Powder Technology 121 Ž2001.

230–238
www.elsevier.comrlocaterpowtec

The design of flights in rotary dryers

D. Revol a , C.L. Briens a,) , J.M. Chabagno b
a
Department of Chemical and Biochemical Engineering, The UniÕersity of Western Ontario, London, Ontario, Canada, N6A 5B9
b
Groupement de Recherches de Lacq, ATO FINA, BP 34, 64170 Lacq, France
Received 28 November 2000; received in revised form 29 March 2001; accepted 3 April 2001

Abstract

Rotary dryers for particulate solids commonly use flights along the length of the shell to lift solids and make them rain across the
dryer section. These flights are critical to dryer performance. Previous models for the prediction of the solids holdup in rotary drums have
assumed that the angle of the solids level in a flight is independent of the flight geometry. They have considered only the case of extended
circular flights and angular flights with two segments.
New equations were derived to predict the solids holdup in flights with three segments. Their predictions were compared to
measurements performed with a pilot plant rotary dryer. Theoretical equations provided estimates of the flight holdup that were accurate
enough to predict the power required to lift the solids. On the other hand, the model is not accurate enough to predict the variations in the
flux of solids over the drum cross-section. This shows that the angle of the solids level in a flight can be affected by the flight geometry.
q 2001 Elsevier Science B.V. All rights reserved.

Keywords: Rotary dryer; Flight; Solids holdup; Dynamic coefficient of friction

1. Introduction function of its dynamic coefficient of friction. If a powder

is poured on a flat surface, it will form a pile whose angle
Rotary dryers are commonly used to dry particulate with the horizontal plane is called the static angle of
solids w1–5x. They are made of a long cylindrical shell that repose. This angle of repose is affected by the powder
is rotated. The shell is usually slightly inclined to the cohesivity. Particles within a flight will also display an
horizontal to induce solids flow from one end of the dryer angle of repose, with a horizontal plane, which will depend
to the other. In direct heat rotary dryers, a hot gas flowing on the angular position of the flight. Since the angle of
through the dryer provides the heat required for the vapor- repose is affected by the drum rotation speed, it is called
ization of the water. To promote gas–solid contact, most the dynamic angle of repose. Schofield and Glikin w6x
direct heat dryers have flights, placed parallel along the provided an equation that relates this dynamic angle of
length of the shell, which lift solids and make them rain repose to the drum rotation speed, the flight angular posi-
across the section of the dryer Fig. 1.. tion and an intrinsic powder characteristic, which they
The flights are critical to dryer performance. The solids called the dynamic coefficient of friction.
should rain down over nearly 1808 of rotation. In many Kelly w7,8x provided a measurement procedure for the
cases, this is achieved by using curved flights of an dynamic friction coefficient. He also reviewed the equa-
appropriate shape that lift and then release gradually a tions that relate the drum loading and the flight loading w7x.
large amount of solids. Good flight design is essential to Baker w9x showed how the dynamic angle of repose
promote the gas–solid contact that is required for rapid could be used to calculate the solids holdup of a flight at
and homogeneous drying. any angular position. He developed equations for several
Several authors have developed design procedures for typical flight geometries. Baker w9x also explained how to
the flights. Schofield and Glikin w6x derived an equation to estimate the optimum number of flights in a rotary dryer to
evaluate the dynamic angle of repose of a powder as a maximize the drum holdup and thus make the most effec-
tive use of the dryer volume.
Kelly w10x extended his theory to the prediction of the
)
Corresponding author. ideal flight geometry that would provide the equal horizon-
E-mail address: cbriens@uwo.ca ŽC.L. Briens.. tal distribution ŽEHD. of solids. The EHD distribution

0032-5910r01r$ - see front matter q 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 0 3 2 - 5 9 1 0 Ž 0 1 . 0 0 3 6 2 - X
 D. ReÕol et al.r Powder Technology 121 (2001) 230–238 231

Each cylinder is half-filled with powder and the drum is

rotated. Photographs of the cylinder at various speeds
provide the dynamic angle of repose for various rotation
speeds and angular positions. Eq. Ž1. can then be solved to
provide the dynamic coefficient of friction Žthe radius R O
is then replaced by the drum radius R ..
Since Baker w9x did not provide equations for three-seg-
ment flights, a new set of equations was developed. This
set of equations can be generalized to any number of
segments. For example, it can also be used for two-seg-
ment flights.
An example of three-segment flight is given in Fig. 2. It
can be characterized by the segment lengths L, LX and LY ,
their angles aA and a B and the radius R O of the circle
described by the flight tip O as the drum rotates Žsee Fig.
2.. Two sets of Cartesian coordinates are considered. The
origin of the Ž x, y . set is at the flight tip, with the x-axis
along the first segment; this set of coordinates moves as
the flight rotates. The origin of the stationary Ž X,Y . set is
Fig. 1. Solids cascade within the pilot plant dryer Žcounterclockwise at the drum axis, the X-axis being horizontal ŽFig. 2..
rotation.. According to Schofield and Glikin w6x, the powder level
in the flight forms an angle with the horizontal direction
that is equal to the dynamic angle of repose, which can be
obtained from Eq. Ž1.. To calculate the volume of powder
would distribute solids evenly on the surface of a horizon- in the flight, the coordinates of points A, B and C are first
tal plate introduced in the middle of a rotating drum. It calculated Žsee Fig. 2., the angle d between the two sets of
would ensure excellent gas–solid contact in a dryer. coordinates is evaluated and the volume of powder is
The flight geometry also affects the residence time finally obtained by considering various types of powder
distribution of solids in rotary dryers. Hatzilyberis and fills.
Androutsopoulos w11x obtained the solids residence time The equations of the three segments can easily be
with three different flight geometries. Cao and Langrish obtained:
w12x reviewed various models for the prediction of solids
residence time distributions. v segment 1:
In spite of the important effect of flight design on rotary
dryer performance, no experimental validation of the mod- y1 s 0 Ž 2.
els from Schofield and Glikin w6x, Kelly w7,8,10x and Baker v segment 2:
w9x has been published. The objective of this publication is
to extend and test these models. y 2 s a2 q b 2 x Ž 3.
with: a 2 s x AtanŽ aA .; b 2 s ytanŽ aA .
v segment 3:
2. Theory
y 3 s a3 q b 3 x Ž 4.
Schofield and Glikin w6x related the dynamic angle of with: a 3 s y B q x B tanŽ aA q a B .; b 3 s ytanŽ aA q
repose f to the dynamic coefficient of friction m of the aB.
powder, the angular position u of the flight tip, the radial
position R O of the flight tip and the rotation speed v of The coordinates of points A, B and C are given by:
the drum:
v2 v point A:
m q RO Ž cos u y m sin u .
g x A s LY
tan f s Ž 1.
v2 yA s 0 Ž 5.
1 y RO Ž sin u y m cos u .
g v point B:
Kelly w8x showed that the dynamic coefficient of friction x B s x A y LX cos Ž aA .
of a powder could be obtained by placing transparent
cylinders on the perimeter of a drum, at equal intervals. y B s LX sin Ž aA . Ž 6.
232 D. ReÕol et al.r Powder Technology 121 (2001) 230–238

with as ordinate:

y 3 s a3 q b 3 x 3 Ž 15 .
The intersection of the solid level line with the drum wall
has the following abscissa:

xw sy
(
Bw " Bw2 y 4 A w C w
Ž 16 .
2 Aw

with: A W s 1 q wtan Ž g .x 2 ; B W s 2 X O wcos Ž d . y

tanŽg .sinŽ d .x q 2YO wtanŽg .cosŽ d . q sinŽ d .x; C W s R O2 y
R 2 ; and its ordinate is given by:

y W s x W tan Ž g . Ž 17 .
Four types of powder fills can occur:
Fig. 2. Calculation of the volume of powder in a flight.
v the powder reaches the wall. This will occur if:
v point C: yC
x C s x B q Lcos Ž aA q a B . Ž 7.
g ) A tan ž / xC
Ž 18 .

yC s y B y Lsin Ž aA q a B . and the cross-sectional area occupied by powder is

then given by:
In the stationary set of coordinates, the position of point C
must satisfy the following equation since it is located on R2 1
Ss b y sin Ž b . q < x A y B q x B yC y x C y B
the wall of the drum of radius R: 2 2
XC2 q YC2 q s R 2 Ž 8. q x C y W y x W yC < Ž 19 .
The two sets of coordinates are related by the following
with:
equations:
XC s XO q x C cos Ž d . y yC sin Ž d .
b s 2 A sin
(Ž x C yxW
2
. q Ž yC y y W .
2

Ž 20 .
s R O cos Ž u . q x C cos Ž d . y yC sin Ž d . Ž 9. 2R
YC s YO q yC cos Ž d . q x C sin Ž d .
v the powder does not reach the wall but reaches the
s R O sin Ž u . q yC cos Ž d . q x C sin Ž d . Ž 10 .
third segment. This will occur if:
Substituting Eqs. Ž9. and Ž10. into Eq. Ž8. yields an
equation which can be solved for d , for any specified yC
angular position u .
g ) A tan ž / xC
and (Ž x y x
3 C
2
. q Ž y 3 y yC .
2

The equation of the powder level line is given by:

-L Ž 21 .
y s x tan Ž g . s x tan Ž f y d . Ž 11 .
Its intersection with the line tracing the second segment and the cross-sectional area occupied by powder is
has the following abscissa: then given by:
a2
x2 s Ž 12 . 1
tan Ž g . y b 2 Ss < xA y B q x B y 3 y x 3 y B < Ž 22 .
2
with as ordinate: v the powder does not reach the third segment but
y 2 s a2 q b 2 x 2 Ž 13 . reaches the second segment. This will occur if:
Its intersection with the line tracing the third segment has yC
the following abscissa: g ) A tan ž / xC
and (Ž x y x
2 B
2
. q Ž y2 y y B .
2

a3
x3 s Ž 14 .
tan Ž g . y b 3 - LX and y2 ) 0 Ž 23 .
 D. ReÕol et al.r Powder Technology 121 (2001) 230–238 233

Table 1 If L z is the flight length and r is the bulk density of

Flight geometry Žsymbols are defined in the Notation and in Fig. 2. the solids in the flight, the potential energies are given by:
Flight Flight Flight
type a1 type a2 type a3 pr2
D E1up s Ž r gRL z . Hypr2S Ž u . cos u d u Ž 29 .
Number of segments 2 3 3
L, mm 31 28 41
X
L , mm – 18 15 3p r2
Y D E1down s Ž r gRL z . Hpr2 S Ž u . cos u d u Ž 30 .
L , mm 26 18 15
aA , 8 162 150 135
aB, 8 – 150 135 Substituting in Eq. Ž28. provides the total power consump-
aC , 8 90 90 90 tion for solids circulation, as the flight completes a full
rotation by going from ypr2 to 3pr2:
Nvr gRL z 3p r2
Ps
2p
Hypr2S Ž u . cos u d u Ž 31 .
and the cross-sectional area occupied by powder is
then given by:
1 3. Equipment and experimental procedures
Ss < xA y 2 < Ž 24 .
2
3.1. Powder characteristics
v the flight is empty. This will occur if:
The powder consisted of adsorbent beads, 0.7 mm in
y2 - 0 Ž 25 . diameter, with a bulk density of 800 kgrm3 a moisture
content of 0.03 kg H 2 Orkg dry solids.
Although most of the experiments used powder with a
The above model works for angles u larger than 08.
moisture of 0.03 kgrkg, some tests were conducted at
The maximum flight holdup is assumed to be reached at
higher moistures. To humidify the powder as uniformly as
u s y908. At angles between y908 and 08 Žypr2 and 0
possible, it was introduced in the dryer and humid air was
radians., the flight holdup can be obtained by assuming a
circulated until steady state was reached. Humid air of a
Amirror imageB w7x, i.e. the amount of solids lost by the
constant humidity was obtained by combining a stream of
flight as it rotates from u Ž- 0. to 0 is equal to the amount
dry air Ž12% humidity. with a stream of air saturated with
lost as it rotates from 0 to < u <:
water vapor in a flooded packed column. The moisture of
S Ž u . s S Ž 0. q S Ž 0. y S Ž < u <. Ž 26 . the powder was varied by adjusting the proportions of dry
and saturated air.
The above model allows the prediction of the variation
with angular position of the solids holdup S in a flight. 3.2. Dryer
The solids mass flux per unit area of the plane going
through the drum axis Žhorizontal plane if the drum is not The rotating drum was 1.21 m long, with an internal
inclined., can then be obtained from Žsee Appendix A.: diameter of 0.387 m. It was made of Plexiglas to facilitate
r Nv dS
Fs s Ž 27 .
2p Rsin u d u
The above model also allows for the prediction of the
power required to rotate the dryer. The following deriva-
tion does not take into account the friction in the bearings.
The net power required to lift the solids is given by:
N Nv
Ps
t1
Ž D E1up y D E1down . s 2p Ž D E1up y D E1down .
Ž 28 .
where N is the number of flights, t 1 is the time required
for a flight to go through a whole rotation, D E1up is the
potential energy required to lift the particles in one flight
from ypr2 to pr2 and the D E1down is the energy
recovered by the particles of the flight coming down from Fig. 3. Dynamic coefficient of friction for the powder with 3 wt.%
pr2 to 3pr2. moisture Ž4 rpm drum rotation speed..
234 D. ReÕol et al.r Powder Technology 121 (2001) 230–238

were taken repeatedly. The dynamic angle of repose f

was determined from the pictures and, together with the
corresponding angular position u of the cylinder, was used
to obtain the dynamic coefficient of friction from Eq. Ž1..

3.4. Measurement of the flight holdup

The flight holdup was determined by taking pictures.

The 12 flights were attached to the drum wall and the
drum was loaded with the optimum amount of solids,
defined as the loading at which powder at the bottom of
the drum just starts flowing over the top of the flights w13x.
The drum rotation was started and pictures were taken
every 58 for flight angular positions ranging from 08 to
Fig. 4. Effect of solids moisture on the dynamic friction coefficient Ž4
1108. For each angular position, the camera was carefully
rpm drum rotation speed..
located to avoid parallax errors.
The volume of bulk solids in a flight, at a particular
angular position, was obtained from the cross-sectional
visual observations. The drum was perfectly horizontal. An area occupied by solids. Integration from the pictures
electric motor with an electronically controlled speed ro- provided this area.
tated the drum through a transmission with a speed reducer
ratio of 1r300. For all the experiments reported herein, the
drum rotation speed was 4 rpm.
4. Experimental results and discussion
The drum was equipped with 12 removable aluminum
flights. Three different flight geometries were tested. Table
1 shows the characteristics of the three flight geometries. 4.1. Dynamic coefficient of friction

3.3. Measurement of the dynamic coefficient of friction Fig. 3 shows the measured coefficient of friction for the
powder with 3 wt.% moisture. The dynamic coefficient of
The dynamic coefficient of friction of the powder was friction was independent of the angular position of the
measured with the method proposed by Kelly w8x. Four cylindrical container, as assumed by the theory. From the
cylindrical containers were half-filled with powder, closed 73 runs shown in Fig. 3, the 95% confidence interval for
to prevent any change in solids moisture and attached to the dynamic coefficient of friction is Ž0.562, 0.575., with
the rotary drum wall. The drum was rotated and pictures an average value of 0.568.

Fig. 5. Effect of solids moisture on the dynamic friction coefficient Ž4 rpm drum rotation speed.: 95% confidence intervals.
 D. ReÕol et al.r Powder Technology 121 (2001) 230–238 235

Fig. 6. Variation of the flight solids holdup, expressed as m3 of bulk solids per m of flight linear length, with the angular position of the flight tip ŽFlights
a1 and 2, 4 rpm, 32 wt.% moisture..

Fig. 4 shows that the dynamic coefficient of friction particles became saturated and liquid water appeared on
remained nearly constant as the solids moisture was in- the outside of the particles, making them sticky.
creased from 0 to 0.35 kgrkg. The 740 values displayed
on Fig. 4 also indicate that there was a significant scatter 4.2. Flight holdup— comparison with model predictions
in the results since values ranged" 10% for each condi-
tion. The large number of measurements, however, greatly The flight holdup, expressed as the volume of bulk
reduced the random errors, as indicated by the small 95% solids in the flight per unit length of flight, decreased with
confidence intervals shown in Fig. 5. the flight angular position: solids steadily emptied out of
Fig. 5 confirms clearly that the dynamic coefficient of the rotating flight. This was confirmed by preliminary
friction remained nearly constant as the solids moisture experiments conducted with flight types a1 and a2 Fig. 6.
was increased from 0 to 0.35 kgrkg and started to rise Fig. 6 also shows that there is a reasonable agreement
quickly as the solids moisture was increased past 0.35 between the measured holdups and the holdups predicted
kgrkg. The porous structure of the solid particles means with the equations developed earlier. More thorough exper-
that, at low solids moisture, there was no liquid water on iments were thus conducted to test the model equations.
the outside of the particles. At moistures above 33%, the

Fig. 7. Variation of the flight solids holdup, expressed as m3 of bulk Fig. 8. Comparison of the measured and predicted flight solids holdups,
solids per m of flight linear length, with the angular position of the flight expressed as m3 of bulk solids per m of flight linear length ŽFlight a3, 4
tip ŽFlight a3, 4 rpm, 3 wt.% moisture.. rpm, 3 wt.% moisture..
236 D. ReÕol et al.r Powder Technology 121 (2001) 230–238

power consumption. The error due to the difference be-

tween predicted and actual holdups is small when com-
pared to other components of the power consumption, such
as bearings friction and motor efficiency.

5. Conclusions

Previous models for the prediction of the solids holdup

in rotary drums have assumed that the angle of the solids
level in a flight is independent of the flight geometry. New
equations were derived from the same assumption to pre-
dict the solids holdup in flights with three segments.
Fig. 9. Comparison of the measured and predicted solids fluxes ŽFlight Theoretical equations provided estimates of the flight
a3, 4 rpm, 3 wt.% moisture.. The solids flux is expressed per unit area of holdup in a pilot plant rotary dryer which were accurate
the horizontal plane including the drum axis. enough to predict the power required to lift the solids.
On the other hand, the model does not seem accurate
enough to predict the variations in the flux of solids over
Fig. 7 shows three series of measurements taken with the drum cross-section. This shows that the angle of the
flight type a3 ŽTable 1. in a drum rotating at 4 rpm, with solids level in a flight can be affected by the flight
3 wt.% moisture solids. Although there was some scatter geometry.
in the raw data, the number of replicates was sufficient to
significantly reduce the width of the 95% confidence
intervals ŽFig. 7.. Notation
Fig. 8 compares the measured flight holdups with the AW coefficient defined in Eq. Ž16.
holdups predicted by the model equations. Measurement a1 intercept of the line traced by the first flight
errors on the dynamic coefficient of friction had a negligi- segment OA Žsee Fig. 2., m
ble effect on the predicted holdup, as shown by performing a2 intercept of the line traced by the second flight
calculations with the two ends of the 95% confidence segment AB Žsee Fig. 2., m
interval on the dynamic coefficient ŽFig. 8.. a3 intercept of the line traced by the third flight
Although the model gave a reasonable prediction of the segment BC Žsee Fig. 2., m
flight holdups ŽFig. 8., it does not seem to predict the BW coefficient defined in Eq. Ž16., m
reduction in solids flowrate observed for angular positions b1 slope of the line traced by the first flight segment
ranging from 208 to 508. This discrepancy appears more OA Žsee Fig. 2., m
clearly when considering the flux of solids raining down b2 slope of the line traced by the second flight
from the flights, expressed per unit of area of the horizon- segment AB Žsee Fig. 2., m
tal plane including the drum axis ŽFig. 9. The actual solids b3 slope of the line traced by the third flight segment
flux was more uniform than the predicted flux. BC Žsee Fig. 2., m
The discrepancy between predicted and measured CW coefficient defined in Eq. Ž16., m2
holdups and fluxes suggests that the dynamic angle of D E1down potential energy recovered as the particles in 1
repose is not completely independent of the shape of the flight go from pr2 to 3pr2, J
flight. In other words, powder at the surface seems to be D E1up potential energy required to lift the particles in 1
affected by interactions between powder and flight wall, flight from ypr2 to qpr2, J
below the surface. Fs solids mass flux per unit area of the horizontal Žif
the drum is not inclined. plane including the drum
4.3. Power consumption axis, kgrsrm2
g gravity constant, mrs 2
Using Eq. Ž31. and the theoretical equations for the L length of third flight segment, nearest the drum
flight holdup SŽ u ., the power consumption for the drum wall Žsee Fig. 2., m
X
equipped with 12 flights of type a3 and rotated at 4 rpm is L length of second flight segment Žsee Fig. 2., m
Y
1.025 W. Using Eq. Ž31. and the measured flight holdups, L length of first flight segment, nearest the flight tip
the power consumption is 1.070 W, i.e. a difference of less Žsee Fig. 2., m
than 5%. Lz flight length, in direction parallel to the drum
The theoretical equations developed above to predict axis, m
the flight holdup can therefore be used to predict the N number of flights
 D. ReÕol et al.r Powder Technology 121 (2001) 230–238 237

P power required to rotate the dryer, W Acknowledgements

R drum radius, m
RO radius of the circle traced by the flight tip as drum The authors gratefully acknowledge the financial contri-
rotates, m butions of ATOFINA and NSERC.
S cross-sectional area occupied by the powder in the
flight Žin plane perpendicular to the drum axis.,
m2 Appendix A
XC abscissa of point C in the coordinate set centered
on drum axis Žsee Fig. 2., m While a flight goes from angular position u to Ž u q d u .,
XO abscissa of flight tip in the coordinate set centered the mass of solids it releases is:
on drum axis Žsee Fig. 2., m r Lzd S
x2 abscissa of points in the second flight segment
When a flight moves from angular position u to qpr2,
AB Žsee Fig. 2., m
the projection of its line of attachment to the wall on the
x3 abscissa of points in the third flight segment BC
central plane delineates the area:
Žsee Fig. 2., m
xA abscissa of point A Žsee Fig. 2., m L z Rcos Ž u .
xB abscissa of point B Žsee Fig. 2., m When a flight moves from angular position u to Ž u q
xC abscissa of point C Žsee Fig. 2., m d u ., the projection of its line of attachment to the wall on
xw abscissa of point W, at intersection of powder the central plane delineates the area:
level with wall Žsee Fig. 2., m
YC ordinate of point C in the coordinate set centered L z Rsin Ž u . d u
on drum axis Žsee Fig. 2., m While a flight goes from angular position u to Ž u q d u .,
YO ordinate of flight tip in the coordinate set centered the mass of solids it releases per unit of central plane area
on drum axis Žsee Fig. 2., m is therefore:
y1 ordinate of points in the first flight segment OA r Lzd S
Žsee Fig. 2., m
y2 ordinate of points in the second flight segment L z Rsin Ž u . d u
AB Žsee Fig. 2., m The time that a flight takes to go from angular position
y3 ordinate of points in the third flight segment BC u to Ž u q d u . is given by:
Žsee Fig. 2., m du
yA ordinate of point A Žsee Fig. 2., m
yB ordinate of point B Žsee Fig. 2., m v
yC ordinate of point C Žsee Fig. 2., m The flux during that time:
yw ordinate of point W, at intersection of powder r dS v
level with wall Žsee Fig. 2., m
Rsin Ž u . d u du
Greek letters Fraction of time that a given flight is between u and
aA angle between the first and second flight segments Ž u q d u .:
Žsee Fig. 2., radian du
aB angle between the second and third flight seg-
2p
ments Žsee Fig. 2., radian
b angle defined by Eq. Ž20., radian Average flux from 1 flight:
d angle of the coordinate set centered on the flight rdS du v
tip with the coordinate set centered on the drum Rsin Ž u . d u 2p d u
axis Žthe angle d shown in Fig. 2 is negative.,
radian Average flux from N flights:
f dynamic angle of repose, radian vN rdS
g angle of the powder level with the first flight 2p Rsin Ž u . d u
segment OA Žsee Fig. 2., radian
m dynamic coefficient of friction,-
t1 time required for a flight to go through a whole
rotation, s References
u angular position of the flight tip Žsee Fig. 2.,
w1x J.J. McKetta, Encyclopedia of Chemical Processing and Design 17
radian Ž1976. 1–29.
r bulk density of the powder, kgrm3 w2x Kirk-Othmer, Encyclopedia of Chemical Technology, 3rd edn., vol.
v angular rotation speed of the drum, radianrs 8 Ž1960. 75–113.
238 D. ReÕol et al.r Powder Technology 121 (2001) 230–238

w3x P.Y. McCormick, Chem. Eng. 95 Ž11. Ž1998. 113–122. w9x C.G.J. Baker, Drying Technol. 6 Ž4. Ž1988. 631–653.
w4x R.B. Keey, Drying: Principles And Practice, Pergamon Press, Ox- w10x J. Kelly, Drying Technol. 10 Ž4. Ž1992. 979–993.
ford, 1972. w11x K.S. Hatzilyberis, G.P. Androutsopoulos, Drying Technol. 17 Ž4–5.
w5x A.S. Mujumdar, Handbook of Industrial Drying, Marcel Dekker, Ž1999. 759–774.
New York, 1987, pp. 133–154. w12x W.F. Cao, T.A.G. Langrish, Drying Technol. 17 Ž4,5. Ž1999. 825–
w6x F.R. Schofield, P.G. Glikin, Trans. Inst. Chem. Eng. 40 Ž1962. 836.
183–190. w13x A.J. Machett, M.S. Sheik, Trans. IChemE 68 ŽPart A. Ž1990.
w7x J.J. Kelly, J.P. O’Donnell, Chem. Eng. Symp. Ser. 29 Ž1968. 34–44. 139–148.
w8x J.J. Kelly, J. Soc. Cosmet. Chem. 21 Ž1970. 37–51.