In T at the critical point 4. Cheairs, M. N., and E. C. Makin, Paper
Data, 5, 269
= activity coefficient
= square root of nonpolar (dis-
persion) pressure =
:( (Y‘ hv, )
1/2
= Hildebrand’s
solubility parameter for non-
polar molecules, atm.’”
= dipole moment, esn. cm.
= dispersion frequency, set.-'
= effective volume fraction
xc VL‘
2x6 VL,
i5
= square root of polar (orienta-
l K \ l 2
tion pressure -
712
- k,)
(3
= polar pressure of
van Arkel, atm.lP
LI1‘ERATURE CITED
1. Barker, J. A., et al., j . Chem. Phys., 19,
1430 (1951); 20, 1526 (1952); 21,
1391 (1953); 22, 375 (1954).
2. Bromley, E. C., and Dorothy Quiggle,
Ind. Eng. Chem., 25, 1136 (1933y.-
3. Carlson, H. C., and A. P. Colburn,
ibid., 34, 581 (1942).
A Theoretical Correlation of
Spray-Dryer Performance
Spray dryers, because of their effec-
tive contact between heating medium
and particles undergoing drying, are
potentially applicable to the widest
variety of materials. However their use
has been retarded by the lack of relia-
ble engineering procedures for scaling
up and analyzing performance. Many
studies have been made of the indi-
vidual factors involved in the spray-
drying process ( 3 , 7, 8 ) . To facilitate
scale up and permit design from basic
fundamentals a unifying theory has
been needed to tie these factors to-
gether.
In this development of an overall
correlation of spray-dryer performance
the author has calculated the capacity
heat transfer or production rate based
on the heat transfer rate to the largest
Page 460
Winkle, J. Chem. Eng.
presented at Am. Chem. SOC.Regional
Meeting at St. Louis, Missouri (Dec.
1-3, 1960).
5. Dreisbach, R. R., Am. Chem. SOC. Adv.
in Chem. Series, No. 15, (June, 1955);
No. 22 (March, 1959).
6. Drickamer, H. G., G. G. Brown, and
R. R. White, Trans. Am. Inst. Chem.
Engrs., 41, 555 (1945).
7. Finch, R. N., M.S. thesis, Univ. Texas,
Austin, Texas ( 1961) .
8. Glasstone, Samuel, “A Textbook of
Physical Chemistry,” 2 ed., Van Nos-
trand, New York (1946).
9. Hildebrand, J. H., and R. L. Scott,
“The Solubility of Non-electrolytes,” 3
ed., Reinhold, New York ( 1955).
10. Kobe, K. A., and R. E. Lynn, Chem.
Reu., 52, 121 (1953).
11. Lyderson, A. L., R. A. Greenkorn, and
0. A. Hougen, Rept. No. 4, Engineer-
ing Experiment Station, Univ. Wiscon-
sin, Madison, Wisconsin ( 1955).
12. Moelwyn-Hughes, E. A., “Physical
Chemistry,” Pergamon Press, New
York (1957).
13. Murti, P. S., and Mathew Van Winkle,
A.1.Ch.E.J o u T ~3,~517, (1957).
13a. Perry, J. H., “Chemical Engineer’s
Handbook,” 3 ed., McGraw-Hill, New
York (1950).
14. Prausnitz, J. M., and Ralph Anderson,
A.1.Ch.E.journal, 1, 96 (1961).
15. Q ~ Ahmad,
~ ~
and ~ ~ Van
Matthew , ber 11, 1961.
E. 1. du Pont de Nemourr and Company, Wilmington, Delaware
spray particles. The production rate,
which also is calculated from heat and
material balances, is limited by the re-
quirement that the largest particles be
dried during their flight from the atom-
izer to the chamber wall. The time of
flight of the largest particles is deter-
mined from a consideration of the fluid
dynamics of the process. Thus, from a
theoretical analysis based on the com-
putation of minimum heat transfer
rates (for this case of gas-phase resist-
ance limiting) at capacity conditions
(as determined by incipient wall depo-
sition), expressions have been devel-
oped which relate overall performance
to the dryer geometry, drying temper-
ature, atomizer type, and particle size
of the material produced.
A.1.Ch.E. Journal
(1960).
16. Quiggle, Dorothy, and M. R. Fenske,
J . Am. Chem. SOC., 59, 1829 (1937).
17. Reid, R. C., and T. K. Sherwood, “The
Properties of Gases and Liquids,” Mc-
Graw-Hill, New York (1958).
18. Riedel, L., Chem. Eng. Tech., 26, 83
( 1954).
19. Robinson, C. S., and E. R. Gilliland,
“Elements of Fractional Distillation,”
4 ed., pp. 285-289, McGraw-Hill, New
York ( 1950).
20. Steinhauser, H. H., and R. R. White,
lnd. Eng. Chem., 41,2912 (1949).
2 1. Union Carbide Chemical Company,
“Physical Properties of Synthetic Or-
ganic Compounds” ( 1958).
22. van Arkel, A. E., Trans. Faraday Soc.,
42B, 81 (1946).
23. , and S . E. Vles, Rec. trav.
Chim., 55,407 ( 1936).
24. van Laar, J. J., 2. physik. Chem., 72,
723 (1910); 83, 599 (1913).
25. , and R. Z. Lorentz, anorg. all-
gem. Chem., 146,42 (1925).
26. Wesson, L. G., “Tables of Electric Di-
pole Moments,” The Technology Press,
Mass. Inst. Technol., Cambridge, Mas-
sachusetts ( 1948).
27. Yang, C. P., and Matthew Van Winkle,
Ind. Eng. Chem., 47, 293 (1935).
Manuscripts received June 5, 1961; revision re-
ceiued December 8, 1961; paper accepted Decem-
F. A. GLUCKERT
With these expressions agreement
between predicted and measured heat
transfer rates has been obtained for
drying of sodium sulfate solutions in
1-, 2-, and 3-ft. diam. dryers and cal-
cium carbonate slurries in the 2-ft.
diam. dryer. The obtainable useful
rates however are 52% above the mini-
mum rates predicted by these expres-
sions. This agrees with the observation
that drops of sodium sulfate solution
and calcium carbonate slurries do not
require complete drying if they are to
be free flowing and are not to stick to
the dryer walls. Materials which exhibit
less stickiness than sodium sulfate when
partially dried have been processed at
greater rates than sodium sulfate,
whereas materials which exhibit greater
stickiness have not.
September, 1962
 The developed relationships will be by Friedman, Gluckert, and Marshall 8. The performance of a spray dryer
useful for scale up, estimating the ef- ( 5 ) are employed to determine the is determined by the heat transfer rate
fect of small changes in operating con- maximum particle size produced by q that can be achieved between the
ditions, or preliminary design for cost centrifugal disk atomizers. drying gas and the spray without depo-
estimates. Extension of the techniques 4. A droplet Nusselt number equal sition or buildup on the dryer wall.
presented here should allow more ef- to 2, corresponding to pure conduction The inherent stickiness of partially
fective designs of other spray processes, to infinity, is used for evaluating the dried material will determine to a large
including spray combustion for fur- coefficient of heat transfer. Kesler ( 6 ) extent the dryer performance, because
nace, jet engine, or rocket design. has studied experimentally the evapo- many materials vary in stickiness with
ration of drops and found that the degree of dryness. The performance
THEORY evaporation proceeded approximately of the dryer is as if a characteristic
The theoretical development has as predicted, based on the assumption drop were being dried. If the material
been based on the following premises: that a Nusselt number of 2 is applica- is not sticky with a given residual
ble. This is further substantiated by moisture content, complete drying is
1. The rate of heat transfer to the the analysis of So0 ( l l ) ,who as noted not required and flight time to com-
entire spray equals the weight rate of before, has shown that the relative ve- pletely dry the largest drop is not
flow of the spray times the rate of heat locity between the particles and the needed. Instead time must be provided
transfer per unit weight of the slowest gas stream is negligible despite the to dry the largest drop to a nonsticky
drying portion of the spray multiplied turbulent eddies, and by the experi- condition or to completely dry a some-
by its time of flight.* The largest drop- mental work of Dlouhy and Gauviri what smaller or characteristic drop.
lets, which dry most slowly are the ( 3 ) and Manning and Gauvin ( 7 ) . These premises have been employed
limiting portion of the spray and are 5. Drying conditions are uniform to calculate the heat transfer y under
used therefore for this evaluation. throughout the drying chamber. Tlie capacity conditions in a spray dryer for
2. The time of flight of the largest spray jet entrains several times its volu- the following three types of atomizers
drops before they impinge against the metric flow rate of the surrounding gas commonly used.
dryer wall is determined by integrating for each jet diameter away from its
the equation for velocity decay of a source; thus tremendous recirculation Two-Fluid Atomizers
free jet. So0 (11) has shown that mo- currents are set up in the dryer. This A two-fluid atomizer produces an air
mentum transfer in a two-phase stream, recirculation is so great that the entire jet which forms and carries the drop-
consisting of small particles carried by chamber is essentially at the chamber lets with it. The trajectory of this spray
a fluid, very rapidly reduces the rela- outlet temperature. Temperature meas- has been computed from data obtained
tive velocity between the solid particles urements made in a drying chamber with jets of similar geometry (1) . The
and the other phase of the stream until indicated that the inlet temperature falloff in center-line velocity V of a jet
the average velocity of the solid parti- persists only a few jet diameters im- having an initial velocity in the neigh-
cles is essentially that of the fluid mediately below the gas inlet. borhood of 1,000 ft./sec., which is
stream. The behavior of jets has been 6. Drying occurs from the drop sur- typical of two-fluid atomizers, is given
amply treated by Albertson et al. (I), face which consists of a liquid satu- by
Alexander et al. ( 2 ) , and Taylor et al. rated with the solutes. The work of
(12). The location of wall impinge- Duffie and Marshall ( 4 ) and Ranz and -V
=6.2(+) (1)
ment in spray dryers with sprays Marshall (10) on the drying of drops V,
directed axially has been observed at 3 indicated that a saturated liquid film
to 4 chamber diameters from the noz- is established very rapidly at the drop Since V = ddd8 (2)
zle. This wall impingement stems from surface and that the rate of evapora-
the expansion of the spray jet. The tion corresponds to that of a saturated
expansion has been reported by the liquid. Thus in view of premise 5
dx
- _ - 6.2 V,
dB
(%) (3)
above investigators to occur at a 15-to- above, the temperature driving force Separating variables one gets
20-deg. angle. The larger spray angles for drying is the difference between the
given in manufacturers' catalogues for drying gas outlet temperature and the x dx
d8 = (4)
pressure-nozzle atomizers have been adiabatic saturation temperature of a 6.2 V,D,
found only to apply close to the atom- saturated solution of the material being
izer. dried. Integrating to obtain the total resi-
3. The largest droplets in a spray 7. The drop diameter is constant dence time in the jet of length xt one
population are assumed to be three during drying. The drying from the obtains
times the surface per unit volume surface mentioned above tends to form Xt2

average size, D,, = Z n,Dt"/S nlD:, as a surface shell which maintains the 0, = (5)
3 j
12.4 V9,,Ve
drop diameter essentially unchanged as
suggested from available data for cen-
drying proceeds. As has been borne The total volume 0 of the cylindrical
trifugal disk atomizers. The empirical
out by experience with many systems, dryer to the end of the spray jet will he
correlation presented by Nukiyama and
Tanasawa (9) is employed to deter- hollow particle beads are produced. If
on the other hand it is assumed that
mine the droplet size produced by two-
the drop diameter decreases to zero,
fluid atomizers. The rule of thumb for
the time required for drying will be
drop size, D< = 5OO/f/Ap, where Ap greater by a factor of 1.5 and the cal- The full cylindrical volume is useful
is the pressure across a single-fluid culated performance correspondingly because the volume outside the jet is
nozzle in pounds per square inches, is reduced. * actually required to recirculate air. The
employed to determine the average _ _ atomized material is projected down-
drop size from single-fluid nozzle atom- * Integration of the equation for the rate of
12krAtrB ward in the chamber as a jet expand-
izers, and the relationships developed transfer to a drop gives Q' = ___ if the
heat
drop diameter is constant. If the
p=
rop diameter
ing at an angle of approximately 20
* All of the material sprayed is dried. Because decreases with evaporahon, deg. This jet entrains the surrounding
of differences in particle size some portlons of the
spray are dried more rapidly than others, hut each 8krAttB gas in order to expand, and it expands
Q' =
portion requires the same quantity of heat per
pound for drying. p1 ( D ) % n i t i a i - (D)er~na~l to the full chamber diameter at ap-

Vol. 8, No. 4 A.1.Ch.E. Journal Page 461

proximately 3 chamber diameters from The rate of heat transfer to the largest
its source. The source may be con- drop is
sidered a point because the nozzle
diameter is small compared with -dQ
_ - h A Att = 2 ~ k D,,,
, At,
the chamber diameter. The loca- d8
tion of wall deposition which oc- (21)
curs from operation of the dryer at Substituting the value of D, from where A = m(Dmar)*,h = 2kf/D,.,,
greater than design capacity is 3 to 4 Equation (15) in Equation (16) one and Atr is the difference between the
chamber diameters from the atomizer. gets drying gas and drop surface tempera-
The center line of a spray dryer of ture.
high length-to-diameter ratio (length/
diam. = approximately 8) has been
tjlI = (
T>'I8
v2/a

The rate of heat transfer per unit

1 initial mass of largest drop is
probed while the dryer operated at (24.8) V,w, -
capacity conditions. It was found that TW,,V,P,, dQ' 1 2 kf~ D,,, Att - 12 k, At,
material moving directly down the (171 - = --
dB ps T (D,,,,x)'i p,(DmJ
chamber of this dryer operating at When one rearranges and collects terms
capacity conditions is dry after moving (22)
4 or 5 chamber diameters from the since the weight of the largest drop is
atomizer. Thus a desirable minimum
number of chamber lengths is 4 -C vz
8, = 0.532 v2/" -f& (18) p. TT (D,,,,,) "6. In the event that a solid
filmlike structure is formed upon dry-
diam. This uncertainty has introduced Because the liquid contributes little to ing no appreciable change in drop di-
an uncertainty of 2 15% in the coef- the volume of the spray-gas mixture ameter will be experienced during dry-
ficients of the correlating equations: ing, and Equation (22) may be inte-
grated to
D,
-=-
1
(7)
XI 4 Equation (18) can be written
which gives 8,= 0.532 v2/' The composition of the largest drop is
assumed to be representative of the
entire spray, and consequently the
Substituting for D,' [Equation (8)1 quantity of heat transferred per initial
in Equation (6) one gets
TXtJ
v=- (9)
64
from which

Substituting this value for x I y into

Equation (5) one obtains

The total mass flow rate through the

atomizer will be

w,, + w, = V,nD,'p.,,
4
(12)

The liquid being sprayed is accelerated >IS

by transfer of momentum from the
atomizing gas and momentum is con- >-
served. The initial momentum of the k
liquid is negligible: 0 1
0
-J
+
V,W, = ( w , w,)V, (13) w
>
Replacing V , in Equation (12) by V,, -I
from Equation (13) one gets 5
k 0.1
TD,' z
-
w, + w , = v, ( w . w+, w . ) -4
p,,,
LI
0
(14)
Solving for D , z
-_- 0
F 0.01
2 0.1 I 10 1000 10,000
I00
d X X
(15) LL NOZZLE DIAMETERS FROM SOURCE,
Do
OR -
Dl
-
and replacing V,,&in Equation (11) by
V , from Equation (13) one obtains Fig. 1. Axial velocity of induced air from pressure nozzles.

Page 462 A.1.Ch.E. Journal September, 1962

 s - 3 -0 "*L.

0 saucer 9' 1800

0 "one 8' ,800
a Vane 5' 7890
e Yaw 8" moo
a "am 5' iew
0 "ane 3.5" 7800
A "am 5' 3500
B Yam 5" 3500
V Vane ,.5" 1500

0.01

0.00I
0.I I 10 I00 I000 I0,OOO

FUNCTION OF SLOT-WIDTHS FROM SOURCE, 7. OR -

bl
R
R-r . -

Fig. 2. Velocity of induced air from atomizing disks.

weight of spray also equals Q'. The transfer rates somewhat higher than
rate of heat transfer per unit time for this calculated value can be achieved
the entire spray therefore is given b y because it is not necessary to dry the
largest drop completely but only to
dry it until the drop is not sticky, that
4 =dQ
d e =Q'w. (24)
is dry the characteristic drop.
or combining Equations (23) and Pressure-Nozzle Atomizer
(24) one gets For pressure-nozzle-atomizer-spray
12 k, Att w. ,,O., drying the trajectory of the spray has
(25) been computed from measurements of
= p. (Dl,,nx)2 air caused to flow by the momentum
transfer to the air from the spray. The
Let Or equal Omax and substitute the
value of 8, [Equation ( 2 0 ) l in Equa- velocity of the air entrained by the
spray, which by extension of the work
tion (21):
by Kesler ( 6 ) and So0 (11) has been
6.38 k, v2/' A t t taken as a good approximation of the
4= -
W,
spray velocity, was measured in sprays
(Dm..)2 P"
from pressure nozzles and centrifugal
disks along the axis of the spray jet.
The measurements were made with a
pitot constructed from a piece of Y4-in.
The heat transfer rate y given by copper tubing. A trap was employed
Equation (26) is the maximum rate between the pilot and registering ma-
which can be achieved if the largest nometer to prevent spray caught by the
particles in the spray from a two-fluid pitot from reaching the manometer and
atomizer are dried as they reach the interfering with the pressure measure-
dryer wall or bottom. Actually heat ments. From the manometer readings
of impact head the velocity of the gas
moving with the spray was computed
in the usual fashion. The data ob-
tained for the axial decay of the air
velocity in a pressure-nozzle spray jet
is shown in Figure 1. The behavior is
similar to that which has been shown
for free air jets. The constants how-
ever have required some revision
-= 3 4 3 (la)
v,
From Equation ( l a ) the time required
for the spray to travel from the pres-
sure-nozzle atomizer to a position 4
chamber diameters from the atomizer
has been calculated in a similar man-
ner to that Fhown for two-fluid atom-
izers:
Do is the initial diameter of a jet of
exit composition gas having the same
velocity and momentum as the liquid
jet which originates it. Since both the
momentum and velocity of the initial

Vol. 8, No. 4 A.1.Ch.E. Journal Page 463

liquid jet and this fictive initial gas jet
are equal, the mass rates of flow are
equal and
v, = W,
=v= W.
7r 7r
-
4
p 8( D s ) -p t (Do)’
4 sum of the radial and tangential veloc-
(3a)
Solving Equation ( 3 a ) for D oone gets
I/’
D, = D, (e)
PI
(4a )
When one substitutes the values of V ,
and D , [Equations ( 3 a ) and ( 4 a ) l in
Equation ( 2 a ) and applies the result
to Equation ( 2 5 ) , the rate of heat
transfer per unit time for the entire
spray is as was shown for two-fluid
atomizers :
The heat transfer rate q given by
Equation ( 5 a ) is the maximum rate
which can be achieved if the largest
particles in the spray from a pressure-
iiozzle atomizer are dried as they reach
the dryer wall or bottom. The same
qualifications on rate as discussed un-
der two-fluid atomizers apply in this
case.
Centrifugal-Disk Atomizers
The theory for operation of a cen-
trifugal-disk-atomizer-spray dryer has
been developed in a similar fashion to
that for two-fluid and pressure-nozzle-
atomizer-spray dryers, except that the
trajectory of the spray is outward from
the disk. Measurements of air caused
to flow by the spray from a disk, made
in the manner described for pressure-
nozzle atomizers, are shown in Figure
2. I t will be noted that for a disk sys-
tem the fraction of initial velocity of
induced air is lower than it is for a jet
alone from a slot orifice ( a peripheral
slot in a right cylinder) and that the
difference is greater close to the disk.
A portion of this difference is due to
the interrupted pattern of the spray
produced by vaned disks. (Very few
data are available on saucer-shaped
disks.) Because the difference tends to
become less important at greater dis-
tances from the disk, more weight has
been given to the impact head meas-
urements made at greater distances
from the disk than close to the disk in
specifying the falloff in velocity of the
sprays from a centrifugal disk. As
shown in Figure 2 the observed tra-
jectories of centrifugal-disk-atomizer-
spray jets are represented by
Page 464
where b, the width of the annular jet
from a centrifugal disk of exit com-
position gas having the same velocity
and momentum as the liquid jet, is
evaluated as follows:
The liquid jet velocity is the vector
ities
V , = dz 27rr N
This equals the velocity of a fictive jet
(width b ) of exit composition having
the same velocity and momentum
(thus the same initial mass flow rate) :
- W*
V,=\/2 2m N=V=-
pi 2 ~ br
Solving for b one gets
W,
b=
~ ‘ 2( - 2 ~ Np,
) tal phase of the study was based princi-
Since V = dR/dB ( 5 b ) , one obtains
from Equation ( l b )
dR (br)lp
-- =
1.2 v,
d8 [ ( R - r ) R]Ifl
Separating variables one obtains
dB
1.2 V, (br)lPn
= [(R-r)(R)]”2dR
Integrating between limits of e = 0 to
8, and R = r to 1% = R, one gets
1.2 V,(br)”’ Or
- (2R,-r)[(R,-r)(R,)]’1’
- of heat transfer to sprays of calcium car-
4 bonate slurries was measured in the 2-ft.
For cases of interest R, > 5r. Thus the
approximation
( R , - r ) + R,
d ( R c- r ) ( R , ) = - 2
will introduce an error of less than
1% % . Similarly the logarithmic term
may be neglected without introducing
more than a 5% error.
Making these approximations in
Equation ( S b ) , collecting the terms,
and solving for et one gets
When one substitutes Equation ( 4 b )
for b and Equation ( 2 b ) for V , in
Equation ( 9 b ) and applies the result
to Equation ( 2 5 ) , the rate of heat
transfer per unit time for the entire
A.1.Ch.E. Journal
spray is as was shown for two-fluid
atomizers:
4 =
4.19 k, ( R , - G)‘ Att w. p t
-
rN
(Dnlax)2
p.
(lob)
The heat transfer rate q given by
Equation (12b) is the maximum rate
(2b)
which can be achieved if the largest
particles in the spray from a centrifu-
gal-disk atomizer are dried as they
reach the dryer wall. The same quali-
fications on rate as discussed under
two-fluid atomizers apply in this case.
EXPERIMENTAL PROCEDURE*
A series of tests in spray dryers of dif-
ferent sizes and with different atomizers
was made to give a definitive test to the
(4b) theories developed above. The experimen-
pally 011 a series of runs drying sodium
sulfate solutions to allow a cross compari-
son of the tests. Sodium sulfate solutions
were selected as a test material because
( 6 b1
they ai‘e relatively difficult to dry and
were easy to clean from the drying cham-
ber when the capacity was exceeded.
Calcium carbonate slurries proved impos-
sible to feed to the 1-ft. diam. dryer
through the mall lines with which it was
(7b) equipped.
The rate of heat transfer to sprays of
sodium sulfate solution was measured in
spray dryers 1, 2, and 3 ft. in diameter,
employing two-fluid atomizers; 2 and 3
ft. in diameter, employing pressure-nozzle
atomizers; and 5 ft. in diameter, employ-
ing a centrifugal-disk atomizer. The rate
diam. dryer with a two-fluid atomizer.
The I-ft. diam. two-fluid atomizer spray
dryer consisted of a vertical, cylindrical
chamber 1 ft. in diameter and 5 ft. high.
Feed slurry or solution was introduced in
in a downward direction centrally at the
top of the chamber. Atomizing gas (whch
supplies part of the heat for drying) was
discharged through an orifice concentri-
cally around the feed point. Additional
heat was supplied by lower-pressure gas
introduced concentrically around the atom-
izing gas and feed. The moisture carrying
drying gas and the dryed product were
taken from the coned bottom of the dryer
and separated in a bag filter.
The 2-ft. diam. two-fluid atomizer spray
dryer consisted of a vertical, cylindrical
chamber 2 ft. in diameter and 10 ft. high.
Feed slurry or solution was introduced in
a downward direction centrally at the top
of the chamber. Atomizing gas (which
supplied part of the heat for drying) was
discharged through an orifice concentri-
cally around the feed point. Additional
heat was supplied by lower-pressure gas
introduced concentrically around the atom-
izing gas and feed, The moisture carrying
drying gas and the dryed product were
__
Summaries of the experimental data have been
Q
deposited as document 7158 with the American
Documentation Institute, Photoduplication Service,
Library of Congress, Washington 25, D. C., and
may be obtained for $1.25 for photoprints or for
35-mm. microfilm.
September, 1962
 Atomizer Conditions including a 60 deg. cone bottom. The ceil-
0 1-ft. d i m . 2-fluid ing of this chamber was essentially flat.
V 1 - f t . diam. 2-fluid The feed was introduced from a cen-
0 1-ft. d l m . 2-rluld Cold stomizlng D i p
trifugal-disk atomizer located centrally.
0 1-et. d i m . 2-fluid L o w atomizing a i r ppe@sure
The drying gas was introduced concentri-
A 2-fluld Medium LtomiEing a i r pre8SuPB
1 - f t . Dlam.
cally to the disk through vanes imparting
2-rt. d i m . Z-fluld
0 a swirl to the gas and optionally through
A 2-rt. dim. Prcasure-nozzlt
four inlets at the periphery of the chamber.
m 2-ft. dim. 2-fluld
The performance, as defined under
Q 2-ft. dim. 2 - flu id
theory, of these several experimental spray
e 3-ft. d i m . 2-fluid
dryers was determined from the maximum
A 3-ft. d i m . Pressurs-nozrl.
Prea8ure-nozrle Lo" preas"m
evaporation or heat transfer rate q which
Q 3-ft. dim.
could be achieved without deposition of
R 5-ft. d l m . BOwen CentrlNgal-disk NO s i d e air
partially dried material in the drying
@ 5-et. d i m . wwen Centrlfwal-diek
chamber. The specific value of dryer per-
formance was determined as follows. The
dryer was operated at selected tempera-
tures with a given low value of the drying
gas flow rate, and the evaporation rate or
heat transfer rate q measured. Then the
drying gas flow rate was increased, and
again the evaporation rate or heat transfer
rate q was measured. The increase in feed
rate required to maintain the same tempera-
tures resulted in a larger maximum drop
size from two-fluid and centrifugal-disk
atomizers. To obtain the necessary increase
in feed rate with a pressure-nozzle atomizer
a larger nozzle was required which, if
operated at a reduced pressure, produced
a larger maximum drop size. When the
proper conditions were selected for the test
the larger drops produced upon increasing
flow rates did not dry. Alternatively the
drying-chamber temperature was reduced
in steps until there was insufficient Ate to
I00 1000 10,000 100,000 1,000,000 dry the larger drops. Inspection of the dry-
ing chamber between runs identified the
PREDICTED HEAT TRANSFER RATE, B.T.U./HR. heat transfer rate for which negligible wall
deposition was experienced (under capac-
TWO-FLUID ATOM1ZER: ity) and another greater rate which pro-
duced wall deposition (over capacity). In
6.38 k f v 2 I 3 A t , w, many instances it was difficult to determine
4= whether or not the dryer was operating at
(Dmax.J2
less than capacity conditions. The runs for
which this condition existed have been
PRESSURE-NOZZLE ATOMIZER: designated borderline.
The measured rates of heat transfer for
10.98 k f v " ~A t , the above described runs with several dry-
4=
(Dmax.)2 ers and atomizers at essentially capacity
conditions have been compared in Figure
CENTRIFUGAL-DISK ATOMIZER: 3 with the heat transfer rates predicted by

4 =
( D m s X .)
dq
4.19 k f ( R , - r/2)' A t ,
P.
the theory presented. The points show a
deviation of from +41% to -25% from
their average which is 52% above the theo-
retical line for minimum heat transfer rate
Fig. 3. Comparison of measured and predicted transfer rate in spray
at capacity conditions. It should be noted
dryers.
that the points on Figure 3 cannot be ex-
pected to fall exactly on a line. They might
fall more or less lower (under capacity) or
higher (over capacity) than a line repre-
taken from the coned bottom of the dryer The 3-ft. diam. pressure-nozzle-atom- senting the optimum heat transfer rate set
and separated in a bag filter. izer-spray dryer which employed the same by the precise rate at which the largest
The 3-ft. diam. two-fluid atomizer spray chamber as the two-fluid atomizer dryer drops of the material being sprayed are just
dryer consisted of a vertical, cylindrical was 27 ft. high. However the spray noz- dried to a nonsticky condition or the rate at
chamber 3 ft. in diameter and 27 ft. high. zle was mounted on the end of an ex- which the characteristic drop is dried. The
Feed slurry or solution was introduced in tended feed pipe 15 ft. from the bottom predicted heat transfer rate is a minimum
a downward direction centrally at the top of the dryer. rate; in the case of sodium sulfate solution
of the chamber. Atomizing gas (which The 2-ft. diam. pressure-nozzle-atom- drying it is 52% low. However the diame-
supplies part of the heat for drying) was izer-spray dryer employed the same cham- ter of the spray dryer chamber necessary
discharged through an orifice concentri- ber as the two-fluid atomizer dryer. The for a given drying rate with material such
cally around the feed point. Additional pressure nozzle replaced the central feed as sodium sulfate could be specified from
heat was supplied by lower-pressure gas and atomizing gas pipes and discharged the theoretically predicted minimum rate to
introduced concentrically around the atom- downward concentrically within the dry- within about 20% because the chamber
izing gas and feed. The moisture carry- ing gas supply orifice. diameter varies approximately as the square
ing drying gas and the dryed product was The 5-ft. diam centrifugal-disk-atomizer- root of the heat transfer rate. A limited
taken from the coned bottom of the dryer spray dryer consisted of a 5-ft. diam. number of runs (not reported here) of pro-
and separated in a cyclone and bag filter. vertical, cylindrical chamber 7 ft. high prietary materials such as dyes, pigments,

Vol. 8, No. 4 A.1.Ch.E. Journal Page 465

surface active agents, etc. have indicated
that the theory is generally applicable.
CONCLUSIONS
An overall correlation of spray-dryer
performance has been developed. Sub-
ject to the limitations discussed under
theory, a spray-dryer design can be
specified by substituting the values
corresponding to the required duty of
the dryer and selected values of drop
size D,,, and the temperature driving
force A t , in the appropriate equation
[Equation (26) for two-fluid atomizers,
( 5 a ) for pressure-nozzle atomizers,
and ( l o b ) for centrifugal-disk atom-
izers] and solving for the dryer volume
or dimensions. In an analogous manner
the maximum capacity achievable in
an existing dryer can be determined for
various conditions. It will be noted
that for a specified duty only At, and
D,,, can be freely selected. Tempera-
ture driving force Atr is a function of
drying air temperatures, drying air
rate, and spray rate. For two-fluid-
atomizer-spray dryers D,,,, is a func-
tion of atomizing air quantity w , and
velocity V,; for pressure-nozzle atom-
izer spray dryers DmaX is a function of
nozzle orifice size D,; and for centrifu-
gal-disk-atomizer-spray dryers Dmaxis
a function of disk radius r, speed N ,
and spray rate w,.
The degree to which the largest
drop must be dried to render it non-
sticky and the puffing, shrinking, or
disintegrating of the drop during dry-
ing will cause differences in the drying
rate of various materials. The effect of
these factors is best determined by
experiment. Thus the developed rela-
tionships will be useful principally for
scaleup, for estimating the effect of
changes in operating conditions, and
for first design for cost estimates. The
experiments required can be carried
out in small equipment and scaled up
to large production equipment with
confidence.
ACKNOWLEDGMENT
The results reported here are based on
investigations carried out by co-workers at
the Engineering Research Laboratory of the
Engineering Department of E. I. du Pont
de Nemours and Company over a period of
years. Particular credit should be given to
Eno Bagnoli, M. T. Cichelli, S. J. Friedman,
R. C. Holmes, R. L. Menegus, and E. T.
Ruehl for work in developing the theory
presented and to M. A. DiMatteo for the
actual operating of the dryers.
NOTATION
A = area for heat transfer, sq.ft.
W,
h = , fictive width
dF(ZTT) a N pt
of annular jet from centrifu-
gal-disk atomizer of exit-com-
Page 466
position gas having same ve-
locity and momentum as the
initial liquid jet, ft.
27, = width of annular-slot air ori-
fice, ft.
D = diameter of spray drop, ft.
D, = diameter of air nozzle or ori-
fice, ft.
D , = diameter of dryer chamber
with pressure nozzle or two-
fluid atomizer, ft.
D , = effective diameter of atomizer
orifice, ft.
D , = diameter of drop of class j , ft.
D,,,,, = maximum drop diameter, ft.
(assumed to be three times
the surface per unit volume
average size)
D , = D , ( P . / P ~ ) "diameter
~, of a jet
of exit-composition gas having
same velocity and momentum
as the liquid jet, ft.
D, = diameter of pressure-nozzle
discharge orifice, ft.
D,, = surface per unit volume aver-
age drop size,
2 n5D,3/C. n,D5*,p
5 J
h = heat transfer coefficient, B.t.u./
(hr.) (sq.ft.) ( O F . )
i = class assigned to drop size
range selected for considera-
tion
k, = thermal conductivity of gas
film surrounding drop, B.t.u./
(hr.) (sq.ft.) ( " F J f t . ) (evalu-
ated at average temperature
between dryer gas and drop
temperature )
N = rate of rotation of centrifugal
disk, rev./hr.
NA7, = Nusselt number, h D,,,,/k,,
based on largest drop
n = number of drops
n, = number of drops of class j
Ap = pressure drop across pressure
nozzle, Ib./sq.in.
Q = heat, B.t.u.
Q' = heat transferred per unit
weight of spray, B.t.u./lb.
9 = rate of heat transfer to spray,
B.t.u./hr.
7 = radius of centrifugal-disk atom-
izer, ft.
R = radial distance from center of
dryer chamber with centrifu-
gal-disk atomizer, ft.
R, = radius of dryer chamber with
centrifugal-disk atomizer, ft.
Atr = temperature driving force, O F .
(temperature difference be-
tween dryer gas at exit con-
ditions and the drop surface)
u = volume of dryer chamber,
cu.ft.
V = axial velocity of air in a jet,
ft./hr.
V, = velocity of atomizing air at
atomizer, ft./hr.
V, = velocity of liquid leaving
atomizer, ft./hr.
A.1.Ch.E. Journal
V, = velocity of the spray jet at the
two-fluid atomizer, ft./hr.
V, = initial velocity of air from ori-
fice or nozzle, ft./hr.
W, = weight rate of flow of atomiz-
ine" air.' lb./hr.
w. = weight rate of flow of liquid,
Ib./hr.
x = distance from atomizer, ft.
xr = distance from atomizer to po-
sition four chamber diameters
away, ft.
Greek Letters
8 = time, hr.
Om,, = time required for maximum
size drop to dry, hr.
0, = time required for spray to
travel from a two-fluid or
pressure-nozzle atomizer to a
position 4 chamber diameters
away from the atomizer or
from a centrifugal-disk atom-
izer to the chamber wall, hr.
pa = density of atomizing air, lb./
cu.ft.
p,, = density of the spray jet at the
atomizer, lb./cu. ft.
p, = density of sprayed material,
Ib./cu.ft.
pt = density of gas at exit condi-
tions, lb./cu.ft.
S = operator indicating summation
LITERATURE CITED
1. Albertson, M. L., Y. B. Dai, R. A.
Jensen, and Hunter Rouse, Proc. Am.
SOC. Civil Engrs., 74, 1577-96 (1948).
2. Alexander, L. G., Arnold Kivnick,
E. W. Comings, and E. D. Henze,
A.1.Ch.E. Journal, 1, 55-73 (1955).
3. Dlouhy, Jan, and W. H. Gauvin, ibid.,
6,29-34 (1960).
4. Duffie, J. A,, and W. R. Marshall,
Chem. Eng. Progr., 49, 417-423, 480-
486 (1953).
5. Friedman, S. J., F. A. Gluckert, and
W. R. Marshall, ibid., 48, 181-191
( 1952).
6. Kesler, G. H., Ph.D. thesis, Mass. Inst.
Technol., Cambridge, Massachusetts
( 1952).
7. Manning, W. P., and W. H. Gauvin,
A.1.Ch.E. Journal, 6, 84-190 (1960).
8. Marshall, W. R., Chem. Engr. Progr.
Monograph Ser. No. 2 , 5 0 (1954).
9. Nukiyama, Shiro, and Yasushi Tana-
sawa, SOC. Mech. Engrs. (Japan),4, 5,
and 6, (1938-49), translated by E.
Hope, Defense Research Board, De-
partment of National Defense, Canada.
10. Ranz, W. E., and W. R. Marshall,
Chem. Eng. Progr., 48, 141-173 (1952).
11. Soo, S. L., Chem. Eng. Sci., 5 , 59-67
( 1956).
12. Taylor, J. E., H. L. Grimmett, and
E. W. Comings, Chem. Eng. Progr.,
47, 175-180 (1951).
Manuscipt received May 24, 1961; revision re-
ceived January 2 1962. paper accepted January
4 , 1962. Paper pAented at A.1.Ch.E. Los Angeles
meeting.
September, 1962