Journal of Food Engineering 59 (2003) 267275

www.elsevier.com/locate/jfoodeng

The role of heat and mass transfer phenomena in atmospheric

freeze-drying of foods in a uidised bed
P. Di Matteo, G. Dons, G. Ferrari *

Department of Chemical and Food Engineering, University of Salerno, Via Ponte Don Melillo, 84084 Fisciano (SA), Italy
Received 29 July 2002; accepted 23 November 2002

Abstract
Freeze-drying under atmospheric pressure in a uidised bed of adsorbent ne particles is proposed for dehydrating food products
at lower cost than the traditional freeze-drying process under vacuum. The aim of this paper is to evaluate the role of heat and mass
transfer coecients between the product and the bed during the process at atmospheric pressure and to analyse the eect of transport
coecients and operating variables on freeze-drying rates. Experiments were performed using potato cylinders as the test material.
The results obtained are discussed also in connection with further developments of this technique and optimal design of the process.

2003 Elsevier Science Ltd. All rights reserved.

Keywords: Heat transfer; Mass transfer; Fluidised bed; Freeze-drying

1. Introduction Heat and mass transfers processes take place simulta-

Freeze-drying is a widespread dehydration technique
consisting of the removal of water from a frozen food by
sublimation. The quality of freeze dried products is very
high in comparison with that of the products dehydrated
using other techniques (King, 1970; Liapis, 1987) due to
the prevention of heat damage. In fact, since freeze-
drying is carried out at low temperature, due to the ne-
cessity of keeping the product in the frozen status
throughout the process, nutritional characteristics are
well preserved. Moreover, due to the direct removal of
water vapour from ice crystals, freeze dried product show
an interconnected porous structure which can be rehy-
drated very eectively. Thus, the sensory properties and
quality prole of fresh foods are easily regained on rehy-
dration.
The process is carried out under vacuum in order to
keep water vapour pressure and temperature in the de-
hydration chamber well below that of the triple point of
the solution permeating the product. Conventional ap-
paratus consists essentially of a vacuum chamber con-
taining a series of trays which accommodate the frozen
product to be dehydrated (Mellor, 1978; Oetjen, 1999).
*
Corresponding author. Tel.: +39-089-96-4134; fax: +39-089-96-
4057.
E-mail address: gferrari@unisa.it (G. Ferrari).
0260-8774/03/$ - see front matter
2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0260-8774(02)00467-3
neously. Heat is transferred to the product by conduc-
tion, through heating plates, and/or by radiation. Water
vapour forming by sublimation condenses on the coils
of a condenser, usually located outside the chamber. A
vacuum pump removes incondensable gases from the
freeze-drying chamber.
The main disadvantage of this technique is the high
xed and operating costs (Wol & Gibert, 1990a,
1990b). The latter are due to the energy intensive ope-
rations in series involved in the process: freezing of the
fresh product, heating of the frozen foods at low tem-
perature to induce sublimation, condensation of water
vapour and mechanical energy consumption to maintain
the vacuum. Moreover, vacuum operations are mainly
carried out batchwise, which represents another addi-
tional cost together with the peculiar requirements of
apparatus operated under vacuum.
This causes limited use of this technique in the food
industry and restricts freeze-drying to dehydration of only
high added value products, such as instant coee or baby
foods. Only recently it has been proposed to freeze dry
fruits and vegetables to be used as high quality ingre-
dients in ready-to-prepare soups, snacks and delicacies.
The limiting stage of the traditional freeze-drying
process under vacuum is the transfer of heat to the
product due to the decrease in thermal conductivity with
decreasing the pressure of the freeze-drying chamber
268 P. Di Matteo et al. / Journal of Food Engineering 59 (2003) 267275
Nomenclature
Symbols
Cp gas phase specic heat (J/kg K)
CpE emulsion phase specic heat (J/kg K)
CpS particle specic heat (J/kg K)
dp particle diameter (m)
D sample diameter (m)
D mass diusivity (m2 /s)
h external heat transfer coecients (W/m2 K)
hGC gas convective component of the heat transfer
coecient (W/m2 K)
Ke internal mass transfer coecient (s/m2 )
0
K reciprocal of the external mass transfer co-
ecient (s/m)
K external mass transfer coecient (m/s)
M0 initial moisture content (kg H2 O/kg d.s.)
Mf nal moisture content (kg H2 O/kg d.s.)
Mw molecular weight of water (kg/kg mol)
NuGC Nusselt number for the gas convective com-
ponent
NuMAX maximum Nusselt number
Pi water vapour partial pressure at the subli-
mation ice front (N/m2 )
PS water vapour partial pressure at the sample
surface (N/m2 )
PE water vapour partial pressure in the uidised
bed (N/m2 )
(King, 1970; Liapis, 1987). To improve heat transfer
rates, two dierent approaches have been followed:
modications of the traditional process (Carn & King,
1977; Goldblith, Rey, & Rothmayr, 1975; King, 1970;
Wol & Gibert, 1988) and operation at atmospheric
pressure utilising a uidised bed of adsorbent particles,
uidised with cold air (Boeh Ocansey, 1985; Lombrana
& Villaran, 1996; Wol & Gibert, 1990a).
The main advantages of this process are simplica-
tion of the apparatus due to the absence of a vacuum
chamber and ancillary equipments and the signicant
reduction in energy costs. A simplication of the process
is also produced by the absence of any device for sup-
plying external energy to the product inside the cham-
ber, as a consequence of the fact that the adsorbent
material plays the double role of capturing the water
vapour forming by sublimation and of providing the
heat of adsorption to the sublimation front. Moreover,
operation in a uidised bed improves the external heat
transfer coecients. As already discussed by Dons and
Ferrari (1995) and by Dons, Ferrari, and De Vita
(1998), heat transfer coecients between a uidised bed
of ne particles and immersed objects are more than one
order of magnitude higher than those between gas and
solids. Moreover, since the proposed process operates at
PrE emulsion phase Prandtl number
Ri cylindrical frozen core radius (m)
RS cylinder radius (m)
ReE emulsion phase Reynolds number
t time (s)
Tf freezing temperature (K)
Ti temperature of the sublimation front (K)
TS sample surface temperature (K)
Tb uidised bed temperature (K)
U gas velocity (m/s)
Umf minimum uidisation velocity (m/s)
V specic volume of solid material per unit
weight of water (m3 /kg)
W residual moisture content (%)
WR product to adsorbent weight ratio (kg/kg)
X water fraction (dry basis) (kg H2 O/kg d.s.)
Greeks
e bed voidage
emf bed voidage at minimum uidisation
k thermal conductivity (W/m K)
DHS latent heat of sublimation of ice (kJ/kg)
l gas viscosity (kg/m s)
q gas density (kg/m3 )
qD bulk density of dry solid (kg/m3 )
qS particle density (kg/m3 )
atmospheric pressure, the internal heat transfer process
by conduction becomes more ecient, as discussed by
many authors (Boeh Ocansey, 1988; Wol & Gibert,
1988, 1990b) due to the increase in gas thermal con-
ductivity with pressure. The main drawback of atmo-
spheric freeze-drying is the increase in the freeze-drying
time, due to the decrease of the freeze-drying rate. This,
in turn, is due to the decrease in water vapour diusivity
with increasing pressure in the chamber.
In this paper the eect of the main processing vari-
ables on freeze-drying rates of potato cylinders is analy-
sed and discussed, with special attention to the role of
heat and mass transfer rates between the product and
the uidised bed of adsorbent particles. A simplied
model for the evaluation of internal and external heat
and mass transfer coecients is proposed as a valuable
tool for preliminary evaluation of process time.
2. Material and methods
2.1. Materials
Fresh potato from the local market were used in the
experiments. Potatoes were peeled, cut into samples of
 P. Di Matteo et al. / Journal of Food Engineering 59 (2003) 267275
cylindrical shape and frozen at )18 C in a blast freezer.
The length of cylinders was always 10 mm while three
dierent diameters (6, 8 and 11 mm) of the samples were
tested to evaluate the eect of this parameter on the
dehydration rate.
2.2. Experimental set up
The sketch of the experimental atmospheric freeze-
dryer is shown in Fig. 1. The freeze-drying chamber was
a stainless steel column, 80 mm ID and 450 mm high,
provided with an external jacket, insulated with glass
wool. The chamber was refrigerated with a 60% solution
of ethylene glycol, kept at controlled temperature in a
cryogenic bath, which circulated in the interstitial space
between the column and the jacket. The uidising gas
was air. The air stream, whose ow rate was measured
with a rotameter, was dehumidied in a xed bed of
silica gel, and cooled in a heat exchanger. Cool air was
mixed with air at ambient temperature to achieve the
test temperature and the ow rate adjusted to the test
value by means of a by-pass downstream of the circu-
lation pump. A distributor of sintered brass provided
the uniform distribution of the uidising medium in the
bed cross section. A xed bed of stainless steel spheres,
above the distributor, provided a further cooling of the
air stream. The uidised bed of absorbent was located
above the xed bed of metal spheres. Characteristics of
absorbent particles are given in Table 1. A transparent
lid, at the top of the uidisation column, allowed con-
tinuous monitoring of the uidisation conditions in the
freeze-drying chamber. The temperature of the air
stream and of the bed was measured with thermocou-
ples. The temperature of the bed was kept in the range
)10 to )5 C. This ensures that the temperature of the
product is always below that of the triple point of the
Fig. 1. Atmospheric freeze-drying experimental set up.
269
solution at the water vapour partial pressure kept in
the bed by the adsorbent. After the start up, the system
operated as a closed circuit. As the freeze-drying cham-
ber reached the test temperature, a metal basket con-
taining the frozen samples was introduced into the bed.
Samples were taken from the bed at stated times,
weighed and dried in a oven at 105 C to determine the
dry solid content.
3. Evaluation of heat and mass transfer coecients
During freeze-drying heat and mass transfer takes
place simultaneously. One of the main advantages of the
use of beds of adsorbent material is the fact that the heat
of adsorption and the heat of sublimation are of the
same order of magnitude, so that no extra energy supply
is necessary to induce sublimation in the atmospheric
freeze-drying process.
Heat is transferred from the adsorbent (the heat
source) to the surface of the product and, by internal
conduction, to the sublimation front. The water vapour
forming by sublimation, in turn, diuses through the
outer dry layer of increasing thickness, moves to the bed
and is captured by the adsorbent.
The use of uidised beds of adsorbent particles
greatly improves external heat and mass transfer rates
compared to other contact techniques. A preliminary
analysis of transport phenomena between the sample to
be dried and the bed is thus carried out in order to
predict the eect of external operating conditions on the
process time.
Heat and mass transfer coecients are evaluated
from the relationship between drying rate, expressed in
terms of the fraction of the residual water content,
driving forces (temperature dierence for heat transfer
270 P. Di Matteo et al. / Journal of Food Engineering 59 (2003) 267275

Table 1 1. 8E +05
Adsorbent materials characteristics 1. 6E +05 y = 1.16 E+05x + 7.55E+03
dC 6 mm dC 8 mm dC 11 mm 1. 4E +05

(Te-Ti)/(-dW/dt)
h (W/ k (W/ h (W/ k (W/ h (W/ k (W/ 1. 2E +05
m2 K) m K) m2 K) m K) m2 K) m K) 1. 0E +05
TE 6 C 490 0.2 420 0.15 400 0.155 8. 0E +04
TE 14 C 520 0.06 510 0.066 490 0.056 6. 0E +04
The particle diameter dp is the diameter corresponding to the 50% of 4. 0E +04
the cumulative distribution measured by laser diractometry. 2. 0E +04
0. 0E +00
0 0.5 1 1.5
and partial pressure for mass transfer) and product -lnW
physical properties.
Fig. 2. Experimental values of left hand side of Eq. (4) as a function of
the logarithm of the residual water content. Potato cylinders (d 11
3.1. Heat transfer
mm, L 1 cm); Adsorbent: Bran, WR 1=20, Tb 6 C, u  umf 7
cm/s.
In the hypothesis that quasi-steady state conditions
hold, the heat balance equation in cylindrical coordi-
nates can be written in the form: ditions were evaluated and data are reported in Table 2.
These data demonstrate that the values of the heat
TS  Ti transfer coecients in atmospheric freeze-drying process
RS hTE  TS k 1
RS are more than one order of magnitude higher than those
ln
Ri measured in conventional vacuum freeze dryers (Boeh
and thus Ocansey, 1988). Experimental heat transfer coecients
are also compared with those predicted on the basis of a
h RS semiempirical correlation proposed by Dons et al.
ln Ti
TE R S
k Ri (1998) describing heat transport phenomena between
TS 2
h RS uidised beds and immersed objects.
RS ln 1
k Ri According to this correlation, the maximum values of
With the further assumption that the heat transferred to the heat transfer coecients are evaluated as a sum of
the product during the freeze-drying is totally utilised two contributions: the particle convective component
for ice sublimation, the following equation can be and the gas convective component. The latter compo-
written: nent is evaluated according to the following equation:
   1=2  
DHS R2S dW hGC D qUD Cp l
 hTE  TS  3 NuGC 2 0:707 5
Mw V 2Ri dt k el k
Substituting Eq. (2) into Eq. (3), the following rela- NuMAX is thus calculated as follows:
tionship is obtained:  b
dp
  NuMAX NuGC a RecE PrEd 6
TE  Ti DHS RS 1 1 D
qD M0  Mf  RS ln W 4
dW Mw 2 h 2k
 where
dt
qS 1  emf U  Umf D
Plotting the values of TE  Ti =dW =dt as a function ReE 7
l
of ( ln W ) a straight line is obtained. The values of heat l
transfer coecients are calculated from the slope and CpE l
PrE CpS 1  emf Cp emf  8
the intercept of this line. In Fig. 2 data for a selected k k
operating condition are presented as an example of this The numerical constants appearing in Eq. (6) assume the
procedure. Following this approach, heat transfer co- values reported below for xed (9) or moving objects
ecients for dierent sample sizes and operating con- (10) respectively

Table 2
Heat transfer coecients in atmospheric freeze-drying process (potato cylinders, L 1 cm)
dC 6 mm dC 8 mm dC 11 mm
K (m/s) D (m2 /s) K (m/s) D (m2 /s) K (m/s) D (m2 /s)
TE 6 C 0.112 6.2 106 0.093 7.8 106 0.073 8 106
TE 14 C 0.143 6 106 0.139 6.2 106 0.127 6.6 106
 P. Di Matteo et al. / Journal of Food Engineering 59 (2003) 267275 271

600 7. 0E+0 7

6. 0E+0 7
y = 4 . 2 E+ 0 7 x + 1 . 7 E+ 0 6

( Pe - P i ) /( - d W / d t )
5. 0E+0 7
500
4. 0E+0 7
h ( W /m 2 C )

3. 0E+0 7
400
2. 0E+0 7

1. 0E+0 7
300
0. 0E+0 0
0 0 .5 1 1 .5

- ln ( W )
200
6 8 10 12 14
Fig. 4. Experimental values of the left hand side of Eq. (14) as a
d (mm)
c function of the logarithm of the residual water content. Potato cylin-
ders (d 11 mm, L 1 cm). Adsorbent: Bran, WR 1=20, Tb 6
Fig. 3. Predicted (- - -) and experimental (s) heat transfer coecients C, u  umf 7 cm/s.
as a function of sample diameter.

ported in Fig. 4. Mass transfer coecients can be eva-

a 353; b 0:422; c 0:275; d 10:8 9
luated from the slope and the intercept of this line.
a 2800; b 0:396; c 0:248; d 1:41 10 The values of mass transfer coecient for dierent
In Fig. 3 heat transfer data as an average of three tests sample size and operating conditions were obtained and
are compared with model predictions (Eq. (6)). A good data are reported in Table 3. It is possible to observe
agreement between experimental and calculated data that the heat and mass transfer coecients decrease with
can be observed. This conrms the validity of the em- increasing the sample size and the bed temperature. This
pirical correlation proposed also in a temperature range is in agreement with literature data on heat and mass
outside that utilised in previous experiments (Dons transfer between uidised beds of ne particles and
et al., 1998). immersed bodies (Dons & Ferrari, 1995; Dons et al.,
1998; Prins, Casteleijn, Draijer, & Van Swaaij, 1985).
3.2. Mass transfer It is important to emphasise that the external heat
and mass transfer processes in atmospheric uidised bed
In the hypothesis that quasi-steady state conditions freeze-drying are much higher than those generally ob-
hold, the mass balance in cylindrical coordinates can be tained in the traditional freeze-drying process under
written as: vacuum (Boeh Ocansey, 1988). This feature of uidised
beds makes such a contact technique promising in this
PS  Pi peculiar application.
K 0 PS  PE RS Ke 11
RS However, the external resistance to heat and mass
ln
Ri transfer play a role only at the initial stage of the pro-
and thus cess. In fact, in atmospheric freeze-drying, the increase
in the operating pressure determines the increase in the
RS eective thermal conductivity but, at the same time, a
Pi Ke K 0 PE RS ln
Ri decrease in the eective diusivity occurs (Liapis, Pikal,
PS 12
RS
Ke K 0 RS ln
Ri
The following relationship between the mass ux and Table 3
the dehydration rates can be written: Mass transfer coecients in atmospheric freeze-drying process (potato
cylinders, L 1 cm)
1 R2S dW Adsorbent Particle True Bulk Experimen-
 K 0 PS  PE 13
V 2Ri dt diameter density qp density qb tal minimum
dp (lm) (g/cm3 ) (g/cm3 ) uidisation
Substituting Eq. (12) into Eq. (13) one obtains: velocity Umf
  (cm/s)
Pi  PE RS 1 1
qD M0  Mf  R S ln W 14 Bran 580 1.4760 0.867 12
dW 2 K 0 2Ke
 Corn our 750 1.4755 0.758 20
dt Starch 30 2.2912 0.987 0.14
From the plot of the ratio Pi  PE =dW =dt as a Zeolites 640 2.2302 0.917 19
Bentonite 658 2.3620 1.191 21
function of ( ln W ) a straight line is obtained, as re-
272 P. Di Matteo et al. / Journal of Food Engineering 59 (2003) 267275
& Bruttini, 1996). As a consequence, the limiting stage
of the process is represented by the internal mass
transfer, while in traditional freeze-drying under va-
cuum the limiting stage is represented by the internal
heat transfer.
4. The eect of operating variables
The eect of the following variables on the atmo-
spheric freeze-drying rates of potato cylinders was also
investigated:
freezing temperature;
uidised bed temperature;
sample size;
uidisation velocity;
nature of adsorbent material;
size of the adsorbent particles;
product/adsorbent weight ratio;
regeneration temperature of the adsorbent.
To evaluate how of the freezing temperature aects
atmospheric freeze-drying rates of potato cylinders,
samples were frozen in a blast freezer at dierent tem-
peratures: )10, )20, )50 C, and in liquid nitrogen.
Fig. 5 demonstrates that the dehydration rate in-
creased with increasing the temperature of the freezing
medium. This is related to the size of ice crystals which
increased at high freezing temperature, when the ice
crystal growth prevails, and decreases at low freezing
temperature, when the prevailing phenomenon is the
nucleation.
During freeze-drying a porous layer of dry solid
forms and the pores size is directly related to the ice
crystal size, the bigger the ice crystals, the bigger the
pore diameter. The water vapour diuses more easily
Fig. 5. Atmospheric freeze-drying rates of potato cylinders (D 6 mm)
as a function of processing time. The eect of freezing temperature.
Adsorbent: Zeolites; WR 1=20; Tb 10 C, u  umf 7 cm/s.
Fig. 6. Atmospheric freeze-drying rates of potato cylinders (D 6 mm)
as a function of processing time. The eect of uidised bed tempera-
ture. Adsorbent: Bran; WR 1=20, u  umf 4 cm/s.
when the pore size in the dry layer of the product in-
creases and, thus, a reduced processing time is detected.
However, the higher freezing temperature might cause
the damage of the cellular tissue of the product which in
turn determines the loss of the nutritional value as well
as of the texture after rehydration. This suggests that
this variable should be carefully chosen according to the
nature of the foodstu and optimised with reference to
the quality of the dry product.
Bed temperature aected strongly atmospheric
freeze-drying rates of potatoes as shown in Fig. 6, in
which data obtained at dierent operating temperatures
are reported. It can be observed that the higher the
uidised bed temperature, the faster the process. How-
ever, the bed temperature cannot be increased arbitra-
rily, as thawing of the product should be avoided.
Increasing the bed temperature, the water vapour pres-
sure in the bed decreases. As a consequence, the driving
force of the mass transfer process, which is the dierence
of the vapour partial pressure at the ice front Pi and in
the bed PE , increases and, consequently, the freeze-
drying rate increases. The increase in the bed tempera-
ture did not aect the heat transfer process as conrmed
by experimental data. In fact, the temperature dierence
between the bed and the sample is of the order of 0.6 C
in all experiments.
Fluidisation velocity had no signicant eect on the
dehydration rates as shown in Fig. 7. This conrms the
conclusion that the limiting stage of the mass transfer
process is represented by the internal diusion and that
the external mass transfer coecients plays a role only
in the initial stage of the process.
The eect of the characteristics of adsorbent material
is also studied. Experiments were carried out using dif-
ferent materials, whose characteristics are reported in
Table 3.
The dehydration rates of potato cylinders freeze dried
in beds of dierent adsorbents is shown in Fig. 8. The
 P. Di Matteo et al. / Journal of Food Engineering 59 (2003) 267275 273

Fig. 7. Atmospheric freeze-drying rates of potato cylinders (D 6 mm) Fig. 9. Atmospheric freeze-drying rates of potato cylinders (D 6 mm)
as a function of processing time. The eect of uidisation velocity. as a function of processing time. The eect of the adsorbent particle
Adsorbent: Bran; WR 1=20; Tb 8 C. size. Adsorbent: Bran; WR 1=20; Tb 8 C.

haviour of the bed of nes during the process. This be-

4
haviour was even more pronounced with increasing the
3.5 starch
bran
processing time, due to the increase of the bed moisture
3 Bentonite content and thus of the stickiness of the particles.
X (kg H2O/kg d.s.)

2.5
Zeolites Also the size of adsorbent particles inuenced the
maize flour
freeze-drying rate as shown in Fig. 9. It can be observed
2
that the lower the particle size the higher the freeze-
1.5
drying rate. This can be explained by considering the
1 number of particles in contact with the product during
0.5 the process. This increases with decreasing the particle
diameter. Thus, the adsorption rate per unit volume of
0
0 500 1000 1500 2000
the bed increases. Fig. 9 refers to data obtained with
time (min)
bran of dierent sizes, but similar results were obtained
with the other adsorbents.
Fig. 8. Atmospheric freeze-drying rates of potato cylinders (D 6 mm) The product to adsorbent weight ratio was also
as a function of processing time. The eect of the nature of the ad- changed during the experiments to verify the eect of
sorbent. WR 1=20; Tb 8 C.
this parameter on the freeze-drying rates. The value of
this parameter was varied in the range 1=8 to 1=80. In
highest freeze-drying rates were detected in a bed of Fig. 10 the dehydration curves obtained with four dif-
zeolites and the quality of dried product obtained is also ferent values of the weight ratio are reported. It can be
very good. Nevertheless, the adsorption eciency was seen that the higher the product to adsorbent weight
not the only criterion to be satised when choosing the ratio the faster the drying process, especially at the very
adsorbent material. In fact, the other criterion to be
taken into account is its compatibility with the foodstu.
4
To full this requirements corn our, bran, bentonite
3.5 WR=(1/8)
and starch were used. It can be observed that corn our
WR=(1/20)
was the more ecient adsorbent, but, due to its irregular 3
X(kg H2O/kg d.s.)

WR=(1/40)
shape and hardness, it caused mechanical damage to the 2.5 WR=(1/80)
product as a consequence of the attrition. Good results 2
were obtained with bran and bentonite. The latter ma-
1.5
terial is widely used in the food industry in wine ltra-
1
tion, fruit juices clarication and oil decolouration. Few
applications are also reported on the utilisation of 0.5

bentonite as a food moisture adsorbent (Barbanti, 0

DOrazio, & Versari, 1997). 0 500 1000 1500 2000

Many problems were detected during the experiments time (min)

with starch. This material was characterised by a mean Fig. 10. Atmospheric freeze-drying rates of potato cylinders (D 6
particle size of 20 lm. During uidisation, starch formed mm) as a function of processing time. The eect of the product to ad-
agglomerates and channels which caused uneven be- sorbent weight ratio. Adsorbent: Bran; Tb 8 C; u  umf 6 cm/s.
274 P. Di Matteo et al. / Journal of Food Engineering 59 (2003) 267275
beginning of the process. This parameter represents a
processing variable to be optimised. In fact very high
values of the product to adsorbent ratio should be
avoided considering that the volume of the freeze-drying
chamber occupied by the adsorbent is unavailable for
product processing. On the other hand, if very low
values of this ratio are used the product segregated at
the bed bottom and a very poor contact between the
product and the adsorbent takes place.
After the atmospheric freeze-drying process it is
necessary to regenerate the adsorbent material to re-
cover its adsorption capacity. Dierent regeneration
temperatures were used to investigate the eect of this
variable on the processing time. Experimental results
obtained with beds of zeolites demonstrate that regene-
ration is eective irrespective of the temperatures used in
the range from 60 to 105 C. Thus, it is not necessary to
use high regeneration temperatures since the eect of
this parameter is not very signicant. Similar conclu-
sions can be drawn for the other adsorbents used in
these experiments.
On the basis of the experimental results presented
above, it can be argued that the key parameter of the
atmospheric freeze-drying process is the size of the
product. This clearly appears by analysing the dehy-
dration curves of potato cylinders of dierent diameters
reported in Fig. 11. Increasing the sample size, the
freeze-drying rate decreased. Bearing in mind that the
limiting stage of this process is represented by the water
vapour diusion through the dry layer of the sample, the
increase of the sample size implies the increase of the
thickness of the dry layer and, consequently, increase of
the water vapour diusion path. This, in turn, decreases
the mass transfer rate and thus increases the processing
time for bigger samples.
These results suggest that the reduction of the pro-
cessing time at atmospheric conditions can be mainly
achieved by reducing the size of the product, while a
limited eect is obtained by enhancing external heat and
mass transfer rates.
4
3.5 D=6mm
X (kg H2O/kg d.s.)
3 D=8mm
2.5 D=11mm
2 Boeh Ocansey, O. (1985). Some factors inuencing the freeze drying of
1.5
1
0.5
0 coecients. Journal of Food Engineering, 7, 127146.
0 500 1000 1500 2000
time (min)
Fig. 11. Atmospheric freeze-drying rates of potato cylinders as a
function of processing time. The eect of sample size. Adsorbent: Bran;
WR 1=20; Tb 8 C; u  umf 2 cm/s.
5. Conclusions
Present results conrm that atmospheric freeze-
drying in uidised beds of adsorbent material represents
an interesting alternative to the traditional freeze-drying
process. This technique, compared to operation under
vacuum, shows a considerable reduction of plant and
energy costs. Moreover, carrying out the operation in
the presence of a gaseous environment improves exter-
nal heat and mass transfer coecients, which are further
enhanced by the utilisation of uidisation as the contact
technique between adsorbent and products.
The experimental analysis of heat and mass transfer
rates in this work yields values of external heat and mass
transfer coecients, which were shown to be orders of
magnitude higher than those expected in vacuum ope-
ration. The simple correlation proposed predicts with
good approximation transport coecients in dierent
operating conditions for cylindrical samples of dierent
products. This analysis also demonstrates that, after the
rst stage of drying, the process under atmospheric
conditions is controlled by internal vapour diusion and
becomes relatively insensitive to external transport
conditions, provided that relevant coecients are kept
suciently high as in uid bed operation.
This explains why dehydration time in this conditions
is of the same order of magnitude, or even longer, than
in vacuum conditions. A wide range analysis of the eect
of operating conditions on freeze-drying rates was per-
formed to determine optimal conditions to reduce pro-
cess time. Among other variables, the sample size is
identied as a key parameter in determining the contact
time required by the process. If product size is reduced,
the processing time is reduced, down to values which
compare well with those detected in the traditional
process. The choice of a proper set of other variables
also contributes to making atmospheric freeze-drying a
technique suitable for wider application in the food in-
dustry.
References
Barbanti, D., DOrazio, M., & Versari, A. (1997). The use of Bentonite
as a moisture regulating system 1. Study on some sorption
properties of bentonites for their potential use in Food Technology.
Journal of Food Engineering, 33, 193206.
carrot discs in vacuo and at atmospheric pressure. Journal of Food
Engineering, 4, 229243.
Boeh Ocansey, O. (1988). Freeze-drying in a uidized bed atmospheric
dryer and in a vacuum dryer: evaluation of external transfer
Carn, R. M., & King, C. J. (1977). Modication of conventional freeze
dryers to accomplish limited freeze drying. AIChe Symposium
Series, 73(163), 103112.
Dons, G., & Ferrari, G. (1995). Heat transfer coecients between gas
uidised beds and immersed spheres: dependence on the sphere
size. Powder Technology, 82, 293299.
 P. Di Matteo et al. / Journal of Food Engineering 59 (2003) 267275
Dons, G., Ferrari, G., & De Vita, A. (1998). Experimental determi-
nation of heat and mass transfer coecients in two component
uidised beds. In Proceedings of the sixth international conference
multiphase ow in industrial plants, Milan, Italy.
Goldblith, S. A., Rey, L., & Rothmayr, W. W. (1975). Freeze drying
and advanced food technology (pp. 345348). NY: Academic Press.
King, C. J. (1970). Freeze-drying of foodstus. CRC Critical Review
Food Technology, 1379.
Liapis, A. I. (1987). Freeze drying. In A. S. Mujumdar (Ed.),
Handbook of industrial drying (pp. 295326). NY: M. Dekker, Inc.
Liapis, A. I., Pikal, M. J., & Bruttini, R. (1996). Research and
development needs and opportunities in freeze-drying. Drying
Technology, 14(6), 12651300.
Lombrana, J. I., & Villaran, M. C. (1996). Interactions of kinetic and
quality aspects during freeze-drying in adsorbent medium. Indus-
trial Engineering Chemistry Research, 35, 19671975.
275
Mellor, J. D. (1978). Fundamentals of freeze drying (pp. 239243).
London: Academic press.
Oetjen, G. W. (1999). Freeze drying (pp. 133173). WILEY-VCH.
Prins, W., Casteleijn, T. P., Draijer, W., & Van Swaaij, W. P. M.
(1985). Mass transfer from a freely moving single sphere to the
dense phase of a gas uidised bed of inert particles. Chemical
Engineering Science, 40(3), 481497.
Wol, E., & Gibert, H. (1988). Developpments technologiques
nouveaux en lyophilisation. Journal of Food Engineering, 8, 91
108.
Wol, E., & Gibert, H. (1990a). Atmospheric freeze-drying part 1:
Design, Experimental investigation and energy saving advantages.
Drying Technology, 8(2), 385404.
Wol, E., & Gibert, H. (1990b). Atmospheric freeze-drying part 2:
Modelling drying kinetics using adsorption isotherms. Drying
Technology, 8(2), 405428.