Effects of "natural" water and "added" water on prediction of moisture content and bulk density of shelled corn from microwave dielectric properties.

Abstract

Dielectric properties of samples of shelled corn of "natural" water content and those prepared by adding water were measured in free space at microwave frequencies and 23 degrees C. Results of measurements of attenuation, phase shift and dielectric constant and loss factor at 9 GHz show no difference between the samples with "natural" water and those in which water was added artificially. Bulk densities and moisture contents predicted from calibration equations expressed in terms of dielectric properties of both natural and added water samples agreed closely, and standard errors were less than 1% for moisture content and relative error for bulk density was less than 5%.

Full text

Page 1

Effects of “Natural” Water and “Added”
Water on Prediction of Moisture Content
and Bulk Density of Shelled Corn from
Microwave Dielectric Properties
Samir Trabelsi, Stuart O. Nelson
U. S. Department of Agriculture, Agricultural Research Service, Richard B. Russell Agricultural
Research Center, Athens, GA 30604-5677 USA
Micah A. Lewis
Department of Agricultural and Biological Engineering, Driftmier Engineering Center
The University of Georgia, Athens, GA 30602 USA
Received: July 24, 2009
Accepted: January 26, 2010
ABSTRACT

prepared by adding water were measured in free space at microwave frequencies and 23ºC.
Results of measurements of attenuation, phase shift and dielectric constant and loss factor
at 9 GHz show no difference between the samples with “natural” water and those in which
water was added artificially. Bulk densities and moisture contents predicted from calibration
equations expressed in terms of dielectric properties of both natural and added water samples
agreed closely, and standard errors were less than 1% for moisture content and relative error for
bulk density was less than 5%.
Dielectric properties of samples of shelled corn of “natural” water content and those
KEY WORDS: Dielectric properties, natural water, added water, free space, bulk density,
moisture content.
INTRODUCTION
It is well known that water content of grain and seed strongly influences their dielectric
properties [Nelson, 1973a; Nelson, 1982; Nelson, 1973b]. Understanding water behavior at
microwave frequencies is key to the success of dielectric-based heating and moisture sensing
applications [Metaxas and Meredith, 1983; Nyfors and Vainikainen, 1989; Kraszewski, 1996]. For
purposes of research, it is customary to artificially add water to grain and seed to investigate
variations of their dielectric properties with frequency, temperature, moisture content and bulk
density [Nelson, 1973a; Trabelsi and Nelson, 2003]. Also, moisture sensors are often calibrated
with data obtained from measurements on grain and seed samples to which water was added.
The question is how well the dielectric properties obtained for grain and seed samples
with “added” water compare to those obtained for grain and seed samples with “natural” water.
And, consequently, how accurate moisture measurements on freshly harvested grain performed
with sensors calibrated with grain samples with “added” water are. It is the objective of this
paper to answer these basic, important questions.
Journal of Microwave Power and Electromagnetic Energy, 44 (2), 2010, pp. 72-80
A Publication of the International Microwave Power Institute
72
International Microwave Power Institute

Page 2

are intrinsic electric properties that describe
the electric field-material interaction. They
are commonly represented by a complex
number, the relative complex permittivity
ε=ε’-jε”, where the real part ε’ (dielectric
constant) is associated with the capability
for energy storage in the electric field in
the material, and the imaginary part ε”
(dielectric loss factor) is associated with
energy dissipation in the material in the
form of heat. Another parameter often used
to describe dielectric materials is the loss
tangent, which is expressed as the ratio of
the dielectric loss factor to the dielectric
constant. Measurement of dielectric properties
at radio frequencies has been widely used
to characterize materials containing water,
particularly in foods and agricultural products
[Kraszewski, 1996; Trabelsi and Nelson,
2006]. This is mainly because of the dipolar
nature of water molecules and associated
relaxations, which usually take place in the
radio-frequency range [Hasted, 1973].
Dielectric properties of liquid water
are well described by the single-relaxation
Debye model [ Hasted, 1973; Kaatze, 1989].
However, those of bound water are not well
understood, and they are often described as
lying between those of ice and liquid water,
depending on the degree of binding of the
water molecules inside the material. This
complicates the modeling of the dielectric
response of materials containing bound water
and therefore, to date, there is no valid
single model that covers the wide spectrum
of the dielectric behavior of bound water in
materials. From an engineering perspective, a
full understanding is not always needed, and
dielectric data collected under controlled
conditions can be used successfully in many
industrial applications [Metaxas and Meredith,
1983; Nyfors and Vainikainen, 1989].
Several techniques can be used for
the measurement of dielectric properties of
materials [Von Hippel, 1954; Bussey, 1967;
Chen et al., 2004]. At microwave frequencies,
these consist of transmission line, impedance
The dielectric properties of materials and cavity measurements. Selection of the
appropriate technique depends on the nature
of the material, the frequency range of
interest, and the desired degree of accuracy.
Among these techniques, free-space
methods have the advantages of requiring
minimum sample preparation and no need
for physical contact with the material. Also,
recent advances in antenna technology make
it possible to determine accurately the
dielectric properties over a broad frequency
range with a single system [Trabelsi and
Nelson, 2003; Chen et al., 2004]. Another
reason for selecting this measurement
technique is that it can be implemented
easily in many industrial applications for on-
line monitoring of physical characteristics of
materials [Trabelsi et al., 1998; Trabelsi et
al., 2001; Trabelsi and Nelson, 2004].
In this study, a free-space transmission
technique was used to perform dielectric
measurements between 2 and 18 GHz at 23
ºC on samples of shelled corn with “natural”
and “added” water. The dielectric properties
of each sample were determined from
measurements of the attenuation and phase
shift caused by a layer of material placed
between two horn-lens antennas facing each
other [Trabelsi and Nelson, 2003]. For purpose
of illustration, results of measurements on
samples of shelled corn with “natural” and
“added” water are compared at 9 GHz and 23
ºC. Also, effectiveness of prediction of bulk
density and moisture content from calibration
equations developed with data obtained
for shelled corn with “added” water are
compared to those obtained from calibration
equations developed with data obtained for
shelled corn with “natural” water.
MATERIALS AND METHODS
Bulk density and moisture content
For granular and particulate materials
containing water, conventionally the bulk
density is determined gravimetrically and
moisture content is determined with oven-
drying techniques. The sample bulk density
was calculated by dividing its weight by
Samir Trabelsi et al., Effects of “Natural” Water and “Added” Water on Prediction of Moisture Content and Bulk ...
Journal of Microwave Power and Electromagnetic Energy, 44 (2), 2010
International Microwave Power Institute
73

Page 3

the volume v of the container in which the
material was placed:
(1)
where mt is the total mass of the sample, mw
is the mass of water and md is the mass of dry
matter. For moisture determination, the a
standard oven-drying method was used [ASAE,
2002]. The moisture content, M, in percent
was determined on a wet basis as:
(2)
Corn samples with “natural” water
Four samples of hybrid shelled corn
grown in Texas were used in this study. Upon
receiving them their moisture contents were
determined by the oven technique [ASAE,
2000].
Their initial
contents were 12.6%, 17.1%, 17.8%, and
21.4%, respectively. To cover a moisture
range, after each dielectric properties
measurement session, the samples were dried
by spreading them on plastic sheets in open
air for a given period of time. Then, they
were placed in sealed plastic bags and stored
at 4°C for a few days to equilibrate before
the following measurement session. Fourteen
moisture levels were obtained. They ranged
from 10.5% to 21.4%. Upon removal from
cold storage, the samples were conditioned
to room temperature for 24 hours before
measurements.
“natural” moisture
Corn samples with “added” water
Samples of shelled corn of different
moisture contents were prepared by spraying
distilled water on samples to bring them to
the desired moisture level. Starting from the
lowest moisture level, the moisture content
was increased gradually in increments of
about 1% until the highest desired moisture
level was reached. After being sprayed whit
the proper amount of water, the sample was
mixed to distribute the water more evenly
throughout the sample. Then, the samples
were placed in sealed plastic bags and stored
for at least 72 hours at 4 ºC to equilibrate.
For corn samples with “added” water, the
moisture content ranged from 10.8% to
25.0%. Samples were conditioned to room
temperature for 24 hours before dielectric
measurements.
Free-space measurement of the dielectric
properties
Microwave dielectric measurements on
corn hybrid samples with “natural” moisture
content and those with added water were
performed with a free-space transmission
system [Trabelsi and Nelson, 2003]. In free
space (transmission mode), a sample is
placed between two antennas facing each
other, and the modulus, │S21│, and phase, φ,
of the scattering transmission coefficient, S21,
are measured. From these measurements and
computation of attenuation and phase shift
caused by the layer of material, the dielectric
properties of each shelled corn sample were
determined [Trabelsi and Nelson, 2003].
The measurement setup shown in
Figure 1 consists of a Hewlett Packard 8510C
Figure 1. Diagram of the measurement setup.
Samir Trabelsi et al., Effects of “Natural” Water and “Added” Water on Prediction of Moisture Content and Bulk ...
Journal of Microwave Power and Electromagnetic Energy, 44 (2), 2010
International Microwave Power Institute
74

Page 4

vector network analyzer (VNA), a computer,
two high quality coaxial cables with APC-7
connectors at their terminations, two linearly
polarized, ultrabroadband (2 to 21 GHz) horn/
lens antennas (BAE SYSTEMS model AHO-2077-
N), a Styrofoam sample holder, and four sheets
of radiation-absorbing material ECCOSORB
AN-79. The horn-lens antennas collimate the
electromagnetic energy in a narrow beam and
provide a plane wave within a short distance
from the transmitting antenna.
The antennas were 37 cm apart. The
sample holder was made of Styrofoam because
it is a lossless material and has a dielectric
constant close to that of air. Finally, radiation–
absorbing material was placed around the
sample to avoid interfering reflections from
surroundings. The measurement system was
calibrated with a response-type calibration
between 2 and 18 GHz. The reference
values for│S21│ and φ were set to zero with
an empty sample holder between the two
antennas. To ensure accurate measurements
the sample thickness and lateral dimensions
were selected to avoid effects of multiple
reflections and edge diffraction, and time-
domain gating was applied to filter out
undesirable post-calibration mismatches and
possible multiple-path transmission. The whole
measurement procedure was automated and
controlled by a computer connected to the
VNA. The measurements were conducted in a
room where the relative humidity was about
40% and temperature was 23 ±1°C.
After calibrating the measurement
system, the sample was poured into the
sample holder and placed inside the radiation-
absorbing enclosure between the antennas.
For each corn sample, measurement of │S21│
and φ were performed between 2 and 18 GHz
with a step of 1 GHz. For a given moisture
content, measurements were carried out at
three different densities ranging from loosely
packed to compacted by settling the sample
on a wooden bench. For samples of corn with
“natural” water content, the bulk density
ranged from 0.738 g/cm3 to 0.891 g/cm3. For
those in which water was added, the bulk density
ranged from 0.687 g/cm3 to 0.877 g/cm3. After
the microwave measurements were completed,
the sample temperature, T, was determined
with a digital thermocouple thermometer. For
moisture determination, for each sample, three
15 g replicates were dried in an oven at 103 °C
for 72 hours [ASAE 2000].
RESULTS
Attenuation and phase shift
The attenuation, A, and phase shift, Φ ,
attributable to each sample, are determined
from the measurement of the modulus
│S21│and phase φ as follows:
(3)
(4)
where n is an integer to be determined. With
a vector network analyzer, φ can only be
measured between –180 and +180 degrees.
Therefore, when the thickness d of the
layer is greater than the wavelength in the
material, an ambiguity in the phase occurs.
Methods to resolve the phase ambiguity were
proposed previously [Musil and Zacek, 1986;
Trabelsi et al., 2000].
Figures 2 and 3 show variations of
attenuation and phase shift each divided by
bulk density and the sample thickness as a
Figure 2. Attenuation divided by bulk density and thickness
as a function of moisture content at 9 Ghz and 23°C.
Samir Trabelsi et al., Effects of “Natural” Water and “Added” Water on Prediction of Moisture Content and Bulk ...
Journal of Microwave Power and Electromagnetic Energy, 44 (2), 2010
International Microwave Power Institute
75

Page 5

function of moisture content.
Results for the attenuation are nearly
identical for both the shelled corn samples
with “added” water and those with “natural”
water. However, a slight difference can be
noticed between the phase shifts.

Dielectric properties
Assuming a plane wave is traversing
a layer of low-loss material, the dielectric
properties, ε’and ε”, were calculated from
attenuation and phase shift as follows
[Trabelsi et al., 1997]:
(5)
(6)
where c is the speed of light in m/s, f is the
frequency in Hz, and d is the thickness of the
layer of material in meters.
Figures 4 and 5 show the variations with
moisture content of the dielectric constant
and dielectric loss factor each divided by
bulk density. The dielectric properties were
divided by bulk density to minimize its effect
and for purpose of clarity in comparing the
two data sets. As observed for the attenuation
and phase shift, the data collected for shelled
corn samples with “added” water are nearly
superimposed on those obtained for samples
of corn with “natural” water.
Predicting bulk density and moisture content
of shelled corn samples
Bulk density and moisture content
can be predicted from measurement of
the dielectric properties [Trabelsi et al.,
1998; Trabelsi et al., 2001]. Bulk density
is determined from a complex-plane
Figure 3. Phase shift divided by bulk density and thickness
as a function of moisture content at 9 GHz and 23°C.
Figure 4. Dielectric constant divided by bulk density
as a function of moisture content at 9 GHz and 23°C.
Figure 5. Dielectric loss factor divided by bulk density
as a function of moisture content at 9 GHz and 23 ºC.
Samir Trabelsi et al., Effects of “Natural” Water and “Added” Water on Prediction of Moisture Content and Bulk ...
Journal of Microwave Power and Electromagnetic Energy, 44 (2), 2010
International Microwave Power Institute
76