Journal of Food Engineering 4 (1985) 157-I 68

Thermal Conductivity of Tomato Paste*

A.E. Drusas and G.D. Saravacosf

Chemical Engineering Department, National Technical University,

GR-106 82 Athens, Greece

ABSTRACT

Experimental measurements of the thermal conductivity and thermal

diffusivity of tomato paste at concentrations of 27-44 Brix are reported.
The thermal conductivity was measured in a guarded hot-plate apparatus
while the diffusivity was estimated by a simplified transient method. The
thermal conductivity (X) values fell in the region of 0.460-0.660 W m-l
K-l, decreasing with increasing solids concentration and increasing as the
temperature was raised from 30 to 5OC. The temperature effect was less
pronounced at higher solids concentration. The thermal diffusivity of
tomato paste at 35 Brix and 20C was estimated as 1.42 x 10e7 mz s-l,
which is in good agreement with data from the steady-state method.

INTRODUCTION

Thermal conductivity is an important thermophysical property of foods

which is useful in the design and operation of food processing equip-
ment and in the processing and storage of food products. Tomato paste
is produced in large quantities by vacuum concentration of tomato
juice and is heat-sterilized in cans. It is markedly non-Newtonian and of
high apparent viscosity, and heat is transmitted mainly by conduction.
Slow heat penetration during heating or cooling may cause undesirable
changes in quality, particularly if the product is heated or cooled in

*Presented at the 44th Annual Meeting of the Institute of Food Technologists,

Anaheim, California, lo-13th June, 1984.
t Present address: Department of Food Science, Rutgers University, New Brunswick,
New Jersey 08903, USA.
157
Journal of Food Engineering 0260-8774/85/$03,30 - 0 Elsevier Applied Science
Publishers Ltd, England, 1985. Printed in Great Britain.
158 A. E. Drusas, G. D. Saravacos

large containers. It is therefore important that the thermal conductivity

or the thermal diffusivity of this product is known with reasonable
accuracy, if more accurate calculations of the heating and cooling pro-
cesses are to be made.
The literature contains insufficient data on the thermal conductivity
and diffusivity of tomato paste. The only two citations of Kolarov
(1969) and of Maslikov and Medwedew (1969) (K.ostaropoulos, 197 1)
give incomplete data which are not in agreement with each other.
The experimental data of the co-operative project COST 90, reported
by Kent et al. (1984), include materials similar to tomato paste, with
which a comparison can be made. The models for predicting thermal
conductivity of liquid foods were reviewed by Cuevas and Cheryan
(1978).

METHODS OF MEASUREMENT

Experimental methods for measurement of thermal properties of foods

have been reviewed by Nesvadba (1982). The transient methods were in
general favoured in the COST 90 project, because they are simpler and
quicker than the steady-state methods, although they involve more
complicated calculations.
The steady-state methods have the advantage of greater accuracy and
simpler calculations, but they require equipment with small dimensional
tolerances and rigorous experimental procedure.
For the purpose of comparison, two experimental methods were
selected: (1) a steady-state method using a guarded hot-plate apparatus:
and (2) a transient method involving measuring the temperature of the
centre of a cylindrical sample as a function of time, while the external
temperature is kept constant (Dickerson, 1965 ; Porsdal-Poulsen, 1982).
The thermal conductivity (X) of a sample of thickness A.x is given by
the equation

where 4 is the heat flux and AT is the temperature difference. In SI

units, X is expressed in W m-l K-l.
The transient heating of a product is represented by the Fourier
eauation
 Thermal conductivity of tomato paste 159

dT a2T
-_=(x-
(2)
at a.2
where (Y= X/PC,, the thermal diffusivity (usually in m2 s-l), which is
assumed to remain constant.
The transient heating time (t) of a product in a cylindrical container
is given by the empirical equation (Ball and Olson, 1957)

(3)

where f = reciprocal of the slope of the heating line, j = the lag factor,
T, = initial centre temperature, T, = external temperature and T =
centre temperature after time t.
For a long cylindrical container, with high surface heat transfer
coefficient, eqn (3) is written as:
r2 Te- To
t = 0.398 ;log 1.6 (4)
T, - T >
where r = internal radius of the cylinder. For a given time-temperature
plot, the inverse slope (f) can be determined and the thermal diffusivity
is obtained from the relationship
01= 0.398 rlf (5)

EXPERIMENTAL

Steady-state measurements

Since there is no commercially available guarded hot-plate instrument

for measuring the thermal conductivity of liquids and pastes, it was
necessary to construct a laboratory apparatus according to specifica-
tions published in the literature. The basic design used was the ASTM
(1968) apparatus, as modified for liquids by Fritz and Poltz (1962)
and Poltz (1965).
The apparatus is shown schematically in Fig. 1. The main unit was
machined from brass slabs to the dimensions specified below. The
homogeneity of the metal was checked by gamma radiography. The
upper (hot) plate (P,) was of 98 mm diameter and 10 mm thick. An
160 A. E. Drusas, G. D. Saravacos

I r 1

Water

Water

Fig. 1. Diagram of guarded hot-plate apparatus: P,, PJ, hot and cold plates; Pz, Pq,
guard rings; R, electrical resistance; 1,2, thermostats; 3, power supply and measure-
ment; 4, data logger.

electrical resistance (R), made of manganin wire of O-376 mm diameter

and specific resistance 3.88 R m-l, was embedded in the plate. The
upper guard ring (I*) of 149 mm diameter and 23 mm thickness
surrounded the hot plate, and a gap was formed between them by
Teflon spacers, 2 mm at the top of the plate and 1 mm at the sides.
The lower (cold) plate (P3) had a diameter of 108 mm and a thick-
ness of 23 mm. It was surrounded by guard ring P4 of 181 mm diameter
and 23 mm thickness, from which it was separated by two l-mm thick
Teflon rings. The surfaces of the plates which contact the sample were
machined to ensure a uniform sample thickness.
Electrical input to the upper plate was controlled by a power
stabilizer (4) and measured with an indicating digital ammeter and
 Thermal conductivity of tomato paste 161

voltmeter. Three copper-constantan thermocouples, insulated with

magnesia, were used in the upper plate, and two in the lower, to measure
the temperature of the plates during the experiments. The thermo-
couples were connected to a data logger, which recorded the tempera-
ture with a precision of +O.OlC. The plate assembly was supported in
a horizontal position on three arms with levelling screws. Two thermo-
statically controlled water baths of 12 litres volumetric and 2 kW
electrical capacity were used to maintain the desired temperature
constant to within +O.lC.
Commercial canned tomato paste from Southern Greece was used in
all experiments. The tomato paste was manufactured by the cold-break
method and was stored for between 6 and 12 months before the
measurements. The concentration of this paste varied from 32 to 36
Brix, as measured with a laboratory refractometer. More dilute samples
were prepared by adding distilled water to the tomato paste and mixing
thoroughly, and more concentrated samples were prepared by vacuum
concentration at 10 torr and 5OC.
Most of the measurements were made using a thickness of tomato
paste of 2 mm between the hot and the cold plates. The two plates
were kept the desired distance apart by Teflon spacers of the appropri-
ate thickness (e.g. 2 mm). Four hard Teflon cubes were placed sym-
metrically on the lower guard ring and one in the centre of the cold
plate. The gap thickness was confirmed by micrometer measurement.
Some measurements were made on 1 mm thickness of tomato paste and
some on aqueous solutions. The tomato paste was spread on the lower
plate in excess, and the upper plate was placed and moved on top of it,
forcing to the sides the excess paste and any trapped air.
The two water baths were then set and maintained at the selected
temperature until the two plates and the sample all attained the same
temperature. Then, the upper plate was heated by the electrical resis-
tance to establish the desired temperature difference (2-5C) between
the two plates. The upper guard ring was maintained at the same tem-
perature as the hot plate by using thermostat 1 (Fig. 1). After a
steady state was established, the temperature difference (A7) and the
heat flux (4) were recorded. Duplicate measurements were normally
made and the average values are reported. For a statistical validation
of the results, five replicate sampies from the same can of the 36
Brix tomato paste were measured at three different temperatures (30,
40 and 500.
162 A. E. Drusas. G. D. Saravacos

Transient measurements

A simple thermal diffusivity apparatus, like that originally used by

Dickerson (1965) and modified by Porsdal-Poulsen (1982), was con-
structed and used in this work. The basic unit consists of a brass tube
of 48 mm inside diameter, 1 mm wall thickness and 200 mm length,
with Teflon caps at the ends (Fig. 2).
The tube, filled with the sample, was placed in a water bath, main-
tained at a constant temperature (+O.lC). The temperature of the
centre of the sample was measured with a copper-constantan thermo-
couple, which was supported in an open glass tube of 1 mm inside
diameter, and inserted along the axis of the cylinder. Another thermo-
couple was used to measure the bath temperature. The two temperatures
were recorded at intervals of 2 min in a data logger.

EXPERIMENTAL RESULTS

Table 1 and Fig. 3 show the measured thermal conductivity (h) of

tomato paste at six concentrations and three temperatures, using the
guarded hot-plate apparatus. The mean value of h (five replicates) of
the 36 Brix paste increased from O-525 to 0.557 W m-l K- as the
temperature was raised from 30 to 5OC. The precision of the measure-
ments was quite satisfactory, the relative standard deviation ranging
from 1 .l to 2.9%.

Food
Sample

1 Water Bath

Fig. 2. Diagram of thermal diffusivity apparatus.

 Thermal conductivity of tomato paste 163

TABLE 1
Thermal Conductivity (X) of Tomato Paste

A(Wm-K-)

oBi-lk JGV 30c 40C 50C

27 0.708 0.595 0.630 0.660

30 0.678 0.560 O-590 0.620
32 0.657 0.550 0.575 0.600
36 0.618 0.525a 0.546 0.557a
39 0.588 0.485 0.505 0.512
44 0.538 0.460 0.478 0.490

a Standard deviations of(h) of the 36 Brix samples: s = 0.0068 (30C), s = 0.0158

(4OC) and s = 0.0062 (5OC).

I
kO.70 - OBrix

o 27
0.65 -
ii% 0
.J)
0.60 - 0 . 0 32
0

0.55- : t t xJ

t
0.50 - 0 0 39
0 l 44
.
.
0.45 -

Fig. 3. Thermal conductivity versus temperature of tomato pastes (27-44 Brix).

The data of Table 1 are plotted in Fig. 3 as h versus T and in Fig. 4 as

X versus X,. The water fractions (X,) of the pastes were calculated
from the Brix values, using the conversion tables of Brix to total
solids for tomato paste (NCA, 1950). Diluted samples at concentrations
below 27 Brix presented problems of phase separation (serum/sus-
164 A. E. Drusas, G. D. Saravacos

0.70
k
+sooc
W0.6E
+ 040C
ix
0.60 + o l 3ooc
0
.
0.5: 2
I
+

cl5c
0

2 l

0.4:

I I I

050 (160 0.70

x,

Fig. 4. Thermal conductivity versus concentration of tomato pastes (30-5OC).

pended solids) during the experimental measurement. As a result,

inconsistent values of thermal conductivity were obtained, particularly
at higher temperatures. Linear regression analysis of the experimental
data for the 36 Brix paste (Fig. 3) resulted in the following equation:

X = 0.482 + 0.0015T (6)

where X is in W m-l K-l and Tin C.
The correlation coefficient (Y = 0.786) was rather poor, suggesting
that a non-linear model might be more appropriate.
Figure 4 shows a sharp increase in X at any temperature with increas-
ing water content (X,). The data appear to deviate from the linear
relationship. More experitiental data are needed for estimating an
appropriate non-linear model.

Transient method

Figure 5 shows the temperature of the tomato paste (35 Brix) in the
centre of the thermal conductivity apparatus. It rises exponentially
with time and there appears to be a change in the slope of the heating
line near 30C. Similar results were obtained with tomato paste of 32
Brix. Diluted samples at concentrations less than 30 Brix gave unsatis-
 Thermal conductivity of tomato paste 165

factory plots, evidently due to the separation of the serum from the
tomato paste.
Considering the lower part of the heating line of Fig. 5, the slope
was found to be f = 1612 s, and the thermal diffusivity of tomato
paste of 35 Brix at about 20C was calculated from eqn (5) to be
(11= 1.42 X lo- m2 Cl.
In order to calculate the thermal conductivity h of a material from
its known thermal diffusivity ar, the density p and the specific heat C,
are needed. The density of the tomato paste at 35 Brix was determined,
by the standard gravimetric method, as p = 1163 kg rne3. The specific
heat was estimated from its water content (X, = O-627) after Miles er
al. (1983) to be C, = 3302 J kg- K-l. From these data, the thermal
conductivity of the tomato paste at 35 Brix was calculated to be
h = 0.546 W m-l K-l.
The extrapolated value of h for the tomato paste at 35 Brix and
20C, as determined by the hot-plate method (Fig. 3), is O-520 W m-l
K-i. The values of X, determined by the two different methods, differ

t, min

Fig. 5. Centre temperature of tomato paste (3.5 Brix) in the thermal diffusivity
apparatus versus time.
166 A. E. Drusas, G. D. Saravacos

by about S%, which can be considered as a satisfactory agreement for

engineering purposes.

DISCUSSION AND CONCLUSIONS

The values determined for the thermal conductivity of tomato paste at

concentrations 32-44 Brix and 30-50C (0.460-0.660 W m-l K-l) are
similar to those for aqueous sucrose solutions of similar concentration
(0.473-0.574 W m-l K-l) reported in the literature (Kostaropoulos,
1971). At lower concentrations, the values for tomato paste were in
general higher than the sucrose solutions and they approached the
thermal conductivity of water. The values of X for tomato concentrates
reported by Kolarov and by Maslikov and Medwedew, as cited by
Kostaropoulos (197 l), are smaller than the present results by nearly
20%.
The correlation of thermal conductivity for fruits and vegetables
proposed by Sweat (1974) gave significantly lower values than those
determined experimentally for tomato paste. No satisfactory agreement
could be found between the present data and the correlations of h with
water content and temperature compiled by Miles et al. (1983).
The thermal conductivity of tomato paste is similar to the values
obtained for apple pulp, meat paste and fish paste (Kent et al., 1984).
The X of these materials increased slightly with temperature, following
a pattern similar to that for tomato paste at high concentrations (Fig. 3).
The experimentally determined thermal diffusivity of tomato paste
(CX= 1.42 X lo- m2 s-l at 35 Brix and 20C) is similar to the values
reported in the literature for water, _7% carrageenan gels and apple pulp
(Kent et al., 1984), and some other fruits and vegetables (Kostaropoulos
et al., 1975). The break in the heating line of tomato paste (Fig. 5) is
similar to those observed in other food materials (Ball and Olson, 1957).
Tomato paste is markedly non-Newtonian (Saravacos et al., 1967;
Fito et al., 1983), and temperature has a small effect on its thermal
conductivity and apparent viscosity. Heat transfer in tomato paste can
be improved by high shear rates (e.g. by agitation) which results in
reduced apparent viscosities (Saravacos, 1970, 1974).
In measurements on diluted tomato paste and aqueous solutions with
the guarded hot-plate apparatus, convection may cause apparently
greater thermal conductivities. This effect may also explain the high
thermal conductivity of water observed in this apparatus.
 Thermal conductivity of tomato paste 167

COMPUTER PREDICTION

The experimental thermal conductivity data of tomato paste at 27, 30,

32, 39 and 44 Brix were compared with the computer predictions of
the COSTHERM program, developed in the COST 90 project (Miles
et al., 1983). The tomato paste was assumed to consist of carbohydrates
(% Brix) and water. A difference of 6% between the predicted and the
experimental values was found for thermal conductivity, and the agree-
ment can be considered as satisfactory. The values of thermal diffusivity
differed by 16%, and this difference is attributed to the experimental
method. The measurements in the cylindrical apparatus were influenced
by the physical state of the semi-fluid sample (tomato paste), and a
significant variation in the results was noticed.

ACKNOWLEDGEMENT

We acknowledge the help of Ir. G. van Beek of the Sprenger Institute,

Wageningen, The Netherlands, who ran the COSTHERM computer
calculation.

REFERENCES

ASTM (1968). Standard method of test for thermal conductivity of materials by

means of the guarded hot plate. American Society for Testing and Materials.
ASTM Standard C 177-63.
Ball, C. 0. and Olson, F. C. W. (1957). Sterilization in Food Technology, Academic
Press. New York.
Cuevas, R. and Cheryan, M. (1978). Thermal conductivity of liquid foods - A
review. J. Food Process Engineering, 2, 283-306.
Dickerson, R. W. (1965). An apparatus for the measurement of thermal properties
of foods. Food Technology, 19, 198-204.
Fito, P. J.. Clemente, G. and Sanz, F. J. (1983). Rheological behaviour of tomato
concentrate (hot break and cold break). J. Food Engineering, 2, 51-62.
Fritz, W. and Poltz, H. (1962). Absolutbestimmung der Warmeleitfahigkeit von
Fliissigkeiten - I. Kritische Versuche in einer neuen Plattenapparatur. Znt. J.
Heat Mass Transfer, 5, 307-16.
Kent, M., Christiansen, K., van Haneghem, 1. A., Holtz, E., Morley, M. J., Nesvadba,
P. and Poulsen, K. P. (1984). COST 90 collaborative measurements of thermal
properties of foods. J. Food Engineering, 3. 117-50.
168 A. E. Drusas, G. D. Saravacos

Kolarov, K. (1969). Konserwnaja i. Owostsches Promischl., 24 (S), 34-6.

Kostaropoulos, A. E. (197 1). Wtirmeleitzahlen von Lebensmitteln und Methoden
zu deren Bestimmung. VDMA, No. 16, Maschinenbau Verlag, Frankfurt.
Kostaropoulos, A. E., Spiess, W. E. L. and Wolf, W. (1975). Anhaltswerte fur die
Temperaturleitfahigkeit von Lebensmitteln. Lebensm.-Wiss. u. Technol., 8,
108-10.
Maslikov, W. A. and Medwedew, 0. K. (1969). Pistschewaja Technol., 64 (6), 69-70.
Miles, C. A., van Beek, G. and Veerkamp, C. H. (1983). Calculation of Thermo-
physical Properties of Foods. In: Physical Properties of Foods, eds R. Jowitt
et al., Elsevier Applied Science Publishers Ltd, London.
NCA (1950). Tomato Products. National Canners Association, Research Bulletin
No. 27-L, Washington, DC.
Nesvadba, P. (1982). Methods for the measurement of thermal conductivity and
diffusivity of foods. J. Food Engineering, 1,93- 113.
Poltz, H. (1965). Die Warmeleitfahigkeit von Fliissigkeiten III. Abhangigkeit der
Wlrmeleitfahigkeit von der Schichtdicke bei organ&hen Flussigkeiten. ht. J.
Heat Mass Transfer, 8,609-20.
Porsdal-Poulsen, K. (1982). Thermal diffusivity of foods measured by simple equip-
ment. J. Food Engineering, 1, 115-22.
Saravacos, G. D. (1970). Effect of temperature on viscosity of fruit juices and
purees. J. Food Sci., 35, 122-5.
Saravacos, G. D. (1974). Rheological Aspects of Fruit luice Evaporation. In:
Advances in Preconcentration and Dehydration ojFoods. ed. A. Spicer, Elsevier
Applied Science Publishers Ltd, London.
Saravacos, G. D., Oda. Y., and Moyer, J. C. (1967). Tube viscometry of tomato
juice and concentrates. Report. NY State Agric. Exp. Station. Geneva, New
York, 11 pp.
Sweat, V. E. (1974). Experimental values of thermal conductivity of selected fruits
and vegetables. J. Food Sci.. 39, 1080-3.