Multidimensional Modelling of a Toxical Molecule Formation during Food Frying

Excerpt from the Proceedings of the COMSOL Users Conference 2007 Grenoble
Multidimensional Modelling
of a Toxical Molecule Formation during Food Frying

G. Carrieri, M.V. De Bonis and G. Ruocco*
CFDfood – DITEC, Università degli studi della Basilicata, Potenza Italy
*Corresponding author: Campus Macchia Romana, Potenza 85100 Italy, gianpaolo.ruocco@unibas.it


Abstract: Heat-related processing can have
occupies an important role: this molecule can be
various effects on food and food ingredients.
formed in the heat processing of food, as first
Besides the intended effects to improve safety
postulated in Sweden in 2000 (Williams, 2005).
and quality of the products, undesired changes,
In April 2002, the “Swedish National Food
nutritional or toxicological, can also occur. In
Administration” (SNFA) highlighted the AA
this circle, the acrylamide formation occupies an
presence in different foodstuffs (Claeys, De
important role, as this molecule has been
Vleeschouwer. & Hendrickx, 2005a; Claeys, De
considered as a potential cause of cancer.
Vleeschouwer. & Hendrickx, 2005b; Mestdagh,
The possibility that this molecule may be
De Meulenaer & Van Peteghem, 2007). AA
formed in the heat processing of food was first
forms during heating of certain foods, rich of
postulated in 2000. Acrylamide forms during
carbohydrates (100-400 mg/kg) and relatively
heating of foods, rich of carbohydrates
poor of proteins (<100 mg/kg), at high
associated with frying, baking, or roasting. In
temperatures and low moisture conditions as
this work, acrylamide formation during frying of
associated with frying, baking, and roasting
potatoes sticks with alteration of food
(Claeys, De Vleeschouwer & Hendrickx, 2005a).
composition is analyzed through the CFD and a
AA has been considered as a potential cause
gas-chromatography measurement.
of cancer and has been classified by the
Characteristic biochemical notations have
International Agency for Research on Cancer as
been joined into a distributed parameter model to
probably carcinogenic to humans (Group 2A),
enforce the interdependent transport phenomena.
whose neurotoxic action depends of quantities
Complete multidimensional distributions of
assumed. In fact, an average AA daily intake of
acrylamide concentration are reported in food
100 mg could represent a non-negligible dose for
during cooking and the associated experiments
risk of cancer in population (Tareke, Rydberg,
allowed to validate the simulations.
Eriksson & Tornqvist, 2002; Claeys, De

Vleeschouwer & Hendrickx, 2005a).
Keywords: Acrylamide, Frying, Transport
A number of theoretical mechanisms have
phenomena, Kinetics, Computational Fluid
been proposed for AA formation; most probably,
Dynamics
it results from the browning process by the

Maillard reaction of reducing sugars with
1. Introduction
asparagine (Claeys, De Vleeschouwer. &

Hendrickx, 2005; Gokmen & Senyuva, 2006;
Acrylamide (AA) is an industrial chemical
Mestdagh, De Meulenaer & Van Peteghem,
used in the production of polyacrylamides for
2007). Other probably minor pathways including
different technical applications. Recently, high
acrolein and acrylic acid. Oxidation of acrolein
levels of AA have been measured in different
to acrylic acid and subsequent reaction of acrylic
food; the highest levels have been found in
acid with ammonia generated from the pyrolysis
French fries, potato chips, and other fried, deep-
of nitrogen-containing compounds present in
fat fried, or oven cooked potato products,
food, results in the formation of acrylamide.
together with some crisp bread, biscuits,
Alternatively, AA could be formed from
crackers, and breakfast cereals.
rearrangement of nitrogen–containing
Processing, in particular heat-related, can
compounds present in food without involving
have various effects on food and food
acrolein (Claeys, De Vleeschouwer. &
ingredients. Besides the intended effects to
Hendrickx, 2005a). Potato products have been
improve safety and quality of the products,
associated with some of the highest levels of
undesired changes, nutritional or toxicological,
acrylamide, partly due to relatively high levels of
can also occur. In this framework, AA formation
suspected acrylamide precursors, while raw

---

Excerpt from the Proceedings of the COMSOL Users Conference 2007 Grenoble
potatoes do not contain AA (Williams, 2005;
the chemical kinetics to solve differential
Becalski et al., 2004).
equations that govern the transport phenomena
The study on the thermal process is justified
involved by means of CFD. The study of the
from the fact that AA starts forming at 120 °C
performance of a prototype process is possible,
(Gokmen & Senyuva, 2006; Pedreschi et al.,
with no costs related to the experimental trials.
2007; Mestdagh, De Meulenaer & Van
Complete multidimensional distributions of AA
Peteghem, 2007) and it is maximum in a system
concentration are described during food
asparagine/glucose (1:3) heated to 170 °C for 30
evolution and, from the associated experiments,
minutes (2629 mg/g) (Ehling & Shibamoto,
the simulations are validated.
2005). The fact that AA is not formed during
This work has focussed on the following
boiling indicates that higher temperatures and/or
important aspects: 1) distribution of AA in the
low moisture conditions are necessary for its
fried products (therefore distinguishing between
formation. During heating at atmospheric
superficial and internal formation, and related
conditions, higher temperatures can be reached
diffusion); 2) the dependence of AA formation
only if simultaneous drying takes place, which is
on the temperature; 3) frying’s combination
the case in frying. However, the temperature of
time-temperature; 4) potato’s variety (and
frying oil was reported of the change of local
therefore the content of reagents) used for frying.
internal temperatures during frying and since a

temperature distribution in the product occurs
2. Experimental assessment
due to conductive resistance to heat transfer and

evaporation of water from the product, the rate of
Potato (cultivar Agata) was cut into strips (4
AA formation will not be the same for layers
cm x 1.5 cm) and the strip length was adjusted to
from surface to core of product (Gokmen &
1.5 cm. Frying of three strips was performed
Senyuva, 2006).
using sunflower oil (300 ml) in a common
Among the heat-related processes, frying is a
stainless steel pot, placed over an electric heater
procedure widely used by both the food industry
equipped of a thermostat switch. Frying
and consumers, so the problem of AA formation
temperatures and times were monitored such that
pertains to both industrial products and
3 different configurations were employed (at 170
household food preparations. Therefore, the
°C at 6 min, 180 and 190 °C at 4 min).
optimization of industrial processes and the
Temperatures were acquired with two K-type
creation of products destined to private domestic
thermocouples, one placed inside the stick to
economies that produce the smaller possible
measure the core temperature and the other into
quantity of acrylamide during the roasting, the
oil, carefully avoiding contact with steel
cooking in the oven and the frying, is necessary.
surfaces, at initial, mid-and end process times.
The knowledge of such simultaneous heat and
For every frying configuration, five samples
mass transfer, with their related critical
were analyzed using a GC-MS method with
processing variables, is paramount for the
derivatisation of AA (Williams, 2005): 1) the
improved product safety and quality. Frying
strip at end frying; 2) the strip’s crust at end
processes may be designed through the use of
frying; 3) the strip’s core at end frying; 4) the
mathematical models and a reliable simulation of
strip’s crust at half frying; 5) the strip’s core at
the process may be decisive for process
half frying.
optimization and control. In addition, models

seldom allow for distributed analyses (such as
3. Model formulation
concentration and temperature fields), and their

applicability is further limited in that superficial
3.1 Governing equations
heat and mass transfer coefficients are only

empirically assumed.
In this work, frying of potato strips has been
In this work, AA formation during frying of
studied, by including AA kinetic mechanism
small sticks of potatoes is analyzed by both the
formation, and its evolution for each adopted
CFD (Computational Fluid Dynamics), including
configuration.
alteration of food's composition and the
The following assumptions have been
experimental investigation. CFD is an innovative
adopted:
computational approach that can be coupled to

---

Excerpt from the Proceedings of the COMSOL Users Conference 2007 Grenoble
1. the oil is incompressible with constant

properties and subject to laminar flow;
In Eqs. (4) and (5) Kw is a kinetics constant for
2. the geometry is two-dimensional (only a
water evaporation (Krokida, Oreopoulos &
transversal section of the pot is
Maroulis, 2000).
considered for sake of symmetry);

3. the potato’s properties, except density,
Reactant:
vary with temperature and composition;
? c
?
c
?
c
? ?
4. the viscous heat dissipation neglected;
R
R
R
+ u
+ v
= ?K c
?
?
?
?
f R (6)
5. no-slip enforced at every solid surface;
? t
?
x
?
y
? ?
as usual, velocity is zero in the solid;

6. perfect conductibility through the
AA:
metallic walls of the frying pan;
? c
?
c
?
c
?
?
7. diffusive transport of AA and reagents
AA
AA
AA
+ u
+ v
= K
?
?
?
?
AA (7)
is negligible;
? t
?
x
?
y
? ?
8. diffusive transport of water is

negligible.
Initial conditions: u = 0, v = 0, cAA = 0, for t = 0,
With reference to such considerations, the
x,y; To = Top, Tp = Tamb for t = 0.
transport equations that drive the phenomena, in

rectangular coordinates, are the following (Bird,
Boundary conditions:
Stewart & Lightfoot, 2002):
To the free (horizontal) oil surface (slip

condition, contact with atmosphere):
Continuity:
dT
h
v = 0,
= ? hor (T ? ?
T ) , c
dx
k
AA = 0,
o
?u ?
+ v =
for 0 ? x ? L, y = H, t > 0.
0
?
(1)
x
?y
To the bottom surface of oil (no-slip condition,
perfect contact with electric heater):

u = 0, v = 0, T = Top, cAA = 0, for 0 ? x ? L, y = 0,
Momentum equation in natural convection

:
t > 0.
1. in horizontal direction:
To the side plate (no-slip condition, contact whit
? ?u
?u
?u ?
? ?2u ?2u ?
?
+
vertical plate of plant):
u
+ v
=
?
?
?
? µ??
+
(2)
2
2 ?
?
? ?t
?x
?y ?
? ?x
?y ?
dT
h
u = 0, v = 0,
= ? vrt (T ? ?
T ) , c
dx
k
AA = 0, for
2. in vertical direction:
o
? ?v
?v
?v ?
? ?2v ?2v ?
x = 0, 0 ? y ? H; x = L, 0 ? y ? H, t > 0.
?
+ u
+ v
= µ
+
+
?
?
?
?
?
?

To the superior surface of potato (no-slip
2
2 ?
?
? ?t
?x
?y ?
? ?x
?y ?
condition, dimensions of potato sticks (?xp, ?yp):
+ ?g? (T ? T
u = 0, v = 0, T = T
(3)
o.
0 )
The mechanisms of formation and elimination of

AA can be described with first order kinetics.
Energy:
AA content of in food can be assumed to result
??T
?T
?T? k ??2T ?2T ?
from two consecutive reactions, moving from R
+u
+v
=
+
?
?
?
?
?
?
?
(reactant) to D (AA-protein complex/AA
2
2 ?
?
? ?t
?x
?y ? c
?

p ? ?x
?y ?
degradation product), with Kf and Kel the first
? K
?
?
order formation and elimination rate constants at
w (c
c
(4)
0
e ) h
the given temperature:


Water:
Kf Kel
? ?c
?c
?c ?
? ?2c
?2c ?
R AA D (8)
w + u
w + v
w
= D
w +
w
?
?
?
?
?
w ??
2
2 ?
?
? ?t
?x
?y ?

? ?x
?y ?
The effect of temperature can be expressed by
? K c ? c
Arrhenius relation (Claeys, De Vleeschouwer &
w (
(5)
0
e )

---

Excerpt from the Proceedings of the COMSOL Users Conference 2007 Grenoble
Hendrickx, 2005a; Claeys, De Vleeschouwer &
temperature and the poor sensibility of switch
Hendrickx, 2005b):
thermostat.

One of the interesting outcomes of the
? E ?
??
present model is the ability to perform
?
K = K
a ? 1
exp
? 1 ??
parametric analysis to assess the AA formation
ref
?
?
? (9)
R T
T ?
?
?
as dependent on the initial carbohydrates content.
ref
??
On the basis of the present AA measurements,

(9)
the model provides for a good prediction of AA
A Finite Element solver has been employed to
formation evolution when the initial contents of
integrated the system of partial differential
two glucose and fructose are assumed as close to
equations (COMSOL Multiphysics™, 2007).
the admissible least quantity (Claeys, De

Vleeschouwer. & Hendrickx, 2005a) in the
3.1 Potato properties
whole product. AA appears more sensitive to

glucose content, due to the higher kinetics values
The thermal properties of the potato used in
(Eaf and Kfref, Claeys, De Vleeschouwer &
present simulation are:
Hendrickx, 2005b). AA be found in larger
1. Density ? (kg/m3) = 1090 (Yildiz,
quantity where the thermal stress has been
Palazoglu & Erdogdu, 2007)
greater (on the product crust). The agreement
2. Specific heat cp (J/kgK) = 4.180xw +
with experimental data is excellent up to
1.711xp + 1.928xf + 1.547xc + 0.908xa
intermediate process duration, whereas for
where xw, xp, xf, xc and xa denote the
longer times the model underestimates the
fraction of water, protein, fat,
experiment at the crust, while the opposite is true
carbohydrates and ash of the product
at the product core.
(Gonzalez-Martinez, Ahrné, Gekas &

Sjoholm, 2004)
4.1 Temperature fields
3. Thermal conductivity, k (W/mK)=

0.4392T0.0746 (Gonzalez-Martinez,
The present multidimensional simulation
Ahrné, Gekas & Sjoholm, 2004)
allows visualizing several additional

information, as for example the local distribution
4. Model validation and results
of temperature and AA concentration, which are

important to infer on the process viability and the
The thermal evolution is scrutinized first in
available optimization choices.
order to evaluate the model ability to predict the
The oil temperature field for Configuration
experiment evolution, which is important as the
n.3 is presented at first in Figure 1, where the
thermal regime is the driving factor for the AA
process time is reported on a vertical axis. As the
formation. A relatively small departure between
process at hand is characterized by natural
predicted and experimental temperature is
convection, the temperature distribution is
observed, for every frying configuration, to be
affected by the flow field. The toroidal flow field
particularly attributed to the empirical values of
is influenced, in the present geometry, by the
h assumed in this study. Then, for the higher
perturbing potato sticks; the hot oil is subject to
temperature (190 °C) a certain overestimation of
buoyancy near the walls and in the available
temperature is due to the very active fluid
space at pot center, forming a complex flow
dynamic regime that allowed distributing more
pattern, which is reflected in the thermal field.
uniformly the heat transported from the heated
However, a mere 20 °C increase in process
walls. Finally, the model should account for the
temperature favours an oil velocity increase of
variation of thermodynamic properties of frying
approximately 20% (not shown).
oil with temperature.
As the working fluid mass is just of one
Besides the thermal evolution of stick’s
order of magnitude different that the processed
centroid is reported and compared with the
mass, the oil temperature decreases due to the
associated measurements. At the end of process,
evaporative cooling of the moisture content,
the agreement is very good, while there is a
especially for the highest temperature.
relatively large difference at process beginning,

due to material’s properties variation with

---

Excerpt from the Proceedings of the COMSOL Users Conference 2007 Grenoble
confirms the adequacy of the model to simulate
crust formation and growth, as water is depleted
for temperature above 100 °C.

4.2 Concentration Field

Finally, in Figure 3 the AA quantity (on the
vertical axis) is correlated with the temperature
distribution, in the flow and in the product, for
the highest process temperature, showing that the
molecule generation and accumulation is tightly
associated to the temperature increase in the
product, which in turn is connected to the flow
field. In all explored configurations AA have
been formed and accumulated similarly, the
greatest quantity to be found in corners, as to
confirm what speculated for temperatures in

Figure 2.

Figure 1. Oil temperature distribution for Conf. n.3.

In Figure 2 the thermal evolution of one
processed stick is reported with time.



Figure 3. Variation of AA concentration with
temperature distribution for Conf. n. 3 at 240 s.


The maximum AA local concentration is
Figure 2. Potato temperature distribution for Conf. n.
1.1×10-4 mol/m3, two order of magnitude of the
3.
calculated average, whereas for the lowest

temperature (not shown) the maximum AA local
concentration is 4.4×10-5 mol/m3. Therefore, it is
A maximum temperature of approximately
inferred that to a modest 10% temperature
160 °C has been detected at the stick’s corners,
corresponds a 60% AA concentration increase. It
correspondent to the minimum heat resistance. In
is then concluded, as expected, that the AA
contrast, along the stick’s surfaces, the substrate
generation kinetics is very sensitive to the
is more affected by the evaporative cooling
thermal driving factor: a small increase of
underneath and therefore the temperature is
temperature involves a strong increase of such
limited to approximately 100 °C. Figure 2 also
molecule.

---

Excerpt from the Proceedings of the COMSOL Users Conference 2007 Grenoble


5. Conclusions
Subscripts

0 initial
A multidimensional model has been
a air
presented to predict acrylamide formation and
amb. environment
transport, its temporal evolution, and its
AA
acrylamide
dependence on the process conditions. The
e equilibrium
model allows for assessment of each critical
el elimination
parameters involved as local temperature and
f formation
concentration of carbohydrates.
hor horizontal
The frying temperature and the thermal
o oil
gradient found in the product influences the
op operational
formation of the toxic molecule, that is found
p potato
and predicted in larger quantities on the product
R reactant
surface and for higher process temperatures. It is
vrt vertical
then confirmed that the biochemistry at hand is
w water
strongly and non-linearly dependent from the
? surrounding environment
conjugate transport phenomena. The use of

various kinetics for disaggregated carbohydrate
7. References
reactants allowed for a better insight, with

respect to the available literature, of their
1. Bird, R.B., Stewart, W.E. & Lightfoot, E.N.,
influence. The generality of the model brings the
Transport Phenomena, John Wiley and Sons,
potential of implementation for a variety of
(2002)
biotechnology and food products, and process
2. Claeys, W.L., De Vleeschouwer, K., &
configurations.
Hendrickx, M.E., Quantifying the Formation of

Carcinogens during Food processing:
6. Nomenclature
Acrylamide, Trends in Food Science &

Technology, 16, 181-193 (2005a)
c concentration, mg/g or mol/m3
3. Claeys, W.L., De Vleeschouwer, K., &
cp specific heat, KJ/kgK
Hendrickx, M.E., Kinetics of Acrylamide
D mass diffusivity, m2/s
Formation and Elimination during Heating of an
Ea activation energy, kJ/mol
Asparagine-Sugar Model System, Journal of
g gravitational acceleration, m/s2
Agricultural and Food Chemistry, 53, 9999-
h convective heat transfer coefficient, W/m2K
10005 (2005b)
H height, m
4. Gonzalez-Martinez, G., Ahrné, L., Gekas, V.,
?h entalpia variation, kJ/kg
& Sjoholm, I., Analysis of Temperature
k thermal conductivity, W/mK
Distribution in Potato Tissue during Blanching
K reaction rate constant, s-1
and its Effect on the Absolute Residual Pectin
Kref pre-exponential factor, s-1
Methylesterase Activity, Journal of Food
L length,
m
Engineering, 65, 433-441 (2004)
p pressure, Pa
5. Gokmen, V., & Senyuva, H.Z., Study of Color
R universal gas constant, J/molK
and Acrylamide Formation in Coffee, Wheat
S surface,
m2
Flour and Potato Chips during Heating, Food
t time,
s
Chemistry, 99, 238-243 (2006)
T temperature, K
6. Krokida, M.K., Oreopoulos, V. & Maroulis,
u,v x-component and y-component velocity, m/s
Z.B., Water Loss and Oil Uptake as a Function
x, y coordinate, m
of Frying Time, Journal of Food Engineering,

44, 39-46 (2000)
Greek
7. Mestdagh, F., De Meulenaer, B., & Van
? thermal expansion coefficient, 1/K
Peteghem, C., Influence of Oil Degradation on
µ dynamic viscosity, Pa/s
the Amounts of Acrylamide Generated in a
? density, kg/m2

---

Excerpt from the Proceedings of the COMSOL Users Conference 2007 Grenoble
Model System and in French Fries, Food
Chemistry, 100, 1153-1159 (2007)
8. Pedreschi, F., Leon, J., Mery, D., Moyano, P.,
Pedreschi, R., Kaack, K., & Granby, K., Color
Development and Acrylamide Content of Pre-
Dried Potato Chips, Journal of Food
Engineering, 79, 786-793 (2007)
9. Tareke, E., Rydberg, P., Eriksson, S., &
Tornqvist, M., Analysis of Acrylamide, a
Carcinogen Formed in Heated Foodstuff,
Journal of Agricultural and Food Chemistry, 50,
4998-5006 (2002)
10. Williams, J.S.E., Influence of Variety and
Processing Conditions on Acrylamide Levels in
Fried Potato Crisps, Food Chemistry, 90, 875-
881 (2005)
11. Yildiz, A., Palazoglu, T.K., & Erdogdu, F.,
Determination of Heat and Mass Transfer
Parameters during Frying of Potato Slices,
Journal of Food Engineering, 79, 11-17 (2007)

---