ASAE Journal Article
Nondestructive Measurement of Moisture Content of Different Varieties of Wheat Using a Single Calibration with a Parallel-Plate Capacitance Sensor
C. V. K. Kandala, J. Sundaram, N. Puppala, V. Settaluri
Published in Transactions of the ASABE Vol. 55(4): 1583-1587 ( 2012 American Society of Agricultural and Biological Engineers ).
Submitted for review in October 2011 as manuscript number FPE 9442; approved for publication by the Food & Process Engineering Institute Division of ASABE in June 2012.
Mention of company or trade names is for purpose of description only and does not imply endorsement by the USDA. The USDA is an equal opportunity provider and employer.
The authors are Chari V. K. Kandala, ASABE Member, Research Agricultural Engineer, USDA-ARS National Peanut Research Laboratory, Dawson, Georgia; Jaya Sundaram, ASABE Member, Food Technologist, USDA-ARS Russell Research Center, Athens, Georgia; Naveen Puppala, Associate Professor, and Vijayasaradhi Settaluri, Post-Doctoral Fellow, College of Agriculture and Home Economics, New Mexico State University, Clovis, New Mexico. Corresponding author: Chari V. K. Kandala, USDA-ARS NPRL, P.O. Box 509, Dawson, GA 39842; phone: 229-995-7400; e-mail: Chari.email@example.com.
Abstract. A simple, low-cost instrument that measures impedance and phase angle was used along with a parallel-plate capacitance system to estimate the moisture content (MC) of six varieties of wheat. Moisture content of grain is important and is measured at various stages of processing and storage. A sample of about 150 g of wheat was placed separately between a set of parallel plate electrodes, and the impedance and phase angle of the system were measured at frequencies 1 and 5 MHz. A semi-empirical equation was developed using the measured impedance and phase angle values, the computed capacitance, and the MC values obtained by standard air-oven method. Multiple linear regression (MLR) was used for the empirical equation development using statistical software. In the present work, a low-cost impedance analyzer, designed and assembled in our laboratory, was used to measure the impedance and phase angles. MC values of wheat samples in the moisture range of 9% to 25%, not used in the calibration, were predicted by the equations and compared with their standard air-oven values. For over 97% of the samples tested from the six varieties of wheat, the predicted MC values were within 1% of the air-oven values. This method, being nondestructive and rapid, will have considerable application in the drying and storage processes of wheat and similar field crops.
Keywords. Capacitance, Impedance analyzer, Moisture content, Parallel-plate electrodes, Phase angle, Wheat.
Wheat is an important crop that is used as a staple food in almost all parts of the world. As soon as wheat is harvested, measuring its moisture content (MC) before drying is a crucial part of maintaining its quality. Wheat should be dried to 13.5% MC for immediate sale (Missouri Extension, 2003). It is recommended that wheat should be dried with natural air, but using hot air is not uncommon depending on the climatic conditions. During the drying process, the MC of the wheat is measured periodically to determine if the required drying has been achieved. Some of the electronic moisture meters presently used require either fixed or large quantities of wheat samples each time a measurement is made, while others are destructive of the sample. Using the high correlation between the dielectric properties of grain and its MC, a method was developed to determine the MC of single kernels of corn (Kandala et al., 1989).The technique was extended to measurements on small samples (15 to 30 kernels) of wheat (Kandala et al., 1996) using a commercial impedance meter. While commercial impedance meters performed well in the establishment of the impedance method for moisture determination, they have several extra features that make them expensive and are not needed for this work. Thus, a low-cost meter that measures impedance parameters at the required frequencies on a moderately large sample of grain would be useful in the estimation of moisture content and would give a better average value of the MC of bulk quantities. A low-cost impedance meter (CI meter) was designed and used along with a parallel-plate capacitance system to estimate the MC of in-shell peanuts and yellow-dent field corn samples of about 150 g (Kandala and Sundaram, 2010). The impedance meter measured the impedance and phase angle of the parallel-plate capacitance system holding the sample at two frequencies, 1 and 5 MHz, and from the measured values computed the MC of the sample. In this work, the CI meter was calibrated using samples from five wheat varieties, and the calibration equation generated was used to predict the MC of samples, not used in the calibration, of three of these varieties and another variety grown in New Mexico.
Materials and Methods
It was found earlier that the high correlation between dielectric properties of aqueous materials and their MC can be used in predicting the MC of these materials. From earlier documented work (Nelson, 1978), the variation in dielectric constant with MC for shelled yellow field corn was found to be more pronounced between 1 and 5 MHz. Because the degree of change in the dielectric constant with the change in moisture content decreases with increasing frequency (Nelson, 1981), the difference in the dielectric constants of the parallel-plate capacitor and kernels measured at two frequencies ( f 1 and f 2 ) should be an estimator of moisture content. Since capacitance is a function of dielectric constant, the capacitance difference at these two frequencies should also be a good indicator of the moisture content (Nelson et al., 1992) Thus, the difference in the dielectric constants at 1 and 5 MHz, or any other higher frequency (e r 1 – e r 2 ), should be a good indicator of the moisture present in the material. The difference in capacitance of a parallel-plate system of plate thickness A and separation d at two frequencies can be written as:
where e r 1 and e r 2 are the dielectric constants of the material between the plates at the two frequencies, and e 0 is the permittivity of free space (8.854 × 10 -12 farad m -1 ). It was found earlier that although ( C 1 – C 2 ) was a good estimate of the MC, it alone was not able to predict the MC of the material with sufficient accuracy (Kandala and Nelson, 1990). This was partially because the volume of space that a sample of odd-shaped material, such as grain, occupies between two parallel plates would vary each time the material is placed between the plates. Air gaps between the grain kernels and between the kernels and the capacitor walls would occur differently, introducing errors. To compensate for these errors, two other related electrical parameters, phase angle (?) and impedance ( Z ) were also measured at these two frequencies using the CI meter. The capacitance of the parallel-plate system was computed from the values of ? and Z . Using the differences in the values of C , ?, and Z at the two frequencies minimized the errors due to air gaps. While the capacitance change represents the dielectric variation, the phase angle change accounts for the loss factor, and the impedance values represent the quality factor of the sample material. Using these differences and their square terms in an empirical equation, the average MC of a ~150 g sample of wheat was determined within 1% of the air-oven values. The calibration equation is shown below:
The CI Meter
The design and operation of the CI meter was described previously (Kandala et al., 2008). Three frequencies (1, 5, and 9 MHz) are generated by three crystal oscillators and applied alternately to a parallel-plate system, which acts as the impedance load ( Z ), by switching through a multiplexer. The values of impedance ( Z ) and phase angle (?) at 1 and 5 MHz are read from the instrument, and the real and imaginary components of Z at each frequency are calculated as R = | Z |cos? and X = | Z |sin?. The capacitance of the parallel-plate system with a wheat sample between the plates is obtained as:
The measurement system set up is shown in figure 1. The CI meter is provided with a regulated power supply that can be plugged into a 110 VAC line. It can also be operated on two 12 VDC rechargeable batteries provided for field operations. A laptop computer was used to control the operations, measure and register the data, and calculate the moisture contents.
A cylindrical acrylic tube, fitted with a set of parallel-plate electrodes (fig. 1), served as the sample holder and sensor, as described earlier (Kandala et al., 2008). The cylinder is 190 mm long with an internal diameter of 50 mm and a wall thickness of 7 mm. An electrode assembly consisting of two rectangular aluminum plates, 140 mm long and 50 mm wide, was fitted inside the cylinder about 25 mm from the ends of the cylinder. The gap between the parallel plates is 42 mm and is filled with the sample, as shown in figure 1. Except for the electrodes, no metal parts were used in the assembly of the electrode system nor in the sample collecting system to prevent any interaction with the RF signal used in the measurements. With the drawer below the cylinder pushed all the way in, the cylinder was filled with the wheat sample, and the impedance measurements were taken. After completion of the measurements, the drawer was pulled out slowly, allowing the sample to fall into the drawer. The drawer was emptied before another sample was placed in the cylinder for measurement. With a wheat sample occupying the space between the electrodes, the analyzer measured the impedance and phase angle of this electrode system at 1 and 5 MHz, and a computer controlled and collected the data from the analyzer. Using these measurements in an empirical equation, the computer was programmed to calculate the moisture content of each sample.
Figure 1. RF impedance measuring system: 1 = CI meter, 2 = cylinder with electrodes, and 3 = computer.
Six varieties of wheat, planted and harvested around the Texas panhandle and at the New Mexico State University Station near Clovis, were used in this study (Bean, 2011). The wheat varieties were Tam111, Duster, Scoutt66, Endurance, Jagger, and Hatcher, planted during October 2010 and harvested during July 2011. All sample lots were stored at 4°C and 40% relative humidity. When the samples were received at the USDA-ARS National Peanut Research Laboratory (NPRL), their MC was about 9% (all moisture contents are expressed in percent wet basis in this article). These lots were divided into 22 sublots and placed in glass jars. Leaving the contents of one jar of each variety at its original MC level, appropriate amounts of water were added to the other jars to develop 22 moisture levels between 9% and 25%. Thus, from the available quantities, six MC levels of Tam111, five of Duster, four of Hatcher, five of Jagger, and one each of Scoutt66 and Endurance were developed for calibration and validation. The moisture content of each subsample was determined by the standard air-oven method ( ASAE Standards , 1994) by drying triplicate samples of 10 g each for 19 h at 130°C. Samples were weighed before and after drying, and the MC of each sample was obtained as the percentage ratio of the weight loss to the original wet weight of the sample. Three replicates were selected randomly at each moisture level and subjected to the oven test. The resulting MC values were averaged and are shown as the “oven values” in tables 1 and 3 for that moisture level.
Impedance measurements were made on 30 samples from each sublot. Each wheat sample was transferred from the jar into the cylinder fitted with the electrode system until the space between the two parallel plates was completely filled. The cylinder accommodated about 150 g of wheat sample. The room temperature during the measurements was maintained at 21°C ±1°C. With a sample in the cylinder, the impedance ( Z ) and phase angle (?) were measured with the CI meter at 1 and 5 MHz. The computer was programmed to repeat each measurement 30 times, compute the average value, and save it to an Excel spreadsheet. The sample was then collected in the drawer below the cylinder by gently pulling the drawer out and tapping on the cylinder for the sample to drop down. The drawer was emptied and reset in its box. This procedure was repeated for all wheat samples (sublots) from the rest of the jars.
Analysis of Measurement Data
The samples for calibration were selected randomly from the 22 sublots, making sure that the moisture groups with the lowest and highest MC values were included and the moisture range in between was adequately covered. Using the MC values and the measured impedance values of the calibration group, which consisted of two moisture levels each of Tam111 and Duster, and one level each of Jagger, Scoutt66, and Endurance varieties, values of the calibration constants of equation 2 were determined by multiple linear regression (MLR) analysis using Unscrambler software (version 9.7, Camo Software, Inc., Woodbridge, N.J.). Substituting the values of the constants, the measured values of ? and Z , and the computed value of C (eq. 3) of the validation samples into equation 2, the MC of each wheat sample in the validation lot was calculated and compared with the values determined by the air-oven method.
Results and Discussion
From the measured values of C , ?, and Z on the calibration sublots with the CI meter, and using Unscrambler procedures for regression analysis, the values obtained for the constants A 0 to A 6 in equation 2 were:
A 0 = -27.235, A 1 = 0.0032, A 2 = 4.734, A 3 = 20.476,
A 4 = -0.000088, A 5 = -0.405, and A 6 = -2.694
This calibration had an R 2 value of 0.99, and all the terms used in equation 2 had a probability of a greater absolute t value (Pr > | t |) under the null hypothesis for the variables (Moore, 2000) of less than 0.0001. These constants along with the values of impedance, phase angle, and capacitance (obtained from eq. 3) were used in equation 2 to calculate the MC of each of the 30 samples from the calibration sublots. The calculated values were averaged over the 30 samples in each moisture group and compared with their air-oven values, and the results are shown in table 1 along with the standard deviations, differences, and predictability. Predictability is defined as the percentage of samples in each moisture group for which the predicted MC values were within 1% of their air-oven values.
Table 1. Calibration sublots: Comparison of CI meter and air-oven MC measurements.
Moisture Content (%)
Predicted by Equation 2
[a] All samples were from the 2010 harvest.
[b] Mean of 30 sample measurements; SD = standard deviation.
[c] Diff. = difference between predicted and oven MC values.
[d] Pred. = predictability, the percentage of samples in each moisture group for which the predicted MC is within 1% of the oven value.
Shown in table 2 are the fitness measures for the calibration lots. The SEC was 0.49% MC (with p = 6 and n = 210):
Table 2. Fitness measures for the calibration set of wheat samples.
where n is the number of observations, p is the number of variables in the regression equation with which the calibration is performed, and e i is the difference between the observed and reference values for the i th observation. The predictability was good at all the MC levels. The calculated values averaged over 30 samples from each moisture group agreed well with their air-oven values, and the predictability ranged from 97% to 100%. The standard deviation for 30 samples at any moisture level was under 0.5%. An R 2 value of 0.99, an SEC of 0.49, a low bias value, and a high rate of predictability suggest that this model is suitable for MC predictions. This was further verified by the fitness measures obtained for the validation sets.
Shown in table 3 are the MC values for 15 validation sets consisting of three varieties used in the calibration, and an additional variety, Hatcher, not used in the calibration. The moisture levels of the three varieties used in the validation are different from the levels used in the calibration. The MC values were predicted using the measured values of impedance and phase angle, and the calibration constants in equation 2, averaged over 30 samples in each group, and compared with their air-oven values. Also shown are the standard deviations and predictability of each moisture group. The predictability was 87% or better for any moisture level and averaged about 98% over the 15 validation levels.
Table 3. Validation sublots: Comparison of CI meter and air-oven MC measurements.
Moisture Content (%)
Predicted by Equation 2
[a] Mean of 30 samples in each moisture group; SD = standard deviation.
[b] Diff. = difference between predicted and oven MC values.
[c] Pred. = predictability, the percentage of samples in each moisture group for which the predicted MC is within 1% of the oven value.
Shown in table 4 are the fitness measures for the validation group of samples. The validation set also came up with a good R 2 value of 0.98, and the standard error of prediction (SEP) was 0.60:
where n is the number of observations, e i is the difference between the moisture content predicted and that determined by the reference method for the i th sample, and is the mean of e i for all of the samples. The SEP value is close to the SEC value, and the R 2 value compares well with that of the calibration group. A low bias indicates the closeness of the mean calculated values to the air-oven values, confirming that the prediction model is dependable. The SEC value of 0.49 and the SEP value of 0.60 were better than the 0.88 for SEC and 0.91 for SEP obtained earlier with small wheat samples (Kandala et al., 1996).
Table 4. Fitness measures for the validation sublots.
Shown in figure 2 is a comparison of the predicted and air-oven values of the seven calibration wheat lots. An R 2 value of 0.99 further indicates the fitness of the calibration equation. Similarly, in figure 3, the MC values of the validation samples of wheat are compared with their respective air-oven values. The predicted MC values compare well with their corresponding air-oven values, with an R 2 value of 0.99. The calibration equation predicted the MC values of different varieties quite well with a single calibration equation developed with randomly selected MC levels from the different available varieties. Moreover, the measurement is rapid and nondestructive, and there was no need to measure the volume or weight of the wheat samples.
Figure 2. Comparison of oven and predicted MC values for the calibration lots.
Figure 3. Comparison of oven and predicted MC values for the validation lots.
From the measurements of impedance and phase angle using a low-cost impedance meter, the moisture content of 150 g to 200 g wheat samples could be determined. The weight or volume of the samples need not be measured. The moisture range of the samples used was between 9% and 25%. Calibration was done with five different wheat varieties, and validation was done on three of these varieties and one additional variety. For 98% of the samples tested from the four varieties, the predicted MC values were within 1% of their air-oven values. The meter performed well in predicting the MC for all four varieties of wheat with a single calibration equation. Using impedance measurements taken at two frequencies helped in eliminating the errors due to air gaps that would occur randomly when the sample is placed between the plates. Moisture content being an important parameter to measure and monitor at various stages of wheat production and storage, this low-cost instrument would be useful for the grain industry.
The authors would like to thank Manuel Hall, Engineering Technician at the USDA-ARS National Peanut Research Laboratory, Dawson, Georgia, for his help during this work.
ASAE Standards . 1994. ASAE S352.2: Moisture Measurement – Unground grain and seeds. St. Joseph, Mich. ASAE.
Bean, B. 2011. 2011 Wheat variety trials conducted in the Texas and New Mexico High Plains. College Station, Tex.: Texas AgriLife Extension. Available at: http://clovissc.nmsu.edu/ documents/wheat-variety-trial-summary-final2011.
Kandala, C. V. K., and S. O. Nelson. 1990. Measurement of moisture content in single kernels of peanuts: A nondestructive electrical method. Trans. ASAE 33(2): 567-572.
Kandala, C. V. K., and J. Sundaram. 2010. Nondestructive measurement of moisture content using a parallel-plate capacitance sensor for grain and nuts. IEEE Sensors J. 10(7): 1282-1287.
Kandala, C. V. K., S. O. Nelson, and K. C. Lawrence. 1989. Non-destructive electrical measurement of moisture content in single kernels of corn. J. Agric. Eng. Res. 44: 125-132.
Kandala, C. V. K., S. O. Nelson, and K. C. Lawrence. 1996. Determining moisture content in small wheat samples by dual-frequency RF impedance sensing. Trans. ASAE 39(3): 1185-1188.
Kandala, C. V. K., C. L. Butts, and M. C. Lamb. 2008. Moisture content determination for in-shell peanuts with a low-cost impedance analyzer and capacitor sensor. Trans. ASABE 51(4): 1377-1381.
Missouri Extension. 2003. Management of soft red winter wheat. Publication IPM1022. Columbia, Mo.: University of Missouri Extension. Available at: http://extension.missouri.edu/p/ IPM1022.
Moore, D. S. 2000. The Basic Practice of Statistics. 2nd ed. New York, N.Y.: W. H. Freeman and Co.
Nelson, S. O. 1978. Frequency and moisture dependence of the dielectric properties of high-moisture corn. J. Microwave Power 13(2): 213-218.
Nelson, S. O. 1981. Frequency and moisture dependence of the dielectric properties of chopped pecans. Trans. ASAE 24(6): 1573-1576.
Nelson, S., K. Lawrence, and C. Kandala. 1992. Sensing moisture in peanut and pecan kernels by RF impedance measurements. J. Microwave Power 27(3): 171-174.