5.3. Graphical Analysis of Heat Penetration Data 151
deteriorates with repeated use. Tong and Lenz (1993). Have determined the
dielectric properties of bentonite pastes at different temperatures.
For studying convection, a wide range of materials have been used: water, solu-
tions of ethylene glycol, sugar solution, oils with thermal properties independent
of temperature and various solutions of known viscosity.
For conduction heating a range of suspensions, including agar, and starch-based
materials are available. More recently, silicone elastomers (Dow Coming Lid.,
Barry, South Glamorgan) have been used, e.g. Sylgard 170 and 184; although
these are expensive materials, they give very reproducible results when used for
simulating conduction-heating packs. The thermal diffusivity is, however, some-
what lower than that of water and water-based food products: 1.02 x 10-7m? s~!
compared with 1.4-1.6 x 10~7m? s~! (Bown et al. 1985).
The thermal diffusivities for some simulants are given in Table 2.2.
5.3. Graphical Analysis of Heat Penetration Data
5.3.1. The Linear Plot
‘The simplest method of plotting the results of a heat penetration experiment is
to use linear graph paper. Figure 5.1 shows a typical heating and cooling curve,
Retort temperature
ie mp
Product temperature
measured at the slowest
‘haating point
Temperature
I temperature of product
Time
FIGURE 5.1. Typical temperature profile for steam heating of canned foods in a batch
retort.152 5, Heat Penetration in Packaged Foods
with the retort temperature profile and the corresponding temperature measured
at the critical point. This type of plot is useful for observing the come-up of the
retort and for converting the temperatures to a lethality curve for determining the
Fo-value.
2. The Semi-Logarithmic Plot
If the temperature-time results from a heat penetration experiment are plotted on
multi-cycle logarithmic-ordinary graph paper, itis possible to construct a straight
line through the points. The data for conduction-heating products usually have
a lag period before linear heating is achieved, represented by a curved portion
to the graph. The extent of this depends on the external heat-transfer coefficient
and tends to increase in curvilinearity with increasingly slower heating media or
thicker packaging material. The data for convection-heating packs often follow a
slightly sigmoidal pattern, and are usually less uniform than those for conduction
heating products.
Three methods of representing the results are in common use. The first is
log(Tr — T) v. time ¢ (Figure 5.2). When the logarithm of the temperature dif-
ference between the retort {emperature and the critical point temperature, known
as the temperature deficit, (Te — 7), is plotted against the time, then a plot is
obtained with curvilinear and linear portions with negative slope. When the deficit
is greater than one then three-cycle logarithmic paper is satisfactory; however,
when the deficit is less than one four cycles will be required. This method is
mainly used in theoretical work.
The second method is logl(Te — T)/(Tr ~ To)] v. time ¢ (Figure 5.3), where
(Tr —T)/(Tp — To) is the reduced temperature and Tp the initial temperature of
the food in the can; in this case a graph similar to Figure 5.2 is obtained. This
method is mainly used in theoretical work.
‘The third method is log 7 v. ¢ (Figure 5.4). If the logarithmic scale is arranged
to run in the opposite direction to the previous two (inverted scale), and the top
line of the graph is numbered 1° below the retort heating temperature, the next
line is numbered 2° below, and so forth, then a linear plot with positive slope is
obtained. Using this method the lines are numbered with the actual temperatures
and no conversion is necessary.
‘The cooling curves may be represented in a similar manner,
5.3
A straight line can be obtained from the log temperature-time or semi-log plots by
drawing the asymptote to the curve (Olson & Jackson 1942; Ball & Olson 1957).
‘The equation of this line is then fully defined by its intercept and slope.
The intercept is obtained by extrapolating the asymptotic line to the axi
defining a temperature T4 such that
3. Analysis of Heat Penetration Graphs
and
Tr — Ta = j(Tr- Ta) (S.1)5.3. Graphical Analysis of Heat Penetration Data 153
Ta-T (°C)
L 1 1 L 1
° 0 2 go 40 50
Time (min)
FIGURE 5.2. Temperature deficit plot.
or
Tr-Ts
Tot (62)
i
where j is known as the lag factor, since it measures the lag in establishing a
uniform heating rate; it is also designated j, to represent the heating part of the
process, and j, is the corresponding value for cooling. Part of the lag is due
to the slow come-up of the retort, and this is accounted for by determining a
new zero time for the process. Ball and Olson (1957) used 58% of the come-up
time as making a useful contribution to the process and this is widely accepted.
It corresponds to adding 42% of the come-up time (o process time at retort
temperature, The j-value for this process is found by using the intercept of the154 5, Heat Penetration in Packaged Foods
2
08
06
0s
os
03
o2
tet
TanTo,
0.4
0.08
0.05
0.04
0.03
0.02
oot L L L L L
° a a ne a |)
Time (min)
FiGurE 5.3, Reduced temperature plot.
line with the new zero axis, T; (see Figure 5.5). For this purpose
Tri
ja tit (5.3)
Tr-To
‘The slope of the line is given by the tangent of the angle between the line and
the t-axis, although it is best obtained as the time for the curve to traverse one
logarithmic cycle. ‘This time is known as the f-value and is the time to reduce
a given temperature to one-tenth of its value. The f-value is the reciprocal of
the slope and cotangent of the angle between the line and the time axis. For the
heating phase it is designated f,, and for the cooling phase fo.
‘Combining these two heat penetration factors in the straight-line equation y =
mx +c gives the equation
logu = -1/f +log j, (4)
where w is the reduced temperature, (Tr — T)/(TR — To), the slope is —1/f and
the intercept log j. This equation may also be re-expressed as
= flog(i/u)5.3. Graphical Analysis of Heat Penetration Data 155
114 Tact?
Tere
Ta?
| tea
TS
Ta
F Tato"
Tr-20°
+ Ta-30°
4 Taso"
+ Ta-50°
Tp 60°
Ta-70°
F Tatoo
E L L L L L 3
0 jo 2 30 40 50 60
Time (min)
FiGuRE 5.4. Inverted scale plot for Tx = 115°C and Ty = 62°C.
or
J MOUE = je 00 66)
The value of j ranges from approximately 1 to 2, although both smaller and larger
values are found in practice. The value, as can be shown theoretically, depends on
the position at which the temperature is measured, the shape of the container or
body being heated and cooled, and the initial temperature distribution.
The value of f is expressed in minutes, and depends on the thermal properties,
in particular the thermal diffusivity, and the dimensions of the object being heated.
‘The faster the rate of heat penetration, the steeper the slope and the smaller the156 5. Heat Penetration in Packaged Foods
120 Trt
Retort temperature = 121°C
n9 Tre
18 Te3
7 Te
116 TeS
15 Tr8
13 Tr8
1" Tr=10
tot Tr-20
Sor Taoo
at Ta-40
n Ta-50
Ta60
Tr-80
Tr=100
Corrected intial temperature T; = 48°C
Teh
5
~79 Tr-209
179 Tr-00
}— Retort temperature up to “|
y2rec = 12min 4
L 1 L |
2 30 +40 «50 «0
fa) Time (min)
FIGURE 5.5. Heat penetration curves: (a) heating phase.
‘f-value. For most conduction-heating packs fy is greater than 20 min, and for
convection-heating packs values of between 4 and 11 min for the common sizes
of cylindrical container are obtained. Table 5.3 shows some typical values of fx
for static steam-heating processes.
‘The values for f- tend to be larger because of the slower heat transfer with water
compared to steam. An empirical guideline is that fe = 1.3 fi, which is confirmed
from the work of Hurwicz and Tischer (1956) for the processing of beef. These
workers also found a greater variance in the values of f. compared with fy
For broken-heating curves (see Figure 5.6) there are two fy-values, usually
designated f, and f>, and a break point, Tp, at time tp. The retort temperature
tends to influence the value of f and the break point parameters tp and Tp. The
closing temperature also affects the break point. The can size affects both fi and
‘fr. For a 3.5% suspension of bentonite in water in No. 2 cans the values of j varied5.3. Graphical Analysis of Heat Penetration Data 157
300
220 4 7.4200
Tinintercept
T-inital temperature atstartot cooing Tet 100
20 Ts70
80 T.s60
70 150
a” 4 1440S
2 5
= 50 4790
40 4 1.120
30h rH10
28
28
25
28 Te
23 Te
2 Te?
Test
ay sp
OO) Time (rin)
FIGURE 5.5. Heat penetration curves: (b) cooling phase.
TABLE 5.3. Some typical Jf,-values for products processed in
static retorts using steam heating,
Conduction heating Convection heating
Can size (ain) (cain)
211 x 202507 25 40
211 x 301 picnic 34 45
211 x 400 AT 9 50
300 x 20787 u 45
300 x 4083 UT a 45
301 x 409162 2 35
307 x 408 A2 8 60
401 x si1A2d 8 10
603 700 A10 198 110158 5. Heat Penetration in Packaged Foods
120 Tet
19
ne
7
16 al
{= 56-19 = 43min
a5 f (conduction
A (conduction) q
M3 4
mr Tp, tp breakdown qi
f=18-6=7min
101 {convection}
Tato)
Corracted zero rate 6.0 min 4
A ty-100
eh ages
ja C5 = Ht won
Tae” ae aes
pg
01020 40 3060
Time (rin)
FIGURE 5.6, Example of broken heating curve.
from 0.99 to 1.72; fi from 4.87 to 6.84 min; fz from 44.6 to 62.0, and fp from 6.7
to 8.3 min (Jackson 1940; Jackson & Olson 1940). Baked beans in tomato sauce
exhibit a broken heating curve; initially the pack shows typical convection heating
behavior and then as the tomato thickens by gelation of the starch leached from the
beans, the pack shows increasing conduction behavior. For an A10 can processed
al 121.1°C, and heated for 35 mins typical values for the heat penetration factors
are fi = 8.7 min and fy = 39.8 min and fy = 10 min. A typical broken heating
curve for this type of product is shown in Figure 5.6. (Tucker 2002). The problems
associated with this type of process are dealt with in Section 6.4.2.3.
Giannoni-Succar and Hayakawa (1982) reported values fi; = 29.37 and fe =
34.45 min for 8%w/w bentonite in can size 307 x 115, with corresponding j
values of 2.13 and 1.56 respectively. The ratio f/ fi, Was 1.17, which is in general
agreement with the quoted value of 1.3.54, Theoretical Analysis of Heat Penetration Curves 159
Bichier et al, (1995) found that for tomato sauce in 300 x 407 cans in a batch
retort the fy = 44.4-45.2 min, fe = 45.2 min, jy = 1.69 & je = 1.69, whereas
for an agitated cook in a Steritort retort fy, = 3.7-4.45 min 7 jy = 1.44-2.05 with
the fe & je very similar. The value for the stationary process agrees well with the
value in Table
5.4. Theoretical Analysis of Heat Penetration Curves
5.4.1. Conduction-Heating Packs
In Chapter 2 the general equation for determining the temperature distribution in a
cylindrical container of conduction-heating food was presented (Olson & Jackson
1942; Ball & Olson 1957). From this series solution, a first-term approximation
can be derived which applies to any point (r, z) in a cylinder of radius r and
height 2b:
«= Antodo(Rir/a) sin (2S) eeieertanier, 67)
where u = (Tk — T(0))/(Tr — To), Ario is a function of the initial temperature
distribution (equal to 2.03970 for a uniform distribution), Jmr(x) is the mth-
order Bessel function of x, Rn the nth positive root of Jo(x) and a the thermal
diffusivity. The first-term approximation is only valid for long processing times,
and for the purposes of the cooling part it is necessary to examine whether this
criterion applies.
Equation (5.7) is of the same form as the exponential heat penetration equation
(5.6), viz.
30% /f
from which
2.308
aS 5.8)
f [Ri /a? + 1? /4b7] 68)
a+b)
J = Aniodo(Rir/a) sin CS) 69)
Ry = 2.404826,
There are two important conclusions from this comparison: first, from equa~
tion (5.8) the f-value depends on the thermal diffusivity and the container dimen-
sions; it is the same at all points inside the container, since it is independent of
z and r. Second, from equation (5.9) the j-value depends on the initial temper-
ature distribution and the position in the can; for conduction-heating packs the
theoretical value of j at the center will be 2.03970.160 5. Heat Penetration in Packaged Foods
The above analysis applies to containers heated by steam; if there is a finite
heat-transfer coefficient on the surface of the container then the equations given
in Chapter 2 should be used. This is discussed further in Section 5.5.6.
5.4.2. Convection-Heating Packs
Using the uniform internal temperature model, discussed in Chapter 2, the basic
equation for the temperature in the fluid core is
w= 10-04Kt, (5.10)
where w is the reduced temperature (Tx — T(1))/(Tp — To) and K = UA/pV, in
which U is the overall heat-transfer coefficient, A is the internal surface area of
the container, ¢ is the specific heat and p is the density.
‘Comparing equation (5.10) with the heat penetration equation in the same form,
u=j-10-/, GAD
we obtain
f = 2.303/K = 2.303¢pV/UA (6.12)
=1
For many products the value of cp is constant, so that f is directly proportional
to the volume of the system and indirectly proportional to the internal area of the
container and the heat-transfer coefficient. Although in theory j = I, in practice
a wide range of values, some less than I, are observed. The reason put forward
for this is that the U-value is initially high because of the high driving force AT,
and with increasing time this decreases progressively to a constant value. The
fa-value will be small initially and will increase to a constant rate of heating. The
initial low value will favour low values of j, which will be less than the theoretical
value of 1
Datta (1990) examined the case of natural convection in canned foods and con-
cluded that there was no justification for the existence of an asymptotic straight-
line temperature-time relationship, because of the complexity of the temperature
field and the flow patterns, However, he concluded that over short ranges of
processing time the errors would be sufficiently small to use the approximation for
practical purposes. It should be remembered that the method is entirely empirical
and care should be taken in the evaluation of convection heating.
Schmidt and Pflug (1966) studied the effect of solids (glass spheres of varying
sizes) on the heat penetration factor f, for jars of water/spheres. They found that
the fi-value decreased with increasing surface area-volume ratio for the spheres,
but did not vary with sphere size. Jars of water alone had a higher f, than filled
jars,5.5. Factors Affecting Heat Penetration 161
Ikegami (1977) derived an empirical convection equation for heat transfer into
a can of solids with a covering liquid:
Gif tip 513
ape (6.13)
where f and j apply to the solid portion and fz and j’ to the liquid portion.
From various experimental results the j-values for the solid portion were approx-
imately 2.0, the j’-values for the liquid portion were below 0.5. The formula was
rigorously derived in Ikegami (1978).
‘Akterian (1996) developed a model for predicting heat penetration data for
convection heat fruits in syrup and vegetables in brine packed in glass jars. This
‘was based on the uniform internal temperature model (Section 2.3), and used the
used concept of the coefficient of thermal inertia E = fi/loge10. The values for
the fi calculated from E values ranging from 7-25 min are much greater than
those given in Table 5.3 for canned products because of the thickness of the walls
of the glass bottles and the effect of the external heat-transfer coefficient, A table
giving experimental data, including process data, E- & F-values and heat-transfer
coefficients, for twenty products is useful for manufacturers using glass jars.
5.4.3. Computer Modeling
A computer model, NumeriCAL, has been developed by Manson (1992) for
determining heating factors from heat penetration data. The program has the
advantage of not using the Ball 42% come-up time but calculating this effect
from the actual processing conditions. This means that pilot-plant data can be
used more reliably. An extension of this model to determine the fe and je factors
for the cooling phase was proposed.
5.5. Factors Affecting Heat Penetration
5.5.1. Effect of Container Shape and Dimensions
SS... fi-value
The fy for a Cylindrical Container of Conductive-Heating Product is Given by
the Equation
Sh = 0.398 /[a(1/a? + 0.427/4b)], (5.14)
and is obtained by substituting the values for R and 7? in equation (5.8). This
equation can be used for determining the fy-value for a container of different
dimensions, with the same food product. Thus if the first container has a value f;
and dimensions radius a and height 2b and the second container an unknown f2
and dimensions ¢ and 2d, respectively, then the value of f2 can be obtained by