Journal of Crystal Growth 198/199 (1999) 1294—1298

Computer modeling of sugar crystallization during drying

of thin sugar films
E. Ben-Yoseph, R.W. Hartel*
Department of Food Science, University of Wisconsin, 1605 Linden Driven, Madison, WI 53706, USA

Abstract

A computer model was developed to predict drying and crystallization during processing of thin sugar films under
the conditions of importance for sugar-coated cereals. The three-dimensional model analyzed simultaneous heat and
moisture transfer, accounting for shrinkage, with temperature and moisture-dependent transport properties. Crystal
growth was evaluated from crystallization kinetics based on supersaturation and temperature. The influence of growth
on the solution concentration and drying were taken into account. Concentration profiles around the growing crystals
during drying of the thin films were evaluated. The model predicts a concentration reduction around the growing
crystals, and film temperatures that are higher than the drying temperature (as a result of the release of fusion heat).
A phenomenon of case hardening was observed for dried films with a high density of seeds on surface.  1999 Elsevier
Science B.V. All rights reserved.

Keywords: Crystallization; Drying; Sucrose; Sugars; Modeling; Films

1. Introduction parameters that influence these processes, for devel-

Controlling crystallization of sugars in thin films
is critical to development of the proper appearance
of sugar-coated cereal products. In this process,
a thin sugar film is applied to the surface of a cereal
piece, and then allowed to dry. By proper control of
the crystallization and drying processes, the appro-
priate appearance, either frosted or glazed, can
be developed. However, our understanding of the
* Corresponding author. Fax: #1 608 263 1965; e-mail:
hartel@calshp.cals.wisc.edu.
oping a specific appearance, is severely lacking.
Cereal companies have problems getting consistent
results with this process. Usually, the film is frosty
at the end of the process, which is the desired
appearance. However, sometimes the appearance is
glossy, which is undesired.
The objective of this work was to develop a com-
puter model for predicting crystallization kinetics
during drying of thin sugar films under the condi-
tions of importance for sugar-coated cereals. This
includes developing a three-dimensional model
that evaluates changes in all dimensions within the
film, evaluating the concentration profile around

0022-0248/99/$ — see front matter  1999 Elsevier Science B.V. All rights reserved.
PII: S 0 0 2 2 - 0 2 4 8 ( 9 8 ) 0 1 0 1 4 - 8
 E. Ben-Yoseph, R.W. Hartel / Journal of Crystal Growth 198/199 (1999) 1294–1298 1295

the growing crystals, and accounting for crystals z are dimensionless depth into the box in the x, y,
growing together. The simulation program deter- and z directions, respectively, defined as
mined the crystal size, shape and distribution in the
X ½ Z
film. Using this information, the final appearance of x" , y" , z" (3)
X ½ Z
the film, either frosty or glassy, will be predicted in   
the next phase of the study. where X, ½, and Z are the depth into the box in the
x, y, and z directions, respectively, and X , ½ , and
 
Z are the thickness of the box in the x, y, and

2. Theory, modeling and experimental procedure z directions, respectively, as a function of time. D ,
V
D and D are the mutual diffusion coefficients in
W X
2.1. Computer simulation of drying process water-sucrose system in the x, y, and z directions,
respectively. S , S , and S , are the shrinkage magni-
V W X
Okazaki et al. [1] presented the mass transfer tude of the box in the x, y, and z directions.
equation in a one-dimensional, two component sys- The initial conditions describe the temperature
tem with shrinkage, assuming all moisture move- and concentration of the solution at the beginning
ment was by diffusion. This equation was expanded of the process:
to three dimensions, since crystal growth is a
at t"0 for all i, j, or k: C "1, (4)
three-dimensional phenomenon, to get GHI
at t"0: S"1,¹"¹ , (5)

    
jC j D jC x jS jC
" V # V where i, j, and k are subscripts which refer to nodes
jt jx Xj x S j t jx
R V in the grid point system in the x, y, and z directions.

   
2D 1 jC  The boundary conditions supply information on
! V
C#C SX j x external conditions surrounding the solution,
V R which affect the behavior of the surface solution

    
j D jC y jS jC temperature and heat and mass transfer rates dur-
# W # W
jy ½j y S j t jy ing drying. The boundary conditions used were
R W

   
2D 1 jC  for z"0 and t'0:
! W
C#C S½ j y jC
W R "0 (6)
jz
    
j D jC z jS jC
# X # X
jz Zj z S j t jz for z"1 and t'0:
R X

   
2D 1 jC  ¼ D o jC
X , (1) ! X "k (P !P ) (7)
!
C#C SZ j z (C#C)SZ jz E  
X R 
where C is the dimensionless moisture in film and dS k
!o "  (P !P ) (8)
t is time. C, C and ¼ are defined as follows:  dt Z  

[m c #c Z X ½ (o !o )
C"
¼ !¼
  , C "
¼#¼
 ,        
¼ !¼
 
¼ !¼
  d¹
#c Z X ½ (o !o )#c Z X ½ o S ]
           dt
o
¼"  , (2)
o!o dS
  "h X ½ (¹ !¹)#r o Z X ½ , (9)
        dt
where, ¼ is the mass fraction of water, ¼ is
 
the equilibrium water mass fraction, ¼ is the where, o is the solution density, k is the mass
 
initial water mass fraction, o is the density of pure transfer coefficient at the solution surface, P is
 
water, o is the density of pure sucrose. x, y, and the partial pressure of water vapor at film surface,

1296 E. Ben-Yoseph, R.W. Hartel / Journal of Crystal Growth 198/199 (1999) 1294–1298
P is the partial pressure of water vapor in hot air,
 
m is the mass of the flake, c is the heat capacity of
 
the flake, c is the heat capacity of sucrose, c is the
 
heat capacity of water, o is the initial density of

solution, o is the water density at the initial

solution temperature, r is the latent heat of vapor-
 
ization of water, h is the heat transfer coefficient at
 where w and w are weight fractions of component
film surface, ¹ is the solution temperature and ¹ is
the drying air temperature.
Calculating the temperature only as a function
of time, and not as function of location, we assume
that the temperature distribution is negligibly
small because of the film being very thin and there
is no convection flow in the film at this high visco-
sity of the solution. Furthermore, the thermal
diffusivity of sucrose/water system in high concen-
tration of sucrose, as given by Bhowmik and
Hayakawa [2] and by Hayakawa and Bakal [3], is
3 fold higher than the mass diffusivity of the
system.
The resulting system of nonlinear partial differ-
ential equations in time and three dimensions were
solved numerically by applying the finite difference
method. In the classical finite difference applica-
tion, the distance between the nodes is constant.
However, since shrinkage is not negligible during
drying (the film shrinks when dried at a rate de-
pending upon the solution concentration and the
degree of drying) the physical dimension of the film
is changing during drying. Therefore, the variable
grid central finite difference method was applied
[4,5].
The physical parameters used during calcula-
tion were assumed to be functions of local con-
centration and/or temperature. These parameters
were defined explicitly from several literature
sources.
2.2. Predicting the state of solution, and crystal
growth rate, size, shape and distribution
The information about temperature and concen-
tration at any location in the film during the drying
process located each point in the film in the phase
diagram; either it is an undersaturated, metastable,
labile or glassy solution.
A region in the film was considered glassy if its
temperature reached below the glass transition
temperature, ¹ , during the drying process. Roos
and Karel [6] gave a method to calculate ¹ for

sucrose by applying the Gordon and Taylor equa-
tion:
w ¹ #kw ¹
¹ "    , (10)
w #kw
 
 
compounds, ¹ and ¹ are the absolute glass
 
transition temperatures (K) of the component
compounds, and k is a constant. In our case,
¹ "138 K (¹ of water), ¹ "335 K (¹ of pure
   
sucrose), and k for sucrose—water equal to 4.7.
For supersaturated, non-glassy solution, the
model calculated the supersaturation level as
a function of the local concentration and the satu-
ration concentration (which is a function of temper-
ature). The growth rate of sugar crystals as a
function of the supersaturation level and temper-
ature was calculated from a sucrose growth model
based on data given by Hartel et al. [7]. Regression
from this data (R"0.9697), gave the following
equation, valid for temperature range of 44—54°C
and supersaturation ratio above 1.0576:
G"7.0538S!7.46, (11)
where, G is the growth rate for a single crystal
(lm/min) and S is the supersaturation ratio (con-
centrations of grams sucrose per grams of water).
2.3. Evaluation of the crystal growth influence
on the film
All the following phenomena were taken into
account in the simulation program:
(1) Reduction in concentration: when a crystal
grew, sucrose was removed from the solution and
the concentration around the crystal decreased.
The change in concentration was calculated from
mass balance of sucrose and water.
(2) Generation of fusion heat: when a crystal
grew, it generated heat that raised the film temper-
ature. An additional term was added to the heat
transfer equation, relative to the amount of mass
that was built on the crystal [8].
(3) Decrease in diffusivity: the volume of a crys-
tal acted as a barrier for diffusion of sucrose and
water. To model this influence, a mass transfer of
 E. Ben-Yoseph, R.W. Hartel / Journal of Crystal Growth 198/199 (1999) 1294–1298
Fig. 1. Concentration profiles during drying of thin sucrose film without nucleation sites. Drying air was at 50°C, 0% RH and velocity
of 3.0 m/s.
zero was used between two points when a crystal
was present between them.
(4) Decrease in drying rate: crystals that grew on
the surface of the film decreased the surface area for
water migration from the solution to the air. In
order to take this into account, the drying surface
area was multiplied by one minus the ratio of
crystal cross sectional area to film cross sectional
area in the direction of water flux.
3. Results
The concentrations at any time and location in
the film during drying were determined using the
simulation program. The initial conditions of the
film, for all the simulations, were: concentration of
70% TS, temperature of 40°C and thickness of
200 lm. The drying air velocity was 3.0 m/s and its
relative humidity was 0%. The drying air temper-
ature was 50°C.
Simulating a dehydration process without crys-
tallization, we observed a concentration increase of
sucrose as the drying time increased, at any loca-
tion in the film (Fig. 1). The rate of change and the
final concentration were higher for locations closer
to the surface of the film (after 2 min of drying the
concentration was 96% in the surface, compared to
87% near the flake). The concentration reached
1297
a plateau very fast near the surface and then the
change was mild. At the initial stages of drying the
concentration gradient was high, whereas at later
stages the concentration gradient flattened.
Simulating a drying process that includes crystal-
lization showed the concentration fields around
growing crystals (Fig. 2). Here, we placed two
spherical seed crystals (diameter of 5 lm) in the
film, at 24 and 120 lm distance from the surface.
The seed that was closer to the surface grew much
faster than the other. A concentration reduction
around the fast growing crystal can be seen, as well
as interference for diffusion by both crystals. The
film temperature was higher than the drying tem-
perature due to release of fusion heat. After 5 min of
drying the film temperature reached 51.4°C.
When high density of seed crystals were placed
on the surface of the film, the seeds grew very fast
during drying and formed a closed crust after ap-
proximately one minute. This crust inhibited moist-
ure migration from the film to the drying air (case
hardening). In the first 12 s, the sucrose concentra-
tion was high at the surface and decrease toward
the bottom (Fig. 3). Later (25 and 40 s in the chart),
the concentration was high between the growing
crystals, but low in the immediate layer below the
crystals, because of the concentration reduction
during the growth. After 60 s of drying, the film
reached equilibrium, with crystalline material on
1298 E. Ben-Yoseph, R.W. Hartel / Journal of Crystal Growth 198/199 (1999) 1294–1298

Fig. 2. Concentration field around two growing sucrose crystals after 5 min of drying of thin film of sucrose solution. Drying air was at
50°C, 0% RH and velocity of 3.0 m/s.

Fig. 3. Concentration profiles in a thin film of sucrose with high density of seeds on its surface. Drying air was at 50°C, 0% RH and
velocity of 3.0 m/s. The values above the curves are the drying times. After one minute the film temperature was 45.2°C, then, the
saturation concentration was 71.2% w/w.

the surface, and saturated solution (with uniform [3] K.I. Hayakawa, A. Bakal, J. Food Sci. 38 (1973) 623.
concentration) below. [4] B.A. Finlayson, Numerical Methods for Problems with
Moving Fronts, Ravenna Park, Seattle, WA, 1992.
[5] M. Balaban, G. M Pigott, J. Food Sci. 53 (1988) 935.
References [6] Y. Roos, M. Karel, Food Technol. (December 1991) 66.
[7] R.W. Hartel, K.A. Berglund, S.M. Gwynn, P.M. Schierholz,
[1] M. Okazaki, K. Shioda, K. Masuda, R. Toei, J. Chem. Eng. V.G. Murphy, AlChE Symp. Ser. 193 (1980) 65.
Japan 7 (1974) 99. [8] A. Benard, S.G. Advani, Int. J. Heat Mass Transfer 38 (1995)
[2] S.R. Bhowmik, K. Hayakaw, J. Food Sci. 44 (1979) 469. 819.