You are on page 1of 10

388 INDUSTRIAL AND ENGINEERING CHEMISTRY VOL. 30. NO.

denses on the surfaces and causes costly rejections. The obtain comfort a t the lowest possible cost. (c) Those who
lowering of the dew point to a t least 20" below the prevailing have been concerned with adsorbent drying have been en-
dry-bulb temperature, by means of an activated alumina gaged in the standardization of equipment to be used in exist-
dehumidifier, completely eliminated this difficulty. I n an- ing and less competitive lines. It is believed that adsorbent
other plant 5 per cent relative humidity a t 76" F. is being dehumidification, in conjunction with refrigeration or other
maintained for the purpose of testing electrical appara- methods of cooling, has a place in comfort conditioning.
tus under specific conditions. Another activated alumina This place will be fixed by the prevailing costs of power, gas,
installation is used in Conjunction with the drying of cellu- cooling water, and by any future progress which may be made
loid products where the dry air prevents blemishes which in adsorption equipment specially adapted to this application.
form on the product when the moisture concentration becomes Adsorbents have been known for a long time but their com-
too great. These several applications are merely indicative of mercial use is new. The success which has been attained so far
the variety of operations which may be improved by controlled is the basis for predicting a promising future for the greater
humidity. use of complete and partially dried gases.
Comfort Air Conditioning Literature Cited
(1) Bower, Bur. Standards J . Research, 12, 241 (1934).
Much less progress has been made in the comfort air con- (2) Dover and Marden, J . Am. Chem. SOC., 39,1609 (1917).
ditioning of houses and public buildings than in the industrial (3) Johnson, Ibid., 34, 911 (1912).
field. The major reasons are as follows: (a) I n addition to (4) Munro and Johnson, IND.ENQ.CHEM.,17,88 (1925).
dehumidification, some sensible heat must be eliminated. (5) Yoe, Chem. News, 130,340 (1926).
(b) Each large installation must be separately engineered to RXICEIVED
February 7, 1938.

Drying Materials in Trays


Evaporation of Surface Moisture

T HE removal of moisture from a material by passage of


heated air over the surface of the wet material repre-
sents the most common form of industrial drying today.
The mechanism involved in this form of drying has been
C. B. SHEPHERD, C. HADLOCK,
AND R. C. BREWER
E. I. du Pont de Nemours & Company,
Wilmington, Del.
studied experimentally and theoretically by a number of in-
vestigators during the past ten years, but very little advance
has been made in developing the results obtained to the point
of practical application. This is probably due largely to the
difficulty of developing theoretical formulas which will con- The tray drying of surface moisture
sistently cover the entire drying period of different materials.
I n general, the drying of most solids may be divided into from nonhygroscopic materials and the
three distinct periods: a heating period, a constant rate of effects of various drying conditions o n the
evaporation period, and a falling rate of evaporation period. constant drying rate have been studied.
The heating period is usually short in comparison with the Two samples of Ottawa sand (20-30 and
total drying time and hence is of minor practical importance. 50-70 mesh) were employed. The drying
The constant rate period varies in industrial drying proc-
esses from a negligible fraction to a major portion of the dry- variables studied with the ranges covered
ing time. This is the period most susceptible to the applica- were : air temperature, 115-300 O F.; rela-
tion of theoretical considerations, and the influence of most tive humidity of air, 10-60 per cent; air
of the variables affecting the mechanism of drying in this velocity, 150-1375 feet per minute; ma-
period has been fairly well established (9). The primary pur- terial depth, 0.5-2 inches; and insulation
pose of the present paper is to contribute new experimental
data to this field and a t the same time to present a rational of tray, none and 1-inch cork. The
method for the practical application of these data. evaporation of water in trays under simi-
The falling rate period also may vary from a small propor- lar air conditions was also studied for
tion of the drying time to practically the entire drying period. comparison.
Owing to the fact that the rate of drying in this period is con- I n the runs with sand, nearly all the
tinuously decreasing, the falling rate period may constitute
the greater portion of the total drying time even when a large water was removed during the constant
proportion of the total moisture is removed during the con- rate period. The results obtained chiefly
stant rate period. A considerable amount of experimental concern this period. The constant dry-
work has been performed in the study of the falling rate ing rate was found nearly identical for
period, but the results have been so diverse that no great prog-
ress has been made in the development of a general theory.
APRIL, 1938 INDUSTRIAL AND ENGINEERING CHEMISTRY 389

1. EXPERIMENTAL
FIGURE DRYER

Further experimental data are indicated t o be necessary before


a complete picture of the drying mechanism in this period TABLE CHARACTERISTICS
I. PHYSICAL OF OTTAWA
SAND
can be set forth. 20-30 Mesh 50-70 Mesh
Cumulative wt. % retained on:
20 mesh 0.4 ...
Equipment and Procedure 28 mesh
35 mesh
99.6
100.0
...
0.06
48 mesh ... 1.6
The present study has been confined t o the drying of non- 6 5 mesh 99.5
100 mesh 100.0
hygroscopic materials in trays. The materials consisted of Specific gravity 2.66 2.66
Ottawa sand of two different sizes, 20-30 mesh and 50-70 Bulk density, lb./cu. ft. 100 92
Voids, % 40 44
mesh. The individual particles of this sand appeared spheri- Av. particle diam., in. 0.0314 0.0118
cal. The physical characteristics of the two sizes are shown
in Table I. T h e evaporation of water i n trays was also
measured for comparison. of the dryer. The tests were conducted in the horizontal section
of the dryer with air assing over the trays parallel to the tray
The drying tests were performed in a wind-tunnel type of dryer surface. The inside Slimensions of this section were 24 inches
designed to give close control of air temperature, humidity, and high and 18 inches wide. The horizontal length of the section
velocity, and a t the same time to permit these variables to be upstream from the tray was 10 feet.
changed over a wide range. Figure 1 gives the essential features Two alternative methods were used for determining the ~ O S Sin
weight of the wet material during the drying: (a) The tray was
supported from a calibrated spring which was so connected to the
recording mechanism of an ordinary pressure recorder that a
continuous plot of the weight was obtained; and (b) the tray was
supported on a balance pan, and readings of the weight were
sand and water under the same drying taken a t frequent intervals. Both of these methods gave reliable
results.
conditions. This rate has been expressed The temperature of the heated air used for drying was meas-
in terms of either the heat transfer co- ured by two copper-constantan thermocouples located about %
inch upstream and downstream with respect to the tray, and
efficient or the mass transfer coefficient about 1 inch above the level of the tray surface. The dry-bulb
between the wetted material surface and temperature ahead of the tray was also checked by a mercury
thermometer. The wet-bulb temperature was measured at the
air. The results indicate that these co- same point by a mercury thermometer whose bulb was covered
efficients vary with approximately the 0.8 with several folds of light gauze. The gauze was wetted with dis-
tilled water a t approximately the wet-bulb temperature each time
power of the air velocity over the range a reading was to be made.
covered. However, the heat transfer co- Temperatures of the material during drying were read by nine
copper-constantan thermocouples located in three groups of
efficient is preferable from the standpoint three couples each. One of these groups was placed 1 inch from
of reliability and convenience of use. the upstream edge of the tray, one in the center of the tray, and
one 1 inch from the downstream edge. In each group the three
The coefficients for perfectly insulated thermocouples were placed one over the other, one located inch
trays, based on experimental data, may below the top surface, one l / ~inch above the bottom surface, and
one in the center of the material bed. These groups were located
be employed for the prediction of the con- in a line equidistant from the sides of the tray and parallel to the
stant drying rate of any similar material direction of air flow.
The procedure followed in the experimental tests was essen-
under any given drying conditions in an tially as follows: A tared tray, 12 inches square (inside measure-
uninsulated tray. If the critical mois- ments), was filled level with weighed amounts of sand and water
which had been intimately mixed; sufficient water had been
ture content of the material is low, the added to the dry sand to fill the voids completely but not so
rate so found will enable an approximate much that any large pools of water were in evidence. The
calculation of the total drying time.
thermocouples measuring the material temperatures were next
inserted in the wet sand. The tray containing the sand was then
weighed and placed in the dryer. In all tests the dryer was
.
operated at the desired conditions of temperature, humidity, and
390 INDUSTRIAL AND ENGINEERING CHEMISTRY VOL. 30, NO. 4
velocity for at least 1 hour previous t o
the start of a drying test. The loss
in moisture was determined either con-
tinuously by means of the automatic
recorder or at frequent time intervals by
means of the balance until no further
change in weight was observed. (In
all cases the final moisture content was
less than 0.05 per cent, dry basis, and
usually less than 0.01 per cent.) The
tray was then removed and weighed,
and a representative sample of the dried
m a t e r i a l was placed in an oven at
220" F. t o d e t e r m i n e the bone-dry
weight.
Experimental Results
Tests were run on sand to deter-
mine the effect on drying rate of the
following variables: air velocity (125-
1370 feet per minute), temperature
(114-303' F.), and relative humidity
(10-60 per cent); m a t e r i a l depth
( l / ~ 2inches) and particle size (20-30
and 50-70 mesh); and tray insula-
tion. The standard drying condi-
tions w e r e c h o s e n as 150" F. dry
bulb, 30 per cent relative humidity,
and 300 feet per minute velocity,
since these values represent an ap-
proximate average of the conditions
commonly employed in i n d u s t r i a 1
practice. The experimental r e s u 1 t s
obtained are summarized in Table
IIA. The d a t a a r e a l s o s h o w n
graphically in the form of plots of
drying rate us. moisture content of
the material to illustrate the effects of
each of the above variables (Figure 2 ) .
Several tests were a l s o m a d e on
the rate of evaporation of water in
trays in order to compare these rates
with the drying rates obtained with
sand. The results are summarized
in Table IIB and shown graphically on
Figure 3.
I n order to analyze and correlate
the test results reported, the individual
data on each test were corrected and
averaged as follows: The dry-bulb
measurements of the air as recorded
by the upstream thermocouple were
corrected for r a d i a t i o n from the
z Z Z Z Z% ?2?4$%$%262zz%z% dryer walls by the method of Mc-
x 000
000
0.0 o c o o o o o o ~ ~ o c , ~ o
00 00o0ccoococcoo Adams (6). The wet-bulb tempera-
tures as recorded by a wick ther-
mometer were similarly corrected,
although in this case it was necessary
to set up a balance between radia-
24 A tion, convection, and evaporation.
The wet-bulb correction was negligible
in most of the t e s t s b e c a u s e the
temperatures of the dryer walls were
very close to the measured wet-bulb
temperature. The dry- and wet-bulb
temperatures immediately preceding
the drying tray were used as a basis
for all the test data rather t8hanaver-
age values .of the wet- and dry-bulb
temperatures across the tray, since
the former are the values generally
known in practice.
APRIL, 1938 INDUSTRIAL AND ENGINEERING CHEMISTRY 391

PERCENT MOISTURE CONTENT -DRY BASIS

FIGURE
2. DRYING
OF SAND
Experimental conditions were as follows, with variations as noted for each
set of curves: air temperature ljOo F 9relative humidity, 3 0 , p e r cent:
air velocity, 300 feet/minute: katerial a e p t h , 1 inch; particle size, 20-30
mesh; insulated tray.
A. Effect of tray insulation D. Effect of air velocity
B. Effect of air temperature E. Effect of particle size
C. Effect of relative humidity 8'. Effect of material depth
PERCENT MOISTURE CONTENT - D R Y BASIS

Tray Insulation of moisture removal was found to be directly proportional to


the temperature difference between the drying air and the
I n practically all commercial types of dryers in which the
surface of the material (for the same degree of insulation).
drying takes place from only one surface of the material, an
These results are in accord with theoretical predictions.
appreciable amount of heat is generally introduced through
the remaining unlagged surfaces, and this heat serves to in- The influence of relative humidity a t a fixed dry-bulb tem-
crease the rate of moisture removal. When drying takes perature is shown graphically in Figure 2C. I n the constant
rate period the drying rate was directly proportional to the
place from two surfaces, especially when the distance between
pressure difference between the partial pressure of water
the two surfaces is small, the addition of heat through the un-
wetted surfaces is generally negligible. I n the present in- vapor in the air stream and the vapor pressure of water a t the
material surface, except in tests 6 and 6-2 which were made
vestigation two runs were made with sand in uninsulated
trays, and the remainder in trays insulated with a 1-inch thick- a t a very high humidity. I n these cases accurate values of
ness of cork covered with tin foil. A comparison of drying the pressure difference were difficuIt to obtain because of the
rates in insulated and uninsulated trays is shown in Figure large variations in pressure differences due to very small
2A. The increased rate of evaporation obtained in the un- variations i n surface temperature. However, the drying rate
lagged trays is in agreement with that predicted by heat trans- for different relative humidities varied directly with the tem-
fer calculations, as will be shown later. perature difference between the drying air and the material
A similar increase in rate of evaporation was also noted for surface, even in the high humidity tests. The latter also
water in an uninsulated tray as compared with that in an in- held true in the evaporation of water (tests W-2 and W-5).
sulated tray (tests W-3 and W-2). Actually the effects of relative humidity and dry-bulb
temperature are always linked together, inasmuch as these
two variables combine to determine the temperature difference
Air Temperature and Humidity or the partial pressure difference between the material sur-
The effect of air temperature on the rate of drying of sand face and the drying air. The question of whether to use tem-
is shown in Figure 2B. I n the constant rate period the rate perature difference or partial pressure difference as a basis for
392 INDIJSTIIIAL AliD ENGINEERING CHEMISTRY VOL. 30, NO, 4

showing the effect of air temperature and relative humidity the top of the tray until the level reached tlie bottom of tile
isdiscussed later. tray.

Air Velocity Heating Period


The effect of air velocity on drying rate is sliown in Figure In all the tests reported for sand, tlie heating period was
20. In the constant rate period bhe rate varicil as the 0.76 relatively short (approximately 5 per cent of the total drying
time). This is generally
the case when driing mn-
terial c o n t a i n i n g an ap-
preciable amount of mois-
ture. Far materials con-
t a i n i n g 1 to 2 per cent
moist.ure or less, however,
the heating period may ex-
tend over a large portion of
the Ootal drying time. By
g r a p h i c a l l y determining
the mean t e m p e r a t u r e
difference between air and
the material surface dur-
ing the h e a t i n g period,
heat transfer coefficients
were calculated and found
to be approximately the
same as those developed
below for the c o n s t a n t
rate period. The quantity
of tho moisture removed
during the heating period
was only a small fraction
of the moisture removed
during the constant rate
period. For p 11 r p o s e s of
calculation, it may be as-
mmed that the material is
first heated to the wet-
biilb temDeraturc b e f o r e
power of the air velocity witliin the range of velooities in- any evaporation takes place and that theevaporation then
vestigated (125-1370 feet per minute). The effect of velocity proceeds along the constant rate line.
on the rate of evaporation of water in trays checked the re-
sults obtained with sand (tests W-I, W-2, XT--4.and W-6). Falling Rate Period
This is in accordance with the established results on heat
transfer and mass transfer in the turbulent region. At lonm The scope of this paper does not warrant a discussion of the
velocities it would be expected that natural convection would falling rate period of drying in any detail. In the present
control and that the air velocity would have less effect on the tests this period was relatively short, as shown by representa-
drying rate. tive drying curves of moist.ure content vs. drying time in
Figure 4. Owing to the rapid change in rate, the rate data
Particle Size obtained in this period cannot be considered so reliable as
those obtained in the constant rate period, and further data
The effect of the particle size of the material being dried is are desirable before an accurate analysis of the drying of sand
shown graphically in Figure 2E. Very little difference was in this period is attempted. However, the falling rate curves
observed in the rate of evaporation during the constant rate for sand all followed approximately the ssnie shape, as shown
period between the two sizes of sand and water alone. These in Figure 2.
results do not confirm those reported by Ceaglske and Hougen
(1) which indicated a definite difference in drying rate between Critical Moisture Content
water, coarse sand, and fine sand, the rate decreasing in this
order. In all the tests employing a 1-inch depth of 20-30 mesh sand
in an uninsulated tray, tlie critical moisture'content (i. e., the
Depth of Bed moisture content. at which the constant rate period ends and
the falling rate period commences) occurred within a range of
The influenceof the depth of the material tied 011 thc drying 2 to 4 per cent water. Ceaglske and Horigen reported critical
rate is shown graphically in Fiigurc 2 F . The rate of drying iooistures of the order of 10 per cent and Sherwood and Com-
in the co&ant rato period v;as essentially the same for ings (IO) reported 5.5 per cent for sand within the same
I-, and 2-inch depths of sand, altlioiigli theoretically the in- particle size range. A range of air temperature, humidity,
creased insulating effect of the greater depth should slightly and velocity was covesed so as to result in a variation of 700
reduce the amount of heat transferred from the bottom of the pcr cent in tlie absolute drying rate, and yet no definite effect
tray and hence the over-all drying rate. I n the evaporation on the critical moisture content mas observed. Within the
of water in trays, tlie rate remained practically constant in limit of experimental error this substantiates Ceaglske and
two test.s from the time when the wmter surface was flush with IIougen's rcsults, which showed a variation in absolute drying
APRIL, 1938 INDUSTRIAL AND ENGINEERING CHEMISTRY 393

rate of 50 per cent, and indicates that the movement of mois- rw = latent heat of vaporization of water a t surface
ture from the interior to the surface of the sand bed may well temperature
d W / d s = instantaneous rate of evaporation of water
be due to capillary forces.
The particle size of sand over the range studied in this in-
vestigation apparently does not have a considerable effect The equilibrium temperature (tu) reached under these con-
on the critical moisture content, but the value appears to in- ditions is the wet-bulb temperature. The value of h/kr, de-
crease slightly for the finer sand'(Figure 2E). Ceaglske and pends on the temperature and humidity of the air and, accord-
ing to psychrometric charts, it should vary from 0.29 to 0.24
for the conditions employed with water in air in the present
investigation (when the temperature is expressed in degrees
Fahrenheit and the partial pressure in millimeters of mercury).
If heat is supplied to the wet surface by radiation as well as
convection from the surroundings, h in Equation 1is replaced
by (h, + h,.), and a slightly higher surface temperature is
attained with a resultant increase in the rate of evaporation.
If heat is supplied to the material by convection and conduc-
tion through the unwetted surfaces of the tray, the wetted
surface temperature and the rate of evaporation are correspond-
ingly higher. This effect was noted by Sherwood in his work
on the tray drying of clays (8).
I n order to correlate the present test data, the dry- and wet-
bulb temperatures of the air were corrected as previously de-
scribed and, together with the measured surface temperature,
were used as a basis for computing the partial pressure dif-
EVAPORATION TIME - HOURS ference of water in the air and a t the drying surface and also
FIGURE3. E~APORATIOX
OF WATERFROM FREESURFACE the temperature difference between the air and drying sur-
(Standard experimental conditions with exceptions a s noted)
face. Mass transfer and over-all heat transfer coefficients
were then computed b y the application of Equation 1. These
coefficients are given in Table I1 as pounds of water/(hour)
Hougen's data also indicated a negligible increase in critical (square foot) (mm. mercury) and B. t. u./ (hour) (square foot)
moisture with increase in fineness of the sand. Over the same ( O F.), respectively.

range of particle size, the data of Sherwood and Comings


check this conclusion. However, with particle sizes less than
"26
200 mesh these authors observed a substantial increase in
critical moisture c o n t e n t n a m e l y , 10 per cent for the 200-
325 mesh sand and 21 per cent for sand finer than 325 mesh.
(Subsequent tests by the present authors on three sizes of glass
beads-3-, 6-, and 12.7-mm. diameter-showed a decided in-
crease in critical moisture content with increase in particle
size of the beads; the values obtained were 14, 15, and 21 per g I2
cent, respectively.)
The depth of sand in the tray had a definite increasing ef-
fect on the critical moisture content, as shown in Figure 2F. 5 6
Ceaglske and Hougen also observed this increase with depth 2 4
and explained it satisfactorily on the basis of the capillary E 2
movement of moisture within the material. A method for '0 I 2 3 4 5 6 7 8 9 10 II 12 13 14 15 16 17 18 19 x) 21 22
predicting the critical moisture content of granular materials - HOURS
DRYING TIME
was advanced by Ceaglske and Hougen, but no attempt was FIGURE
4. MOISTURE
CONTENT TIMEAT VARIOUS
vs. DRYING
made to predict the critical moisture content of the sands in AIR VELOCITIES
the present investigation by this method. (Air temperature 150' F . ; relative humidity, 30 per cent. material
depth, 1 ihch; particle size, 20-30 mesh; insulated trky)

Correlation of Results
The mass transfer coefficients represent the mass transfer
I n a drying process where heated air or gas supplies all the a t the wetted surface and need not be corrected for heat in-
heat for the evaporation of water directly to the wetted sur- put through unwetted surfaces, since this had already been
face, a dynamic equilibrium is set up between the rate of heat taken into account by the use of the actual surface tempera-
input to the material and the rate of moisture removal from
tures. The heat transfer coefficient, however, is an over-all
the wet surface. This equilibrium may for practical purposes coefficient which includes the heat transferred through all sur-
be expressed as follows :
faces by both convection and radiation. In order to obtain a
heat transfer coefficient representing heat transfer to the
r,dW/d@ = hA(t, - t,) = kAr,(p, - pa) (1) wetted surface due to convection only, the over-all coefficients
or Pw - Pa - -h were corrected for heat transfer by radiation and conduction
t, - t, k rw through the unwetted surfaces and radiation to the wetted
where h, k = heat transfer and mass transfer Coefficients, surface. The resulting coefficients, h,, are given in Table I1
respectively and represent the coefficients which would be obtained if the
A = area of wetted surface trays were perfectly insulated and the radiation to the wetted
t,, t , = temperatures of air and surface (in this case surface was negligible.
equivalent to wet-bulb), respectively
pa, p , = partial pressures of water in'air and at wetted A brief discussion of the heat transfer considerations in-
surface, respectively volved in the drying of material in trays is desirable a t this
394 INDUSTRIAL AND ENGINEERING CHEMISTRY VOL. 30, NO. 4

point, since in many plant installations the heat input t o the the thermal conductivity of the material appeared to be of the
unwetted surfaces of the trays is often as large as that to the order of 2 B. t. u./(hour) (square foot) ( O F./foot), whereas
wetted surfaces, and this directly affects the rate of drying that of dry sand is only 0.2 and that of water about 0.4. This
from the wetted surfaces. This is shown by the high drying surprisingly high value appears to be real, since it is well out-
rates obtained in tests 1 and 11 using unlagged trays. Such side the range of any experimental error that might occur in
a discussion will serve to indicate the method used for correct- the temperature measurements. One possible explanation is
ing the over-all heat transfer coefficients obtained in the pres- that an evaporation-condensation phenomenon occurs within
ent detailed investigation and a t the same time to point out a the bed of the material which materially assists in the transfer
logical approach to the solution of practical problems. of heat from a lower to a higher stratum. This matter war-
When material is dried in a tray in a stream of air, heat may rants further investigation.
be transferred to the wetted surface in three ways: (a) by With the value of ICl known or estimated, the only unknown
direct convection from the air stream, ( b ) by radiation from factor in the above expressions becomes hl. Both Equations
the dryer walls or the surrounding trays, and ( c ) by heat trans- 3 and 4 may then be expressed as follows :
fer through the material bed (which in turn is comprised of
convection and radiation to the unwetted tray surfaces and (5)
conduction through the trays and the material being dried).
This may be expressed as follows:
hi = ho + + hr ha (2) where C is a constant depending on the thermal conductivities
of both material and tray; then in the general case,
where ht =e over-all heat transfer coefficient for total heat to
wetted surface
h, = heat transfer coefficient for convection to wetted
surface
h, = heat transfer coefficient for radiation to wetted sur-
face [equals actual h', as predicted by radiation
formulas corrected to same basis as h, (multiplied
by ta - t a / t w - L)l
h, = over-all heat transfer coefficient for heat to wetted
surface through unwetted surfaces

As an approximation, it may be assumed that the over-all


heat transfer coefficient t o the wetted surface from the sides
of the trays is the same as that from the bottom of the trays.
(The average convection coefficients should be about the
same, and, although the heat entering the sides has only half
as far to go on an average, it may be considered as only half
as effective owing to interference with heat flow from the
bottom.) The over-all heat transfer through the unwetted
surfaces may then be broken down as follows:
1
ha! = (3)
LI La
hX+I;,a,+k,s,
where h, = heat transfer coefficient by convection and radia-
tion to outer unwetted surfaces
kl = thermal conductivity of material being dried
k* = thermal conductivity of tray and insulation (if any)
L1 = thickness of material bed, feet
Lp = thickness of tray and insulation, feet
A, = ratio of outside unwetted surface t o wetted surface
Ai = ratio of unwetted inner tray surface to wetted sur-
face
Az = ratio of average tray surface to wetted surface
( A a = !2L&h)

For an unlagged tray such as is generally used in plant


practice, A , = A1 = AB. Moreover, the resistance to con-
duction of heat through the tray itself when made of metal is
negligible, and the above expression simplifies to :

(4) /
u - I
B
I n the above expressions for h, the thicknesses (L1 and Lz) I
and the area ratios (Au,A I ,and A*) are known for any given 10 500 1000 ZOO0 3000 5000
case. The thermal conductivity of the tray itself and of any
insulation can be readily obtained from available data. The MASS
VELOCITY
(G)-LBs/(SQ. FT.)(HR3
thermal conductivity of the wet material is less readily de- 5. EVAPORATION
FIGURE OF WATER
termined. On the basis of the temperatures recorded by the
A. Effect of air velocity on mass transfer Coefficients
thermocouples located in the wet sand in the present tests, B. Effect of air velocity on heat transfer coeffioients
Substituting for h, in Equation 2, the following
equation is obtained:

The coefficient h, includes both convection and


radiation to the unwetted tray surfaces. It is
reasonable to assume that the convection coefficient
will be the same to both wetted and unwetted sur-
faces. Moreover, in most industrial dryers the net
radiation to and from a tray of material sur-
rounded by other trays is practically zero. I n
this case, the above equation simplifies to:

This equation shows that, in order to obtain the


total heat transfer coefficient, it is necessary to
multiply the coefficient for c o n v e c ti o n to the
wetted surface by a factor of (1 .+ m).A, Courtesy, Proctor & Schwartr, Inc.
A COMMERCIAL
TRAYDRYER
Values of A , and C are known, as previously pointed
out. Similarly, this same factor may be used to de-
termine the increase in rate to be obtained with an unin- determine the value of h, by means of Equation 8. Sufficient
sulated tray as compared with a perfectly insulated tray. experimental data were available from each run to compute
If radiation to the tray is an appreciable factor, an approxi- all the values except h,; hence h, could be obtained directly.
mation of this effect may be obtained by assuming the same
value of h, for both wetted and unwetted surfaces; in this Correlation of Present and Previous Experi-
case Equation 7 becomes : mental Data
I n Table 11,the mass transfer coefficients and the corrected
heat transfer coefficients are reported for each of the experi-
As an example of the use of Equation 8, consider the cor- mental runs of this investigation. These apply only to the
rections t o be applied to ht for run 1 .in order to obtain h,: constant rate period of drying. (Each of these coefficients as
reported is the average of the coefficients calculated from tem-
ht = 6.77B. t.u./(hr.)(sq.ft.)('F.) perature data taken periodically during the runs, Hence
h, = 0 (since temperatures of dryer walls and of material these average values should not necessarily agree with those
surface are practically the same) calculated from the average temperature and pressure data,
A , = 1.33
A1 = 1.33 also reported in Table 11.) A close examination of the data
L1 = 1 in. = 0.0834 ft. shows that the mass transfer coefficients are considerably more
kt = 2 B. t. u./(hr.) (sq. ft.) ( O F./ft.) erratic than the heat transfer coefficients. I n fact, in two
cases (tests 6 and 7) some of the individual data showed nega-
Substituting these values in Equation 8: tive mass transfer coefficients. This wider variation in the
mass transfer coefficients is due both to the inaccuracy in the
6'77 = hc (' +1 + 1.33 X h,
0.042 ) determination of the partial pressures of water in air by means
of psychrometric charts and also to the large errors in the
h, = 3.12 B. t. u./(hr.)(sq. ft.)(OF.)
calculations of pressure differences resulting from small errors
Actually, h, may be slightly different for the unwetted sur- in surface temperature measurements. The error involved
faces and for the wetted surfaces, since the temperature a t the in computing temperature differences and from them, heat
former surfaces will be the higher, depending on the thermal transfer coefficients, is considerably smaller. Hence in all
conductivity of the material, tray, and lagging (if any). For the tests the heat transfer coefficients were reasonably uni-
unlagged metal trays, the difference in radiation coefficient form, both for individual readings during the same test and
will not be sufficient to affect the value of ht appreciably, and for two check tests under the same conditions.
Equation 8 should be satisfactory. The mass transfer coefficients computed from the present
A knowledge of the mechanism of heat transfer to a ma- data are plotted in Figure 5A against the mass velocity of the
terial drying in an air stream is very important in the applica- drying air, together with coefficients calculated from other
tion of theory to industrial practice or in the application of available data in the literature on the evaporation of water or
small-scale drying tests to a large-scale dryer. Equations 7 on the constant rate period of drying solids. The best line
and 8 show that a prediction of the over-all drying rate of the through the present data is a straight line on log-log paper
material in trays, during the constant rate period, may be with a slope of 0.75. I n other words, the mass transfer co-
made provided the coefficient of heat transfer by convection efficient varies with the 0.75power of the mass velocity between
to the wetted surface is known. the values of 500 and 6000 pounds/(hour) (square foot),
In the present tests on sand and water, the over-all values equivalent to 125-1370 feet/minute a t 150' F. and normal
of ht obtarined from the experimental data were corrected to barometer.
395
396 IKDUSTRIAL AND ENGINEERING CHEMISTRY VOL. 30, NO. 4
The data of the other investigators indicate lower coeffi- developed on the falling rate period of drying and the critical
cients than the present data, although the variation with moisture content before any successful and practical solution
velocity is closely similar. The results of Powell and Griffiths can be obtained for the complete drying mechanism.
(6) and Hinchley and Himus (8) are low in both cases. I n The data reported above on the constant rate period have
these investigations the evaporation of water was being served to establish absolute values for the rate of moisture re-
studied, the temperature of the water being higher than that moval from sand during this period (expressed in the form of
of the air. This fact may explain the low values of the mass either mass transfer or heat transfer coefficients). I n addi-
transfer coefficients obtained, since the liquid surface film re- tion, the similarity between the rates of evaporation from
sistance may cause the surface temperature, and hence the sand and from water surfaces lends considerable weight to the
vapor pressure difference, to be lower than that actually assumption that the evaporation rate will not vary greatly
measured and used in the caIculation. I n the tests of Lurie in the constant rate period for most materials. Sherwood and
and Michailoff (4) and of Thiesenhusen ( I l ) , the evaporation Comings (10) also reported that the rates of evaporation of
of water was carried out adiabatically as in drying; the water water alone and from various materials are closely similar.
assumed a temperature close to that of the wet bulb. The The results reported from Kamei’s work on clay (Figure 5B)
water evaporation data reported by Rohwer (7) were con- confirm this assumption.
sidered to be too widely divergent among themselves to war-
rant inclusion in the plot. The data of Kamei, Mizuno, and
Shiomi (3) were obtained during the constant rate period of TABLE
111. COMPARISONOF HEATTRANSFER
AND MASSTRANS-
the drying of clay, and those of Ceaglske and Hougen ( I ) FER COEFFICIENTS,
CALCELATED
ON VARIOES
BASES
during the constant rate period of the drying of sand. All of Heat Transfer Coefficients Mass Transfer Coefficients
Based on: Based on:
these data agree reasonably well with the present data. -
Readine: Readine
When the same data, obtained either under conditions of Run from from-
No. (to - t r ) (to - t w ) Fig. 5B ( p a - PO) (PW - p4) Fig. 5A
adiabatic drying or evaporation of water, are plotted as heat F. F. Mm. Hg M m . Hg
transfer coefficients (Figure 5B), the agreement with each 1 3.12 2.58 3.4 0.0131 0.0207 0.0125
other and with the present data is much more pronounced. 2
3
3.56
3.22
3.31
3.21
3.4
3.35
0.0109
0.0122
0.0135
0.0130
0.0125
0,0120
This fact demonstrates one of the reasons for preferring the 3-2 3.42 3.44 3.4 0,0144 0.0140 0,0125
4 3.47 3.77 3.4 0.0347 0,0180 0,0125
use of heat transfer coefficients to that of mass transfer coef- 4-2 3.39 3.46 3.4 0.0159 0.0143 0.0125
ficients in interpreting the constant drying rate of water from 5 1.53
2.21
1.46
2.13
1.7
2.0
0.0057
0.0074
0,0070
0,0089
0.0064
0.0072
5-2
wet material. As with mass transfer coefficients, the heat 6 3.73 3.87 3.3 0,0840 0.0192 0,0120
transfer coefficients based on the present data vary with the 6-2 3.80 4.16 3.2 .... 0.0256 0,0120
7 6.38 7.74 6.0 0.0302 0.0225
0.76 power of the mass velocity. 7-2 6.49 6.21 6.0 o’oib 0.0241 0.0225
8 2.88 2.65 3.2 0.0089 0.0130 0.0120
9 3.79 3.55 3.8 0.0134 0.0157 0.0140
Practical Application of Results 10
11
3.40
2.99
3.27
2.33
3.35
3.4
0.0122
0.0120
0,0109
0,0208
0.0120
0.0125
No reliable method has yet been advanced for the predic- 12 3.55 3.46 3.4 0.0130 0.0118 0.0125
13 3.63 3.65 3.4 0.0153 0,0147 0.0125
tion of commercial drying times and dryer performance which 14 10.94 11.33 11.0 0.0615 0.0490 0.037
14-2 10.68 10.59 10.5 0.0376 0.0396 0.037
is generally applicable to a variety of materials. Ceaglske
and Hougen made a notable start in their recent work on
granular materials, but their method of attack requires
verification for materials other than sand, and it will probably The computation of drying problems on the basis of a heat
not prove applicable to a wide range of materials. Sherwood transfer mechanism rather than a mass transfer mechanism
offered the following relation between moisture content and is definitely to be preferred in the constant rate period. Heat
drying time as a possible approach to the estimation of drying transfer coefficients calculated from the present experimental
times : data have been shown to be much more consistent than mass
transfer coefficients determined from the same data. Con-
sequently, the appIication of heat transfer coefficients in de-
termining drying rates would be more reliable than the use of
where ec, OFtotal drying time, drying time in constant
= mass transfer coefficients. Furthermore, unless the tempera-
rate period, and drying time in falling rate ture of the drying surface is measured, it must be calculated
period, respectively
To,Tc,T E = initial, critical, and equilibrium moisture by means of heat transfer considerations before it is possible
content, respectively to apply mass transfer coefficients for drying rate predictions.
T = moisture content at time e, dry basis I n practice, surface temperature measurements are rarely
a = aconstant made, and i t is usually assumed that the surface is essentially
The application of this equation requires a knowledge of the a t the wet-bulb temperature of the air during the constant
critical moisture content and the constant a, neither of which rate period. I n Table I11 both heat transfer and mass transfer
is usually obtainable except by means of actual drying tests. coefficients have been calculated for the present tests based
Moreover, the logarithmic relation applying to the falling on (a) actual material surface temperatures (as in Table II),
rate period holds only for certain types of falling rate curves (b) measured wet-bulb temperature of air corrected for radia-
and only approximately in these cases. Another disadvan- tion, and (c) average curves of the coefficients shown in
tage of Equation 9 is that it does not readily show the rela- Figure 5 .
tive influence of the variables affecting the over-all drying As shown in Table 111, the assumption that the surface is
time. a t the wet-bulb temperature of the air introduces a more
I n the present paper it is proposed primarily to develop the serious error in the computation of mass transfer than of
available knowledge on the constant rate period of drying to heat transfer coefficients. For an unlagged tray, however, a
. the point of practical application to dryer design and perform- significant error is introduced in both coefficients.
ance. This is considered as merely the first step in the I n the application of the present results to practical drying
problem of developing a rational basis for the estimation of problems, it is recommended that heat transfer coefficients
commercial drying times and optimum drying conditions for be used as the basis for calculations of drying rate. Figure
a given material. It appears evident, however, that a con- 6 permits a ready estimate to be made of the relative rates of
siderable amount of fundamental information must still be drying for various air temperatures and relative humidities.
APRIL, 1938 INDUSTRIAL AND ENGINEERING CHEMISTRY 397

The chart is based on the difference between the dry-bulb and When the drying takes place in unlagged solid bottom trays,
wet-bulb temperatures of the entering stream of air. The the surface temperature of the material may be calculated
absolute drying rates shown are based on a heat transfer coef- reasonably accurately by the application of the same heat
ficient of 3.1 B. t. u./(hour) (square foot) (” F.) for air a t 300 transfer methods and the chart readings corrected by the
feet/minute velocity, 150” F., and 30 per cent relative hu-
midity. They also take into account the change in air den-
factor (-)
t o - t,
where t, = dry-bulb temperature, t, = surface
sity with temperature. This value was obtained from Figure temperature, and t, = wet-bulb temperature.
5B and was chosen slightly (i. e., 10 per cent) below that given The fact that the proposed method for computing drying
by the smooth curve representing the present data in order rates is applicable only to the constant rate period of drying
to give some weight to the other data studied. Until further definitely limits its scope. There are a number of possible
data on other materials are available, this chart may be as- uses for the method, however, which were not investigated in
sumed approximately correct for any material in the constant the present study. The over-all drying time of some ma-
rate period of drying. terials exhibiting a distinct falling rate period has been found
to be influenced by velocity, humidity, and temperature in
approximately the same degree as is the case for the constant
rate period. For these materials a knowledge of the drying
time under any given set of conditions can be used to predict
the drying time under other conditions by analogy to the con-
stant rate period.
Again, for processes where air is blown up or down through
a bed of a granular material, the drying is often nearly 100
per cent within the constant rate period. I n this case the
equivalent area can be determined by a single test, or possibly
by particle size distribution data, and the drying can then be
predicted from constant rate data. The same possibility
applies to the drying of materials in a rotary dryer. Finally,
the above correlation of data for the constant rate drying
period makes it possible to use an over-all drying expression
of the type proposed by Sherwood without the necessity of a
number of drying tests, provided Tc is known. For instance,
the constant in Sherwood’s Equation 9 could be determined
from constant rate data as follows:

I n this expression only Tc is of uncertain value for any


given material, since the equilibrium moisture content ( T E ) ,
the wetted surface area ( A w ) the
, dry weight, and the latent
heat (T,) are likely to be known, and the temperature dif-
ference (t, - tJ and over-all heat transfer coefficient (h,) are
calculable from Figure 6 and Equation 8.
6. RATEOF EVAPORATION
FIGURE CHART
Acknowledgment
A correction curve t o correct for the air velocity in any given The authors wish t o acknowledge the cooperation of A, P.
problem is incorporated in Figure 6. This curve is based on Colburn in analyzing and calculating the heat transfer through
the variation of drying rate with the 0.8 power of the velocity. the Unwetted surfaces of the trays. They also are indebted to
There is some question as t o the shape of this curve below a W. H. McAdams, T. H. Chilton, and T. K. Sherwood for
velocity of about 150 feet/minute, where natural convection their constructive reviews of the manuscript.
begins to be significant, but the dotted curve as drawn should
be conservative. Literature Cited
The use of this chart in practical drying problems is as Ceaglske, N. H., and Hougen, 0. A., IND.ENQ.CHEM., 29,805-
follows: Since the drying conditions of temperature and rela- 13 (1937): Trans. Am. Inst. Chem. Engrs., 33,283-312 (1937).
Hinchley, J. W., and Himus, G. W., Trans. Inst. Chem. Engrs.
tive humidity are fixed, the corresponding absolute drying (London), 2,57-64 (1924).
rate is read from Figure 6. This value is then multiplied by Kamei, S., Mizuno, S., and Shiomi, S., J . SOC.Chem. Ind. Japan,
the correction factor corresponding to the air velocity em- 38,460-3B (1935).
ployed (also given on Figure 6). The rate so obtained, how- Lurie, M., and Miohailoff, N., IND. ENG.CHEM.,28, 345-9
(1936).
ever, does not include any effects of radiation or of conduc- McAdams, W. H.,“Heat Transmission,” 1st ed., pp. 222-4,
tion through unwetted surfaces. I n the case of a material New York, McGraw-Hill Book Co., 1933.
drying from two sides, the latter effect is generally negligible, Powell, R. W., and Griffiths, E., Trans. Inst. Chem. Engrs.
and the radiation can be readily computed. The rate de- (London), 13,175-92 (1935).
termined by Figure 6 is then corrected by the factor (h, + Rohwer, C., U . S. Dept. Agr., Tech. Bull. 271 (1931).
Sherwood, T. K., IND ENQ.CHEM., 21,976-80 (1929).
h,)/h,. If drying takes place from only one surface, as in Sherwood, T. K.,Trans. Am. Inst. Chem. Engrs., 32, 150-68
most cases of industrial tray drying, a reasonably approxi- (1936).
mate correction can be made for heat transfer through the Sherwood, T.K.,and Comings, E. W., Ibid., 27,118-33 (1932).
Thiesenhusen, H., Gesundh.-Ing., 53, 113-19 (1930).
unwetted surfaces, as indicated earlier in the paper in the dis-
cussion of heat transfer to material drying in trays. RECEIVED
February 7, 1938.

Additional papers presented a t this sym-


posium will appear in subsequent issues. 1

You might also like