instead is funnel-
shaped [9.63]. The spouted bed is used for the drying of
suspensions of fine particles on inert coarser particles
as heat carrier, and for the drying of large particles in
millimetre range with narrow particle size distributions,
which are difficult to fluidize. The cone angle should
have a size of at least 40 [9.64], while the diameter-
ratio between nozzle and device should lie below of
0.35 [9.65].
To achieve, that through all holes of a gas distribution
plate an equal amount of gas is flowing, must a certain
pressure loss be generated by the plate. 15 to 40 percent
of the pressure drop in the fluidized bed is being
considered sufficient. The upper value applies to flat

Fig. 9.29
Abrasion behaviour of fibre precursor granules using dif-
ferent gas distribution plates

Fig. 9.30
Design of gas distributor floors for fluidized bed dryers

Fig. 9.31
Two-stage circular fluidized bed with disc for the material
feeding, air-broom and heated internals

59
fluidized beds, in which the influence of the approach
flow conditions is particularly great. The distribution
plate of a dryer usually has pressure losses between 5
and 10 mbar.
During drying of sticky and highly agglomerated filter
cakes, the formation of channels can be avoided by
distributing the wet material with the help of a stirrer
into already dry material, or by feeding the product
with the help of a rotating disk. Clumps, which are
contained in the wet material, can be crushed by using
built-in grinding discs.
Mainly in multistage fluidized beds, the clogging of
the holes in the gas distribution plate can be a problem.
This often can be prevented with the help of an air-
broom. This device rotates below the plate and blows
out the holes with the help of sharp air jets.
In some cases, the considerable horizontal equilibration
of concentration and temperature, which occur in con-
tinuously operated fluidized beds, can be of disadvan-
tage. To achieve a greater average dwell time for the
main proportion of the material, a larger length/width
ratio is to be aimed for. In a compact dryer one can
meet this requirement due to the installation of spiral-
shaped baffles in a round bed. Fig. 9.31 shows a two-
stage fluidized bed with disc-product feeding, air-
broom and spiral-shaped internals [9.66].
9.9.3 Multi-stage fludized beds
For substances that have a low moisture content, but
due to the moisture-transport from the inside of the
product to its surface larger dwell times needing, can
convection dryers like tower dryer or multistage fluid-
ized bed be used. Through multistage fluidized beds
can a better balanced dwell time be realized, which of-
ten is connected with a positive effect on the quality of
the dried product. With the use of this kind of dryers,
the formation of channels in the product can be
avoided, which in case of cohesive bulk solids in
chamber dryers often occurs, so that the largest part of
the bulk solid has no contact with the drying gas. Com-
pared to the single-stage fluidized bed, the multi-stage
requires a far lower amount of drying gas, and the
dwell-time distribution is tighter. Moreover, counter
current flow of product and gas is realized, so that es-
pecially in the drying of hygroscopic products, heat and
mass transfer are positively affected.
Fig. 9.32 shows schematically the two different types
of multistage fluidized beds. In both cases, a separate
stage consists of a sieve plate with a fluidized bed on it.
The material moves from top to bottom, the drying gas
from bottom to top.
The multistage fluidized bed with downpipe matches
largely with its structure the sieve plate columns from
the rectification and absorption. In this type of con-
struction flows the drying gas through the sieve plate of
each stage, while the material reaches the next sieve
plate through an overflow pipe. With no cellular wheel
sluice in the overflow pipe, the pipe must immerse into
the fluidized bed on the next plate. The filling height of
this pipe must be such, that the pressure loss, that is
caused by the gas flow through the sieve plate and the
associated material bed, is smaller than the pressure
loss, that is generated by the unwanted flow through the
overflow pipe. These conditions require a relatively
large ground clearance. Furthermore, the start up of the
facility is problematic, because the overflow tubes are
not yet being filled. Such problems can however be cir-
cumvented by using locks and shutoff devices, as they
are specified in Perry's Chemical Engineer Hand-book
[9.67]. So the tube can remain for example using a
slider when starting so long closed, until a sufficiently
high product column exists.
Particularly useful and reliable are rotary feeder as
shut-off devices. There is no difficulty during start-up
of the facility because the dryer without special meas-
ures can be filled. The cell wheel can be arranged close
to the gas distribution plate, and at low fill level is also
the jamming of particles in the cell-wheel to get around.

Fig. 9.32
Multistage fluidized bed

60
The mode of action in a simpler type of fluidized bed,
which is known as 'trickle-stage fluidized bed' (Riesel-
boden-Wirbelschicht), is less problematic. As gas dis-
tribution plates serve so-called Dual-flow sheets, which
are equipped with such large holes, that both gas and
material can pass [9.68 and 9.69]. Overflow pipes are
therefore not required. For technically interesting di-
mensions of such an apparatus are, however, difficul-
ties are expected, because the pressure loss, that is re-
quired for an even distribution of gas by the sieve plate,
is difficult to reach. This applies especially to fine ma-
terials and such goods, which due to varying moisture
distribution, varyingly agglomerate, so that they are
particularly reliant on an even gas distribution by the
plate.
The multi-stage 'trickle-stage fluidized bed' will re-
main limited for this reason on the drying of same-sized
particles in the millimetre range, for example plastic
granules.
Multistage fluidized beds can be understood as
chained ideally mixed single-stage fluidized beds. Fig.
9.33 depicts the temperature profile along a six-stage
apparatus [9.70]. There are significant temperature
variations between the stages. In every single stage, the
temperatures are, however, constant over the height.
With increasing gas velocity and growing free cross-
sectional area of the sieve plate rises but also the pro-
portion of material, which is thrown from the lower flu-
idized bed through the holes onto the overlying plate.
This material arrives on a stage, which it already has
passed through. Fig. 9.34 depicts this process quantita-
tively: depicted is the axial mixing coefficient depend-
ing on the gas velocity. As parameter serve the free
cross section of hole plates (13 and 24 percent), as well
as that of a slit plate (52 percent).
Multistage fluidized bed dryers are only used for
products with large second drying sections. Its drying
time is dependent on complex diffusion processes.
Therefore, are for the design of the drying apparatus
investigations necessary. Results from such tests are the
inlet and outlet temperatures of the drying gas, the re-
quired gas-solid ratio and the appropriate gas velocity,
from which the diameter of the apparatus for the re-
quired material throughput is determined. Finally, the
experimentally determined drying time is used, to spec-
ify the number of stages and the fluidized bed height.
The bed height can be adjusted with the height of the
overflow pipe.
9.9.4 Heating surfaces internals
For drying of material with low softening point, the
temperature difference between the incoming and the
outgoing gas is relatively small, so that relatively large
amounts of gas and thus large areas for fluidization are
required. It is much more economic to provide a portion
of the required drying heat with the help of heating sur-
faces, which are inbuilt into the fluidized bed.
For internals there is a distinction between plates and
bundles of round or oval tubes. Neither the shape of the

Fig 9.33
Temperature profile along a multistage fluidized bed

Fig. 9.34
Axial mixing coefficient in multistage fluidized beds

61
inbuilt bodies [9.71] nor the use of pipes and their ar-
rangement (aligned, offset or crossed) [9.72] causes a
significant change in the amount of the average heat
transfer. The heat transfer is also being influenced
hardly by the diameter of the pipe, as long as this is lar-
ger than 15 mm [9.73]. Finned tubes can be used for the
increase the transferred amount of heat in proportion to
the volume of the pipe. Excessive obstruction of parti-
cles between the ribs can be avoided by using tubes
with only two longitudinal ribs, which are arranged in
the direction of flow [9.71]. Due to the increased risk of
incrustation in the corners between the tubes and the
ribs is such a solution for dryers not fully to recom-
mend and has to be carefully checked in long-term
tests.
For the undisturbed fluidization in beds with dived
tube bundles, some authors recommend optimum tube
spacing t (pipe midst distance to diameter ratio), which
from t = 2 [9.74] to t = 4...6 [9.75] vary. More sense
seems to make, however the use of the ratio between
clear distance of the tubes and particle diameter, to take
the mutual influence into account. According to Petrie
[9.76] lies this distance at a value of about 43 times the
largest particle diameter. This value corresponds with
the size ratio for the undisturbed bulk solid discharge
from silos. In addition to the unimpeded moving of the
material, the chosen tube spacing should the possibility
of the cleaning ensure.
The distance between internals and gas distributor
plate must be larger than the penetration depth of the
gas jets, which can be estimated by Eq. (9-22). Fur-
thermore, internals cause an increased pressure loss in
the fluidized bed, if their resistance against the gas flow
is greater than the weight loss of the fluidized bed due
to the reduced volume of the bed. The increase in the
pressure loss, however, is not significant. According to
Neukirchen [9.77], the increase amounts in the worst
case about 18 percent of the total pressure loss (gas dis-
tribution, fluidized bed, internals).
9.9.5 Spray fluidized bed drying
A further development in the field of fluid bed drying
is the spray fluidized bed. This technique is applied for
solutions and suspensions, for which in former times
mostly spray dryers and roller dryers were used.
The principle of the spray fluidized bed dryer consists
therein, that via two-component nozzles, rarely via sin-
gle-component nozzles, flowable product is sprayed
into a fluidized bed, which contains already dried prod-
uct. The moist product is during this process distributed
over the surface of the individual particles. The drying
of the so generated thin liquid layer takes place in the
first drying section, and is therefore extremely fast. The
energy for evaporation of the liquid is delivered
through the drying gas.
If the procedure is operated continuously, two, some-
times three steps are required: firstly a drying and
granulating stage, secondly a crushing stage, which the
grain growth during the drying process counteracts and
new granulation germs creates, and thirdly, if neces-
sary, a classification stage, where the desired end prod-
uct is separated.
The basic design of a spray fluidized bed facility is
depicted in Fig. 9.35. It consists essentially of the fluid-
ized bed with gas distributor plate, a nozzle for the
product infeed, a device such for example a worm con-
veyor for the material withdrawal, a gas outlet and a
downstream textile filter, which is mounted above the
fluidized bed [9.78]. For the classifying of the desired
particle size [9.78 up 9.80] is in many cases a sieve or
sifter responsible, and a mill creates the necessary
granulation germs.
Fundamental prerequisite for the functioning of this
procedure is the combination of the three steps of
granulating, crushing and grading. As expected, the
particle size increases in the drying and granulating
stages according to an exponential growth law [9.81]:


Fig. 9.35
Spray fluidized bed drying facility

62
d
d
M
M
Z
p
p
s
s
0
3
=
|
\

|
.
|
exp
&
K
(9-43)

If the ratio
&
/ M M
s s
as well as the temporal increase of
the layer thickness per particle surface unit are con-
stant, if the particles are approximately spherical, and if
the actual particle size distribution can be expressed by
an equivalent particle diameter, the constant K has the
experimentally confirmed value of 1 [9.62]. It is obvi-
ous, that in the limiting case, the fluidized bed would
consist of only a few large particles. So, a size reduc-
tion and nucleation process must superimpose the
growth process, to maintain the fluidization process.
Crushing can be done in three ways: mechanically
within or outside of the fluidized bed, or taking advan-
tage of the fluidized bed abrasion.
The easiest way is the exploitation of the particle abra-
sion that occurs in the fluidized bed. Apart from prod-
uct properties as strength and shape of the granules, are
the following influence factors important: design of the
gas distribution, fluidization number as well as height
of fluidized bed. More important than the gas distribu-
tion are the fluidization number and the height of the
fluidized bed. These effects are depicted in Fig. 9.36.
With growing fluidization and bed height, the abrasion
rate increases significantly [9.62]. As an additional re-
quirement however, the Reynolds number, formed with
the gas velocity and the particle diameter, must be lar-
ger than 100 [9.61].
Finally the significant crushing during the atomisation
should be mentioned. This effect is in case of two-
substance nozzles especially strongly noticeable. Their
effect complies almost with that of a fluidized bed jet
mill [9.82].
A targeted control of the particle size is possible, us-
ing a mill and a classifying device outside the fluidized
bed. For instance can a screening machine be used for
this task, but the installation of a classifier in the entire
process is much easier. The fine-grain proportion is be-
ing normally by augers or injectors recycled into the
fluidized bed [9.79 and 9.80].
The nozzles can be so arranged, that they spray from
above [9.78 and 9.79], from the side [9.80 and 9.84] or
from below [9.85].
Orientation values about the immediate sphere of in-
fluence of a two-component nozzle, while spraying
horizontally into a fluidized bed, was gained during
drying and granulating of fertilisers by measuring the
temperature distribution in the spray zone [9.86]. The
liquid loading of the nozzle lied between 45 and 240
kg/h, thus at values, which are in technical facilities
common. The amount of atomising air was maintained
at a constant value of 270 kg / h, at gas pressures be-
tween 1.9 and 3.4 bar. First, the temperature in the axis
of the spray cone became with increasing liquid load
smaller, but reached at the highest liquid load in a dis-
tance of 350 to 400 mm from the nozzle the tempera-
ture of the fluidized bed. The temperature profile in ra-
dial direction resembles parabolic curves with a mini-
mum in the spray axis. Even at the highest loads was in
a distance of 100 to 150 mm from the nozzle nearly the
bed temperature reached.
Due to the small extent of the spray cone, the liquid
distribution over a larger area of fluidized bed is only
achievable by the moisture-laden particles migrate out
of the jet region. This process leads to a significant ex-

Fig 9.36
Abrasion behaviour of fibre precursor pellets depending
on the fluidization number and the bulk height

Fig. 9.37
Mixing coefficients in fluidized beds depending
on the fluidization number and the bed height

63
change of energy inside the bed, because the volumetric
heat capacity of the material is about 1000 times as
large as that of the gas.
The experience shows, that the intensity of the mate-
rial movement of a fluidized bed is influenced by the
condition of the bed, by the fluidization number, and by
the bed height. As Fig. 9.37 shows, the mixing coeffi-
cient starts with the value zero at the minimum fluidiza-
tion velocity and increases with growing fluidization
[9.87]. The significant influence of the fluidized bed
height is by the graphic likewise attested [9.88]. So,
high spray rates require large bed heights and large flu-
idization numbers.
The advantage of the spray fluidized bed dryer, com-
pared to competing processes, lies except in the often
higher economic efficiency, in the significant im-
provement of the product properties. Depending on the
intended use particle sizes between 1 and 6 mm can be
produced. The products are free of dust and consist of
compact particles with mostly higher bulk density and
better flow behaviour than they can be produced in any
other technique. Nevertheless, the solubility in liquids
is often better, than that of dust-free products, in which
the primary particles often clump together. Prerequisite
for the application of the spray fluidized bed technol-
ogy is, however, the granulation ability of the material
to be dried. This is first and foremost a product specific
property, which however can also be influenced by the
process conditions, such as gas outlet temperature,
moisture, and kind of atomisation [9.78].


Literature of chapter 9.
[9.1] Wilhelm, R.H.; und M. Kwauk: Fluidization of solid particles. Chem.
Engng. Prozess 44(1948)3, S.201-218.
[9.2] Saxena, S.C.; und G.J. Vogel: The measurements of incipient fluidiza-
tion velocities in a bed of coarse dolomite at temperature and pres-
sure. Trans. IChemE 55(1977), S.184-189.
[9.3] Rowe P.N.; und C.X.R. Yacono: The bubbling behaviour of fine pow-
ders when fluidized. Chem. Engng. Sci. 31(1976)12, S.1179-1192.
[9.4] Goroschenko, W.D.; R.B. Rosenbaum und O.M. Todes: Nachr. d.
hh. Lehranst. d MWO d. UDSSR. Erdl und Gas 1(1958)25.
[9.5] Leva, M.: Fluidization. New York: McGraw-Hill Book Company,
1959. [9.6] Godard, K.E.; und J.F. Richardson: Proceedings of the
Tripartite Chemical Engineering Conference, Montreal, 1968.
[9.7] Davies, L.; und J.F. Richardson: Trans. Inst. Chem. Eng. 44(1966),
S.T293.
[9.8] Rowe, P.N.; und C.X.R. Yacono: The bubbling behavior of fine pow-
ders when fluidized. Chem. Engng. Sci. 31(1976)12, S.1179-1192.
[9.9] Heyde, M.; und H.J. Klocke: Wrmebergang zwischen Wirbelschich-
ten und Einbauten - ein Problem des Wrmebergangs bei kurzfristi-
gem Kontakt. Chem.-Ing.-Tech. 51(1979)4, S.318-319; MS 679/79.
[9.10] Heyde, M.; und H.J. Klocke: Wrmebergang zwischen Wirbel-
schicht und Wrmetauschereinbauten. vt "verfahrenstechnik"
13(1979)11, S.886-892.
[9.11] de Groot, J.H.: Proceedings of the international Symposium on Flu-
idization. Eindhoven: 6.-9. Juni 1967.
[9.12] Huisung, T.H.; und C. Thodos: Expansion characteristics of gas-
fluidized beds. The Canadian J of Chem. Engng. 55(1977)4, S.221-
226.
[9.13] Lewis, W.K.; E.R. Gilliland und W.C. Bauer: Characteristics of fluid-
ized particles. Ind. Engng. Chem. 41(1949), S.1104-1117.
[9.14] de Vries, R.J.; W.P.M. van Swaaij, C. Mantovani und A. Heijkoop:
Design chriteria and performance of the commercial reactor for Shell
Chlorine Process. Proc. 5th Europ. Symp. Reaction Engng., Amster-
dam, 1972, S.B9.-59/69.
[9.15] Xavier, A.M., D.A. Lewis und J.F. Davidson: The expansion of bub-
bling fluidized beds. Trans. Instit. Chem. Eng. 56(1978)4, S.274-280.
[9.16] Werther, J.: Effekt of gas distributor on the hydrodynamics of gas
fluidized beds. Germ. Chem. Engng. 1(1978)3, S.166-174.
[9.17] Geldart, D.: The effect of particle size and size distribution an the
behavior of gas-fluidized beds. Powder Technol. 6(1972), S.201-215.
[9.18] Park, W.H.; W.K. Kang, C.E. Capes und G.L. Osberg: The properties
of bubbles in fluidized beds of contacting particles as measured by an
electroresistivity probe. Chem. Eng. Sci. 24(1969), S.851.
[9.19] Rowe, P.N.; und D.J. Everett: Fluidized bed bubbles viewed by x-
rays, Part III. Trans. Instn. Chem. Engrs. 50(1972), S.55-60.
[9.20] Whitehead, A.B.; und A.D. Young: Fluidization performance in a
large-scale equipment, Part I. Proc. Intern. Symp. on Fluidization,
Eindhoven, 1967, S.284.
[9.21] Fryer, C.; und O.E. Potter: AIChEJ 22(1976)1, S.88.
[9.22] Reuter, H.: Steiggeschwindigkeit von Blasen im Gas-Feststoff-
Fliebett. Chem.-Ing.-Tech. 37(1965)10, S.1062-1066.
[9.23] Moroka, S.; K. Kawazuishi und Y. Kato: Holdup and flow pattern of
solid particles in freeboard of gas-solid- fluidized bed with fine parti-
cles. Powder Technol. 26(1980)1, S.75-82.
[9.24] Fournol, A.B.; M.A. Bergougnou und C.G. Baker: Solids entrainment
in a large gas fluidized bed. The Can. J of Chem. Engng. 51(1973),
S.401-404.
[9.25] Demmich, J.; und M. Bohnet: Feststoffaustrag aus Wirbelschichten.
vt "verfahrenstechnik" 12(1978)7, S.430-435.
[9.26] Merry, J.M.D.: Penetration of a horizontal gas jet into a fluidized bed.
Tran. Instn. Chem. Engrs 49(1971), S.189-195.
[9.27] Merry, J.M.D.: Penetration of vertical jets into fluidized beds. AIChE
Journal 21(1975)3, S.507-510.
[9.28] Heyde, M.: Durchstrmen, Aufwirbeln und Frdern von Schttgtern.
Fortschrittberichte der VDI-Z. Reihe 3, Nr. 66, 1982.
[9.29] Baskakov, A.P.; et al.: Heat transfer to objects immersed in fluidized
beds. Powder Technol. 8(1973), S.273-282.
[9.30] Wicke, E.; und F. Fetting: Wrmebergang in Gaswirbelschichten.
Chem.-Ing.-Tech. 26(1954)6, S.301-309.
[9.31] Knuth, M.; und P.M. Weinspach: Experimentelle Untersuchungen des
Wrme- und Stoffbergangs an die Partikeln einer Wirbelschicht bei
der Desublimation. Chem.-Ing.-Tech. 48(1976)10, S.893; MS
413/76.
[9.32] Karchenko, N.V.; und K.E. Makhorin: The rate of heat transfer be-
tween a fluidized bed and an immersed body at high temperatures.
Int. Chem. Engng. 4(1964)4. S.650-654.
[9.33] Botterill, J.S.M.; und M. Desai: Limiting factors in gas-fluidized bed
heat transfer. Powder Technol. 6(1972), S.231-238.
[9.34] Yamazaki, R.; und G. Jimbo: Heat transfer between fludized beds and
heated surfaces. J of Chem. Engng. of Japan 3(1970)1, S.44-48.
[9.35] Mersmann, A.: Bestimmung der Lockerungsgeschwindigkeit in Wir-
belschichten durch Wrmebergangsmessungen. Chem.-Ing.-Tech.
38(1966)10, S.1095-1098.
[9.36] Mersmann, A.: Zum Wrmebergang zwischen dispersen Zweipha-
sensystemen und senkrechten Heizflchen im Erdschwerefeld. vt
"verfahrenstechnik" 10(1976)10, S.641-645.
[9.37] Mersmann, A.: Zum Wrmebergang in Wirbelschichten. Chem.-
Ing.-Tech. 39(1967)5/6, S.349-353.
[9.38] Kim, K.J.; D.J. Kim, S. Chun und S.S. Choo: Heat and mass transfer
in fixed and fluidized bed reactors. Int. Chem. Engng. 8(1968)3,
S.205-222.

64
[9.39] Werther, J.: Strmungsmechanische Grundlagen der Wirbelschicht-
technik. Chem.-Ing.-Tech. 49(1977)3, S.193-202.
[9.40] Wright, S.J.; R. Hickmann und H.C. Kethely: Heat transfer in fluid-
ized beds of wide size spectrum at elevated temperatures. British
Chem. Engng. 15(1970)12, S.1551-1554.
[9.41] Howe, W.C.; C. Alusio, Pope, Evans und Robbins: Control variables
in fluidized bed steam generation. Chem. Engng. Progr. 73(1977)7,
S.69-73.
[9.42] Betriebsmessung
[9.43] Subramanian, D.; H. Martin und E.U. Schlnder: Stoffbertragung
zwischen Gas und Feststoff in Wirbelschichten. vt "verfahrenstech-
nik" 11(1977)12, S.748-750.
[9.44] Donnadieu, G.: Transmission de la chaleur dans les milieux granulai-
res. Revue Inst. F. Ptrole 16(1961), S.1330.
[9.45] Rowe, P.N.; und K.T. Claxton: Heat and mass transfer from a single
spehre to fluid through an array. Trans. Instn. Chem. Engrs.
43(1965), S.T321-331.
[9.46] Zabescheck, G.: Experimentelle Bestimmung und analytische
Beschreibung der Trocknungsgeschwindigkeit rieselfhiger,
kapillarporser Gter in der Wirbelschicht. Diss. Universitt
Karlsruhe, 1977.
[9.47] Petrovic, V.; und G. Thodos: Effectiveness factor for mass transfer in
fluidized systems. Proc. Int. Symp. on Fluidization, Eindhoven, 1967.
[9.48] Mosberger, E.: ber den Wrme- und Stoffaustausch zwischen Par-
tikeln und Luft in Wirbelschichten. Diss. TH Darmstadt, 1964.
[9.49] Richardson, J.F.; und P. Ayers: Heat transfer between particles and
gas in a fluidized bed. Trans. Instn. Chem. Engrs. 37(1959)6, S.314-
322.
[9.50] Yanata, J.; K.E. Makhorin und A.M. Glukhomanyuk: Inverstigation
and modelling of the combustion of natural gas in a fluidized bed of
inert heat carrier. Int. Chem. Engng. 15(1975)1, S.68-72.
[9.51] Heyde, M.: Strmungsmechanik und katalytische Reaktionen in Wir-
belschichten. Maschinenmarkt 88(1982)78, S.1596-1599.
[9.52] Orcutt, J.C.; J.F. Davidson und R.L. Pigford: Reaction time distribu-
tions in fluidized catalytic reactors. Chem. Engng. Progr. Symp. Ser.
58(1962)38, S.1-15.
[9.53] Massimilla, L.; und U.H.F. Johnstone: Reaction kinetics in fluidized
beds. Chem. Eng. Sci. 16(1961), S.105-112.
[9.54] Werther, J.: Mathematische Modellierung von Wirbelschichtreak-
toren. Chem.-Ing.-Tech. 52(1980)2, S.106-113.
[9.55] Lewis, W.K.; E.R. Gilliland und W. Glass: Solid-catalyzed reaction in
a fluidized bed. AIChE 5(1959)4, S.419-426.
[9.56] Heidel, K.; K. Schgerl, F. Fetting und G. Schiemann: Einflu von
Mischungsvorgngen auf den Umsatz bei der thylenhydrierung im
Fliebett. Chem. Eng. Sci. 20(1965), S.557-585.
[9.57] Shen, C.Y.; und H.F. Johnstone: Gas-solid contact in fluidized beds.
AIChI Journal 1(1955)3, S.349-354.
[9.58] Van Swaaij, W.P.M.; und F.J. Zuiderweg: Investigations of ozone
decomposition in fluidized beds on the Basic of a two-phase model.
Proc. 5th Europ. Symp. Reaction Engng., Amsterdamm, 1972, S.B9-
25/36.
[9.59] Hovmand, S.; W. Freedman und J.F. Davidson: Chemical reactions in
a pilot-scale fluidized bed. Trans. Instn. Chem. Engrs. 49(1971),
S.149-162.
[9.60] Baskakov, A.P.; und I.V. Gubin: Khim. Prom. (russ.) 7(1968), S.
54/56.
[9.61] Reh, L.: Wirbelschichtreaktoren fr nicht katalytische Reaktionen.
Ullmann Enzyklopdie der techn. Chemie, 3. Weinheim: Verlag
Chemie, 1973, S.433-460.
[9.62] Stockburger, D.: Fortschritte und Entwicklungstendenzen in der
Trocknungstechnik bei der Trocknung formloser Gter. Chem.-Ing.-
Tech. 48(1976)3, S.199-205.
[9.63] Brauer, H.: Grundlagen der Einphasen- und Mehrphasenstrmung,
Kapitel 9: Wirbelschicht. Aarau und Frankfurt am Main: Sauerlnder
Verlag, 1971.
[9.64] Hunt, C.H.; und D. Brennan: Estimation of spout diameter in an
spouted bed. Austral. Chem. Engng. (1965)3, S.9-18.
[9.65] Becker, H.A.: An investigation of laws governing the spouting of
coarse particles. Chem. Engng. Sci. 13(1961), S.245-262.
[9.66] Klocke, H.J.; und D. Stockburger: Neues aus der Trocknungstechnik.
Chem. Ind. (1973)25, S.708-712.
[9.67] Perry's Chemical Engineer Handbook, 4th e., S. 20-46/49.
[9.68] Brauer, H.; und V. Asbeck: Druckverlust und Feststoffkonzentration
in mehrstufigen Rieselboden-Wirbelschichten. vt "verfahrenstech-
nik" 6(1972)7, S.230-238.
[9.69] Rben, K.; und E. Steffen: Praktische Anwendungen von Mehrstufen-
Rieselboden-Wirbelschicht-Apparaten. Aufbereitungstechnik
15(1974)12, S.665-669.
[9.70] Nishinaka, M.; Sh. Morooka und Y. Kato: Longitudinal dispersion of
solid particles in fluid beds with horizontal baffles. Powder Technol
9(1974)1, S.1-16.
[9.71] Natusch, H.J.; und H. Blenke: Zur Wrmebertragung an horizon-
talen Lngsrippenrohren in Gas-Fliebetten. vt "verfahrenstechnik"
8(1974)10, S.287-293.
[9.72] Reh, L.: Strmungs- und Austauschverhalten von Wirbelschichten.
Chem.-Ing.-Tech. 46(1974)5, S.180-189.
[9.73] Vreedenberg, H.A.: Heat transfer beetween a fluidized bed and an
horizontal tube. Chem. Engng. Sci 9(1958)1, S.52-60.
[9.74] Lese, H.K.: Heat tranfer from a horizontal tube to an fluidized bed in
the presende of unheated tubes. Dissertation, Lexington/USA (1969).
[9.75] Offergeld, E.: Wirbelbetten mit beheizten Einbauten in der Wirbel-
schicht. vt "verfahrenstechnik" 8(1974)12, S.336-338.
[9.76] Petrie, J.C.; W.A. Freeby und J.A. Buckham: In-bed heat exchangers.
Chem. Engng. Progr. 64(1968)7, S.45-51.
[9.77] Neunkirchen, B.: Gestaltung horizontaler Rohrbndel in Gas-
Wirbelschichten nach wrmetechnischen Gesichtspunkten. Disserta-
tion Universitt Stuttgart, 1973.
[9.78] Rosch, M.; und R. Probst: Granulation in der Wirbelschicht. vt "ver-
fahrenstechnik" 9(1975)2, S.59-64.
[9.79] Bean, S.L.; et al: Process for the production of hydrous granular so-
dium silicate. USA Patent 3.748.103.
[9.80] Shakova, N.A.; u.a.: Investigation of the granulation of ammonium
nitrate in a fluidized bed under industrial conditions. Internat. Chem.
Engng. 13(1973), S.658-661.
[9.81] Shakova, N.A.; u.a.: Kinetics of granule formation in a fluidized bed.
Theor. Foundations Chem. Engng. (Teor. osnovy khim technol.)
5(1971)5, S.656-661.
[9.82] Kaiser, F.: Die Fliebettstrahlmhle. Chem.-Ing.-Tech. 45(1973)10a,
S.676-680.
[9.83] Fritsch, R.; u.a.: Austragsvorrichtung fr Granulierwirbelapparate.
DBP 2303212 (1973).
[9.84] Christmann, G.: Zerstubungs-Wirbelbettgranulator. Chemie-anlagen
+ Verfahren (1973)1, S.42-43.
[9.85] de Jong, B.: Fliebettanlage frdert Flexibilitt von Prozevorgngen.
Chemie-anlagen + Verfahren (1971)8, S.49-51.

[9.86] Shakova, N.A.; und G.A.Minaev: Investigation of the temperature
field in the spray zone of a granular with a fluidized bed. Internat.
Chem. Engng. 13(1973)1, S.65-68.
[9.87] Mori, Y.; und K. Nakamura: Solid mixing in a fluidized bed. Kagaku
Kogaku 4(1966)1, S.154-157.
[9.88] Pippel, W.; u.a.: ber die Vermischung des Feststoffes in Gas-
Feststoff-Wirbelschichten. Chem. Techn. 20(1968)12, S.750-755.



65
10 Solids removal in cyclone
separators
Bulk solid, which is transported by a gas stream, must
be separated from the gas. An elegant way is the
separation in cyclone separators, which are because of
their simplicity and reliability widely used. For this, the
solids-laden gas is being fed into the centrifugal
separator over tangential or spiral-shaped inlets,
sometimes also with the help of guide vanes. In the
cyclone, the gas moves helically downward, then in
spiral orbits inward, and flows through the dip tube
centrally back off (Fig. 10.1).
The separation process is being described vividly by
Krambrock [10.1]. The circumferential velocity in the
cyclone is constantly increasing from the outside to the
inside up to the dip tube radius, and drops in the vortex
core down to zero. The pressure decreases because of
the curved flow with diminishing radius. In the axis of
the cyclone there is a strong vacuum compared to the
outlet tube. At the technically common circumferential
speeds of 20 to 60 m/s and radiuses from 0.1 to 1 m, the
centrifugal acceleration is a hundred to a thousand
times greater than the acceleration due to gravity, so
that even small particles opposite to the inward directed
flow be driven to the cyclone wall.
The boundary layer on the walls of the cyclone rotates
due to the friction much more slowly than the main
flow. Therefore, the there acting centrifugal forces are
relatively small. Because, however, the pressure drop
of the main flow is impressed on the boundary layer, it
comes to strong, almost radially inward directed
secondary currents on the cover and on the conical
bottom. These secondary flows capture the segregated
particles, and discharge them in the conical lower
section of the apparatus, also when the cyclone lies or
"standing on its head". Characteristic are the product
strands, which are being formed due to disturbances in
the boundary layer in connection with the rotational
flow. On the cyclone bottom transports the secondary
flow likewise larger amounts of dust inward, which,
however, during the flow around the dip tube will be
again thrown to the wall of the cyclone.
Design and construction of a cyclone separator
concerns first and foremost the pressure drop and the
separation rate. Extensive theories of these complexes
were established by Barth [10.2] and Muschelknautz
[10.3]. However, regarding the separation efficiency
are the circumstances in the technical practice mostly
such, that the bulk material reaches the cyclone wall in
unfractionated state, because the gas already in the
conveyor pipe and in particular in the cyclone can carry
only a certain solids portion. In case of solid/gas rates
of = 0.1, as they are often present behind mills,
dryers and Sifters, about 90 percent of the material is
separated in this way [10.1]. The calculation on the
basis of the fractional separation efficiency and the
particle size distribution is valid only for the remaining
portion of material.
For high separating performances are cyclones with
large height and high inflow velocities necessary, with
an adverse effect on facility and operating costs.
Useful separation performances and acceptable
pressure losses are achievable at inflow velocities
between 7 and 20 m/s, in devices with a ratio between
cyclone diameter D and dip tube diameter d
i
of about 3,
and a ratio between height h
i
and dip tube diameter d
i

between 6 and 12. The pressure loss without taking into
account the influence of solid material, can at least for
such flow conditions and dimensions be described by a
simple power law:


2
0 . 5
2
4 . 0
e
g
i
d
i
h
i
e
u
f
f
d
h
p
i
i

(

(10-1)

Starting from the ratio between cyclone wall and dip
tube diameter of 3, Eq. (10-1) describes exactly the
conditions when changing the size of the cross-
sectional area of the inlet. A confirmation of this offers
Fig. 10.2, that depicts the results of studies, carried out
by Gloger and Niendorf [10.4] using a model cyclone,
in which the cross-sectional area of the inlet f
e
and thus
the ratio to the cross sectional area of the dip tube f
i
,
systematically has been varied. With the design data of
a cyclone series with three different diving tube
diameters, which before years was introduced by the
BASF, could be verified, that the variation of the dip

Fig 10.1
Schematic representation of the speed and pressure
conditions in a cyclone separator according to Krambrock
[10.1]

66
tube cross-sectional area causes identical pressure
losses compared to the calculation after the method of
Muschelknautz.
Trouble-free operation of cyclone separators can be
supported by specific constructive measures. Proven
elements are for example collection container at the
discharge of the cyclone. They prevent, that already
secluded product passes into the area of the dip tube
vortex. This effect can in vacuum systems be reinforced
by the use of leak gas shocks. A cover cone in the
collection reservoir removes the material finally
permanently from the influence area of the vortex.
Many of the to be separated materials have a strong
abrasive effect on the cyclone wall. Due to the high
circumferential velocity in the lower area of the
cyclone, this is the first affected part, later then the
overlying area. A cylindrical instead of a conical shape
of the cyclon can counteract this development. Low
flow velocities and linings with cast basalt are likewise
suitable, to hold the abrasive effect of solids in borders.


Literature of chapter 10.
[10.1] Krambrock, W.: Die Berechnung des Zyklonabscheiders und prak-
tische Gesichtspunkte der Auslegung. Aufbereitungstechnik (1971)7,
S.391-401.
[10.2] Barth, W.: Berechnung und Auslegung von Zyklonabscheidern auf-
grund neuerer Untersuchungen. Brennstoff-Wrme-Kraft 8(1956)1,
S.1-8.
[10.3] Muschelknautz, E.: Chemie-Ing.-Tech. 44(1972)1/2, S.63-71.
[10.4] Gloger, J.; und G.Niendorf: Untersuchungen an einem Modellzyklon
ber den Einflu verschiedener geometrischer Parameter auf Ab-
scheidegrad und Druckverlust. Chem. Techn. 22(1970)9, S.525-532.


Fig 10.2
Pressure loss in a cyclone with different inlet cross-
sectional areas depending on the gas inlet velocity:
comparison of measurements [10.4] with the calculation
according to Eq. (10-1)
1
Appendix

Industrial pneumatic conveying installations


A computer program, that includes a database, allows the quick and direct use of the here describt
algorithms for evaluation and design of processes, apparatuses and equipment. This free software is
written in Visual Basic and runs even under Windows 7 (32 bit).The german version can be
downloaded from the following link:

https://skydrive.live.com/redir.aspx?cid=0239e78e2f136dde&resid=239E78E2F136DDE!340
&authkey=s60rZ0M*8fA%24


An english version of the software can be downloaded from the following link:

https://skydrive.live.com/#cid=0239E78E2F136DDE&id=239E78E2F136DDE%21936


Literature researches, as they underlie this brochure, are despite the ubiquity of the Internet even today
only possible with the help of technical-scientific libraries. A godsend in this respect, is the for more
than 10 years existing bulk-online portal. In the forum on the subject of pneumatic conveying exists
meanwhile a real fund, that however can only with great effort be utilized. These data are the base for
the following considerations, which are based on my universal state diagram (Fig. 7.2) and my
software. The following link leads to the bulk-online portal:

http://forum.bulk-online.com/forumdisplay.php?11-Pneumatic-Conveying






2

( ) = calculated values, = solid /gas ratio

Table 1: Operating conditions of industrial installations



(1) data set coming from the following bulk-online thread:
http://forum.bulk-online.com/showthread.php?4754-Pneumatic-Conveying-of-PP

In this installation is the measured pressure loss of 0.3 bar about 0.1 bar bigger than the calculated
value for the minimum pressure loss in dilute phase. This surplus is surely caused by the 8 (not
specified) pipe bends of the conveying line.


(2) data set coming from the following bulk-online thread:
http://forum.bulk-online.com/showthread.php?3348-Conveying-of-LLDPE-Granules

These operating data fit extremely well in the scheme of the dense phase, although the particle size of
this material, in contrast to the fine-grained solids, which are commonly conveyed under such
conditions, lies in millimeter range. There were observed no general operational problems, except of
vibrations at pipe bends, which lead the flow vertically upwards. In regard to the terminal velocity of
particles with a size of 4 mm is the actual gas velocity too low. So, material can accumulate
themselves in the pipe bends and possibly develop a behaviour like a pulsating spouting bed, and thus
generate vibrations.
Nr. Product dp50
m
L
m
h
m
D
mm
&
M
s

t/h
&
M
g

Nm
3
/h
p
bar
u
m/s

-
1 PP pellets (1) (4000) 36.5 15.2 150 14 ca 2010 0.3 (30) (6)
2 LLDPE Granules (2) (mm) 100 -- 300 60 -- (0.41) (10.3) 20
3 Cement (3) -- 216 40 ca 285 85 ca 6000 1.6 -- --
4 Polycarbonat:
Powder & Pellets (4)
-- / 2500 50 + 60 15 + 20 100/125 6 / 7 ca 1700 1(0.73) (38) (3)
5 Expanded Perlite (5) 28.5 35.2 28.3 83 3,28 ca 280 0.25 -- (9.35)
6 PTA (6) 130 120 40 150 90 3000 2.66 (27.5) (24)
7 PP Pellets (7) -- 140 22 (187) 30 ca 2700 0.66 (21) (10)
8 Rapeseed (8) 2500 230 10 206 12 (2700) 0.4 (21.5) (3.6)
9 PET Virgin Chips (9) (>12700) 40 20 80 1.5 (ca 570) 0.15 (32) (2)
10 Cement (10) -- 176 -- 50 7.63 (233) 3.15 (18) (26)
11 Spent Cell Liner (11) -- 85 -- 77 2.5 (230) (0.3) (13) (9)
12 Cement (12) -- 132 55 250 120 ca 4200 1.7 (14.5) (24)
3


(3) data set coming from the following bulk-online thread:
http://forum.bulk-online.com/showthread.php?7057-Pneumatic-Transport-Cement-Pipeline-Pigging

The operating data of this system offered the opportunity to determine the boundary of dense flow
more accurately, because the operating conditions are lying exactly on this line. For a gas temperature
of 125 C and a bulk solid throughput of 85 t / h, the calculation on the basis of the improved
coordinates (no longer equal to those of the original state diagram) yields the following values: 7060
kg / h for the gas flow and 1.66 bar for the pressure loss. In this thread was also an increase to 100 t / h
solids throughput discussed; according to my calculations, however, this requires a gas flow of 7860
kg / h at a pressure loss of 1.92 bar, for which the installed compressor capacity does not suffice.

(4) data set coming from the following bulk-online thread:
http://forum.bulk-online.com/showthread.php?21119-Increasing-the-Conveying-Capacity


(5) data set coming from the following bulk-online thread:
http://forum.bulk-online.com/showthread.php?16367-Optimisation-of-a-Pneumatic-Conveying-Line

After the program's calculation lies the operating point of this installation on the boundary of dilute
phase, the so-called saltation line.


(6) data set coming from the following bulk-online thread:
http://forum.bulk-online.com/showthread.php?7259-Conveying-of-PTA

After the calculation lies the operating point of this installation also on the boundary of dilute phase,
the so-called saltation line. The operational pressure loss is equal to the calculated value, when a
(unconfirmed) gas temperature of 90 C is used.


4
(7) data set coming from the following bulk-online thread:
http://forum.bulk-online.com/showthread.php?4797-PP-Pellet-Conveying-Problem

This operating point lies close to the so-called saltation line of the dilute phase. For this operation
conditions yields the calculation a slightly smaller gas flow rate and looks as follows:




The average gas velocity is much smaller than for the conveying of PP pellets in the installation No. 1
of Table 1.

(8) data set coming from the following bulk-online thread:
http://forum.bulk-online.com/showthread.php?5655-Pneumatic-Conveying-of-Rapeseed

This is another good example, to verify the calculated results. For a bulk solid throughput of 12 t / h
yields the calculation for the condition of minimum pressure loss in dilute phase a value of 0.36 bar.
This value lies near the pressure difference of 0.4 bar of the blower, what the possible solids
throughput accordingly limits.


5
(9) Data set coming from the following bulk-online thread:
http://forum.bulk-online.com/showthread.php?4794-Why-Is-Throughput-Reduced

As usual for bulk materials with large particle diameters, the gas velocity is much higher than the
calculated value for the minimum pressure loss in dilute phase: 32 m/s instead of 14 m/s. However, the
bulk solid throughput is too small for such a gas velocity, and I am also of the opinion, that the reason
for this is the poorly functioning material feed. The fan is designed for pressures up to 0.5 bar at a gas
flow rate of 700 m3 / h and would be able, to handle a much higher bulk solid throughput. The
calculation result for the current bulk solid throughput of 1.5 t / h is as follows:





(10) data set coming from the following bulk-online thread:
http://forum.bulk-online.com/showthread.php?5597-Typical-Solids-Friction-Factor-for-Cement

The calculation results for the saltation condition on the boundary of the dilute phase for a
(unconfirmed) gas temperature of 110 C are identical to the operating data:



6
(11) For completeness, here are added the operating conditions for the conveying of used, very
abrasive cell liner:
http://forum.bulk-online.com/showthread.php?5437-Lean-Phase-Conveying-of-Abrasive-Cell-Liner

The information for this system are not complete, However, the calculated pressure loss of 30 kPa for
the saltation conditions fits quite well to the blower specification of 40 or 60 kPa.


(12) data set coming from the following bulk-online thread:
http://forum.bulk-online.com/showthread.php?3709-Determinining-Pneumatic-Conveying-Parameters

The operating point lies a little left hand from the boundary of the dense phase. The estimated values
for the parameters K and Re of the state diagram are 450 and 450,000. The calculation results for gas
flow rate and the pressure drop confirm the assumptions.









7
State diagram



The main concern of the state diagram is the universal and clear presentation of the operating
data of industrial conveying installations with their large number of parameters. For this
purpose a large number of published data from industry and research was added into the
coordinate system. They form the skeleton of the state diagram and reveal the areas of dilute
phase and dense phase conveying as well as their boundaries. There are three characteristic
conveying states, that lie in the diagram on particular lines. Two of the states are the pressure
loss minimum in dilute phase as well as the boundary to the transition area, the so-called
saltation (choking) line. The third one is the boundary of dense phase on the other side of the
transition area, the position of which was determined by the help of the operational data No. 3
in Table 1 more accurately. From the data of dense phase conveying was derived in addition
an estimation for frequently used pipe diameters as ratio to the bulk material throughput.

In order to facilitate the estimation of the sizes of conveying lines, the software helps, as
mentioned, in determining a usable pipe diameter. Furthermore, can be determined on the
basis of the design data, the relevant parameters for the three characteristic conveying states.
The respective values are being calculated by means of simple mathematical equations for
the respective straight lines. For all other possible operating points, the respective parameters
can be read from the state diagram. However, the left border line of the display, which
represents the boundary of the bulk solid movement, is only indicated, but may be precised, as
described above.

For the calculations of conditions in dilute phase uses the program its default pipe diameter
and corrects the calculation result in the case of deviations from the actual diameter. This
correction has been found to be necessary during the recalculation of the used literature data, ,
but it is surely not particularly sophisticated. For bulk solids with big particles is the gas
8
velocity in some cases bigger than the velocity at the point of minimum pressure loss in dilute
phase: for example, 38 m / s instead of 21 m / s for the installation No. 4 (2.5 mm) in Table 1.

In dense phase the operating points are distributed over the entire range. This opens up the
possibility to draw lines of constant bulk solids throughput. Originally, the maximum bulk
solid throughput was 19.5 t / h. Fortunately, the installation No. 3 offered the opportunity for
an expansion of the throughputs and the accurate determination of the position of the
boundary of the dense phase region, because this operating point lies exactly on this line. As
already described above, yields the calculation on the basis of the new coordinates (which are
no longer those of the original in the state diagram registered correspond) for a gas
temperature of 125 C and a bulk solid flow of 85 t / h the following result: 7060 kg / h for
the gas flow and 1.66 bar for the pressure loss.

Once again here is the attention being focused on the installation No. 2 in Table 1. These
operating data fit extremely well in the scheme of the dense phase, although the particle size
of this material, in contrast to the fine-grained solids, which are commonly conveyed under
such conditions, lies in millimetre range. There were observed no general operational
problems, except of vibrations at pipe bends, which lead the flow vertically upwards. In
regard to the terminal velocity of particles with a size of 4 mm is the actual gas velocity too
low. So, material can accumulate themselves in the pipe bends and possibly develop a
behaviour like a pulsating spouting bed, and thus generate vibrations.















*) In the case of horizontal conveyor lines with vertical pipe sections for the steady conveying
state an equivalent length must be used. For this, the value for the whole vertical pipe length
is being enlarged by a factor between 1.7 and 2.0.

**) In the evaluated data of industrial installations was no evidence of additional influences,
which are dependent on the conveyed bulk solids. The only relevant parameter is the material
throughput. Also no other physical similarities could be discovered, which are connected with
the bulk solid properties in some way and possibly exist in form of hidden parameters. Even
the oft-cited solid /gas ratio turned out to be unusable. A factor, which is in this context
closest to a parameter, is the bulk solid throughput related to the cross-sectional area of the
pipe. This fact possibly shows a connection with the Froude number according to Eq. (5-4).


9
This is the scetch for the dense phase area of the state diagram, which was improved by using
the here analyzed operating data . The on the border to the transition area lying endpoints of
the lines of equal bulk solid throughput were calculated by the appropriate modified software
program.






10
Pneumatic conveying at high temperatures

Operating data for industrial pneumatic conveyor lines that are operated at high gas
temperatures are to be found at the following Internet address:

http://www.enviro-engineering.de/pdf/InjektorenBauformenVariantenUndAnwendungen.pdf

( ) = Calculated values, = solid / gas ratio

Table 2: Operating data of industrial installations with high gas temperatures


(1) As in other installations for the transport of bulk materials with large particles, the gas
velocity in this conveying line is by a factor of 1.6 higher than those, calculated for the
minimum pressure loss in dilute phase. The extrapolation from the calculated operating point
of minimum pressure loss gives values of 100 000 and 25 for the parameters Re and K. The
calculation results look like this.





Nr. Product dp50
m
L
m
h
m
D
mm
&
M
s

t/h
&
M
g

Nm
3
/h
p
bar
u
m/s

-
1 Ash (1) / 80 C 0-3000 20 25 80 0.6 ca 350 (0.14) (23.6) (ca 1.5)
2 Ash (2) / 250 C 0-1000 11 15 178 6.5 ca 750 (0.133) (16.5) 6.5
3 Cement (3) / 125 C -- 216 40 ca 285 85 ca 6000 1.6 -- --
11
(2) The data of this conveying system describe the operating condition at a gas temperature
of 250 C, which due to a longer cooling period of the ash under the actual operating
temperature of 450 C lies. The measured ash throughput of 6.5 t / h is assessed as being
surprisingly high. The calculation for this condition results in a pressure loss of 0.133 bar at a
gas flow of approximately 1000 kg / h.


The following calculation results describe now the situation at the actual operating
temperature of 450 C and at the same pressure loss. The gas flow rate lies at about 900 kg /
h, while the ash throughput at this gas temperature is reduced to a value of 3500 kg / h; this
value corresponds exactly to the design conditions of the conveying line.




(3) This is once again the installation No. 3 of Table 1 with the corresponding interpretation: The
operating data of this system offered the opportunity to determine the boundary of dense flow more
accurately, because the operating conditions are lying exactly on this line. For a gas temperature of
125 C and a bulk solid throughput of 85 t / h, the calculation on the basis of the improved coordinates
(no longer equal to those of the original state diagram) yields the following values: 7060 kg / h for the
gas flow and 1.66 bar for the pressure loss. In this thread was also an increase to 100 t / h solids
throughput discussed; according to my calculations, however, this requires a gas flow of 7860 kg / h at
a pressure loss of 1.92 bar, for which the installed compressor capacity does not suffice.