International Journal of Food Properties

ISSN: 1094-2912 (Print) 1532-2386 (Online) Journal homepage: https://www.tandfonline.com/loi/ljfp20

Thermal Properties of Sweet Potato with its

Moisture Content and Temperature

Adefemi Farinu & Oon-Doo Baik

To cite this article: Adefemi Farinu & Oon-Doo Baik (2007) Thermal Properties of Sweet Potato
with its Moisture Content and Temperature, International Journal of Food Properties, 10:4, 703-719,
DOI: 10.1080/10942910601137482

To link to this article: https://doi.org/10.1080/10942910601137482

Copyright Taylor and Francis Group, LLC

Published online: 30 Oct 2007.

Submit your article to this journal

Article views: 3306

View related articles

Citing articles: 19 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=ljfp20
International Journal of Food Properties, 10: 703–719, 2007
Copyright © Taylor & Francis Group, LLC
ISSN: 1094-2912 print / 1532-2386 online
DOI: 10.1080/10942910601137482

THERMAL PROPERTIES OF SWEET POTATO WITH ITS

MOISTURE CONTENT AND TEMPERATURE

Adefemi Farinu and Oon-Doo Baik

Department of Agricultural and Bioresource Engineering, University of
Saskatchewan, Saskatoon, Canada

Thermal properties of sweet potato were experimentally determined and modeled as a func-
tion of temperature and moisture content. The purpose is to develop empirical correlations
that could predict thermal properties during sweet potato processing. Thermal conductivity
from the study was 0.49 ± 0.038 Wm-1K-1 (mean ± s.d.), thermal diffusivity was 1.2 ´ 10-7 ±
9.05 ´ 10-9 m2s-1, specific heat was 3660 ± 477.4 Jkg-1K-1, and density was 1212 ± 73.5
kgm-3. All properties were determined within temperature range of 20 to 60°C and moisture
content range of 0.45 to 0.75 w.b. Prediction models for the thermal properties of sweet
potato were developed as a function of product temperature and moisture content with
experimental data from this study. Mechanistic models were also developed for thermo-
physical properties of sweet potato using major food components of the product. Developed
models were all presented and compared.

Keywords: Thermal conductivity, Thermal diffusivity, Specific heat, Material density,

Mechanistic model.

INTRODUCTION
Sweet potato (Ipomoea batatas) is an important vegetable. It is not particularly
related to the regular potato that is sometimes called the Irish potato to distinguish it. The
two crops are from different families. Sweet potato is very nutritious. In rating 58 vegeta-
bles by adding up the percentages of recommended daily allowance (RDA) for 6 nutrients
(Vitamin A and C, folate, iron, copper and calcium) plus fiber, the Nutrition Action Health
Letter rated sweet potato the highest with a whopping 582 points. Its nearest competitor, a
raw carrot came in at 434.[1] The Centre for Science in the Public Interest also rated the
relative nutritional value of common vegetables and once again, the sweet potato came out
on top with a score of 184, as compared with a similarly prepared regular potato, which
scored only 83 points.[1] Sweet potato is a popular staple food universally. It enjoys
immense and increasing popularity in North and South America, Asia, and Africa.[2]
The knowledge of thermal and physical properties of the food is essential for ther-
modynamic research and modeling the heat treatment of foods including vegetables. This
is because the properties, to a large extent, determine the rate of the heat transfer process.
Previously, constant average values of thermal properties were used in analyses of food

Received 18 January 2006; accepted 27 November 2006.

Address correspondence to Dr. Oon-Doo Baik, Department of Agricultural and Bioresource Engineering,
University of Saskatchewan, Saskatoon S7N 5A9, Canada. E-mail: oon-doo.baik@usask.ca

703
704 FARINU AND BAIK

processes which has lead to inaccurate analyses since these properties actually change
during the process. However, modern analytical techniques have made it possible to
accommodate these dynamic changes. The most important thermophysical properties in
food processing, specific heat capacity, thermal conductivity, density, and thermal diffusiv-
ity of food materials depend mostly on the food’s composition, temperature and density.[3]
Differential scanning calorimetry was used to determine the specific heat of Tortilla
chips as a function of frying time. At temperature range of 150 to 190°C, moisture content
range of 1.3 to 36.1% and oil content, 0 to 27.7%, the Cp of tortilla chip was 2560 to 3360
Jkg−1K−1.[4] Specific heat of garlic was also determined as a function of temperature at dif-
ferent moisture content levels. Specific heat was found to increase with increase in tem-
perature ranging from 2400 to 4100 Jkg−1K−1.[5]
A hot wire probe was used to determine thermal conductivity of starch gel with gel
concentration varying from 5 to 30% and within temperature range of 5 to 35°C. The
range of k data from the experiment was 0.5 to 0.62 Wm−1K−1.[6] Line heat probe was used
to measure the effective thermal conductivity of deffated soy flour at room temperature.
The results showed that thermal conductivity increased with increasing bulk density vary-
ing from 0.03 to 0.09 Wm−1K−1.[7] Thermal conductivity of deep fat fried potatoes was
determined from heat penetration into cylindrical samples between 50 and 100°C.[8] The
authors employed numerical solution of the heat transfer equation for an infinite cylinder
with convective boundary conditions to fit the experimental data. Thermal conductivity of
potato from the study was 0.5 to 0.9 Wm−1K−1.[8]
Thermal conductivity has been found to be generally larger in wet materials.[9] At
25°C, tomato and potato were reported to have thermal conductivities of 0.68 and 0.62
Wm−1K−1 at moisture content of 5 kg/kg (dry basis) as compared to thermal conductivities
of 0.3 and 0.15 Wm−1K−1 at moisture content oft0.3 kg/kg (dry basis), respectively.[9]
Another study also reported thermal conductivity value of 0.63 Wm−1K−1 for potato at
moisture content level of 6.1 kg/kg (dry basis) as compared to 0.36 Wm−1K−1 at moisture
content of 2.0 kg/kg (dry basis).[10]
Most of the methods for determining thermal diffusivity reported in literature were
indirect in the sense that they were calculated from thermal conductivity, density, and spe-
cific heat. Direct (probe method) and indirect method was used to determine the thermal dif-
fusivity of granular starch and the result showed that indirect method yielded more accurate
values than the direct measurement as concluded from F-test at 5% significance level.[11]
Despite the great potential of sweet potato, research works on its thermal properties
are scarce in literature. A study of this nature is therefore important. The objective of this
study was to investigate the effect of temperature and moisture content on thermal proper-
ties of sweet potato and to develop empirical models for thermal and physical properties
of sweet potatoes as a function of product temperature and moisture content. Prediction
models based on the thermophysical properties of major components of sweet potato and
their corresponding mass or volume fractions were also developed and compared.

MATERIALS AND METHODS

Sample Preparation
Sunny II sweet potatoes grown in California (USA) were bought in 20 kg lots from
the same store to reduce errors due to variation in sample. Sample sweet potatoes were
used within one week of purchase. They were stored in a cooling chamber at a temperature
 THERMAL PROPERTIES OF SWEET POTATO 705

range of 13 to 16°C and relative humidity of 85 to 90% to maintain its quality.[2] The
tubers were manually peeled with a hand peeler and then cut into cylinders using a cylin-
drical bore and a knife. Samples diameter was 4 cm and 6 cm height. A cylindrical sample
holder was designed for this experiment. It is made up of two parts: the cylindrical con-
tainer in which the sample is to be placed which is made of stainless steel and the cap to
cover the container which is made of Teflon. A small hole was drilled on top of the cap
through which the probe was inserted into the sample.

Thermal Conductivity and Thermal Diffusivity

Thermal conductivity and thermal diffusivity were measured using KD2 thermal
properties analyzer (Decagon Devices Inc., Pullman, USA). It was based on a transient
method of thermal properties determination. The equation describing the working princi-
ple of the KD2 is described below:

q ⎛ ⎛ r2 ⎞ ⎞
T − T0 ≅ ⎜ ln (t )− g − ln ⎜ ⎟ ⎟ , (1)
4p k ⎝ ⎝ 4a ⎠ ⎠

where T is final temperature (°C); T0 is initial temperature before heating (°C); q is heat
produced per unit length per unit time (Wm−1); k is the thermal conductivity of the edium
(Wm−1°C−1); t is time (s); g is Euler’s constant (0.5772…); r is radial distance from the probe
to the point where temperature is being measured (m); and a is thermal diffusivity (m2s−1).
Sweet potato sample was placed inside the sample holder and covered with the cap.
The holder was then held by a retort stand inside an agitated water bath (Thermomix 1441,
B. Braun Instrument San Francisco, CA) that was preset at 20°C (Figure 1). The sample
was held this way for about 15 min to equilibrate. A thermocouple probe was used to

Figure 1 Experimental set up for determining thermal conductivity and thermal diffusivity.
706 FARINU AND BAIK

check the temperature of the sample until a stable reading was obtained for the sample.
The KD2 was inserted into the sample and the sample was allowed about 10 more minutes
to equilibrate. The KD2 was switched on at this stage, and thermal conductivity, thermal
diffusivity, and product temperature were recorded. The process was repeated for 30, 40,
50, and 60°C and four other sets of results were recorded for thermal conductivity, thermal
diffusivity and product temperature. For higher temperatures (40 to 60°C), the packed
sample was first heated in a microwave oven (Panasonic NN-5553C, Matsushita Corpora-
tion, Franklin Park, IL) for 20 to 40 seconds depending on the desired temperature. The
power setting of the oven was adjusted to the maximum, 800 W. The sample was then
transferred to the water bath and allowed about 20 to 40 min to equilibrate before the read-
ing was taken. Moisture content of the sample was determined by oven method after each
experiment to confirm the previous value measured. To simulate an infinite medium,
using a specific heat estimate of 3500 Jkg−1K−1 for sweet potato at 20°C, the radial dis-
tance r, at which (T-T0) from Eq. (1) is zero over a test time of 100 s is 0.51 cm. The radius
of our sample was 2.0 cm which is therefore adequate for this purpose. This is also greater
than the least value of 0.5 cm required to minimize error due to finite sample diameter
using our probe.[4] Prior to determining k of sweet potato, KD2 was calibrated by deter-
mining the k of agar at room temperature. Agar is a polysaccharide that has very similar
thermal properties with water. The ratio of determined k for agar to the theoretical k value
found in literature was used as the calibration factor for our experiment.

Material Density
Material density of samples was measured with a Gas-pycnometer (Multipycnome-
ter, Quantachrome Corporation, Boynton Beach, FL), which uses gas of very small atomic
size as the displacement medium. The very small atomic size of the gas ensures total pen-
etration of the material pores. The pycnometer measures the volume of the sample which
was then adjusted with the mass of the sample to give the material density of the product.
The operational equation of the pycnometer is given below:

⎛⎛ P ⎞ ⎞
VP = VC − VR ⎜ ⎜ 1 ⎟ − 1⎟ , (2)
⎝ ⎝ P2 ⎠ ⎠

where VP is volume of sample to be determined (cm3); VC is volume of sample cell (cm3); P1

is pressure after pressurizing the reference volume only (Pa); P2 is pressure after including
VC. (Pa); and VR is reference volume (cm3). Prior to determining the material density of
sweet potato, the pycnometer was calibrated by determining the reference volume, VR, and
the sample cell volume, VC. This was done by supplying 20 psi of gas to the sample cell
when it contained a steel sphere of known volume and also when it was empty (contained no
sphere) and then measuring two pressure values each from the sample cell and from the ref-
erence volume. Calibrated VR and VC were given by the following equations:

VS
VR = ,
⎛⎛ P′ ⎞ ⎛P ⎞⎞ (3)
⎜ ⎜ 1 − 1⎟ − ⎜ 1 − 1⎟ ⎟
⎜⎝ ⎜⎝ P ′ ⎟⎠ ⎝ P2 ⎠ ⎟⎠
2
 THERMAL PROPERTIES OF SWEET POTATO 707

and

⎛P ⎞
VC = VS + VR ⎜ 1 − 1⎟ , (4)
⎝ P2 ⎠

where VC =volume of sample cell; VR = reference volume; VS =volume of sphere (cm3),

P1′=pressure from the reference volume without steel sphere inside (Pa); P2′=pressure
from the reference volume and sample cell volume without steel sphere inside (Pa);
P1 =pressure from the reference volume with steel sphere inside (Pa); and P2 =pressure
from the reference volume and sample cell volume with steel sphere inside (Pa). The
weight of the sample was measured by a mass balance and the ratio of the mass to the vol-
ume given by the pycnometer represents the material density of the sample. The material
density was determined within a product moisture content range of 0.45–0.75 wb.

Specific Heat
The specific heat for this study was determined indirectly from the thermal conductiv-
ity, thermal diffusivity and the material density previously determined. From the relation:

k
Cp = , (5)
ra

where Cp is specific heat (Jkg−1K−1); k is thermal conductivity (Wm−1K−1); r is density

(kgm−3); and a is thermal diffusivity (m2s−1).

Moisture Content
Moisture content of the sample was varied by drying in a microwave oven at 800 W.
Preliminary test was performed to evaluate moisture loss of the sample versus microwave
drying time. Afterwards, samples were heated for the period required to achieve the
desired sample moisture content. After heating, moisture content was again determined by
oven dry method[12] to confirm the value. Moisture content range used for the study was
0.45–0.75 wb.

Initial Product Temperature

The KD2 thermal properties analyzer simultaneously provides temperature of the
sample while determining its thermal properties. Also the water bath has a temperature
controller which regulates fluctuations within ± 0.1°C to keep the water and the sample at
the required temperature. Initial water bath temperature values used were 20, 30, 40, 50,
and 60°C. Product temperatures were measured and used in developing our empirical
models.

Data Analysis and Thermal Properties Modeling

Thermal conductivity and thermal diffusivity measurements were made with
the KD2, while density measurement was taken with the gas-pycnometer. The data were
708 FARINU AND BAIK

analyzed using PROC GLM of SAS for the effect of individual factors and interactions
between them. Type 1 error rate was set at a=0.05. We have 2 replications, 4 levels of
moisture content and 5 levels of temperature for thermal conductivity and thermal diffu-
sivity experiments (Table 1). Density was not included in the model to avoid multicol-
linearity among the variables since density was related to moisture content. The
correlation between the density of sweat potato and moisture content was:

r = 1.5571 − 0.5743m R 2 = 0.998, (6)

where r is density (kg/m3), and m is moisture content, decimal (wet basis).

Comparison with Mechanistic Models with Major Food Components

Thermal and physical properties of sweet potato were determined from the thermal
properties of its major components namely carbohydrate, protein, fat, fibre, ash, and
water. A previous work[3] was utilised in determining the thermal conductivity, thermal
diffusivity, material density and specific heat capacity of pure protein, carbohydrate, fat,
fibre, ash and water. Choi and Okos[3] used a modified probe method to determine the
thermal conductivity and thermal diffusivity of pure component suspension of each major
food component for a solid range of 10 to 60% and a temperature range of 40 to 150°C.
The authors also used volumetric pycnometer and differential scanning calorimeter to
measure density and specific heat, respectively. The experimental values obtained were
then used by them to develop general mathematical models with the weight fractions or
volume fractions of major components of foods. Group models were developed for indi-
vidual properties.
The maximum and minimum of thermal conductivity of a two-phase system as
series (two phases are thermally in series with respect to the direction of heat flow) and
parallel distribution (two phases are thermally in parallel with respect to the direction of
heat flow) was determined[13] and.[14] Series model is represented below:

1 n
∈
= ∑ i, (7)
k i =1 ki

where ∈i the volume fraction of ith component phase; ki is the thermal conductivity of ith
phase; and k is the effective thermal conductivity of the composite medium predicted by
the series model.

Table 1 Levels of factors for thermal conductivity and thermal

diffusivity experiments.

Temperature Moisture content

Replication (°C) (% wb)

I 20.0 0.45
II 30.0 0.55
40.0 0.64
50.0 0.75
60.0
 THERMAL PROPERTIES OF SWEET POTATO 709

Parallel model is represented by:

n
k = ∑ ∈i ki . (8)
i =1

Both the perpendicular (series) and parallel models were tested in our present study
however the parallel model consistently gave higher values, which were also closer to val-
ues found in literature for all three of specific heat, thermal conductivity and thermal dif-
fusivity of sweet potato and similar products. This is in agreement with a reported study
that the parallel model applies better to non-fibrous materials where thermal properties are
not dependent on direction of heat flow.[15]
For specific heat, the prediction model in terms of food components[3] presented
below was used in our work:

n
C p = ∑ C pi Xi , (9)
i =1

where Xi is the weight fraction of ith component phase; Cpi is the specific heat capacity of
ith phase; and Cp is the effective specific heat capacity. Prediction model used for
density[3] is:

1
r= n
,
X (10)
∑ ri
i =1 i

where Xi is the weight fraction of ith component phase; ri is the density of ith phase; and r
is the density. Thermal diffusivity for this study was evaluated indirectly from values of
thermal conductivity, density and specific heat:

k
a= , (11)
rC p

The models developed by the authors[3] were used to predict the variation of food com-
ponents properties with temperature, while the moisture content in the product was varied
within 0.45 to 0.75 w.b. to predict the variation of properties with moisture content of the food.
Mass ratio of major food components in sweet potato was obtained from literature[16] and is
presented in Table 2. Volume fractions of the components were developed as a function of
temperature and moisture content from the weight fraction and density of the components.

RESULTS AND DISCUSSIONS

Thermal Conductivity
From the experiment, thermal conductivity of sweet potato was 0.49 ± 0.038 Wm− 1
−1
K (mean ± s.d) for temperature range of 20.5–60°C and moisture content range of 0.45
to 0.75 w.b. (Table 3). For statistical analysis, moisture content, m, temperature, T and
710 FARINU AND BAIK

Table 2 Percentage composition of major food

components in sweet potato (adapted from USDA, 1963).

Percentage by mass
Component (%)

Protein 1.7
Fat 0.4
Carbohydrate 26.3
Fibre 0.7
Ash 1.0
Water 69.9

Table 3 Thermal conductivity of sweet potato determined by KD2.

MC (w.b.) 0.75 0.65 0.55 0.45

Trials I II I II I II I II
−1 −1
Temperature (°C) (Wm K )

20 0.51 0.52 0.50 0.50 0.48 0.49 0.41 0.43

30 0.50 0.50 0.50 0.52 0.47 0.47 0.46 0.47
40 0.53 0.54 0.53 0.54 0.52 0.52 0.47 0.47
50 0.56 0.56 0.57 0.55 0.52 0.51 0.50 0.49
60 0.57 0.56 0.56 0.56 0.56 0.56 0.53 0.53

interaction between the factors, m*T were considered. Statistical information for models
tested is presented in Table 4. For moisture content, both linear and quadratic effects on
thermal conductivity of sweet potato were significant (p < 0.05) but only linear effect was

Table 4 Statistical information on models tested.

Model DFEa Pr > F R2

k=y0+aT+bm+cm2 35 < 0.0001 0.88

a=y0+aT+bm 35 < 0.0001 0.74
Cp =y0+aT+bm 35 < 0.0001 0.89
k=y0+a×m×exp(bT) 37 < 0.0001 0.21
a=y0+a×m×exp(bT) 37 < 0.0001 0.73
Cp =y0+a×m×exp(bT) 37 < 0.0001 0.39

⎛ −b ⎞
k = a × exp ⎜ + c × m⎟ 37 < 0.0001 0.76
⎝T ⎠

⎛ −b ⎞
a = a × exp ⎜ + c × m⎟ 37 < 0.0001 0.23
⎝T ⎠

⎛ −b ⎞
C p = a × exp ⎜ + c × m⎟ 37 < 0.0001 0.82
⎝T ⎠

a
DFE: Degree of freedom of error.
 THERMAL PROPERTIES OF SWEET POTATO 711

significant for product temperature. There was no interaction effect. The model developed
was:

k = 0.0397 + 1.0695m + 0.003T − 0.649m 2 R 2 = 0.88. (12)

Thermal conductivity of sweet potato from our mechanistic models with food com-
ponent mass[3] was 0.30 to 0.41 Wm−1K−1 for temperature range of 20–60°C and moisture
content range of 0.45 to 0.75 w.b. (Table 5). Thermal conductivity of sweet potato
increased with moisture content and temperature. Both the parallel and perpendicular
models from the food component model resulted in a lower value of k when compared
with the experimental data but the parallel model gave predictions closer to experimental
values. The RMSE (root mean square deviation) between predictive values from the two
models and our experimental data were 0.094 for parallel model and 0.115 for perpendic-
ular model. RMSE between the measured data and the empirical model developed from it
was 0.005. The simplified regression model with temperature and moisture content devel-
oped from parallel model and food composition of sweet potato was:

k = 0.1613 + 0.0014T + 0.2924 m + 3.4839 × 10 −7 T 2 . (13)

Three dimensional surface mesh plots of thermal conductivity from both empirical
and mechanistic (parallel) models were made for visual comparison (Figure 2 and Figure 3).

Thermal Diffusivity
Thermal diffusivity of sweat potato in this study was 1.0×10−7 to 1.3×10−7 m2s−1
(Table 6). This was comparable with 1.1×10−7 m2s−1 reported for thermal diffusivity of
sweet potato at 40°C.[17] Linear effects of moisture content, m, and temperature, T on ther-
mal diffusivity of sweet potato (p < 0.05) were significant. The detailed statistical infor-
mation for the model component is given in Table 3. The empirical model developed was:

a = 1.1033 × 10 −7 + 3.2572 × 10 −8 m − 3.8765 × 10 −10 T . R 2 = 0.74 (14)

Thermal diffusivity of sweet potato from the food components mass prediction
model was 8.7×10−8 to 1.09×10−7 m2s−1 for temperature range of 20–60°C and moisture
content range of 0.45 to 0.75 w.b. (Table 7). RMSE between experimental data and pre-
dictive model was 2.72×10−8. RMSE between the measured data and empirical model
developed from it was 1.75×10−8. The mechanistic models with the food components

Table 5 Thermal conductivity of sweet potato by prediction model (Parallel).

MC (w.b.) 0.75 0.65 0.55 0.45

Temperature (°C) (Wm−1K−1)

20 0.338 0.327 0.316 0.304

30 0.355 0.336 0.330 0.318
40 0.369 0.356 0.345 0.331
50 0.384 0.371 0.359 0.345
60 0.410 0.386 0.373 0.359
712 FARINU AND BAIK

0.58

0.56
tivity (W/m/K)

0.54

0.52
Thermal Conduc

0.50

0.48

0.46 60
55
50
0.44

)
(°C
45
40

re
0.42

tu
35

ra
0.70

pe
0.65 30

m
0.60

Te
0.55 25
Mois
ture c 0.50 20
onten
t (%) 0.45
wb

Figure 2 Thermal conductivity of sweet potato calculated from experimental model as a function of moisture
content and temperature.

0.40
Thermal Conductivity (W/m/K)

0.38

0.36

0.34

60
0.32 55
50
)

45
(°C

40
re

0.30
tu

35
ra

0.65
pe

30
0.60
m

Moistu 25
Te

0.55
re con
tent ( 0.50 20
%) wb

Figure 3 Thermal conductivity of sweet potato as a function of moisture content and temperature calculated
from food component mass fraction model (parallel).
 THERMAL PROPERTIES OF SWEET POTATO 713

Table 6 Thermal diffusivity of sweet potato determined by KD2.

MC (w.b.) 0.75 0.65 0.55 0.45

Trials I II I II I II I II
−7 2 −1
Temperature (°C) (× 10 m s )

20 1.3 1.3 1.3 1.3 1.2 1.2 1.1 1.2

30 1.3 1.2 1.2 1.1 1.2 1.2 1.2 1.2
40 1.2 1.2 1.1 1.2 1.1 1.1 1.1 1.1
50 1.2 1.1 1.2 1.2 1.0 1.0 1.0 1.1
60 1.1 1.1 1.1 1.1 1.1 1.0 1.0 1.0

Table 7 Thermal diffusivity of sweet potato from prediction model (Parallel).

MC (w.b.) 0.75 0.65 0.55 0.45

Temperature (°C) (×10−8 m2/s)

20 8.89 8.79 8.76 8.72

30 9.27 9.17 9.13 9.11
40 9.66 9.54 9.82 9.50
50 10.5 9.96 9.93 9.91
60 10.9 10.4 10.3 10.3

were simplified using regression modeling approach to relate the thermal diffusivity with
moisture content and product temperature. The simplified model developed was:

a = 7.3248 × 10 −8 + 3.4094 × 10 −10 T + 2.091 × 10 −8 m + 5.3151 × 10 −13 T 2 . (15)

Three dimensional surface mesh plots of thermal diffusivity from both models were
made (Figure 4 and 5). The effect of temperature was predominant as could be seen from
the plot and the model.

Specific Heat
For the same range of temperature and moisture content as for thermal conduc-
tivity, specific heat of sweat potato was 3660 ± 477 Jkg−1K −1 (mean ± sd.) results are
presented in Table 8. From statistical analysis, linear effects of moisture content, m,
and temperature, T were significant on specific heat of sweet potato (p < 0.05). The
detailed statistical result is presented in Table 3. The empirical model developed
was:

C p = 1304.9 + 2300.4 m + 24.662T R 2 = 0.89. (16)

The range of specific heat of sweet potato from the food component mass fraction
model was 2763 to 3558 Jkg−1K−1 for the temperature range of 20 to 60°C and moisture
714 FARINU AND BAIK

1.35e-7

1.30e-7
2 /s)

1.25e-7
ity (m

1.20e-7
Thermal diffusiv

1.15e-7

1.10e-7

1.05e-7
0.75
1.00e-7 0.70

b
)w
0.65

%
9.50e-8

t(
0.60

en
25

nt
30 0.55

co
35

e
40

ur
Tem 45 0.50

st
pera

oi
ture 50

M
( °C ) 55 0.45
60

Figure 4 Thermal diffusivity of sweet potato from an experimental model as a function of moisture content and
temperature.

content range of 0.45 to 0.75 w.b. (Table 9). The simplified mass fraction model with tem-
perature and moisture content developed is:

C p = 1602.4 + 0.9185T + 2543.2 m − 9.7876 × 10 −5 T 2 . (17)

Two 3-dimensional surface mesh plots of specific heat from both models were plotted
(Figures 6 and 7). Observations from the models and the plots suggest that the effect of
moisture content is predominant in determining Cp of sweet potato. RMSE between exper-
imental data and the mass fraction model for Cp was 838.2. RMSE between the measured
data and empirical model developed from it was 550.

Material Density
Material density of sweet potato was determined and modeled as a function of mois-
ture content. Four moisture content levels of 0.75, 0.65, 0.55, and 0.45 were used. The
result from the experiment is presented in Figure 8 and Table 10. Density was modeled as
a function of moisture content, w.b. (Eq. 18). Coefficient of determination (R2) was 0.97
and mean square error (MSE) was 156.1.
Material density of sweet potato from the food component mass fraction model was
1082 to 1218 kgm−3 for the temperature range of 20 to 60°C and moisture content range of
 THERMAL PROPERTIES OF SWEET POTATO 715

1.06e-7

1.04e-7

1.02e-7
ity (m /s)
2

1.00e-7
9.80e-8
Thermal diffusiv

9.60e-8
9.40e-8
9.20e-8 60
9.00e-8 55
50

)
°C
8.80e-8 45

e(
40

tur
8.60e-8
35

era
0.65
30

mp
0.60
Mois

Te
ture c 0.55 25
onten 0.50 20
t (%)
wb

Figure 5 Thermal diffusivity of sweet potato as a function of moisture content and temperature calculated from
food component mass fraction model.

Table 8 Calculated specific heat of sweet potato from experimental k, a, and r.

MC (w.b.) 0.75 0.65 0.55 0.45

Trials I II I II I II I II

Temperature (°C) (Jkg−1K−1)

20 3483 3360 3233 3233 3222 3289 2870 2759

30 3414 3562 3502 3440 3454 3155 2951 2815
40 3921 3995 3750 3782 3622 3508 3290 3294
50 4143 4239 4155 4105 3993 3853 3505 3430
60 4600 4519 4279 4280 4022 4189 4081 4063

Table 9 Specific heat of sweet potato by prediction model (Parallel).

MC (w.b.) 0.75 0.65 0.55 0.45

Temperature (°C) (Jkg−1K−1)

20 3473 3277 3020 2763

30 3484 3284 3029 2773
40 3491 3291 3037 2783
50 3550 3299 3046 2793
60 3558 3307 3054 2802
716 FARINU AND BAIK

4600

4400

4200
Specific heat (J/kg/K)

4000

3800
3600
3400
3200 60
3000 55
50

)
(°C
2800 45

re
40

tu
2600

ra
35

pe
0.70

m
0.65 30

Te
0.60
Moistu 0.55 25
re con 0.50
tent (%
) wb 0.45

Figure 6 Specific heat of sweet potato from an empirical model as a function of moisture content and temperature.

0.45 to 0.75 w.b. (Table 11). The simplified mass fraction model with temperature and
moisture content developed is:

r = 1491.5 − 0.573T − 496.8m, (18)

where r is density (kg/m3); T is temperature (°C); and m is moisture content (w.b.). Obser-
vation from the result showed that material density of sweet potato reduces with increase
in both moisture content and temperature of the product.
Regression modeling of our results was first done using exponential functions. We
believed this should fit our data best since the kinetics of energy, mass and momentum
transfer and reactions is usually in exponential fashion in nature. However the polynomial
equation fits better with the data (Table 3) and this probably supports why most thermal
properties models in literature were polynomial. There could be variation with source in
the properties of food component.[3] This means that there could be a difference between
the thermal conductivity of carbohydrate extracted from beef and that extracted from
potato. Even same crop specie (horse beans) that were grown in different locations exhibit
difference in qualitative and quantitative carbohydrate composition.[18] Thermal properties
of food materials also depend on any factors that affect the heat flow paths through the
material such as percent void spaces, shape, size, and arrangement of void spaces, homo-
geneity and orientation of fibers.[19] The variation between the thermal properties models
derived from major food compositions and the empirical models could be due to any or all
of the reasons given above.
 THERMAL PROPERTIES OF SWEET POTATO 717

3500

3400
Specific heat (J/kg/K)

3300

3200

3100

3000
60
2900
55
50
2800

)
(°C
45
40

re
2700

tu
35

ra
pe
0.65
30

m
0.60

Te
Moistu 0.55 25
re con 0.50
tent (% 20
) wb

Figure 7 Specific heat of sweet potato as a function of moisture content and temperature calculated from food
component mass fraction model.

1600

1400
Average density (kg/m3)

1200

1000

800

600

400

200

0
0.3 0.4 0.5 0.6 0.7 0.8
Moisture content (w.b)

Figure 8 Average density of sweet potato as a function of moisture content at 95% confidence level determined
using a multipycnometer.
718 FARINU AND BAIK

Table 10 Density of sweet potato determined by Multipycnometer.

Trials I II III IV V
−3
MC (w.b.) (kgm )

0.75 1126 1122 1125 1118 1124

0.65 1189 1171 1182 1194 1201
0.55 1241 1218 1263 1233 1255
0.45 1298 1312 1284 1301 1277

Table 11 Material density of sweet potato by prediction model (Parallel).

M.C. 0.75 0.65 0.55 0.45

Temperature (°C) (kgm−3)

20 1099 1141 1191 1218

30 1095 1139 1189 1215
40 1091 1136 1186 1212
50 1086 1133 1182 1209
60 1082 1129 1178 1205

CONCLUSIONS
A study of thermal properties of sweet potato was carried out. Prediction models
were developed for thermal properties of sweet potato over various ranges of moisture
content and temperature both from experiments and mechanistic models with major food
components of sweet potato. For thermal conductivity, thermal diffusivity, and specific
heat, the parallel model gave better predictions than the perpendicular (series) model.
Thermal properties values from models of major components of sweet potato were gener-
ally lower than values from our experiments. We speculate this could be largely attributed
to variation between sweet potato and the sources of components used in determining
thermal properties models used in this study and also the factors that affect heat flow pat-
terns in both products. The derived models would provide a reasonable estimate of ther-
mal conductivity, thermal diffusivity, and specific heat of sweet potato. It is recommended
that more work be done to further investigate the effect of moisture content and tempera-
ture on thermal properties of sweet potato at higher temperature values at which point
there might be significant moisture migration from the product to environment or mea-
surement sensors during heating and measurement.

REFERENCES
1. Cummings, B. Growing Sweet Potato in East Texas. Van Zandt County Extension Bulletin, 2005.
2. Kotecha, P.M.; Kadam, S.S. Sweet potato. In Handbook of Vegetable Science and Technology:
Production, Composition, Storage and Processing; D.K. Salunkhe; S. S. Kadam.; Eds. Marcel
Dekker; New York, 1998, 71–98.
3. Choi, Y.; Okos, M.R. Effects of Temperature and Composition on the Thermal Properties of
Foods. In Food Engineering and Process Applications, Vol. 1; Lemaguer, M.; Jelen, P.; Eds.;
Elsevier, London, 1985; 93.
 THERMAL PROPERTIES OF SWEET POTATO 719

4. Moreira, R.G.; Palau, J.; Sweat, V.E.; Sun, X. Thermal and Physical Properties of Tortilla Chips
as a Function of Frying Time. Journal of Food Processing and Preservation 1995, 19, 175–189.
5. Kramkowski, R.; Kaminski, E.; Serowik, M. Effect of Moisture and Temperature of Garlic on
its Specific Heat. Electronic Journal of Polish Agricultural Universities. 2001, 4(2), 1. http://
www.ejpau.media.pl/volume4
6. Renaud, T.; Briery, P.; Andrieu, J.; Laurent, M. Thermal Properties of Model Foods in the Fro-
zen State. Journal of Food Engineering 1992, 15, 83–97.
7. Wallapapan, K.; Sweat, V.E. Thermal Conductivity of Deffated Soy Flour. Trans. ASAE 1982,
26, 1440–1444.
8. Califano, A.N.; Calvelo, A. Thermal Conductivity of Potato Between 50 and 100°C. Journal of
Food Science 1991, 56, 586–589.
9. Maroulis, Z.B.; Saravacos, G.D.; Krokida, M.K.; Panagiotou, N.M. Thermal Conductivity Pre-
diction for Foodstuffs: Effect of Moisture Content and Temperature. International Journal of
Food Properties 2002, 5(1), 231–245.
10. Krokida, M.K.; Michailidis, P.A.; Maroulis, Z.B.; Saravacos, G.D. Literature Data of Thermal
Conductivity of Foodstuffs. International Journal of Food Properties 2002, 5 (1), 63–111.
11. Drouzas, A.E.; Maroulis, Z.B.; Karathanos, V.T.; Saravacos, G.D. Direct and Indirect Determi-
nation of the Effective Thermal Diffusivity of Granular Starch. Journal of Food Engineering
1991, 13(2), 91–101.
12. Association of Official Analytical Chemists. Official Methods of Analysis of AOAC Interna-
tional, 17th ed. Vol. 2, AOAC: Gaithersburg, MD, 2002.
13. Woodside, W.; Messmer, J.H. Thermal Conductivity of Porous Media. I. Unconsolidated Sands.
Journal of Applied Physics 1961, 1688.
14. Brailsford, A.D.; Major, K.G. Thermal Conductivities of Aggregates of Several Phases Includ-
ing Porous Materials. British Journal of Applied Physics 1964, 15, 313.
15. Stroshine, R. Physical Properties of Agricultural Materials and Food Products. Purdue Univer-
sity: West Lafayette, IN, 1998.
16. United States Department of Agriculture. Agriculture Handbook No 8; Composition of Foods;
Raw, Processed, Prepared; USDA: Washington, DC, 1963.
17. Wadsworth, J.I.; Spadaro, J.J. Transient Temperature Distribution in Whole Sweet Potato Roots
during Immersion Heating. Food Technology, 1969, 23, 219.
18. Cerning, J.; Saposnik, A.; Guilbot, A. Carbohydrate Composition of Horse Beans of Different
Origins. Cereal Chemistry 1975, 52(2), 125–138.
19. Sweat, V.E. Thermal Properties of Food. In Engineering Properties of Food; Rao, M.A.; Rizvi,
S.S.H. Marcel Dekker: New York, 1995; 99–138.