Вы находитесь на странице: 1из 10

Modeling of mass and heat transfer during combined processes of osmotic

dehydration and freezing (Osmo-Dehydro-Freezing)


Athanasia M. Goula
n
, Harris N. Lazarides
Department of Food Science and Technology, Faculty of Agriculture, Aristotle University of Thessaloniki, Thessaloniki, Greece
H I G H L I G H T S
c A new methodology was developed to describe Osmo-Dehydro-Freezing (ODF).
c The proposed model was not empirical but was based mainly on structural data.
c The proposed model allowed accurate prediction of temperature and phase changes.
c The proposed model allowed accurate prediction of solute and water concentrations.
c The model presents excellent prediction ability in contrast to an empirical model.
a r t i c l e i n f o
Article history:
Received 24 April 2012
Received in revised form
29 June 2012
Accepted 13 July 2012
Available online 24 July 2012
Keywords:
Drying
Food processing
Heat transfer
Mass transfer
Freezing
Osmotic dehydration
a b s t r a c t
A successive mass and heat transfer modeling approach was developed to describe the Osmo-Dehydro-
Freezing (ODF) process. The dehydration process was described by an osmotic diffusion model based on
mass transfer through cellular membranes and the diffusion of different species through intercellular
spaces. Modeling of the freezing step was carried out by means of a numerical method, which includes
the phase change phenomena in thermal balance equations through the value of enthalpy. Thus, the
equations were solved in the whole system as if it were constituted by a single phase. As a result, the
balance was transformed into a transient problem with temperature dependent properties. The
simulation of the ODF process was successfully validated with experimental results obtained on
tomato cubes (R
2
0.9970.999; RMSE0.1560.252). The predicted temperature proles almost
coincided with experimental data. This was not the case with a selected, regularly used empirical
model, which was used for comparison reasons. The prediction deviations of the tested empirical model
were not only large, but they were constantly and signicantly increasing with increased
dehydration times.
& 2012 Elsevier Ltd. All rights reserved.
1. Introduction
The increase in consumer preference for minimally processed
fruits and vegetables has prompted researchers to study combined
methods as a preservation technique. The synergistic or additive
effect of combined inhibiting factors may permit the production of
food with improved quality over that preserved by only one techni-
que, e.g. heat treatment, dehydration or freezing (Bunger et al., 2004).
Freezing is a suitable way to preserve certain quality parameters
of foods, like color, avor, and nutrients. During freezing, part of the
aqueous fraction freezes out and forms ice crystals that damage the
integrity of the cellular compartments. The cellular membranes lose
their osmotic properties and their semi-permeability (Tregunno and
Goff, 1996). The metabolic activity of the plant tissue is interrupted,
inactivation of the enzymatic system occurs and the cell loses its
turgor. Besides the change in texture, bio-chemical deterioration
reactions are probable (Talens et al., 2003). Dehydration can be
applied as a pre-treatment before freezing, to remove part of the
water in order to facilitate the freezing process. As a result, damage
of the cellular membranes is minimal and better conservation of the
food properties is achieved.
In this sense, osmotic dehydration was reported as a highly
benecial pre-treatment (Forni et al., 1987; Robbers et al., 1997).
During osmotic processing, the product is in contact with a low
water activity solution (i.e. salt and/or sugar solutions), in an
oxygen free environment, resulting in protection from oxidizing
reactions. Under these conditions a two-way mass transfer is
established: (i) water is transferred from the product to the
solution, often accompanied by limited leaching of natural sub-
stances (i.e. sugars, vitamins, pigments, avors) (Lazarides et al.,
Contents lists available at SciVerse ScienceDirect
journal homepage: www.elsevier.com/locate/ces
Chemical Engineering Science
0009-2509/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ces.2012.07.023
n
Corresponding author. Tel./fax: 30 2310 991658.
E-mail address: athgou@agro.auth.gr (A.M. Goula).
Chemical Engineering Science 82 (2012) 5261
1997), and (ii) osmotic solute is transferred from the solution to
the product. While solute transfer is assumed to be of the
diffusional type, the fact that water loss is greater than solid gain
is attributed to a selective osmotic transport across semipermeable
cellular membranes (Spiazzi and Mascheroni, 1997). According to
previous studies, solute penetration is conned to extracellular
spaces (Marcotte, 1988). Isse and Schubert (1991) showed that
sucrose passes through the cell wall and accumulates between the
cell wall and the cellular membrane, where it forms a hypertonic
solution leading to water outux through the cellular membrane.
Other authors have suggested that water loss is greater than solute
gain only because of differences between the diffusion coefcient
of water and solute in the product (Karel, 1975). The existence of
these simultaneous and opposite mass uxes is one of the main
difculties arising when modeling mass transfer in osmotic dehy-
dration. In addition, these uxes take place in non-equilibrium
conditions and they are accompanied by product shrinkage and
deformation, as well as interactions between the two different
ows (Biswal and Le Maguer, 1989; Floury et al., 2008).
The process of freezing osmotically dehydrated foods is known as
Osmo-Dehydro-Freezing (ODF). Depending on process conditions,
up to 50% of initial moisture can be removed in liquid form, at
ambient temperatures within a fairly short process time (i.e. 23 h).
The drastic reduction in moisture content results in major process
and product benets, including (1) better preservation of structural
and quality characteristics; (2) lower cost of packaging, distribution,
and storage; and (3) large reduction in refrigeration load and energy
demands during freezing of the remaining water. Despite these
benets, industrial exploitation of ODF is still rather limited, due to
difculties regarding osmotic solution management as well as
modeling and predicting transport phenomena, a requirement for
effective process control (Agnelli et al., 2005).
Process simulation during ODF is a difcult task, as two
different transport phenomena are involved: (1) mass transfer of
solids and water between the product and the osmotic solution
during osmotic dehydration and (2) heat transfer with phase
change during freezing. The mass exchange prole in the dehydra-
tion step naturally affects thermophysical properties, with a strong
impact on heat transfer and the phase change prole during the
subsequent freezing step. A number of studies have tried to model
mass transfer during OD and different types of models have been
proposed (Park et al., 2002; Waliszewski et al., 2002; Giraldo et al.,
2003; Moreira and Murr, 2004; Ochoa-Martnez and Ayala-Aponte,
2005). Two main approaches can be identied: mechanistic models
based on Ficks law (Crank, 1975) and empirical or semi-empirical
models, such as those suggested by Raoult-Wack et al. (1991),
Azuara et al. (1992), Palou et al. (1994) and Parjoko et al. (1996);
and polynomial ttings (i.e. Rahman et al., 2001). On the other
hand, many studies have focused on modeling of the freezing
process (e.g. Cleland et al., 1984; Califano and Zaritzky, 1997;
Delgado and Sun, 2001; Pham, 2006; Wang et al., 2007).
However, limited research has been achieved with regard to
the process simulation during ODF (Agnelli et al., 2005). The
objective of this work was to develop a successive mass and heat
transfer modeling methodology to describe both transport phe-
nomena that are successively taking place during ODF. Such an
approach could prove useful to similar process combinations
involving heat and mass transfer.
2. Materials and methods
2.1. ODF experiments
Raw fresh tomato fruits (Cherokee variety) were purchased at
local markets and stored at 5 1C before use. Fresh tomatoes were
cut into 1.5 cm cubes. Solutions of 55 w/w% carbohydrate in
distilled water were prepared. The carbohydrates selected were
sucrose and 21 DE maltodextrin (Glucidex1, Roquette, France).
Osmotic treatment was conducted in 1 L containers placed in a
temperature controlled water bath (Memmert, Model WNE 7,
Memmert GmbH Co. KG, Schwabach FRG, Germany) at 3571 1C.
The ratio of food to osmotic medium was maintained at 1:25
(35 g of tomato sample:875 g of osmotic medium) in order to
ensure that the concentration of the osmotic solution did not
change signicantly during the experiment. During the 24 h
osmotic dehydration process, triplicate samples were taken after
0, 5, 15, 30, 60, 90, 120, 150, 180, 240, 300, 360, 480, 600, 840,
1080, and 1410 min. The samples were rinsed with distilled
water, blotted with absorbent paper and weighed. The evolution
of mass exchange was followed by monitoring water loss (WL)
and solid gain (SG)
WL
M
0
m
0
Mm
m
0
1
SG
mm
0
m
0
2
where M
0
is the initial mass of fresh tomatoes; M is the mass of
tomatoes after time t of osmotic treatment; m is the dry mass of
tomatoes after time t of osmotic treatment; and m
0
is the dry
mass of fresh tomatoes.
The dry mass of tomatoes was determined by drying in a
vacuum oven at 80 1C until consecutive weighings, made at 2-h
intervals, showed less than 0.3% variation.
After the dehydration step, the samples, suspended on a wire,
were frozen at 40 1C (Sanyo MIR 553, Sanyo Electric Co.,
Ora-Gun, Gunma, Japan). The temperature measurement was pro-
vided by a type T thermocouple (GG-TI-28, Omega Engineering,
Stamford, CT), 0.5 mm in diameter, with an accuracy of 70.1 1C.
All experiments were done in triplicate and average values
were used. Additional (parallel) samples were analyzed if a single
value deviated more than 2% from the triplicate mean.
2.2. Microstructure analysis
Samples of tomato cubes tissue, about 10 mm thick, were
obtained using a microtome. They were placed on a glass plate
and dehydrated for one day at room temperature. Samples were
poured onto microscope slides, stained with methylene blue
solution (1 w/v%), covered with glass cover slips, and observed
using an optical microscope (Axiolab A, Carl Zeiss GmbH, Jena,
Germany) with 40 and 100 objective lenses. Images of each
sample were analyzed with the public domain software Image J
v1.36b (http://rsb.info.nih.gov/ij/).
2.3. Mass transfer modeling
According to Floury et al. (2008), during osmotic treatments mass
transfer follows three different pathways: (a) transport through the
cellular volume (symplastic transport); (b) transport within the
extracellular volume (free-space transport); (c) transport across the
cell membrane (apoplastic transport). In this work, a diffusional
model was developed taking into account mass transfer through
cellular membranes and diffusion of different species through inter-
cellular spaces, based on the microstructural properties of the product
tissue. This model distinguishes between cellular, intercellular and
trans-membrane diffusion, and also taking volume changes into
proper account. Its application is not free of tting, but the number
of tted parameters was reduced to a reasonably selected minimum
A.M. Goula, H.N. Lazarides / Chemical Engineering Science 82 (2012) 5261 53
by use of structural data, correlations from the literature and well
founded estimations, wherever possible.
The mass balance for each species (i) (i w for water, i s/m for
sucrose/maltodextrin) within the product can be expressed as
@ V
c
c
i
int
_ _
@t
r V
c
D
int
ef f
rc
int
i
_ _
A
m
J
i
in the cellular volume 3
@ V
e
c
ext
i
_ _
@t
r
V
e
t
D
ext
ws=m
rc
ext
i
_ _
A
m
J
i
in the extracellular volume
4
where V
c
is the cellular volume per m
3
initial total volume; c
i
int
is
the cellular molar concentration of species (i); D
eff
int
is the effective
diffusion coefcient in the cellular volume; A
m
is the surface area
per m
3
initial total volume; J
i
is the molar ux of species (i) from
the cellular volume into the extracellular volume through the cell
membrane; V
e
is the extracellular volume per m
3
initial total
volume; c
i
ext
is the extracellular molar concentration of species (i);
t is the tortuosity factor; and D
ext
ws=m
is the binary diffusion
coefcient in the extracellular volume.
During osmotic dehydration, the cellular and extracellular
volumes change with both time and position. It was assumed
that all the cellular volume changes are due to the ux across the
membrane (Eq. (5)), whereas the extracellular volume was
considered proportional to the sum of sucrose/maltodextrin and
water volumes (Eq. (6))
@V
c
@t
A
m
v
w
J
w
v
s=m
J
s=m
_ _
5
V
e

c
ext
w
v
w
c
ext
s=m
v
s=m
c
ext
w0
v
w
c
ext
s=m0
v
s=m
V
e0
6
where v is the molar volume of water (v
w
1.810
5
m
3
mol
1
),
sucrose and maltodextrin (v
s
21.110
5
m
3
mol
1
, v
m
120.5
10
5
m
3
mol
1
).
Dimensional changes in the solid will affect concentration and
temperature gradients and this was taken into account by using a
grid based on the original position within the solid. The effect of
expansion was taken into account by dividing the gradients by a
linear expansion factor S (Floury et al., 2008):
S V
T
1=3
V
c
V
e

1=3
7
Thus, Eqs. (3) and (4) become
V
c
@c
int
i
@t

1
S
r
V
c
S
D
int
ef f
rc
int
i
_ _
A
m
J
i
c
int
i
@V
c
@t
8
V
e
@c
ext
i
@t

1
S
r
V
e
S
D
ext
ws=m
t
rc
ext
i
_ _
A
m
J
i
c
ext
i
@V
e
@t
9
The initial concentrations of species (i) in cellular and extra-
cellular volumes were assumed to be equal to
c
int
i,0
c
ext
i,0

x
i,0
r
M
i
10
where x
i
,
0
is the initial mass fraction of species (i), M
i
is the
molecular weight of species (i), and r is the initial density of the
tomato cube (1017 kg mol
1
).
The cellular volume fraction V
c
of the material was determined
by digitally measuring the area occupied by the void phase in the
image of the plane section. The darker regions in the micrograph
were mainly the cytoplamic membrane and the cell walls,
whereas the brighter regions were holes where cell contents
had been. The surface area A
m
was estimated based on the
average diameter of the cells, d, measured from the micrographs
(d83 mm):
A
m

6V
c
d
11
The resistance to mass transfer of sucrose or maltodextrin
from cell to cell is extremely large (Spiazzi and Mascheroni,
1997); thus, its effective diffusion coefcient in the cellular space
was considered negligible. The binary diffusion coefcient can be
estimated from the StokesEinstein equation (Eq. (12)), knowing
the apparent hydrodynamic radius and the viscosity of the
osmotic solution (Saravacos, 2005):
D
ws=m

k
B
T
6pr
B
n
12
where k
B
is Boltzmanns constant (1.3810
23
J mol
1
K
1
); r
B
is
the apparent hydrodynamic radius (4.910
10
m); and n is the
solution viscosity, which is evaluated using the following empiri-
cal equation (Chenlo et al., 2002; Floury et al., 2008):
n
995:20:1284c
s=m
_ _
Un
w
r
w
10:73
c
s=m
995:20:1284c
s=m
_
exp
c
s=m
=995:20:1284c
s=m

_ _
1:10
8:345 T=273:1
_ _
7:402
_ __
13
where c
s
/
m
is the molar concentration of sucrose or maltodextrin;
n
w
is the water viscosity; and r
w
is the water density.
The effective diffusion coefcient of water was calculated as
D
int
ef f

1
D
ws=m

A
m
15 10
7
_ _
1
14
The driving force for the apoplastic transport of solutes and
water across the cell membrane is the chemical potential differ-
ence between the solutions in the cellular and extracellular
spaces:
J
i
L
m,i
UDm
i
15
The macroscopic phenomenological coefcient L
m
,
i
, which
describes the permeability characteristics of the membrane to
solute (i), and the difference in chemical potential of water and
sucrose or maltodextrin across the membrane, Dm
i
, was calcu-
lated as
L
m,i
P
m,i
U~ c
i
16
Dm
i
RTDlna
w
17
where P
m
,
i
is a macroscopic permeability coefcient for species (i),
assumed to be concentration independent; ~ c
i
is the average
concentration of solute (i) across the membrane for which a
chemical potential difference is considered (Eq. (18)); and a
w
is
the water activity estimated in both the cellular and extracellular
spaces using the Norrish equation (Eq. (19)) (Floury et al., 2008):
~ c
i
V
c
UC
int
i
V
e
UC
ext
i
18
a
w
X
w
e
kX
s=m
2

19
where X
w
and X
s/m
are the mole fractions of water and sucrose or
maltodextrin, respectively, and k is a constant (k2.6).
Since there was no data for the permeability of the membrane
to sucrose and maltodextrin, their trans-membrane ux was
assumed to be a xed ratio of the water ux across the membrane
as follows:
J
s=m
r
s=m
v
w
v
s=m
J
w
20
where r
s/m
is the volume ux ratio for sucrose or maltodextrin.
The tortuosity factor, t, was dened in terms of the effective
path length (L
e
) that a migrating molecule has to travel on average
A.M. Goula, H.N. Lazarides / Chemical Engineering Science 82 (2012) 5261 54
in the interstitium compared with the distance covered normal to
the surface (L). This path was estimated from image processing of
microscopic views of tomato cells. A ruler available in the image
processing software was used to measure L
e
from micrographs.
There are two unknown parameters in the model; the mem-
brane permeability, P
m,w
, and the volume ux ratio, r
s/m
. Their
values, ranging from 5.610
15
to 5.610
7
mol N
1
s
1
and
from 0 to1, respectively (Maourel, 1997; Floury et al., 2008), were
determined by empirical curve tting in order to improve the
agreement between predicted and experimental data of WL and
SG. They were found equal to 5.610
10
mol N
1
s
1
(P
m,w
), 0.5
(r
s
), and 0.4 (r
m
). However, using these values the model did not
describe adequately the early part of the WL/SGt curves. This
may be due to the fact that the tortuosity has not been estimated
properly. Thus, its value was adjusted and it was found that a t of
1.1 gave a better t than the t of 1.8 derived from image analysis.
2.4. Heat transfer modeling
At the end of the dehydration process, the concentration
prole of each species inside the product was successively
introduced in the heat transfer balance equation in order to solve
the successive mass and heat transfer problem. The numerical
model was developed using the enthalpy balance equation. The
thermal balance was expressed by
r
@H
@t
r krT 21
where H is the product enthalpy and k is the thermal conductivity.
This equation was solved with the enthalpy as the unknown
variable on a xed grid. Consequently, the knowledge of H allows
calculating the value of the temperature in each point of the grid.
Then, the thermal conductivity and the position of the freezing
front can be deduced. This calculation approach presents the
advantage of reducing sensitivity to the abrupt variation of the
thermal properties within the phase change regime (Agnelli and
Mascheroni, 2001).
The thermal balance was solved using an explicit scheme. The
discretization of the equation was carried out by the nite volume
method in such a way that thermal uxes are conservative (Tocci
and Mascheroni, 1995). The following initial and boundary con-
ditions were used
T T
0
for t 0 22
h T
a
T
s
krT for z L=2 23
where h is the heat transfer coefcient; T
a
is the freezing air
temperature; T
s
is the temperature of the product surface; z is the
position variable; and L is the product thickness. The temperature
dependence of enthalpy was expressed by the correlations pro-
posed by Chang and Tao (1981). Specically, for temperatures
above freezing, the specic heat capacity is a weighed average
based on the composition of the unfrozen product
Cp x
w
UCp
w
x
s
UCp
s
24
where x
w
and x
s
are the mass fractions of water and solids,
respectively, and Cp
w
and Cp
s
are the specic heat capacities of
water and solids, respectively, which are functions of temperature
(Choi and Okos, 1986):
Cp
w
41760:091UT 0:00547UT
2
25
Cp
s
15491:962UT 0:00594UT
2
26
where T is in 1C and Cp is in J kg
1
K
1
.
For temperatures below freezing, enthalpy was calculated by
the following equations (Chang and Tao, 1981):
H H
f
aUT
r
1a T
r
b
_ _
27
H
f
9792:46405096Ux
w
28
T
r

T227:6
T
f
227:6
29
a 0:3620:0498U x
w
0:73 3:465U x
w
0:73
2
30
b 27:2129:04U a0:23 481:46U a0:23
2
31
T
f
287:5649:19Ux
w
37:07Ux
w
2
32
where H
f
is the freezing enthalpy (in J kg
1
); a and b are
constants; T
r
is a ratio of temperature differences; and T
f
is the
freezing point (in K). The unfrozen water in the product was
calculated by Eq. (32) (Moore, 1972)
l
R
1
T
f w

1
T
_ _
lnX
w
33
where l is the heat of fusion per mole of pure water
(6003 J mole
1
); R is the universal gas constant (8.314 J mole
1
K),
and T
fw
is the freezing point of pure water.
The conductivity and density of the mixtures were obtained by
equations in which the contributions of water, ice, and soluble
solids were considered (Heldman, 1982)
1
r

x
w
r
w

x
s
r
s

x
I
r
I
34
k r k
w
x
w
r
w
k
s
x
s
r
s
k
I
x
I
r
I
_ _
35
where r is the density; k is the thermal conductivity; x is the mass
fraction; and the subscripts w, s, and I refer to water, solids, and
ice, respectively. The temperature dependence of components
densities and thermal conductivities was expressed by the equa-
tions proposed by Choi and Okos (1986):
r
w
9970:00314UT0:00376UT
2
36
r
s
16000:366UT 37
r
I
9170:131UT 38
k
w
0:5710:00176UT6:70U10
6
UT
2
39
k
s
0:2010:00139UT4:33U10
6
UT
2
40
k
I
2:2190:00625UT 1:01U10
4
UT
2
41
where T is in 1C, r is in kg m
3
, and k is in W m
1
K
1
.
The heat transfer coefcient was experimentally determined
from the thermal history of an aluminum cube. The cube was
pierced at mid-height along a side in order to place a thermo-
couple in the center of the body. The h value was calculated
solving the heat transfer balance, assuming a constant tempera-
ture in the interior of the body. This is true due to the high
thermal conductivity for aluminum. Thus, the thermal ux of the
aluminum body is given by
m
al
Cp
al
dT
al
dt
hA
al
T
a
T
al
42
where m
al
is the mass of the aluminum body; Cp
al
is its heat
capacity; T
al
is its temperature; and A
al
is its surface area.
A.M. Goula, H.N. Lazarides / Chemical Engineering Science 82 (2012) 5261 55
Integrating, we obtain
ln
T
a
T
al
T
a
T
al0
lnT
n

hA
al
m
al
Cp
al
t 43
Plotting ln T* versus t and calculating the slope, the h value
was found to be equal to 47 W m
2
K
1
.
The model was implemented using a three-dimensional cubic
geometry. All the above equations are solved numerically using
the Matlab program. The mass balance was solved using an
explicit scheme. The discretization of the equation was carried
out by the nite volume method.
2.5. Empirical model
The temperature proles that were predicted by the developed
mechanistic model were compared to those obtained by separate
modeling of mass transfer during OD using a selected empirical
model. Many empirical models have been developed to describe
OD. According to Ochoa-Martinez et al. (2007), who compared
various empirical models using an extensive group of experi-
mental conditions from 25 studies, Azuaras model ts the mass
transfer kinetics data better than any other model. A similar
observation was reported by many other authors (Rahman and
Lamb, 1990; Kar and Gupta, 2001; Singh et al., 2007). According to
Azuaras model (Azuara et al., 1992), a mass balance for water is
given by
WL WL
p
M
w
m
44
where WL
p
is the water loss at equilibrium and M
w
m
is the mass of
water that is able to diffuse but remains inside the product at
time t. For constant temperature and osmotic solution concentra-
tion, the moisture loss is a function of the time and the amount of
water that can diffuse
WL s
1
UtUM
w
m
45
Substitution of Eq. (45) into Eq. (44) leads to Eq. (46), whereas,
in an analogous way, Eq. (47) is obtained for the solids gain:
WL
s
1
UtUWL
p
1s
1
Ut
46
SG
s
2
UtUSG
p
1s
2
Ut
47
where s
1
and s
2
are parameters that can be dened as relative rate
constants for water loss and solids gain, respectively. Eqs. (46)
and (47) can be linearized to yield:
t
WL

1
s
1
UWL
p

t
WL
p
48
t
SG

1
s
2
USG
p

t
SG
p
49
Eqs. (48) and (49) constitute the so-called Azuaras model.
2.6. Statistical analysis
In order to evaluate the accuracy of the proposed mechanistic
model, the predicted values of product temperature during
freezing for all ODF experiments were compared to the experi-
mental ones. Linear regression and paired t-test were applied. The
proposed model was evaluated by the correlation coefcient (R
2
)
and by the standard error of prediction (SE). However, SE was
adjusted for systematic difference (bias corrected) in RMSE, which
is the root mean square error.
3. Results and discussion
3.1. Concentration proles
Fig. 1 presents the predicted solute concentration proles
along half sample thickness (z
1/2
) for the specic process time
increments. These proles were obtained by mass balance calcu-
lations during the dehydration process. The x-axis represents
normalized distance (z/z
1/2
) from sample center. Lower solute
diffusion rates can be observed for maltodextrin solution (Fig. 1b)
compared to sucrose (Fig. 1a) for each and every process time.
This is due to higher solute size and it is in agreement with
ndings of previous studies (Lazarides et al., 1995). As it can be
seen, the external zones have higher (than average) contents of
solute; yet, within the given periods of contact, these zones did
not exactly reach equilibrium with the osmotic solution. The
solute content of inner zones increased with process time due to
diffusion from the external layers. Thus, the penetration of solutes
in the product during typically used periods of osmotic dehydra-
tion (about 34 h) is only important for the sub-surface zone.
Higher solid contents in the external zones act as a barrier
to further transport of solutes from the osmotic solution.
Thus, during the rst 3 h, the solute gain for the internal layers
0
10
20
30
40
50
60
0.0
C
o
n
c
e
n
t
r
a
t
i
o
n

(
%
)
60 min
180 min
300 min
480 min
1410 min
0
10
20
30
40
50
60
C
o
n
c
e
n
t
r
a
t
i
o
n

(
%
)
60 min
180 min
300 min
480 min
1410 min
z /z
1/2
z /z
1/2
0.2 0.4 0.6 0.8 1.0 1.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2
Fig. 1. Concentration proles of sucrose (a) and maltodextrin (b) along normalized sample thickness (z/z
1/2
) as a function of dehydration time in sucrose (a)
and maltodextrin (b) solution.
A.M. Goula, H.N. Lazarides / Chemical Engineering Science 82 (2012) 5261 56
(z/z
1/2
o0.5) is practically negligible, due to the low rate of mass
transfer by diffusion in the bulk of the solid.
The above observation is similar to that reported by Chenlo
et al. (2006, 2007), who developed a logistic doseresponse model
to describe the osmotic kinetics and to obtain the pseudo-
equilibrium moisture and solute content during osmotic dehy-
dration of chestnut. In addition, the proles presented in Fig. 1
can be compared to those obtained for apple halves treated in 50%
sucrose solutions (Monnerat et al., 2006). In that case, the data
analysis showed a quickly equilibrated layer close to the surface
and a constantly advancing disturbance front, dened by the
authors as an imaginary plane located between the outer zone,
which is affected by the mass transfer, and the inner zone,
without substantial concentration changes. Yao and Le Maguer
(1997) also observed the sucrose front moving as a sharp
concentration ridge, which was more pronounced in proles
encountered with high solute concentrations (50%). It is possi-
ble that the high concentrations caused additional structural
changes due to the shrinking tissue. The conjunction of these
factors leads to higher effective diffusion coefcients in the
surface region and lower coefcients in the inner region, an
inverse behavior when compared to that occurring in the diffu-
sivity in pure sucrosewater solutions (Henrion, 1964).
The experimental (means of three replicates) and calculated
values of water loss and solid gain during osmotic dehydration of
tomato cubes are given in Figs. 2 and 3. The repeatability for the
WL and SG measurements, expressed as the average standard
deviation of all triplicate calculations, were 0.67% and 0.28% for
WL and SG, respectively. The WL and SG considerably increased
with increasing contact time. High rates of water removal and
solids uptake during the rst hours were followed by slower rates
in later stages. This may be due to the progressive decrease in the
driving potentials for moisture and solute transfer as a result of
extensive mass exchange within the rst hours. In addition,
progressive solids uptake may result in the formation of a high
solids sub-surface layer, which interferes with the concentration
gradient across the productsolution interface, acting as a barrier
against further removal of water and uptake of solids. The water
loss in the rst six hours was 37% and 46% for solutions of sucrose
and maltodextrin, respectively; these values are much lower than
the ones reported by other researchers (Andrade et al., 2007).
Equilibrium was practically reached after 10 and 18 h of osmotic
processing for sucrose and maltodextrin, respectively. Beyond
equilibrium, water loss and solids gain remained literally con-
stant. Askar et al. (1996) and Panagiotou et al. (1999) reported
shorter equilibrium times (4 h) in osmotic dehydration of
several fruits and vegetables. This difference may be due to the
type of tomato membrane, which is characterized as differentially
permeable and not semipermeable (Andrade et al., 2007). In
addition, the osmotic dehydration in sucrose, which has a lower
molecular weight, caused higher solids gain and lower water loss
compared to OD in maltodextrin. This observation may be
attributed to the fact that solutes of low molecular weight favor
solute impregnation rather than water loss (Lazarides et al., 1995;
Dermesonlouoglu et al., 2007).
0
10
20
30
40
50
60
0
Osmotic dehydration time (min)
sucrose
maltodextrin
mechanistic model
empirical model
W
L

(
%
)
300 600 900 1200 1500
Fig. 2. Water loss (WL) as a function of dehydration time in sucrose and
maltodextrin solution; square symbols correspond to experimental values, while
lines correspond to predicted values using two different models (a proposed
mechanistic and a selected empirical).
0
2
4
6
8
10
12
0
Osmotic dehydration time (min)
sucrose
maltodextrin
mechanistic model
empirical model
S
G

(
%
)
300 600 900 1200 1500
Fig. 3. Solids gain (SG) as a function of dehydration time in sucrose and
maltodextrin solution; square symbols correspond to experimental values, while
lines correspond to predicted values using two different models (a proposed
mechanistic and a selected empirical).
0
20
40
60
80
100
120
140
160
0
WL -sucrose
WL -maltodextrin
SG -sucrose
SG -maltodextrin
t
/
W
L

o
r

t
/
S
G
Osmotic dehydration time, t (min)
300 600 900 1200 1500
Fig. 4. Linear plots of Azuaras model during osmotic dehydration in sucrose and
maltodextrin solutions.
A.M. Goula, H.N. Lazarides / Chemical Engineering Science 82 (2012) 5261 57
Figs. 2 and 3 show that there is a close agreement between
experimental and predicted values of WL and SG by both the
proposed mechanistic and the selected empirical model. The
determination coefcient values were in the range of 0.976
0.997 showing a good t of both models to the experimental
data. The lack of t test applied to the replicated data showed that
the models adequately described the moisture loss and solid gain
in the tomato cubes. The L-values, which were calculated from
independent and dependent estimators of residual standard
deviations for both moisture loss and solid gain, were less than
the upper tail cut-off value obtained from F-distribution, con-
rming the validity of the proposed models for an osmotic
dehydration process.
The values of t/WL and t/SG as calculated from the moisture
loss and solid gain data for the two different carbohydrate
solutions were plotted against time (Fig. 4). A linear trend was
observed in all cases. Straight lines were, therefore, tted by
performing linear regression and the parameter values were
determined. The values of s
1
and WL
N
for moisture loss kinetics
and of s
2
and SG
N
for solid gain kinetics are given in Table 1.
The residual analysis presented in Fig. 5 clearly shows that, for
most of the investigated processing times, much smaller differ-
ences between predicted and experimental values were observed
for the proposed mechanistic model compared to the empirical
one. This can also be veried through the statistical parameters R
2
and RMSE presented in Table 1. Generally speaking, the selected
empirical model seems to be quite accurate for certain processing
times, but it is not accurate enough over the entire range of
processing times. This may be the case because the empirical
model does not take into account the size, shape, and structure of
the material and it does not consider crucial process variables,
such as temperature, concentration, solution-to-fruit mass ratio
or agitation.
3.2. Temperature proles
As mentioned earlier, the concentration prole of each species
inside the product was successively introduced into the heat
transfer balance equation. Fig. 6 presents the experimental
(means of three replicates) and predicted temperature proles
of the product center during freezing of osmo-dehydrated tomato
cubes for specic dehydration times. The repeatability for the
temperature measurement, expressed as the average standard
deviation of all triplicate calculations, was 0.7 1C. As it can be
seen, each temperature curve presented three stages. The rst
stage corresponds to a rapid temperature decrease due to the
removal of sensible heat. This was followed by a period of much
slower temperature decrease, attributed to prevailing latent heat
removal for water crystallization. According to Wang et al. (2007),
the heat transfer effect mainly contributes to ice formation with a
minor simultaneous impact on temperature change. After the
center is frozen, there is no more latent heat to be released, the
apparent specic heat decreases, the coefcient of heat conduc-
tivity increases, and the center temperature decreases rapidly,
due to the removal of sensible heat once again.
The kinetic curves of the temperature change in the product
center predicted by the specic mechanistic model are in good
agreement with the experimental results. R
2
and RMSE ranged
from 0.997 to 0.999 and from 0.156 to 0.252, respectively. In
addition, the proposed model predicts the freezing times more
accurately than the approximate solution methods usually used
Table 1
Parameters of the proposed mechanistic and the selected empirical model tted to experimental data of osmotic dehydration in sucrose and maltodextrin solutions.
Parameter Sucrose solution Maltodextrin solution
Water loss Solid gain Water loss Solid gain
Mechanistic Empirical Mechanistic Empirical Mechanistic Empirical Mechanistic Empirical
R
2
0.996 0.985 0.995 0.991 0.999 0.994 0.997 0.992
RMSE 0.313 0.489 0.086 0.091 0.102 0.372 0.048 0.072
WL
N
or SG
N
(%) 48.077 11.947 57.804 10.152
s
1
or s
2
(min
1
) 0.0084 0.0085 0.0092 0.0086
-3
-2
-1
0
1
2
0
R
e
s
i
d
u
a
l


(
%
)
Osmotic dehydration time (min)
WL -mechanistic model
WL -empirical model
SG -mechanistic model
SG -empirical model
-3
-2
-1
0
1
2
0
R
e
s
i
d
u
a
l


(
%
)
Osmotic dehydration time (min)
WL -mechanistic model
WL -empirical model
SG -mechanistic model
SG -empirical model
300 600 900 1200 1500
300 600 900 1200 1500
Fig. 5. Residuals between predicted (by the proposed mechanistic and the
selected empirical model) and experimental values of water loss (WL) and solids
gain (SG) during osmotic dehydration in sucrose (a) and maltodextrin (b) solution.
A.M. Goula, H.N. Lazarides / Chemical Engineering Science 82 (2012) 5261 58
to simulate freezing processes (Pham, 1996; Sanz et al., 1996,
1999; Lopez-Leiva and Hallstrom, 2003). This can be attributed to
the fact that approximate methods are based on revised Plank
equations; therefore they present three main disadvantages.
Firstly, they assume the existence of a two-phase interface, which
is, time-dependent and can be characterized as a moving
boundary zone. Secondly, the temperatures in approximate
solution methods are often average values; and thirdly, these
methods do not consider that freezing mainly occurs within a
short time-zone below the ice point and latent heat is mostly
released during this period.
In addition, as it can be seen in Fig. 6, the differences between
predicted and experimental data were much smaller when
modeling was based on the proposed mechanistic model, com-
pared to the empirical Azuaras model. This can also be veried
through the statistical parameters R
2
and RMSE, which ranged
from 0.997 to 0.999 and from 0.156 to 0.252, respectively, for the
mechanistic model, and from 0.875 to 0.952 and from 1.197 to
2.066, respectively, for the empirical model. As it follows from
Fig. 6(a)(c), the deviation between predicted and experimental
values for Azuaras model not only large, but it is constantly and
signicantly increasing with increased dehydration times. Thus,
the predictions of Azuaras model for WL and SG cannot be used to
accurately describe the freezing process following the osmotic
drying step.
According to Ispir and Togrul (2009), the empirical and semi-
empirical models may give a reasonable t to the experimental
data, but their use is quite limited, because they are strongly
dependent on the specic experimental conditions under which
they were produced. On the other hand, mechanistic models that
are strictly based on the diffusion approach use a number of
assumptions, which are difcult to fulll (Kaymak-Ertekin and
Sultanoglu, 2000), and effective diffusivity becomes an adjustable
kinetic parameter that strongly depends on the experimental
conditions and the physical properties of the product (Salvatori
et al., 1999). In addition, during osmotic dehydration, the cellular
and extracellular volumes change with both time and position
and these dimensional changes affect signicantly both concen-
tration and temperature gradients (Spiazzi and Mascheroni, 1997;
Li, 2006). In order to overcome these limitations, the proposed
mechanistic model is based on a cellular physiology approach
calculating the evolution of cellular and extracellular volumes and
the resulting (different) concentration values for the two volumes
(cellular and extracellular). For example, during dehydration in
sucrose solution for 60 min, in a region close to product/solution
interface, the cellular volume decreased to 54% its initial value,
while the extracellular volume increased by 32%. After 240 min,
these values were 16% and 67%, respectively, while in a region
closer to the center the respective values were 30% and 58%. After
about 600 min, conditions close to equilibrium were obtained; in
almost all regions of the cube, the cellular volume reaches a
minimum value (9% of the initial), while the extracellular volume
doubles. As a result the percentage of cellular volume with
respect to the total volume reduces to 29%, from an initial value
of 80% in the fresh product. The accurate calculation of volume
and concentration changes for both cellular and extracellular
volumes naturally leads to an accurate prediction of temperature
gradients (Floury et al., 2008). On the contrary, using Azuaras
model, a at concentration prole for water and solids is assumed
and the difference between cellular and extracellular volumes is
not considered. Due to these limitations, despite its accurate WL
and SG predictions, Azuaras model seems unsuitable for simulat-
ing transport phenomena in a successive mass and heat transfer
methodology.
-40
-30
-20
-10
0
10
20
30
0
Freezing time (min)
sucrose
maltodextrin
mechanistic model
empirical model
sucrose
maltodextrin
mechanistic model
empirical model
sucrose
maltodextrin
mechanistic model
empirical model
T

(

C
)
-40
-30
-20
-10
0
10
20
30
T

(

C
)
-40
-30
-20
-10
10
0
20
30
2 4 6 8 10 12 14 16 0
Freezing time (min)
2 4 6 8 10 12 14 16
0
Freezing time (min)
2 4 6 8 10 12 14 16
T

(

C
)
Fig. 6. Experimental (symbols) and model predicted (lines) center temperature proles during freezing of tomato cubes after osmotic dehydration for 60 min (a), 300 min
(b), and 1410 min (c) in sucrose and maltodextrin solutions.
A.M. Goula, H.N. Lazarides / Chemical Engineering Science 82 (2012) 5261 59
Linear regression and paired t-test were applied in order to
compare the experimental and simulated (by the proposed
mechanistic model) values of product temperature during freez-
ing for all ODF experiments. When calculating the slope and
intercept of simulated versus experimental values, no statistical
differences were found from the respective theoretical values of
1.00 and 0.00 (p0.05). In addition, the calculated t-values were
lower than the theoretical t-values (p0.05). Therefore the null
hypothesis was retained: the simulation data on product tem-
perature agreed well with the experimental data conrming the
validity of the proposed model for the ODF process.
4. Conclusions
A new modeling methodology has been developed to describe
the ODF process based on the successive mass and heat
transfer.
The proposed mechanistic model is based on a cellular
physiology approach, thus overcoming major limitations asso-
ciated with both empirical models and mechanistic models
strictly based on diffusion.
The proposed model allowed accurate prediction of tempera-
ture and phase changes as well as solute and water concen-
trations as a function of processing (dehydration or
freezing) time.
The prediction capacity of the proposed model was compared
to that of a selected, regularly used empirical model (Azuaras
model). The latter gave large deviations between predicted
and experimental values, which were signicantly increasing
with increased dehydration times; thus, the tested empirical
model does not appear to be suitable for modeling ODF
processes.
Accurate, successive mass and heat transfer simulation of the
ODF process allows for effective process control, an absolute
requirement for industrial exploitation in the production of
high quality, low cost dehydrofrozen food products.
References
Andrade, S.A.C., Neto, B.B., Nobrega, A.C., Azoubel, P.M., Guerra, N.B., 2007.
Evaluation of water and sucrose diffusion coefcients during osmotic dehy-
dration of Jenipapo (Genipa americana L.). J. Food Eng. 78, 551555.
Agnelli, M.E., Marani, C.M., Mascheroni, R.H., 2005. Modelling of heat and mass
transfer during (osmo) dehydrofreezing of fruits. J. Food Eng. 69, 415424.
Agnelli, M.E., Mascheroni, R.H., 2001. Cryomechanical freezing. A model for the
heat transfer process. J. Food Eng. 47, 263270.
Askar, A., Abdel-Fadeel, M.G., Ghonaim, S.M., AbdeL-Gaid, I.O., Ali, A.M., 1996.
Osmotic and solar dehydration of peach fruits. Fruit Process. 9 (1), 258262.
Azuara, E., Beristain, C.I., Garcia, H.S., 1992. Development of a mathematical model
to predict kinetics of osmotic dehydration. J. Food Sci. Technol. 29, 239242.
Biswal, R.N., Le Maguer, M., 1989. Mass transfer in plant material in contact with
aqueous solutions of ethanol and sodium chloride equilibrium data. J. Food
Process. Eng. 11, 159176.
Bunger, A., Moyano, P.C., Vega, R.E., Guerrero, P., Osorio, F., 2004. Osmotic
dehydration and freezing as combined processes on apple preservation. Food
Sci. Technol. Int. 10 (3), 163170.
Califano, A.N., Zaritzky, N.E., 1997. Simulation of freezing or thawing heat
conduction in irregular two dimensional domains by a boundary tted grid
method. Lebens. Wiss. Technol. 30, 7076.
Chang, H.D., Tao, L.C., 1981. Correlations of enthalpies of food systems. J. Food Sci.
46, 14931497.
Chenlo, F., Moreira, R., Pereira, G., Ampudia, A., 2002. Viscosities of aqueous
solutions of sucrose and sodium chloride of interest in osmotic dehydration
processes. J. Food Eng. 54, 347352.
Chenlo, F., Moreira, R., Fernandez-Herrero, C., Vazquez, G., 2006. Experimental
results and modeling of the osmotic dehydration kinetics of chestnut with
glucose solutions. J. Food Eng. 74, 324334.
Chenlo, F., Moreira, R., Fernandez-Herrero, C., Vazquez, G., 2007. Osmotic dehy-
dration of chestnut with sucrose: mass transfer processes and global kinetics
modelling. J. Food Eng. 78, 765774.
Choi, Y., Okos, M.R., 1986. Effect of temperature and composition on thermal
properties of foods. In: Le Maguer, M., Jelen, P. (Eds.), Food Engineering and
Process Applications. Transport Phenomena. Elsevier, New York.
Cleland, D.J., Cleland, A.C., Earle, R.W., Byrne, S.J., 1984. Prediction of rates of
freezing, thawing and cooling in solids of arbitrary shape using the nite
element method. Int. J. Refrig. 7, 613.
Crank, J., 1975. The Mathematics of Diffusion. Clarendon Press, Oxford.
Delgado, A.E., Sun, D.W., 2001. Heat and mass transfer models for predicting
freezing processesa review. J. Food Eng. 47, 157174.
Dermesonlouoglu, E.K., Giannakourou, M.C., Taoukis, P., 2007. Stability of dehy-
drofrozen tomatoes pretreated with alternatives osmotic solutes. J. Food Eng.
78, 272280.
Floury, J., Le Bail, A., Pham, Q.T., 2008. A three-dimensional numerical simulation
of the osmotic dehydration of mango and effect of freezing on the mass
transfer rates. J. Food Eng. 85, 111.
Forni, E., Torreggiani, K., Crivelli, G., Maestrelle, A., Bertolo, G., Santelli, F., 1987.
Inuence of osmosis time on the quality of dehydrofrozen Kiwi fruit. Acta
Hortic. 282, 425434.
Giraldo, G., Talens, P., Fito, P., Chiralt, A., 2003. Inuence of sucrose solution
concentration on kinetics and yield during osmotic dehydration of mango. J.
Food Eng. 58, 3343.
Heldman, D.R., 1982. Food properties during freezing. Food Technol. 36 (2), 92.
Henrion, P.N., 1964. Diffusion in the sucrosewater system. Trans. Faraday Soc.
60, 7274.
Ispir, A., Togrul, I.T., 2009. Osmotic dehydration of apricot: kinetics and the effect
of process parameters. Chem. Eng. Res. Des. 87, 166180.
Isse, M.G., & Schubert, H. (1991). Osmotic dehydration of mango: mass transfer
between mango and syrup. In: Fourth World Congress of Chemical Engineer-
ing, Karlsruhe, Germany.
Kar, A., Gupta, D.K., 2001. Osmotic dehydration characteristics of button mush-
rooms. J. Food Sci. Technol. 38, 352357.
Karel, M., 1975. Osmotic drying. In: Fennema, O. (Ed.), Principles of Food Science,
Part II. Marcel Dekker, New York and Basel.
Kaymak-Ertekin, F., Sultanoglu, M., 2000. Modelling of mass transfer during
osmotic dehydration of apples. J. Food Eng. 46, 243250.
Lazarides, H.N., Katsanidis, E., Nicolaidis, A., 1995. Mass transfer kinetics during
osmotic preconcentration aiming at minimal solid uptake. J. Food Eng. 25,
151166.
Lazarides, H.N., Gekas, V., Mavroudis, N., 1997. Apparent mass diffusivities in fruit
and vegetable tissues undergoing osmotic processing. J. Food Eng. 31,
315324.
Li, L., 2006. Numerical simulation of mass transfer during the osmotic dehydration
of biological tissues. Comput. Mater. Sci. 35, 7583.
Lopez-Leiva, M., Hallstrom, B., 2003. The original Plank equation and its use in the
development of food freezing rate predictions. J. Food Eng. 58, 267275.
Maourel, C., 1997. Aquaporins and water permeability of plant membranes. Annu.
Rev. Plant Physiol. Plant Mol. Biol. 48, 399429.
Marcotte, M., 1988. Mass Transfer Phenomena in Osmotic Processes. Experimental
Measurements and Theoretical Considerations. Ph.D. Thesis, University of Alberta.
Monnerat, S.M., Pizzi, T.R.M., Mauro, M.A., Menegalli, F.C., 2006. Concentration
proles and effective diffusion coefcients of sucrose and water in osmo-
dehydrated apples. Food Res. Int. 39, 739748.
Moore, W.J., 1972. Physical Chemistry. Prentice-Hall, Englewood Cliffs, NJ.
Moreira, P., Murr, F., 2004. Mass transfer kinetics of osmotic dehydration of cherry
tomato. J. Food Eng. 61, 291295.
Ochoa-Martnez, C.I., Ayala-Aponte, A.A., 2005. Modelos matematicos de transfer-
encia de masa en deshidratacion osmotica. Cienc. Tecnol. Aliment. 4 (5),
330342.
Ochoa-Martinez, C.I., Ramaswamy, H.S., Ayala-Aponte, A.A., 2007. A comparison of
some mathematical models used for the prediction of mass transfer kinetics in
osmotic dehydration of fruits. Dry Technol. 25, 16131620.
Palou, E., Lopez-Malo, A., Argaiz, A., Welti, J., 1994. The use of Pelegs equation to
model osmotic concentration of papaya. Dry Technol. 12, 965978.
Panagiotou, N.M., Karathanos, V.T., Maroulis, Z.B., 1999. Effect of osmotic agent on
osmotic dehydration of fruits. Dry Technol. 17 (1/2), 175189.
Parjoko, K.A., Rahman, M.S., Buckle, K.A., Perera, C.O., 1996. Osmotic dehydration
kinetics of pineapple wedges using palm sugar. Lebens. Wiss. Technol. 29,
452459.
Park, K.J., Bin, A., Brod, F.P.R., Park, T.H.K.B., 2002. Osmotic dehydration kinetics of
pear Danjou. J. Food Eng. 52, 293298.
Pham, Q.T., 1996. Prediction of calorimetric properties and freezing time of foods
form composition data. J. Food Eng. 30 (12), 267275.
Pham, Q.T., 2006. Modelling heat and mass transfer in frozen foods: a review. Int.
J. Refrig. 29, 876888.
Rahman, M.S., Lamb, J., 1990. Osmotic dehydration of pineapple. J. Food Sci.
Technol. 27, 150152.
Rahman, M.S., Sablani, S.S., Al-Ibrahin, M.A., 2001. Osmotic dehydration of potato:
equilibrium kinetics. Dry Technol. 19, 11631176.
Raoult-Wack, A., Guilbert, S., Le Maguer, M., Rios, G., 1991. Simultaneous water
and solute transport in shrinking media. Part 1. Application to dewatering and
impregnation soaking process analysis. Dry Technol. 9, 589612.
Robbers, M., Singh, R.P., Cunha, L.M., 1997. Osmotic-convective dehydrofreezing
process for drying Kiwifruit. J. Food Sci. 62 (5), 10391047.
Salvatori, D., Andre s, A., Chiralt, A., Fito, P., 1999. Osmotic dehydration progression
in apple tissue I: spatial distribution of solutes and moisture content. J. Food
Eng. 42 (3), 125132.
A.M. Goula, H.N. Lazarides / Chemical Engineering Science 82 (2012) 5261 60
Sanz, P.D., Ramos, M., Mascheroni, R.H., 1996. Using equivalent volumetric
enthalpy variation to determine the freezing time in foods. J. Food Eng. 27
(2), 177190.
Sanz, P.D., Ramos, M., Aguirre-Puente, J., 1999. One-stage model of foods freezing.
J. Food Eng. 40 (4), 233239.
Saravacos, G.D., 2005. Mass transfer properties of foods. In: Rao, M.A., Rizvi, S.S.H.,
Datta, A.K. (Eds.), Engineering Properties of Foods, 3rd ed. CRC Press, Taylor
and Francis Group.
Singh, B., Kumar, A., Gupta, A.K., 2007. Study of mass transfer kinetics and effective
diffusivity during osmotic dehydration of carrot cubes. J. Food Eng. 79,
471480.
Spiazzi, E., Mascheroni, R., 1997. Mass transfer model for osmotic dehydration of
fruits and vegetablesI. Development of the simulation model. J. Food Eng. 34,
387410.
Talens, P., Escriche, I., Martnez-Navarrete, N., Chiralt, A., 2003. Inuence of
osmotic dehydration and freezing on the volatile prole of Kiwi fruit. Food
Res. Int. 36, 635642.
Tocci, A.M., Mascheroni, R.H., 1995. Numerical models for the simulation of the
simultaneous heat and mass transfer during food freezing and storage. Int.
Commun. Heat Mass Transfer 22 (2), 251260.
Tregunno, N.B., Goff, H.D., 1996. Osmodehydrofreezing of apples: structural and
textural effects. Food Res. Int. 29, 471479.
Waliszewski, K.N., Delgado, J.L., Garca, M.A., 2002. Equilibrium concentration and
water and sucrose diffusivity in osmotic dehydration of pineapple slabs. Dry
Technol. 20 (2), 527538.
Wang, Z., Wu, H., Zhao, G., Liao, X., Chen, F., Wu, J., Hu, X., 2007. One-dimensional
nite difference modeling on temperature history and freezing time of
individual food. J. Food Eng. 79, 502510.
Yao, Z., Le Maguer, M., 1997. Mathematical modelling and simulation of mass
transfer in osmotic dehydration processes. Part II: simulation and model
verication. J. Food Eng. 32, 2132.
A.M. Goula, H.N. Lazarides / Chemical Engineering Science 82 (2012) 5261 61

Вам также может понравиться