Professional Documents
Culture Documents
R. H. MASCMERONId~ A. CALVELO
SUMMARY
One of the most suitable parameters for relating the freezing rate to the volume of
drip produced during the thawing of meat is the characteristic time, defined as the time
necessary to reduce the temperature of the sample from - 1 . 1 °C (initial freezing
point in bee/') to - 7 ° C (80% of the water frozen).
However, as the freezing of beef in factories takes place with important
temperature gradients, distributions of these characteristic times must be expected
along the pieces of frozen meat.
In order to relate these characteristic time distributions to heat transfer
parameters under industrial freezing conditions, a mathematical model which
simulates the freezing of beef is developed in this paper.
The model establishes the heat transfer equations with simultaneous change of
phase, taking into account the dependence of the thermal properties with the ice content
and considering the anisotropy of the thermal conductivity according to the direction
of the fibres.
Boundary conditions include the possibility of thermal resistances in the
refrigerated interphase.
The model developed was compared with laboratory experiments performed under
factory freezing conditions and showed a satisfactory agreement between theory and
experiment.
NOMENCLATURE
a P a r a m e t e r d e f i n e d in eqn. (26).
Bi Biot n u m b e r Bi = hL/k o.
267
Meat Science 0309-1740/80/0004-0267/$02-25 © Applied Science Publishers Ltd, England, 1980
Printed in Great Britain
268 R . H . MASCHERONI, A. CALVELO
Greek symbols
ct Parameter defined in eqn. (18).
ao Thermal diffusivity o f unfrozen meat ct = ko/Po~p.-
fl Variable defined in eqn. (19).
Parameter defined in eqn. (9).
6 Thickness of the partially frozen zone.
F/ Dimensionless temperature ~/= (Ti -- T)/(Ti -- TI).
r/s Dimensionless parameter defined in eqn. (31) ~I = (To ~TI)/To.
r/i Dimensionless parameter defined in eqn. (30) ~ll = (Ti - T.)/To.
r/l Dimensionless temperature for the beginning of freezing.
H E A T T R A N S F E R P A R A M E T E R S A N D D A M A G E V A R I A B L E S F O R BEEF F R E E Z I N G 269
Subscripts
a Water.
c Continuous matrix.
h Ice.
o Unfrozen meat.
t Tissue.
INTRODUCTION
The damage produced by freezing in meat tissues has been the subject of many
papers (Hiner et al., 1945; Deatherage & Hamm, 1960; Partmann, 1963; Love,
1966; Penny, 1974; Afi6n and Calvelo, 1980). In general, these studies were aimed at
finding a relationship between the volume and/or composition of the exudate
obtained on thawing and a characteristic parameter ~able to describe, in only one
value, the thermal history o f the analysed sample.
Several characteristic parameters have been proposed (Meryman, 1966; Fennema
& Powrie, 1973), one o f the most frequently used being the time elapsed when the
sample changes its temperature through a given range.
Recent studies (Afirn & Calvelo, 1980) show that the exudate volume produced
when frozen meat is thawed depends on the 'characteristic time' (t~), defined as the
time taken by the sample to change its temperature from - 1.1 °C (beginning o f
freezing) to - 7 °C (80 ~ of the initial water frozen).
However, even though the characteristic time can be directly used in small
samples, which can be frozen without significant temperature gradients, in respect of
pieces o f meat frozen under industrial conditions, a whole distribution of
characteristic times must be expected along the frozen product.
Thus, in the vicinity of the refrigerated border, characteristic times will be shorter
than inside the meat.
In order to relate the damage studies with the industrial behaviour of frozen meat,
the characteristic time distributions must be analysed in terms o f the heat transfer
parameters which govern the process. For this purpose a mathematical model which
simulates the freezing of meat has to be available and the temperature profiles for
different times analysed.
270 R. H. MASCHERONI, A. CALVELO
Even though many models have been proposed for the freezing of meat, in general
they are more oriented towards the calculation of freezing times than to the analysis
of temperature distributions. Moreover, in many of them, there are important
restrictions. Thus, the well-known moving boundary models (London & Seban,
1943; Carslaw & Jaeger, 1959; Hrycak, 1963; Tien & Churchill, 1965; Westphal,
1967; Tao, 1967; Cho & Sunderland, 1969) are unsuitable for meat in which freezing
still continues at temperatures as low as - 6 0 ° C (Riedel, 1957).
On the other hand, models which involve a range of freezing temperatures (Keller
& Ballard, 1956; Tien & Geiger, 1967, 1968; Cho & Sunderland, 1969; Bakal &
Hayakawa, 1973) usually include constant properties or other restrictions which
make difficult their application in analysing meat freezing problems.
Although Joshi & Tao (1974) developed a numerical method that gives a good
approximation for meat freezing, the published information is restricted to several
temperature profiles or freezing times and no calculations of characteristic times (as
defined) were performed.
In the present paper, a mathematical model which simulates the freezing of meat
under industrial conditions (plate freezers) is proposed and the effect of the different
thermal parameters on the characteristic time distributions is analysed.
MATHEMATICAL MODEL
~/.7/(~) ~3 £:T
p(~) ~ - ~k(~)~-~x (1)
where, as can be seen, the density, p, the e n t h a l p y , / t , and the thermal conductivity,
k, are assumed to be a function of the fraction of frozen water, co. A scheme for the
system is shown in Fig. 1.
The initial and boundary conditions which simulate the operation in plate freezers
are:
where Ti is the initial temperature, 7": is the temperature of the refrigerating medium
HEAT TRANSFER PARAMETERS AND DAMAGE VARIABLES FOR BEEF FREEZING 271
If
x-o x= L
Fig. I. Scheme of the meat slab system.
and h is the heat transfer coefficient in the interphase between the meat and the
refrigerated plate.
To solve eqn. (1) it is necessary to dispose of functions for p, kand k in terms of o.
A relationship between o and Tis also required. These functions will be developed in
the next paragraphs.
FT
o=E-- (5)
TO - T
where :
E= 1 -X,/X (6)
and :
(7)
In eqns. (5), (6) and (7), A/” is the heat of crystallisation at 0°C X is the initial
content of water on a dry basis, T is the absolute temperature and TOis the melting
temperature of pure water (273.16 K).
The constants X, (bound water content on a dry basis) and D (moles of solutes per
kilogramme of dry substance) were evaluated by regression as X, = O-1965 and
D = 1.455.
272 R . H . MASCHERONI, A. CALVELO
Densi O'
By considering the partially frozen meat constituted by ice, water and dry tissue,
the density, p, will be:
p(og) = po/[1 + og~(r - 1)] (8)
where:
~'= YoPo/P. (9)
and:
r = O,/Ph (10)
In the above equations po is the density of the unfrozen meat, Yo is the water
content on a wet basis (Yo = X/(I + X)), Oh is the density of ice and p,, is that of
water.
where Cp. is the specific heat of the unfrozen meat. F r o m eqns. (12) and (13) one
obtains:
,o do9 (14)
C,(og) = C , . - ogYoAdp- Yo[/.: + A C p ( T - To)ld T
where A~Tp = Cp. - C~, C~ and Cp, are the specific heats of ice and water (which
were assumed to be constant) and the bracket indicates the temperature dependence
of the latent heat of crystallisation.
The first two terms on the right-hand side ofeqn. (14) correspond to the change in
the specific heat of meat when the water is converted into ice while the last term
represents the heat of crystallisation involved in the freezing of meat.
HEAT TRANSFER PARAMETERS AND DAMAGE VARIABLES FOR BEEF FREEZING 273
Thermal conductit'ity
In the case of the thermal conductivity for partially frozen meat, the relationship
to the fraction of frozen water must be established from the distribution of the ice in
the meat tissue.
In a previous paper (Mascheroni et al., 1977) a model for the thermal conductivity
of frozen meat was proposed in which the partially frozen tissue was considered to be
composed of an arrangement o f partially dehydrated fibres surrounded by ice
(Fig. 2).
tibre
• ICE
..toa ©
.i
heat f l u x heat f l u x
(b) (c)
Fig. 2. Model of fibre arrangement in frozen meat.
where:
2 = l -- x / l - MYop/p,, (16)
1 - (1 - ~ko/k,)l~
k, = k . (17)
1 + C a - 1)#
= 3 k d ( 2 k ,. + k . ) (18)
# = (l - a~)-¢/(l - co-~,) (19)
In eqns. (15) to (19) k~ and k~ are the thermal conductivities of ice and water,
respectively.
The thermal conductivity of the matrix inside the fibres, k,., was evaluated from
that o f the unfrozen meat, ko, as k , = 0.3539 W/m K.
7- = k* (20)
and:
f, = a + b/Irl(rl , + fly) - r/i[2 (25)
d- g/Inl(nl + nf) - n,I
HEAT TRANSFER PARAMETERS AND DAMAGE VARIABLES FOR BEEF FREEZING 275
The dimensionless form of the equations leads to the introduction of the following
parameters obtained by rearrangement of the terms in eqn. (20):
F r o m these parameters only those defined in eqns. (30), (31) and (32) depend on
the operating conditions during the freezing of meat while the others depend on
meat properties.
Equations (20) to (23) involve a non-linear partial differential problem without
analytical solution. The numerical solution was performed by dividing the closed
range of the variable ~ into equal intervals and applying the same procedure to the
open range variable, z. An explicit step-by-step finite difference method was applied
to this mesh where a forward difference was used for the time derivative (first order
accuracy) and central differences were introduced for the space derivatives of
eqn. (20) (second order accuracy) (Jenson & Jeffreys, 1963; Lapidus, 1962).
The non-linear problem was linearised evaluating the variable properties at the
conditions of the previous time step (Joshi & Tao, 1974). The sizes of the space and
time intervals were Ax = I mm and At = 1 sec, respectively. Calculations were
performed using an IBM/360 computer.
MODEL RESULTS
Restricting the analysis to the parameters related to the operating conditions in plate
freezers, the solution of eqn. (20) with initial and boundary conditions 421), (22) and
(23) will be:
~/= r/(~, ~, r/;, r/i, Bi) (33)
The rest of the parameters defined in eqns. (26) to (29) depend, in general, on the
physico-chemical properties of the meat. The values used for feeding the computer
program are shown in Table 1. The meanings of the parameters included in eqn. (33)
are as follows:
r/; considers the effect of the difference between the initial temperature and To.
276 R.H. M A S C H E R O N I , A. C A L V E L O
TABLE 1
PARAMETERS FOR BEEF FED TO THE COMPUTER PROGRAM
E = 0.92864 2~ = 3 3 4 . 4 k J / k g
F = 0.00359 I kh = el + e : T
Po = 1053 k g / m 3 k.=L +AT
pa= 1000 k g / m 3 eI = 4.749 W/m K
Ph = 930 kg/m 3 e2 = -0-929 x 10-2W/mK 2
TO = 273.16K ft =0.6751 x 10-2W/mK
Yo = 0-7336 f2 = 2.051 x 1 0 - S W / m K 2
Cp, = 3-475 k J / k g K kc = 0.3539 W/m K
A~7v = 2.135 kJ/kg K ko = 0-5057 W / m K
ql takes into account the effect of the difference between the refrigerant
temperature and TO.
Bi is the dimensionless Biot number and accounts for the thermal resistance in
the interphase between the meat and the refrigerated plate with respect to the
heat transfer resistance within the meat.
Examples of the thermal histories (q versus z) predicted by the model for different
positions ( 0 are shown in Fig. 3. In the same figure the values corresponding to the
beginning of freezing (q~) and to the freezing of 80 ~ of the water (r/v) have also been
indicated.
As can be seen, for positions near the refrigerated plate, the temperature changes
rapidly without important distortions because of the change of phase. It is only for
deeper positions (~ --* I) that the typical 'plateau' becomes evident through a change
in the slope of the temperature-time curve.
An important characteristic in Fig. 3 is that, for a constant dimensionless time, ~,
between 0.45 and 0-55, all the curves show a second change in the slope. This change
is not produced for a constant value of r/(freezing of a certain fraction of water for
the analysed point) but for approximately constant values of 3. (Actually the change
7/7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . o-8-
0.4 0.6
o.4
u.~ I- - . s~:o .
/
I I i I I I ~ ""='~
0 0.1 0.2 0.3 0.4 0.5 0.6
Fig. 3. Predicted t i m e - t e m p e r a t u r e curves for different values o f the d i m e n s i o n l e s s p o s i t i o n , ~.
H E A ' I T R A N S F E R P A R A M E T E R S A N D D A M A G E V A R I A B L E S F O R BEEF F R E E Z I N G 277
of slope takes place at slightly shorter times for curves corresponding to deeper
positions within the piece of meat.)
This effect can be explained on the basis that, for the parameters analysed in Fig.
3, the main proportion of the change of phase at the centre of the meat (~ = l) has
been concluded for z values between 0-45 and 0.55. Because the latent heat transfer
along the meat is thereby almost complete, the rate of elimination of sensible heat
from each point is accelerated, producing a simultaneous change in the slope at all
positions.
The same effect will be analysed later when dealing with the position of the moving
boundaries.
Also from Fig. 3 and by using the indicated temperature levels, r/l and r/7, it is
possible to analyse the dimensionless characteristic time z7 = kotT/po(?p.L 2. This
value is a dimensionless form of the time required for a given point of the piece of
meat to pass from the beginning of freezing to 80 9/0 of the water frozen.
As can be observed in Fig. 3, z7 is shorter near the surface, becomes longer for
deeper positions and diminishes again near the centre. This result shows that the
portion of meat which experiences the slower change of phase does not correspond
precisely to the centre.
In order to obtain a better analysis of this phenomenon and to relate the
characteristic time distributions with the thermal parameters of the systems, the
function z~ = z~(~, q,, rlj, Bi) was calculated from the computer model.
0-4 t- ai =34.44 / ~ -
/' / \
/ =o.o8 /
/3 = 0.,o98 /
o.3is =o.,,,
0.2 3 _
0.1
i57 I I I 1 I
~ j = 0.0256
0.4 7tf = 0.o888
I Bi = lo /
2 = 13.05 /
3 =ZO / ,
4 = 34.44 / /
0.3 s =5o / /
0.2
0.1
I
0
0.2 0.4 0.6 0.8 1
Fig. 5. Predicted dependence of z~ with ~ for different values of the Biot number.
0.15
3
T~f = 0.1464
Bi =34.44
/ti =0.0073
=0.0439
=0.0732
1/
1
0.1
0.05
0
0.2 0.4 0.6 0.8 1
The distributions of z7 for different temperatures of the refrigerant (t/j) are shown
in Fig. 4. As can be seen, the higher q; (low values of the refrigerant temperature Ts)
the lower the values of ~7 obtained along the piece of meat (the freezing takes place
faster).
Moreover, in Fig. 4 the existence of a m a x i m u m o f z 7, for a position different from
the centre, can be detected more easily.
The effect of Bi and r/i on the distributions ofzv can also be analysed f r o m Figs. 5
and 6. As can be observed, the influence of the Biot n u m b e r is less important than
that of q r. In any case, the higher the Biot n u m b e r the lower the z7 profile (as
expected). It is also possible to note a tendency for Bi to become irrelevant for high
values (no thermal resistance in the interphase between the meat and the refrigerated
plate).
The effect of r/,. (initial temperature of meat) is small, as can be seen from Fig. 6.
The m a x i m u m of zv inside the piece of meat can be analysed more clearly by
defining two moving boundaries, X 1 and XT, as the points where the temperature is
- 1.1 °C and - 7 °C, respectively for a given time. O f course, it will be X~ = X~ (t)
x:l
.... ~7 ......
0.8 time]CteristicJ
boundary for freezin(
beginning (-1.1 * C ) - - " /
•boundary for
B0% frozen (-7°C)
0.6
0.4
T/i = 0.0256
7tf = 0.~2B~
0.2
B i = 34.44
O| I I I I I
0.1 0.2 0.3 0.4 0.5
Fig. 7. Variation of the calculated dimensionless boundaries for beginning of freezing and for 80 %
frozen as a f u n c t i o n o f d i m e n s i o n l e s s time, z.
280 R. H. MASCHERONI, A. CALVELO
EXPERIMENTAL
Experimental procedure
Experiments were carried out on semimembranous muscle stored at 4 °C for 2 or 3
days after quartering.
The freezing of meat under industrial conditions was simulated by means of a
cylinder 4-5 cm in diameter and varying between 4-5 and 8.4 cm in length which was
axially frozen.
The cylinder of meat was cut in such a way that the heat flow was parallel to the
fibre direction. It was mounted in an acrylic sample-holder permitting the
establishment of a uniform initial temperature by means of a jacket through which
water from a thermostat was circulated. Once the sample temperature was uniform,
the water of the jacket was drained and replaced by air, which contributed to the
radial isolation.
One of the bases of the cylinder rested on a copper slab-shaped exchanger inside
which thermostatised ethanol from a cryostat was circulated. The temperature was
controlled at +__0-05 °C. The other base of the cylinder, as well as the lateral surfaces,
were isolated with polystyrene foam.
Between the exchanger and the meat, an acrylic slab, 1-11 m m thick, was placed to
HEAT TRANSFER PARAMETERS AND DAMAGE VARIABLES FOR BEEF FREEZING 281
Experimental results
Experiments were performed by covering a range for the initial temperature, 7-,.,
from 4-5 ° to 11-8 °C. The temperature of the refrigerating medium, Tj, was changed
between - 2 2 - 2 ° and - 4 5 . 5 °C, while the sample lengths were varied from 4-5 to
8-4 cm.
In Fig. 8 a characteristic 'time distribution obtained experimentally (three
replicated runs) is c o m p a r e d with that predicted by the model. As can be seen, the
agreement is satisfactory with a higher difference in regions far from the exchanger
(probably due to a fault in the isolation of the upper base of the cylinder).
5O
,Ira,hi Ti = 7.2 oC
40 T! = 45.S°C
L = 5.8cm
30
20
calculated
xomexpcrimenlal
10
I I I I I
0 1 2 3 4 5
x[cm]
Fig. 8. Comparision of calculated and experimental variation of t 7 with pbsition (experimental data
refer to three replicated runs with different samples of equal size).
282 R.H. M A S C H E R O N I , A. C A L V E L O
TABLE 2
A v e r a g e e r r o r = (~'~eil)/N = 15.6 % .
HEAT T R A N S F E R P A R A M E T E R S A N D DAMAGE VARIABLES FOR BEEF F R E E Z I N G 283
150
.~1oo
t~
o@ •
0 I i
50 100 150
t7 experimental[rain]
Fig. 9. Comparision of the model with experimental values of characteristic time.
Table 2 shows the theoretical characteristic times and those obtained in the
experiments for several dimensionless positions (( = x/L) in cylinders frozen under
different conditions. The mean absolute value for the percent errors was 15-6 %. The
distribution of the data is shown in Fig. 9.
It is important to emphasise that no parameter of the model was allowed to
remain free to fit the experimental values. The computer p r o g r a m was fed with the
experimental conditions and with the information on beef properties given in the
literature (Table 1).
CONCLUSIONS
(1) The mathematical model developed describes the parameters involved in the
freezing of beef. T h r o u g h a knowledge of the influence of these parameters on the
system, the consequences of changes in the operating conditions of industrial
freezers can be predicted.
(2) The theoretical analysis of the model shows that the physical interpretation of
phenomena in distributed parameter systems is different from that accepted for the
lumped ones.
284 R. H. MASCHERONI, A. CALVELO
Thus, because of the existence of significant temperature gradients, the lowest rate
of freezing (longer time to pass from - 1.1 “C to - 7 “C) does not occur precisely at
the centre of the piece of meat.
The analysis of the moving boundaries allows an easier understanding of this
phenomenon which involves a maximum in the characteristic time distribution.
(3) The established relationship between heat transfer parameters and the
characteristic time profiles shown in Figs. 4, 5 and 6 makes it possible to relate the
operating conditions in freezers to the voluine of exudate produced on thawing.
Thus, a study of the erect of the characteristic time on the exudate volume (Aiibn
& Calvelo, 1980) showed the existence of a maximum for t7 ‘c 17 min. As
characteristic times are lower for higher Bi and qJ (Figs. 4 and 5) the maximum
amount of exudate will be displaced deeper, the higher these parameters. This
circumstance can be useful in detecting the rate of freezing in a piece of meat,
according to the position of the maximum (ArGn et al., 1980).
(4) The experimental results show good agreement with the predictions ,of the
mathematical model, not only with respect to characteristic time distributions, but
also with temperature profiles and thermal histories.
(5) No fitting parameters were used in the proposed model. All the physical
properties and operating conditions fed to the program were calculated from
accepted values in the literature or from direct measurements of the particular
property.
REFERENCES