You are on page 1of 11

0960 – 3085/99/$10.00+0.

00
© Institution of Chemical Engineers
Trans IChem E, Vol 77, Part C, September 1999

MOISTURE, OIL AND ENERGY TRANSPORT DURING


DEEP-FAT FRYING OF FOOD MATERIALS
H. NI and A. K. DATTA
Department of Agricultural and Biological Engineering, Cornell University, USA

A
multiphase porous media model has been developed to predict the moisture migration,
oil uptake and energy transport in a food material such as a semi-dry potato during
deep-fat frying. The model predictions are validated using experimental data from
the literature. Spatial moisture proŽ les show two distinctive regions (dry or crust region near
the surface and wet region in most of the core) with a sharp interface which can be referred to
as the evaporation front. Spatial temperature proŽ les show two distinctive regions — higher but
constant temperature gradient region near the surface, and lower temperature gradient region in
the core. In the crust region, vapour diffusional  ux is comparable with vapour convective  ux.
In the more moist core region, capillary  ux of liquid water is comparable to the convective
 ux of liquid water. Therefore, all three modes of transport — diffusional, capillarity, and
pressure driven (Darcy)  ow are found to be important. Sensitivity of the Ž nal product
moisture and temperature to changes in oil temperature, initial moisture content of the sample,
thickness of the sample, and the surface heat and mass transfer coefŽ cients are discussed.
Keywords: multiphase  ow; heat and mass transfer; porous media; deep-fat frying;
Darcy  ow

6
INTRODUCTION description of a complex process. Ateba and Mittal
considered separate diffusion equations for energy, moist-
Deep-fat frying is a widely used and industrially important ure, and oil phases, without any evaporation term in the
food process. The fried food develops a crispy porous energy or the moisture transport equation. They included
structure in addition to  avour and other chemical changes only surface evaporation as a boundary condition for the
that is desired by the consumer. During deep-fat frying, energy equation, and equated the rate of surface evaporation
the food is immersed into the oil at a high temperature to the rate of diffusive moisture loss at the surface. Thus
(180 2 190°C) that leads to intensive vapourization of the evaporation was excluded from the inside even though
water in the food and transport out through the surface. As temperatures inside reached higher than 100°C. Moreira
water moves out, part of the pore spaces are taken up by et al.7 , on the other hand, used the same approach (i.e.,
the frying oil moving into the material. The oil picked up evaporation included only at the surface and no internal
by the food during the frying process has increasingly evaporation) but considered only diffusion of energy and
become a public health concern and there is a strong moisture. They provided no transport model for the oil
desire to reduce the oil content. A better understanding phase. Ikedalia et al.8 treated evaporation as spatially
of the transport processes and their relationship to various uniform throughout the material and included it as a sink
parameters should provide ways to optim ize the frying term in the energy equation. The transient rate of moisture
process, thus controlling oil pickup, for example. loss was obtained from experimental data. They did not
A mathematical description of frying at different levels consider transport of moisture or oil phases in their model.
of complexity has been reported. The simplest description Farkas et al.9,10 provided a more detailed model of tem-
can be empirical curve Ž ts of experimental data 1 ,2 . A large perature and moisture transport in deep-fat frying. They
number of models have been based on simple diffusion of provided separate equations for two regions — crust and
energy and mass, with various approxim ations to account the core with a moving boundary. They were perhaps the
for evaporation and sometimes ignoring evaporation Ž rst ones to consider pressure-driven  ow, although they
altogether. For example, only the diffusion of energy and restricted such  ow only in the crust region and only for
moisture was considered without evaporation in the works the vapour phase. They ignored any diffusional  ow in the
3 4 5
of Rice and Gamble and Dincer and Yildiz . Dagerskog crust region. They also ignored pressure driven  ow of
calculated temperatures using only the diffusion term in liquid or vapour in the core region. Such non-conside ration
the energy equation and did not acount for the latent heat of  uxes in various regions reduced mathematical compli-
of evaporation . The evaporation interface was tracked by cations, but their possible signiŽ cance was not discussed.
simply knowing the location where temperature exceeded Their model did not include the oil phase. A transport model
11
100°C from the solution to the energy equation. The amount that includes oil phase was reported by Chen and Moreira .
of evaporation was calculated by following the rate of They did not include the vapour phase separately and
movement of the interface. This is obviously a simple therefore the pressure driven mode of transport, which is

194
MOISTURE, OIL AND ENERGY TRANSPORT DURING DEEP-FAT FRYING OF FOOD MATERIALS 195

important and different from diffusion, was absent (see model system is shown in Figure 1. The major assumptions
discussions below). Several oil absorption or uptake studies of this model include:
that do not include transport modelling have been reported.
1) The solid, liquid, and gas phases are continuous;
These include experimentally measured oil uptake 12 ,13 and
2) Local thermal equilibrium exists between the phases;
how capillarity causes the surface oil to migrate inside during
3) Water vapour pressure is a function of water saturation
cooling 14 . A statistical description of oil distribution in the
15 and temperature, oil vapour is not considered in the present
fried product based on percolation theory was also reported .
model;
Deep-fat frying is an intensive heat transfer process that
4) Liquid water transport results from convective  ow
is expected to produce signiŽ cant internal evaporation and
due to the gradient in total gas pressure and from capillary
pressure generation that are spatially-varying and are func-
 ow due to the gradient of capillary force, which is a strong
tions of the porous structure of the material. At high rates
function of moisture content. Oil transport is assumed
of internal evaporation, signiŽ cant pressure driven  ows
resulting from capillary  ow due to the gradient of capillary
can be present for all phases and throughout the material.
force, which is a strong function of oil content;
Pressure driven  ow is also a fundamentally different
5) Vapour and air transport are driven by convective  ow
mode of transport that cannot be lumped into an effective
due to the gradient in total gas pressure and diffusion
diffusivity in a meaningful way, i.e., without making the
due to the concentration gradient;
‘diffusivity’ completely empirical, as was done in some
6) The contribution of convection to energy transport can
studies 3 ,16 ,11 . In fact, the constant diffusivity data that was 17
be ignored, as shown by Ni et al. ;
obtained by Ž tting the experimental curves in Chen and
7) Geometry does not change during frying and overall
Moreira 11 has strong discrepancies between their model
shrinkage is ignored. An equivalent porosity is deŽ ned
and experimental moisture proŽ les, particularly at higher
as the fraction of the total volume occupied by the liquid
oil temperature when pressure driven  ows are likely to be
water, oil, water vapour and air. This equivalent porosity
more important. Also, since oil, water, vapour, and air will
is assumed constant during the frying process and is used
be occupying the same pore space, their amounts are not
to calculate the concentration of each phase — liquid water,
independent of each other, as has been used by most pre-
oil, gas (water vapour and air). Effect of structure change
vious studies. Thus, an acute need exists for a model that
during frying manifests as changed gas porosity and related
would consider all these factors simultaneously to provide
transport properties. For example, loss of water increases
a comprehensive mechanistic understanding of transport
gas porosity and therefore the intrinsic permeability;
phenomena in deep-fat frying, which is the objective of this
8) Moisture removal from the surface consists of two
study. It appears that a multiphase porous media model of
parts — vapour diffuses to the boundary and is convected
the type developed by Ni et al.17 and applied to microwave
away from the surface area occupied by the gas fraction
heating where signiŽ cant pressure driven  ows can be
on the boundary. Liquid water evaporates at the boundary
present, is a type of model that can be developed to study
and is convected away from the surface area occupied by
deep-fat frying. Here, all transport mechanisms (i.e.,
the liquid fraction on the boundary. For a high moisture
molecular diffusion, capillary, and pressure driven  ow)
food or when signiŽ cant water is pushed to the surface from
and all the phases (i.e., oil, water, vapour, air) keep their
inside due to pressure driven  ow, surface evaporation
individual identity. In addition to providing a comprehen-
of liquid dominates. For a drier surface, vapor diffusion
sive description, such a model can give insight into the
dominates.
relative magnitudes of various transport mechanisms. The
9) Oil transport is driven by convective  ow and capillary
speciŽ c objectives of this study are to:
 ow.
1) Develop a multiphase porous media model for transport 10) Oil saturation at the boundary is assumed constant.
of energy, liquid water, vapour, air, and liquid oil during
deep-fat frying;
2) Calculate the spatial and transient proŽ les of the Constant
moisture, oil uptake, temperature and pressure; oil saturation
at surface
3) Compare the magnitudes of capillary and convective
liquid  uxes in the core region, and diffusive and convec-
tive  uxes of vapour in the crust region;
4) Perform sensitivity analysis to Ž nd the effect of oil tem-
Potato
perature, initial moisture content, sample thickness, and Oil
slab
heat and mass transfer coefŽ cients on temperature, moisture
loss, and crust thickness. Symmetry
(impermeable
surface)
DEVELOPMENT OF THE MULTIPHASE Energy,
POROUS MEDIA TRANSPORT MODEL FOR vapor
DEEP-FAT FRYING convected away
Assumptions
The mathematical formulation of the problem, consider-
ing the energy and mass balance of the phases, are 0 1.27 x (cm)
17
developed following the work of Ni et al. with the oil
phase added to the system. A schematic diagram of the Figure 1. Schematic diagram of the 1D system.

Trans IChemE, Vol 77, Part C, September 1999


196 NI and DATTA

Governing Equations Equations (14) –(18) make the complete set of governing
Equivalent porosity of an elemental volume D V is deŽ ned equations with four unknowns Sw , So , T and P. These
as: equations are developed in 1D for the model system shown
in Figure 1.
D Vw + D Vo + D Vg
w = ( 1)
DV
Equivalent saturation values for the various components Boundary and Initial Conditions
are deŽ ned as: The initial conditions are given by:
D Vw Sw= Swi ( 21 )
Sw = ( 2)
w DV
So=0 ( 22 )
D Vo
So = wDV ( 3) T = Ti ( 23 )
P = Pamb ( 24 )
D Vg
Sg = wDV ( 4) The boundary conditions at the frying surface (x = 0) are
given by:

( )
where, pv M v
Sw + So + Sg =1 ( 5) nv + nw = w (Sg + Sw) 2 rv 0 hmv ( 25 )
RT
Mass concentration values of the components are deŽ ned So= So1 ( 26 )
as:
P = Pamb ( 27 )
pv Mv Sg w
cv = ( 6) q = h( T 2 Tamb) + l nw ( 28 )
RT
pa M a Sg w Mass transport on the frying surface can be very com-
ca = ( 7) plicated. Equation (25) simply assumes the vapour to be
RT
convected away by the surrounding oil. Although it is
cw = rw w Sw ( 8) likely that the vapour can be expelled by the internal
co = r o w So ( 9) pressure, its numerical implementation is problematic.
Farkas et al.9 considered that there is almost no resistance
Flux values of the components are deŽ ned as:
for vapour to leave the surface and considered the pressure
p
kg Cg2 driven  ux on the boundary — however, the boundary

v = 2 rv $ P2 Ma Mv Deff ,g $ xv ( 10 ) condition they used by specifying vapour pressure is not
mg rg
p convincing. Instead, a large mass transfer coefŽ cient hmv is
kg C 2g
n®a = 2 ra $ P 2 Ma Mv Deff ,g $ xa ( 11 ) used here to represent the lower resistance to mass transfer
mg rg for the vapour. In equation (28), l nw represents heat  ux due
k pw to the latent heat of surface evaporation. This term
n®w =2 rw $ P2 Dw r w w $ Sw ( 12 ) is relatively large when the surface is wet and becomes
mw
p
insigniŽ cant as the surface gets dry.
ko Equation (26) is a very simple assumption for oil
n®o = 2 ro $ P 2 Do ro w $ So ( 13 )
mo boundary. In fact, the volum e fraction near the surface
occupied by the oil may not be a constant. It could depend
where P = pa + pv is total gas pressure and x = p/P is the
on whether there is bubbling because the bubble can push
molar fraction. The conservation equations for water
out the oil from the surface. In this study, it is assumed
vapour, liquid water, air, oil and energy in the porous
that the oil saturation on the surface is constant with a
medium are written, respectively, as:
time-averaged value of 0.35. Further work is certainly
­ cv needed to reŽ ne the boundary condition for oil at the
+$ ( n®v ) = Id ( 14 )
­ t frying surface.
­ cw At the centre of the slab, symmetry boundary conditions
+$ ( n®w ) =2 Id ( 15 ) are used,
­ t
­ ca nv =0 ( 29 )
+$ ( n®a ) =0 ( 16 )
­ t nw = 0 ( 30 )
­ co no = 0
+$ ( n®o ) =0 ( 17 ) ( 31 )
­ t
q=0 ( 32 )
­ T
( rcp ) eff =$ (keff $ T) 2 l Id ( 18 )
­ t
where,
Input Parameters
( rcp ) eff= w (Sgrg cpg + Sw rwcpw + Soro cpo ) Input parameters are shown in Table 1. Some of the
+ ( 1 2 w ) rs cps ( 19 ) parameters are discussed here and further details are in Ni
et al.17 . The vapour pressure was considered as a function
keff = w ( Sg kg + Swkw + So ko ) + (1 2 w )ks ( 20 ) of both temperature and water saturation of the material,

Trans IChemE, Vol 77, Part C, September 1999


MOISTURE, OIL AND ENERGY TRANSPORT DURING DEEP-FAT FRYING OF FOOD MATERIALS 197

Table 1. Input parameters used in the simulations. saturation, same form of dependence, as given by equation
(34), is used in this work for oil.
Parameter Value Source p p
The total permeability for gas (k g ) and water (k w ) are
cpw (J kg 2 1 K 2 1 ) 4180 calculated by using two different intrinsic permeabilities
p p
Dw (m2 s2 1 ) Eqn. 34 Ni
17
k gi and k wi for gas and water, respectively, as:
Deff ,g (m 2 s2 1 ) 2 .63 10 2
5
Mills
18

h (W m 2 2 K 2 1 ) 250 Miller
19 k pg= k pgik pgr
hmv (m s2 1 ) p ( 35 )
0.015 This work
k pw = k wi k pwr
ks (W m 2 1 K 2 1 ) 0.21 Choi and Okos
20

kw (W m2 1 K 2 1 ) 0.65 Choi and Okos


20
The intrinsic permeability values in the very wet stage and
kg (W m 2 1 K 2 1 ) 0.026 Choi and Okos
20
the very dry stage are different, as shown in Table 1. The
ko (W m 2 1 K 2 1 ) 0.17 Choi and Okos
20 p p 21
relative permeabilities k gr and k wr are given by Bear :
103 10 2
p
k gi (m 2 ) 14
Ni
24
p
k wi (m 2 ) 53 10 2 14
Ni
24 k pgr =12 1 .1 Sw Sw < 1 /1 .1
( 36 )
p
k oi (m 2 ) 53 10 2
14
Ni
24
k pgr =0 Sw > 1/1 .1
Swi 0.5 This work

( )
3
So1 0.35 This work Sw 2 Sir
Sir 0.08 Bear
21 k pwr = Sw > Sir
Tamb (°C) 180 This work
1 2 Sir ( 37 )
Ti (°C)
r o (kg m 2 3 )
20 This work
20
k pwr =0 Sw < Sir
920 Choi and Okos
r v (kg m 2 3 ) 0 Ni
24
where Sir is the irreducible liquid saturation. Equations (35)
p p
r w (kg m 2 3 ) 1000 to (37) are shown graphed in Figure 2, using the kgi , kwi , and
l (J kg 2 1 ) 2 .4353 10
6
Sir values from Table 1. At high water content, the total
24
w 0.88 Ni gas permeability decays linearly to zero. The total liquid
22
m o (Pas) 0.0317 Steffe et al.
permeability is high at high liquid saturation, but not as
1 .83 10 2
5
m g (Pas)
m w (Pas) 5 .4683 10 2
4 large as the total gas permeability at very low liquid
saturation. This is the effect of the change in structure (more
gas pores). In reality, the intrinsic permeability changes
continuously from the very wet stage to the very dry stage.
23 In the absence of such data, the curves in Figure 2 are a
and data from Ratti et al. was used:
pv reasonable description of the effect of structural changes
=2 0 .0267 M 2
1 .656
ln on permeability. Data on permeability values for oil are also
ps (T )
unavailable. The intrinsic permeability of oil is treated the
+ 0.0107 e2 1.287 M
M 1 .513 ln ps (T ) ( 33 ) same as that for water. The relative permeability of oil is
also assumed to vary in the same manner as water and
where ps (T ) and M are the vapour pressure of pure water at equation (37) is used for oil with Sw replaced by So .
temperature T and the moisture content of the material,
respectively. Capillary diffusivity values are assumed to be
dependent on saturation only, as given by: Numerical Method
The above Ž ve governing equations are transformed into
= 1.03 10 2 exp ( 2 2 .8 + 2 .0 M)
8
Dw ( 34 ) four equations with four variables (Sw , So , T and P). A Ž nite
difference method is used with centre difference in space
for water. Details of the formulation of water capillary and fully implicit in time. A non-uniform grid is used to
24
diffusivity is given in Ni . Since there is no literature data solve a slab with a half thickness of 1.27 cm. The minimum
available for oil capillary diffusivity as a function of oil grid size (0.01 cm) is near the surface and the maximum

10

8
m )
2

p
Gas (k g )
-14

6
Permeability (10

p p
Liquid (kw or k o )
4

0
0.0 0.2 0.4 0.6 0.8 1.0
Liquid saturation, S w or So

Figure 2. Permeabilities of liquid and gas phases, calculated from intrinsic and relative permeabilities, plotted vs saturation.

Trans IChemE, Vol 77, Part C, September 1999


198 NI and DATTA

160 Prediction
Surface 0.05 cm 2.6
Experiment of

Moisture content (d.b.)


140
2.4 Farkas et al. (1996)
Temperature (°C)

120
2.2
100
0.42 cm
2.0
80

60
1.8
Prediction
40 Experiment of 1.6
1.27 cm Farkas et al. (1996)
20 0.85 cm
0 2 4 6 8 10
0 2 4 6 8 10 Frying time (min)
Frying time (min)
0.20
Figure 3. Model prediction of temperatures compared with the experi-
4 ,10
mental data of Farkas et al. at various distances from the surface. The

Crust thickness (cm)


surface temperature (predicted only) is shown to indicate its sensitivity 0.15
to the location.
0.10

Prediction
grid size (0.1 cm) near the centre. Detailed numerical 0.05 Experiment of
procedures are given in Ni 24 . Farkas et al. (1996)
0.00

MODEL VALIDATION 0 2 4 6 8 10
Frying time (min)
The model predictions are compared with literature
9
experimental data for a 1D slab of potato tissue with Figure 4. Model prediction of total moisture loss and crust thickness
9 ,10
half thickness of 1.27 cm, oil temperature of 180°C, heat compared with the experimental data of Farkas et al.
transfer coefŽ cient between the oil and the fried surface, h,
of 250 W m 2 K 2
2 1
and mass transfer coefŽ cient, hmv , of
2 1
0 .028 m s . The higher value of mass transfer coefŽ cient is less evaporation in the crust layer, the linear temperature
is chosen to represent minimum surface transport resist- proŽ les are almost the same as for a pseudo-steady state
9
ance to vapor, as was true for Farkas et al. . heat conduction process. Temperature is somewhat uni-
Figure 3 shows that the predicted temperatures compare form in the core region, primarily due to the presence of the
well with experimental data in all four locations. The evaporation zone that acts as a sink for the incoming energy
temperature increases slowly inside due to the low heat from the surface. The temperature proŽ le is qualitatively
penetration while the temperature near the surface rises similar to conventional drying, as in the work of Nasrallah
fast. The region where the curve becomes  at signiŽ es the et al.25 .
presence of strong evaporation. Temperatures eventually The evaporation temperature is about 90°C and is some-
9
approach the oil bath temperature. The moisture content and what lower than experimental data of Farkas et al. which
crust thickness in Figure 4 shows that the calculated trends is about 100°C. The reason might be due to higher mass
are the same as those in the experimental data. The transfer coefŽ cient (hmv = 0 .028 ms2 1 ) used in the model.
Using a smaller value of hmv = 0 .015ms 2
1
difference in the magnitude for the crust thickness data is increases the
probably due to the speciŽ c temperature chosen to deŽ ne the evaporation temperature to 93°C. As discussed earlier, a
crust. In this model, the crust is deŽ ned as the region over large mass transfer coefŽ cient is used here to represent
which the temperature exceeds 100°C (moisture content is the small resistance to surface mass transfer used by other
already quite low here). Experimentally, crust thickness authors. However, based on the decrease in the evaporation
cannot be deŽ ned the same way and discrepancies between temperature, it appears that the no resistance assumption
measured and calculated crust thickness is inevitable. for the vapour to leave the boundary is questionable .
Another reason for the lower evaporation temperature
might be the lower surface heat transfer coefŽ cient
RESULTS AND DISCUSSION (250 W m 2 K 2 ) used here. Although the exact value is
2 1

Spatial and temporal proŽ les of temperature, pressure, not known, it can be as much as twice this value during
7 ,9
moisture, oil saturation, and oil uptake during deep-fat the intense bubbling period . The effect of changes in heat
frying of a slab of potato are discussed here. Relative and mass transfer coefŽ cients is shown under sensitivity
magnitudes of the rates of moisture transport by pressure, analysis.
diffusional, and capillary modes are compared. Sensitivities Under intensive frying conditions, water saturation Sw
of the frying process to changes in oil temperature, initial near the surface reduces rapidly, as shown in Figure 5.
moisture content, and slab thickness are also included here. However, water saturation in the wet region inside
decreases slowly due to a drastically reduced water capillary
diffusivity of 10 2 m s2 in the very dry region near the
9 2 1
Spatial ProŽ les of Temperature, Moisture, Oil, surface from a value of 10 2 m s 2 in the wet region (see
7 2 1
and Pressure equation (34)). Initially, the dry layer thickness increases
Temperature proŽ les in Figure 4 show two distinctive rapidly. However, it slows down as the crust becomes
regions with very different temperature gradients. As there thicker and its low thermal conductivity reduces the rate

Trans IChemE, Vol 77, Part C, September 1999


MOISTURE, OIL AND ENERGY TRANSPORT DURING DEEP-FAT FRYING OF FOOD MATERIALS 199

0.5 1.010 9 min

Total pressure (0.1MPa)


0.4 1.008
Water saturation S w

1.006
0.3 0 min
1 min 1.004
0.2 1 min
9 min
1.002
0.1
1.000

0.0 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
Location x (cm)
Location x (cm)
160
0.35
140
0.30
Temperature (°C)

120

Oil saturation So
1 min 9 min 0.25
100 1 min
0.20
80 0.15 9 min
60 0.10
0 min
40 0.05
20 0.00
0.0 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
Location x (cm) Location x (cm)

Figure 5. Spatial proŽ les of Sw and T in frying of a potato slab at times 1, 3, Figure 6. Spatial proŽ les of P and So in frying of a potato slab at times 1, 3,
5, 7, 9 minutes. 5, 7, 9 minutes.

of heat transfer. If the evaporation line can be deŽ ned as


the location where the largest gradient of water saturation
occurs, it is about 2 mm from the surface after 9 minutes
of frying.
Pressure reaches the maximum value near the evapora-
tion front (see Figure 6), and it increases with frying time
as the front moves further away from the surface, adding
resistance to convective (Darcy)  ow. There is about 1 kPa
pressure increment up to 9 minutes and the effect of pressure
will be discussed in later sections.
ProŽ les of oil saturation So (see Figure 6) is typical of
diffusional (capillarity driven) transport in a semi-inŽ nite
media. It seems that oil is not only absorbed in the crust
layer but can also penetrate to a distance twice the crust
thickness. This study for the Ž rst time uses a mechanistic
model that includes pressure driven  ow and diffusion to
describe the oil saturation proŽ le.

Temporal ProŽ les of Water Saturation and Pressure


As shown in Figure 7, it takes about 30 seconds for the
surface to get dry. However, it takes almost 2 minutes
for the location at 0.05 cm below the surface to get dry.
This large time delay is also due to the larger capillary
diffusivity of water at high initial moisture in the core
region.
Initially the inside pressure is lower than the boundary
pressure (see Figure 7). The internal pressure rises
with evaporation. Near the surface (0.05 m), the pressure
is built up quickly. However, after the pressure reaches the
maximum, it starts decreasing as the surface dries out. Figure 7. Temporal proŽ les of Sw and P in frying of potato slab at times 1,
While for the inside region, the pressure keeps increasing 3, 5, 7, 9 minutes.

Trans IChemE, Vol 77, Part C, September 1999


200 NI and DATTA

0.30 1 min

Vapor diffusion flux (g/m s)


2
0.25 6
Oil content (d.b.)

0.20
9 min
0.15 4

0.10
2
0.05
0.00 0
0 2 4 6 8 10
Frying time (min) 0.0 0.1 0.2 0.3 0.4
Location x (cm)
Figure 8. Oil uptake of potato tissue in frying.

1 min

Vapor convective flux (g/m s)


2.0
within the calculated period due to high moisture. Although

2
the pressure build-up is only 1 kPa, its effect on the moisture 1.5
transport is still important (see later sections).
1.0 9 min
Total Oil Uptake With Time
0.5
The oil uptake is deŽ ned as the ratio of the weight of oil
intake to the weight of dry material. Oil uptake with time 0.0
(see Figure 8) shows that the rate of oil uptake is initially
higher and then slows down, becoming linear with time.
0.0 0.1 0.2 0.3 0.4
The initial higher rate is due to a larger difference of Location x (cm)
oil concentration between the surrounding oil and initial
concentration of oil in the food. As the crust becomes Figure 9. Comparison of diffusive and convective  uxes of vapour in
thicker, the oil uptake also increases proportionall y. After crust layer during frying of potato slab at times 1, 3, 5, 7, 9 minutes.
10 minutes, the oil content reaches about 30% (d.b.), as
shown in Figure 8. The spatial distribution of oil follows a
diffusion proŽ le in a semi-inŽ nite media, as discussed
earlier.

Vapour and Liquid Water Fluxes in the Crust and


Water capillary flux (g/m s)

Core Region 3.0


2

1 min
2.5
Fluxes in the crust region 2.0
As shown in Figure 9, the vapour diffuses from the 1.5
evaporation front to the surface, and the vapour diffusional
 ux occurs only within the crust layer. The maximum  ux 1.0
occurs near the evaporation front and its magnitude 0.5
decreases with the frying time since the vapour concentra- 9 min
0.0
tion decreases with moisture content. The magnitude of the
 ux in Figure 9 is comparable with the maximum  ux of 0.0 0.2 0.4 0.6 0.8 1.0 1.2
Location x (cm)
8 g m 2 2 s2 1 in the work of Farkas et al.9 .
The vapour convective  ux has the same trend as the
diffusional  ux, as shown in Figure 9. The vapour is driven 0.0 9 min
Water convective flux (g/m s)

by the pressure gradient from the evaporation front to the


2

surface. The vapour convective  ux also occurs only within -0.2


the crust layer and its magnitudes are comparable to those -0.4
for diffusional vapour  ux. Therefore, the convection term -0.6
cannot be discarded in describing the total vapour  ux in the
-0.8
crust layer.
-1.0
Fluxes in the core region -1.2
1 min
The capillary diffusional  ux of water in the core region -1.4
is towards the surface, as shown in Figure 10. There is a
region of constant  ux, which starts from the evaporation 0.0 0.2 0.4 0.6 0.8 1.0 1.2
Location x (cm)
front and extends inside to about double the thickness of
the crust. In that region, the water saturation is spatially Figure 10. Comparison of diffusive (capillary) and convective  uxes of
linear and capillary diffusivity is relatively constant. liquid water in frying of potato slab at times 1, 3, 5, 7, 9 minutes.

Trans IChemE, Vol 77, Part C, September 1999


MOISTURE, OIL AND ENERGY TRANSPORT DURING DEEP-FAT FRYING OF FOOD MATERIALS 201

In the core region, water convective  ux due to pressure Table 2. Comparison of model prediction with experimental data for
is toward the centre. The magnitude of the convective  ux, the effect of oil temperature. Two oil temperatures of 160 and 180°C are
used. The variable for comparison is deŽ ned as the absolute magnitude
although smaller, is comparable with capillary diffusional of (value(180) –value(160))/value(160)3 100.
 ux. Therefore, both the convective and the capillary diffu-
4,10
sional mechanisms contribute to the total water  ux Model prediction Experiment
in the core region. Neither transport mechanism can be
Centre temperature 6.6% 12%
ignored. Moisture loss 5% 12.5%
Crust thickness 36% 31%

Effect of Oil Temperature


The sensitivity analysis in this and the following sections
do not include the curves for oil content since the oil accurate material properties and heat and mass transfer
transport into the material was relatively insensitive to the coefŽ cients will be necessary for the mathematical model.
parameter changes in the range studied. The effects of oil
temperature on the centre temperature of the slab, moisture
Effect of Initial Moisture Content
loss and crust thickness are shown in Figure 11. The centre
temperature increases with the oil temperature, but this The effects of initial moisture content on sample
increase is much lower than the increment of oil temperature temperature, moisture loss and crust thickness are shown
itself. In addition to diffusional resistances of the solid, in Figure 12. The centre temperature decreases with initial
particularly the crust with low thermal conductivity , this is moisture content because increased evaporation in a higher
also caused by the internal evaporation, limiting the heat moisture food reduces the energy  ow to the centre.
transfer to the centre. The moisture content decreases with Moisture loss increases signiŽ cantly with initial moisture
the oil temperature, but only slightly. The crust thickness content because both surface evaporation and subsequent
increases with temperature, and it is generally in the range internal evaporation are much higher for a high moisture
of 1 –1.5 mm at the end of 10 minutes of frying. food. The crust thickness increases signiŽ cantly with
The effect of increased oil temperature as predicted from decreasing initial moisture content. For an initial moisture
this work is compared with the experimental data of Farkas content of 1.55 (d.b.), the crust can form shortly after frying.
at al.9 in Table 2. The model predictions generally agree
with the experiment. To make an absolute comparison, more
80
Center temperature (°C)

70
60
80
Center temperature (°C)

70 50
M=2.58 (d.b.)
60 40 M=2.06 (d.b.)
M=1.55 (d.b.)
50 180°C 30
170
40 160 20
30 0 2 4 6 8 10
Frying time (min)
20
0 2 4 6 8 10 0.6 M=2.58 (d.b.)
Moisture loss (dry base)

Frying time (min) M=2.06 (d.b.)


0.5 M=1.55 (d.b.)
Moisture content (dry base)

2.5 180°C 0.4


170
2.4 160 0.3
2.3 0.2
2.2 0.1
2.1
0.0
2.0
0 2 4 6 8 10
Frying time (min)
0 2 4 6 8 10
Frying time (min)
2.5 M=2.58 (d.b.)
M=2.06 (d.b.)
Crust thickness (mm)
Crust thickness (mm)

1.2 2.0 M=1.55 (d.b.)

1.5
0.8
180°C 1.0
0.4 170
160 0.5

0.0 0.0
0 2 4 6 8 10 0 2 4 6 8 10
Frying time (min) Frying time (min)

Figure 11. Transient centre temperature, moisture loss, crust thickness Figure 12. Transient centre temperature, moisture loss, and crust thickness
as affected by oil temperature. as affected by the initial water saturation.

Trans IChemE, Vol 77, Part C, September 1999


202 NI and DATTA

After 10 minutes of frying, the crust thickness increases but the change is not signiŽ cant. Further work is needed
to 2.6 mm, which is almost twice the crust thickness for an to improve predictions for very thin materials. In this
initial moisture content is 2.58 (d.b.), as compared to one work, a comparatively larger thickness (half thickness
with initial moisture of 1.55. Therefore, controlling initial about 1 cm) is used.
moisture content can dramatically affect the crust thickness
of the Ž nal product.
Effect of Heat and Mass Transfer CoefŽ cients
The effect of heat transfer coefŽ cient on surface and
Effect of Thickness centre temperature and moisture loss are shown in
When the sample becomes very thin, such as a potato Figure 14. Heat transfer coefŽ cient has a much stronger
chip (1–2 mm thick), the model prediction for moisture effect on surface temperature than the centre temperature.
loss and surface temperature has some discrepancy com- As heat transfer coefŽ cient increases, the heat  ux from
pared to the experimental data. This can be due to material the oil to the food increases, which leads to a higher surface
inhomogeneity when it becomes very thin. Predicted moisture temperature. In addition, surface temperature shows a
loss does not increase as fast as the experimental values during shorter plateau before 2 minutes, which means that surface
the initial 20 seconds and the predicted surface temperature evaporation is faster. The centre temperature also increases
also cannot increase as fast as the experimental values. with heat transfer coefŽ cient, although not appreciably.
The effect of thickness on the centre temperature, The effect of mass transfer coefŽ cient on surface
moisture loss, and crust thickness and oil uptake are and centre temperature, and moisture loss are shown in
shown in Figure 13. The centre temperature increases Figure 15. Increasing mass transfer coefŽ cient causes
signiŽ cantly with decreasing thickness. This is due to a more surface evaporation initially, therefore increasing
thinner sample having less moisture and a shorter dis- moisture loss and decreasing surface temperature during
tance for heat  ux to reach to the centre. The surface of surface evaporation period within the Ž rst 2 minutes. After
a thinner sample can be quickly dried out, which reduces that period, the surface temperature increases above 100°C.
the surface evaporation and decreases the moisture loss. Since most of the moisture loss comes from internal
Crust thickness increases with decreasing slice thickness evaporation at this stage, increasing mass transfer coefŽ -
cient leads to increased moisture loss and a lower centre
temperature.
100
Center temperature ( C)

80 CONCLUSIONS
1) A multiphase porous media model has been developed
60 to predict temperature, moisture, oil pickup, and crust
9.5 mm thickness during deep-fat frying. The model considers
40 12.7 mm the transport of oil, water, vapour, and air components
14.0 mm
20
0 2 4 6 8 10 2
h=250 W/m K
160
Frying time (min)
140
Moisture loss (dry base)

0.8 Surface
Temperature ( C)

120 150
200 Center
100
0.6 250 200
80
0.4 60
9.5 mm
12.7 mm 40 150
0.2 14.0 mm
20
0.0 0 2 4 6 8 10
0 2 4 6 8 10 Frying time (min)
Frying time (min)
1.4 2.5
Moisture content (d.b.)
Crust thickness (mm)

1.2
2.4
1.0
2.3
0.8
0.6 2.2 150
9.5 mm
0.4 12.7 mm 2.1 200
0.2 14.0 mm 2
2.0 h=250 W/m K
0.0
0 2 4 6 8 10 0 2 4 6 8 10
Frying time (min) Frying time (min)

Figure 13. Transient centre temperature, moisture loss, and crust thickness Figure 14. Transient centre and surface temperature, and moisture loss,
as affected by thickness. as affected by heat transfer coefŽ cient.

Trans IChemE, Vol 77, Part C, September 1999


MOISTURE, OIL AND ENERGY TRANSPORT DURING DEEP-FAT FRYING OF FOOD MATERIALS 203

k pwr , k pgr liquid and gas relative permeability, respectively


140
molecular weight, kg kmol 2 ; moisture content (dry basis or db)
1
M
total  ux, kg m 2 s
0.01 2
120 n
Temperature (°C)

Surface
Center P, p total pressure and partial pressure, respectively, Pa
100
0.01
R universal gas constant, J kmol 2 1 K
80 hmv = 0.02 m/s S saturation
t time
60 T temperature, K
3
40 V volume, m
0.02 x molar fraction; coordinate
20
0 2 4 6 8 10 Greek symbols
intrinsic density, kg m 2
Frying time (min) 3
r
latent heat of vapourization, J kg 2
1
l
w porosity
2.5 m dynamic viscosity, Pa s
Moisture content (d.b.)

2.4
Subscripts
2.3 a air
0.01 amb ambient
2.2
eff effective
2.1 g gas (vapour + air)
i initial
2.0 hmv= 0.02 m/s
o oil
s solid matrix; surface
0 2 4 6 8 10 v vapour
Frying time (min)
w water
Figure 15. Transient centre and surface temperature, and moisture loss,
as affected by mass transfer coefŽ cient.
REFERENCES
1. Baumann, B. and Escher, F., 1995, Mass and heat transfer during
separately. It is validated using available experimental deep-fat frying of potato slices — I. Rate of drying and oil uptake,
data in the literature. Lebensm-Wiss u-Technol, 28: 395– 403.
2) Pressures from internal evaporation produce signiŽ cant 2. Kozempel, M. F. and Tomasula, P. M. and Craig, Jr, J. C., 1991,
Correlation of moisture and oil concentration in french fries, Lebensm-
convective (Darcy)  ow. Diffusional and convective Wiss u-Technol, 24: 445 –448.
 uxes of vapour are comparable in the outer (crust) 3. Rice, P. and Gamble, M. H., 1989, Technical note: modeling
region, while capillary and convective  uxes of liquid are moisture loss during potato slice frying, Int J Food Sci & Tech, 24:
comparable in the core region. Thus, all three modes of 183 – 187.
4. Dincer, I. and Yildiz, M., 1996, Modeling of thermal and moisture
transport — diffusional, convective, and capillary are
diffusions in cylindrically shaped sausages during frying, J Food Eng,
important. 28: 35 – 43.
3) Spatial temperature proŽ les show two distinct regions — 5. Dagerskog, M., 1979, Pan frying of meat patties: A study of heat and
a pseudo steady state region in the drier crust and a mass transfer, Lebensm-Wiss u -Technol, 12: 217– 224.
transient diffusion-like proŽ le in the interior, becoming 6. Ateba, P. and Mittal, G. S., 1994, Modeling the deep-fat frying of
beef meatballs, Int J Food Sci Tech, 29: 429 –440.
spatially uniform with time. Spatial moisture proŽ les also 7. Moreira, R., Palau, J. and Sun, X., 1995, Simultaneous heat and mass
show two distinct regions — a drier region near the surface transfer during the deep fat frying of tortilla chips, J of Food Proc Eng,
and a more wet region in most of the core. A somewhat 18: 307– 320.
sharp interface, which can be referred to as the evaporation 8. Ikediala, J. N., Correia, L. R., Fenton, G. A. and Ben-Abdallah, N.,
1996, Finite element modeling of heat transfer in meat patties during
front, is also seen.
single-sided pan-frying, J Food Sci, 61(4): 796– 802.
4) Increasing oil temperature, reducing initial moisture 9. Farkas, B. E., Singh, R. P. and Rumsey, T. R., 1996a, Modeling heat
content, and reducing thickness can increase the centre and mass transfer in immersion frying: Model development, J Food
temperature, moisture loss and crust thickness. Increasing Eng, 29: 211– 226.
heat transfer coefŽ cient increases surface temperature 10. Farkas, B. E., Singh, R. P. and Rumsey, T. R., 1996b, Modeling heat
and mass transfer in immersion frying: Model solution and veriŽ cation,
signiŽ cantly more than the centre temperature. Increasing J Food Eng, 29: 227 – 248.
mass transfer coefŽ cient decreases centre temperature due 11. Chen, Y. and Moreira, R. G., 1997, Modeling of a batch deep-fat frying
to increased internal evaporation and higher moisture loss. process for tortilla chips, Trans IChemE, Part C, Food Bioprod Proc,
75(C3): 181– 190.
12. Ufheil, G. and Escher, F., 1996, Dynamics of oil uptake during deep-fat
frying of potato slices, Lebensm-Wiss u-Technol, 29: 640 – 644.
NOMENCLATURE 13. Moreira, R. G. Sun, X. and Chen, Y., 1997, Factors affecting oil uptake
speciŽ c heat, J kg 2 K
1
cp in tortilla chips in deep-fat frying, J Food Eng, 31: 485 –498.
mass concentration, kg m2 total volume
3
c 14. Moreira, R. G. and Barrufet, M. A., 1998, A new approach to describe
molar density of gas mixture, kmol m 2
3
C oil absorption in fried foods: a simulation study. J Food Eng, 35: 1 –22.
effective gas diffusivity in moist materials, m s 2
2 1
Deff ,g 15. Moreira, R. G. and Barrufet, M. A., 1995, Spatial distribution of oil
Dw capillary diffusivity, m 2 s2 1 after deep-fat frying of tortilla chips from a stochastic model, J Food
heat transfer coefŽ cient, W m 2 K
2
h Eng, 27: 279– 290.
vapour transfer coefŽ cient, m s 2
1
hmv 16. Ngadi, M. O. and Correia, L. R., 1995, Moisture diffusivity in chicken
Id volumetric evaporation, kg m 2 s 2
3 1
drum muscle during deep-fat frying, Canadian Agric Eng, 37(4):
thermal conductivity, W m 2 K
1
k 339 – 344.
kp total permeability, m
2
17. Ni, H., Datta, A. K. and Torrance, K. E., 1998, Moisture transport
p p
k wi , k gi intrinsic permeability at very wet stage and at very dry stage, in intensive microwave heating of biomaterials: A multiphase porous
2
respectively, m media model, accepted in Int J of Heat Mass Transfer.

Trans IChemE, Vol 77, Part C, September 1999


204 NI and DATTA

18. Mills, A. F., 1995, Basic Heat and Mass Transfer (Irwin, Chicago, 24. Ni, H., 1997. Multiphase moisture transport in porous media under
USA). intensive microwave heating, PhD Dissertation (Cornell University,
19. Miller, K. S., 1992. Physical and thermal properties of edible frying USA).
oils, MSc thesis (University of California, Davis, USA). 25. Nasrallah, S. B. and Perre, P., 1988, Detailed study of a model of
20. Choi, Y. and Okos, M. R., 1986, Thermal properties of liquid foods— heat and mass transfer during convective drying of porous media, Int J
A review, in Physical and Chemical Properties of Liquid Foods, Okos, Heat Mass Transfer, 31(5): 957– 967.
M. R. (ed) (ASAE, St Joseph, Michigan, USA).
21. Bear, J., 1972, Dynamics of Fluids in Porous Media (American
Elsevier, New York).
ADDRESS
22. Steffe, J. F., Mohamed, I. O. and Ford, E. W., 1986, Rheological
properties of liquid foods: data compilation in Physical and Chemical Correspondence concerning this paper should be addressed to Professor
Properties of Liquid Foods, Okos, M. R. (ed) (ASAE, St Joseph, A. K. Datta, Department of Agricultural and Biological Engineering,
Michigan, USA). Cornell University, Riley-Robb Hall, Ithaca, NY 14853-5701, USA.
23. Ratti, C., Crapiste, G. H. and Rotstein, E., 1989, A new water sorption
equilibrium expression for solid foods based on thermodynamic The manuscript was received 6 July 1998 and accepted for publication
considerations, J Food Science, 54(3): 738 – 747. after revision 4 March 1999.

Trans IChemE, Vol 77, Part C, September 1999

You might also like